f>]p |::nrg|..:riPQ ^l../- ^'. J:. . .h./'il. .:,..J;!,„ tJae M'enioiry ol IBIbert llilalvcr A.jb,lstrO'B"i rbe Sympo^wm Vv^as hdd Awgust IS-IS, 1983 SporKKM'w] by tba JuitioDatl jfi-iaiine H;slKde!ii Scrvia? ABiiiCi'itt-Hi', Society «.)';" ]l':;:!i i:hyoloi?j:^ti5 aad HerfK'rtc^loBiBi'j nj _D ■=0 3 r^ L 1— 1 D o m □ c, / QNTOGENY AND SYSTEMATICS ^ OF FISHES Based on An International Symposium Dedicated to the Memory of Elbert Halvor Ahlstrom The Symposium was held August 15-18, 1983, La JoUa, CaUfomia Sponsored by the National Marine Fisheries Service National Oceanic and Atmospheric Administration United States Department of Commerce \9 ■0 4 5:^ Special Publication Number 1 American Society of Ichthyologists and Herpetologists Library of Congress Catalogue Card Number: 84-72702 ISSN No. 0748-0539 © Copyright, 1984, by The American Society of Ichthyologists and Herpetologists Pnnted by Allen Press Inc.. Lawrence, KS 66044 USA Preface The National Marine Fisheries Service organized, supported and conducted an international symposium entitled Ontogeny and Systematics of Fishes, held in La JoUa, California on August 15-18, 1983, and dedicated to the memory of Elbert Halvor Ahlstrom. Dr. R, Lasker served as convener. The papers presented at that symposium form the basis for this book, which is published by the American Society of Ichthyologists and Herpetologists as their Supplement to Copeia, Special Publication Number 1 . Financial support was provided by the National Marine Fisheries Service, National Oceanic and Atmospheric Administration, U.S. Department of Commerce. For many years. Dr. Ahlstrom planned to write a book on larval fishes and ways in which they contributed to systematics. A few years before his untimely death, he and his colleague H. G. Moser outlined such a book and began to work on the initial chapters. Dr. Ahlstrom left a vast store of notes, data, and partly completed manuscnpts. Dr. Moser realized that much of the significance of these unique and important data would be lost unless they were brought to light. He approached colleagues at the Southwest Fisheries Center to gather a group of larval fish workers who had worked closely with Dr. Ahlstrom, and who were given access to his notes, to collaborate on the book. From this initiative a plan developed to conduct a symposium and publish the results in a book to accomplish the original plan of Dr. Ahlstrom and honor his memory as one of the nation's foremost fishery scientists. A symposium steering committee was formed with H. G. Moser as Chairman and consisted of D. M. Cohen, M. P. Fahay, A. W. Kendall, Jr.. W. J. Richards and S. L. Richardson. The steering committee first met in Boulder, Colorado to develop an outline for the symposium and book and invite potential contributors. The aim was to present the current state of knowledge of early life history of fishes and apply that to systematics. Originally it was intended to concentrate solely on the marine groups with which Dr. Ahlstrom had worked, but because of recent advances in freshwater and other early life history work, the plan was expanded to include all but the primitive osteoglossomorphs. Thus, the coverage was to start with the elopomorphs. Following the Boulder meeting, potential contributors were contacted and responded enthusiastically. The Steering Committee met subsequently in Ocean Springs, Mississippi and Miami, Florida to review progress and refine plans. Because of the subject matter it seemed appropriate that the American Society of Ichthyologists and Herpetologists collaborate in publishing the papers resulting from the symposium. C. R. Robins, then President of ASIH, supported this suggestion and assisted in many ways. Subsequent to the symposium, manuscripts were reviewed and edited by the Steering Committee of the Symposium, which served as an editorial committee for this volume. The Steering Committee thanks all of the authors of this volume among whom there was a great exchange of ideas and generous help. Much additional assistance was provided to the authors and is here acknowledged. Institutional support was provided by the National Marine Fisheries Service through contributions from each of the four Fisheries Centers— Southwest, Southeast, Northwest and Alaska and Northeast. Support was provided by the National Science Foundation through grants DEB76-82279, DEB78-26540; the National Geographic Society by grant 2535-82 from the Committee for Research and Exploration; the Robert E. Maytag Fellowship at the University of Miami; Natural History Museum of Los Angeles County; the Australian Museum Trust, the Australian Marine Science and Technologies Advisory Committee, the Commonwealth Science and Industrial Research Organization Science and Industry Endowment Fund, and the employers of the contributors. The following individuals supplied specimens, data, technical assistance, publications, and reviewed drafts of manu- scripts: M. Allen. R. M. Allen, A. Alvarino, D. Ambrose, M. E. Anderson, W. D. Anderson. Jr.. F. Balbontin. C. Baldwin, E. K. Balon, P. Berrien, D. Blood, S. Boardman, S. S. Boggs, E. Bohlke, M. Bradbury, J. Brill, D. Brown, J. Bullock, M. S. Busby. J. A. Cambray, P. Camus, M. H. Carrington, B. Chemoff, T. A. Clarke, M. Culbreth, M. Cluxton, S. Coombs, A. S. Creighton, K. Davis, W. P. Davis. C. E. Dawson. M. Dehaan, N. Demir, A. Desai, H. H. DeWitt, M. DeWitt, Y. Dotsu, S. D'Vincent, B. R. Engstrand, D. Faber, N. R. Foster, P. Fourmanoir, C. Frandsen, H. J. Franke, E. Fridgeirsson, W. George, R. H. Gibbs, G. Gilmore, D. Gittings, W. Gladstone, T. Goh. M. F. Gomon. B. Goldman, A. R. Gosline, W. A. Gosline, A. E. Gosztonyi, P. H. Greenwood, D. Haggner, G. R. Harbison, G. S. Hardy, K. Hartel. R. Hartwick, T. Hecht, E. Hubert. J. M. Humphries. J. C. Hureau. T. Iwamoto. S. Jewett, P. Keener, S. Kelley, F. Kirschbaum, N. Komada, Y. Konishi, D. L. Kramer, J. K. Langhammer. K. Lazara, K. Lee, S. Lincoln, J. Lobon-Cervia, V. J. Loeb, G. Lundy, N. A. Mackintosh, F. Mago-Leccia, A. M. Martinez, D. McAllister, M. McCabe, J. McCosker. R. F. McGinnis. R. McMichael. R. Meier. N. Merrett. J. Michalski, J. Mighell, R. R. Miller, C. Mills, A. Miskiewicz, G. E. E. Moodie, K. H. Moore, K. Mori, J. Moyer, J. A. Musick, T. Nakata, G. Nelson, J. Nelson, J. Nichols, J. Nielsen, T. North, S. Ochman, G. Patchell, L. R. Parenti, K. Peters. T. Pomeranz, S. Poss, L. C. Prescott, J. Quast, J. Randall, K. S. Raymond, B. Remington, C. S. Richards, T. Roberts, D. E. Rosen. R. Schoknecht, A. Sekerak, T. Senta, J. Shapiro. J. Shoemaker, P. L. Shafland, M. Shiogaki, D. L. Schultz, P. H. Skelton, P. E. Smith, J. Song, D. E. Snyder, A. Soeldner, C. Stehr, D. Stein, B. Stender, K. Steward, K. Stoddard, R. E. Strauss, G. Stroud, K. J. Sulak, A. Suzumoto, H. Sweatman, J. N. Taylor, V. R. Thomas, G. Theilacker. R. Thresher, R. Triemer, D. Tweedle, J. C. Tyler, F. Utter, F. Van Dolah, R. Vari, B. Vinter. L. Vlyman, R. Wallus. T. Watanabe. B. A. Watkins, A. Wheeler, P. Whitehead, N. Wilimovsky, A. B. Williams, L. Wood, B. L. Yeager, P. Yuschak. H. Zadoretsky. B. J. Zahuranec. Illustrators deserve special praise and thanks. B. B. Washington illustrated a large majority of the specimens. Other illustrators include G. Mattson who served in this capacity with Dr. Ahlstrom for many years. B. Y. Sumida and H. Orr at the Southwest Fisheries Center. B. Vinter at the Northwest and Alaska Fisheries Center and J. C. Javech at the Southeast Fisheries Center. The original illustrations are archived at the Southeast Fisheries Center. Miami and Southwest Fisheries Center, La Jolla. During the final editorial processes, J. C. Javech and B. B. Washington mounted illustrations and remade many that were of marginal quality. C. Wolf coordinated and reviewed the literature cited section and P. Fisher typed the literature cited section as well as all last minute editorial changes. The Editorial Committee: H. G. Moser, Editor in Chief W. J. Richards, Managing Editor D. M. Cohen M. P. Fahay A. W. Kendall, Jr. S. L. Richardson CONTENTS Welcoming Address. By /. Barrett vii Frontispiece— Photograph of Elbert Halvor Ahlstrom. By /. R. Dunn viii Dr. Ahlstrom. By R. Lasker _ ix Introduction Ontogeny, Systematics and Fisheries. By J. H. S. Blaxter __ __ 1 Ontogeny, Systematics and Phylogeny. By D. M. Cohen 7 Early Life History Stages of Fishes and Their Characters. By A. W. Kendall, Jr.. E. H. Ahlstrom and H. G. Moser 1 1 Techniques and Approaches Early Life History Descriptions. By E. M. Sandknop. B. Y. Sumida and H. G. Moser _ 23 Synopsis of Culture Methods for Marine Fish Larvae. By J. R. Hunter 24 Identification of Fish Eggs. By A. C. Matarese and E. M. Sandknop 27 Identification of Larvae. By H. Powles and D. F. Markle 31 Illustrating Fish Eggs and Larvae. By B. Y. Sumida, B. B. Washington and W. A. Laroche 33 Clearing and Staining Techniques. By T. Potthoff 35 Radiographic Techniques in Studies of Young Fishes. By /. W. Tucker, Jr. and J. L. Laroche 37 Histology. By y. y. Govoni _ 40 Scanning Electron Microscopy. By G. W. Bochlert _ _ 43 Developmental Osteology. By J. R. Dunn 48 Otolith Studies. By E. B. Brothers __ 50 Preservation and Curation. By R. J. Lavcnhcrg. G. E. McGowen and R. E. Woodsum 57 Development and Relationships Elopiformes: Development. By W. J. Richards 60 Notacanthiformes and Anguilliformes: Development. By P. H. J. Castle _ 62 Elopiformes, Notacanthiformes and Anguilliformes: Relationships. By D. G. Smith 94 Ophichthidae: Development and Relationships. By M. M. Leiby 102 Clupeiformes: Development and Relationships. By M. F. McGowan and F. H. Berry 108 Ostariophysi: Development and Relationships. By L. .A. Fuiman 126 Gonorynchiformes: Development and Relationships. By H'. J. Richards 138 Salmoniforms: Introduction. By H '. L. Fink _ 1 39 Esocoidei: Development and Relationships. By F. D. Martin 140 Salmonidae: Development and Relationships. By A. W. Kendall. Jr. and R. J. Behnke 142 Southern Hemisphere Freshwater Salmoniforms: Development and Relationships. By R. M. McDowall _... 150 Osmeridae: Development and Relationships. By M. E. Hcarnc 153 Argentinoidei: Development and Relationships. By E. H. .Ahlstrom. H. G. Moser and D. M. Cohen 155 Stomiatoidea: Development. By A'. Kawaguchi and H. G. Moser 169 Stomiiforms: Relationships. By W. L. Fink 1 8 1 Families Gonostomatidae, Stemoptychidae. and Associated Stomiiform Groups: Development and Relation- ships. By E. H. .Ahlstrom. \V. J. Richards and S. H. Wcitzman 184 Giganturidae: Development and Relationships. By R. K. Johnson _ 199 Basal Euteleosts: Relationships. By W. L. Fink _ 202 Myctophi formes: Development. By M. Okiyama _ 206 Myctophidae: Development. By H. G. Moser. E. H. Ahlstrom and J. R. Paxton 218 Myctophidae: Relationships. By J. R. Pa.xton. E. H. Ahlstrom and //. G. Moser 239 Scopelarchidae: Development and Relationships. By R. K. Johnson 245 Evermannellidae: Development and Relationships. By R. K. Johnson 250 Myctophiformes: Relationships. By M. Okiyama 254 Gadiformes: Overview. By D. M. Cohen 259 Gadiformes: Development and Relationships. By M. P. Fahay and D. F. Markle _ __ 265 Gadidae; Development and Relationships. By J. R. Dunn and A. C. Matarese 283 Bregmacerotidae: Development and Relationships. By E. D. Houde 300 Ophidiiformes: Development and Relationships. By D. J. Gordon. D. F. Markle and J. E. Olney 308 Lophiiformes: Development and Relationships. By T. W. Pietsch _ 320 Ceratioidei: Development and Relationships. By E. Bertelsen __ _ _ 325 Atherinomorpha: Introduction. By B. B. Collette 334 Beloniformes: Development and Relationships. By B B. Collette, G. E. McGowen. N. V. Parin and 5. Mito 335 Atheriniformes: Development and Relationships. By B. N. White, R. J. Lavenberg and G. E. McGowen 355 Cyprinodontiformes: Development. By K. W. Able 362 Lampriformes: Development and Relationships. By J. E. Olney 368 Mirapinnatoidei: Development and Relationships. By E. Bertelsen and TV. B. Marshall 380 Beryciformes: Development and Relationships. By M. J. Keene and A'. A. Tighe 383 Zeiformes: Development and Relationships. By A'. A. Tighe and M. J. Keene 393 Gasterosteiformes: Development and Relationships. By R. A. Fritzsche 398 Scorpaeniformes: Development. By B. B. Washington, H. G. Moser, W. A. Laroche and W. J. Richards 405 Cyclopteridae: Development. By A'. W. Able, D. F. Markle and M. P. Fahay 428 Scorpaeniformes: Relationships. By B. B. Washington, W. N. Eschmeyer and K. M. Howe 438 Tetraodontoidei: Development. By J. A/. Lets 447 Balistoidei: Development. By A. Aboussouan and J. M. Lets _ 450 Tetraodontiformes: Relationships. By J. M. Lets - 459 Percoidei: Development and Relationships. By G. D. Johnson _ 464 Serranidae: Development and Relationships. By A. W. Kendall, Jr. _ 499 Carangidae: Development. By W. A. Laroche, W. F. Smith- Vaniz and S. L. Richardson 510 Carangidae: Relationships. By W. F. Smith- Vaniz 522 Mugiloidei: Development and Relationships. By D. P. de Sylva 530 Sphyraenoidei: Development and Relationships. By D. P. de Sylva 534 Polynemoidei: Development and Relationships. By D. P. de Sylva 540 Labroidei: Development and Relationships. By W. J. Richards and J. M. Leis 542 Acanthuroidei: Development and Relationships. By J. M. Leis and W. J. Richards 547 Blennioidei: Introduction. By R. H. Rosenblatt 551 Schindlerioidei: Development and Relationships. By W. Watson. E. G. Stevens and A. C. Matarese 552 Trachinoidea: Development and Relationships. By W. Watson. A. C. Matarese and E. G. Stevens _ 554 Notothenioidea: Development and Relationships. By E. G. Stevens. W. Watson and A. C. Matarese — 561 Blennioidea: Development and Relationships. By A. C. Matarese. W. Watson and E. G. Stevens 565 Ammodytoidei: Development and Relationships. By E. G. Stevens. A. C. Matarese and W. Watson 574 Icosteoidei: Development and Relationships. By A. C. Matarese. E. G. Stevens and W. Watson 576 Zoarcidae: Development and Relationships. By M. E. Anderson _ 578 Gobioidei: Development. By D. Ruple 582 Gobioidei: Relationships. By D. F. Hoese 588 Scombroidei: Development and Relationships. By B. B. Collette, T. Potthoff, W. J. Richards, S. Ueyanagi, J. L. Russo and Y. Nishikawa 59 1 Stromateoidei: Development and Relationships. By M. H. Horn _ 620 Gobiesociformes: Development and Relationships. By L. G. Allen 629 Callionymidae: Development and Relationships. By E. D. Houde 637 Pleuronectiformes: Development. By E. H. Ahlstrom, K. Amaoka, D. A. Hensley, H. G. Moser and B. Y. Su- mida 640 Pleuronectiformes: Relationships. By D. A. Hensley and E. H. Ahlstrom - 670 Literature Cited _ 688 Index _ - 746 Photograph of Symposium Attendees 760 VI Welcoming Address IzADORE Barrett Director of the Southwest Fisheries Center ON behalf of the National Marine Fisheries Service's Center Directors, sponsors of the Symposium on the Ontogeny and Systematics of Fishes, I am pleased and honored to welcome you to La JoUa. We are here to honor the memory of an outstanding biologist, Elbert Halvor Ahlstrom, known to his friends and colleagues as Ahlie, and his contributions to fisheries science. As fishery biologists we all recognize the vital importance and contributions of systematics and students of evolution to the development of fishery science. Less well known or appreciated is the unique role and interrelationship of the early life history studies of fishes and the assessment of the role of ontogenetic characters in fish systematics. This was, of course, the field of fisheries research to which Ahlie dedicated 40 years of his professional life and where he initially evolved the special methods and techniques which have so greatly influenced the work of fishery biologists around the world. I know that I speak for the Directors of the four fisheries centers— the Northwest and Alaska Fisheries Center in Seattle, the Southwest Fisheries Center in La Jolla, the Northeast Fisheries Center in Woods Hole, and the Southeast Fisheries Center in Miami when I say that I am proud that the National Marine Fisheries Service is the sponsor of this symposium. I believe that this gathering will be a landmark in fisheries science, a unique event which has brought together eminent scientists from 10 countries to present 87 papers reviewing the major fish groups, with particular attention to ontogenetic characters and their utility in assessing phylogenetic relationships. I fully anticipate that the resulting symposium volume which will be based on the papers presented here will stand as a definitive work in larval fish biology for many years to come. Again, a warm welcome to all of you and especially to Marge Ahlstrom who is seated in the audience this morning. I hope that the weather and circumstances will cooperate and that your stay here in one of the most attractive cities of the United States will be pleasant and productive. P.O. Box 271, La Jolla, California 92038. Dr. Ahlstrom Reuben Lasker MY colleagues have entrusted to me the pleasant task and distinct privilege of saying a few words in remembrance of Dr. Elbert H. Ahlstrom, to whom this symposium is dedicated. Like most of you I was his colleague for many years, 23 to be exact. He was also my friend and mentor to whom I could go when I needed advice and where I knew I would be heard as an individual with the bond of common scientific endeavors. For those of you who did not know Dr. Ahlstrom 1 would like to capsulize his enormous contribution to systematics and fishery science by outlining what I believe to be his major scientific contributions. Ahlie realized in the late 40's that the study of eggs and larvae could give us information about fish populations unobtainable from fishery statistics, the mainstay of fishery science at that time. He believed, rightly, that the ease with which eggs and larvae could be caught allowed an assessment of the geographic distribution and the seasonal extent of spawning of pelagic species. He recognized that any assessment of a fish population was dependent on surrounding that population in time and space and that this would require a major effort. He was the first. I believe, to determine the extent of a major pelagic fish population using this technique. The simplicity and thoroughness of the plankton net made an impression on him and, while he sought to improve collecting techniques constantly, he consistently analyzed the errors of the plankton net so that this tool could be used more and more reliably. Today, it is still one of the most powerful collecting and assessment tools we have, largely because of his diligence and persistence. The scope and thoroughness of Dr. Ahlstrom's work was particularly important. His taxonomic skills are attested to in the many papers he wrote and which stand today as mainstays of the systematic and fishery literature. He liked to use the title "Kinds and abundance of fishes" and usually provided taxonomic lists in these of several pages in length. His point, of course, was to detail the complexity and uniqueness of particular oceanic regimes and to set the ground work for ecological research which inevitably followed. Well, what of his other attributes? I used to call him the modem Renaissance Man because I realized whenever I had occasion to meet him socially that he knew almost all there was to know about the arts and the sciences. Of his fabulous classical record collection 1 recall that 1 asked him once if he really listened to all of them. His reply was "we used to hear each one once a year, but now, since the collection has grown so large, it's once every two years." He belonged to the San Diego Great Books Society, and read them all. Engage him in conversation and you would find out quickly he knew literature, fine wines, photography and baseball, to name a few. I would like to sum up this brief eulogy by pointing out an example of one aspect of Ahlie which holds my greatest admiration: that is, his dedication to work. One incident during our relationship illustrates the point I wish to make. When Science Fairs started to become the vogue in San Diego, Dr. Ahlstrom was asked to host a group of young Science Fair participants to teach them something about oceanography. He arranged to take out the old Bureau of Commercial Fisheries ship, the Black Douglas, for a day to illustrate collecting methods at sea. In fact, the day was beautiful, but there was a swell upon the sea and no sooner did we get out of the harbor than almost everyone, except Ahlie and some of the seasoned veterans, felt the effects of a rather pronounced roll for which the Black Douglas was famous, even in the calmest of seas. Dr. Ahlstrom proceeded with his typical dedication to illustrate Nansen bottles, plankton nets, and bathythermographs to the group of Science Fair students who were becoming less and less interested and more and more seasick. Ahlie continued with a single-mindedness of purpose and a dedication that was so characteristic of him. Without his noticing, a caucus was held by these young students and a representative meekly asked, "Dr. Ahlstrom, may we please go home?" Two versions of what happened next were told to me later. The first was that Ahlie responded immediately to the problem and ordered the ship to port. Another version was that Ahlie continued until he was finished, made sure he had a proper sample, and then ordered the ship into port. I'm afraid I can't tell you which is correct— I was in a bunk, seasick! I meant this story as a small illustration of Dr. Ahlstrom's dedication to his work. He was a dedicated scientist who had an insatiable curiosity about the biotic world and who was convinced that what he was doing was important and would advance fishery science. This symposium is one piece of evidence that he was right. Now the question must be asked— how is it that Ahlie could be so dedicated to work and yet have found time to become a true example of a Renaissance man, with a deep knowledge of art, wine, architecture, photography, sports, and much more? I pondered this with admiration for many years and I think I have the answer. He was one of those rare individuals who never cease learning, because he had a true scholar's love for learning. I like Robert Whittenton's description of Sir Thomas More when I think of Ahlie: he was, like More, "a man for all seasons." Southwest Fisheries Center, P.O. Box 271, La Jolla, California 92038. Photograph of Elbert Halvor Ahlstrom, by J. R. Dunn. INTRODUCTION Ontogeny, Systematics and Fisheries J. H. S. Blaxter IN the inter-war years work on fish eggs and larvae was Umited to studies on horizontal and vertical distribution with a view to completing our knowledge of the early life history of different species. Resources for research were then much more limited than they are today and most work was done on the important food fishes. In the 1 950's a great expansion took place as fisheries biologists realised how much a study of early life history would be a key to solving some of their problems. This expansion took place on a broad geographical and mtemational front, but great credit must be given to the foresight and imagination of E. H. Ahlstrom. who built up a team of biologists at La Jolla who then and subsequently, played a major role m leading and de- veloping this field with special reference to the fisheries of the California Current. In the last two decades the output of publications has risen at an exponential rate as evidenced, for example, by the 62 papers in the 1973 Early Life History Symposium held in Oban (Blaxter, 1974) and the 139 papers in the 1979 Symposium at Woods Hole (Lasker and Sherman. 1981). Furthermore, in a selected hibhography of pelagic fish and larva surveys prepared by Smith and Richardson (1979), some 1200 papers are listed, most of them published in the last 30 years. Ahlstrom was certainly a major catalyst in this reaction, but it is sad to record that his obituary appeared in the Proceedings of the 1979 Sym- posium, although he was still alive and present at the meeting itself to impart his wisdom and expertise. It is proposed to discuss the post-war advances in our knowl- edge of early life history stages under five headings: (1) as they impinge on systematics and taxonomy. (2) the success and role o{ experimental work in tanks and of modelling, (3) the scaling- up of tank studies to large enclosures and embayments, (4) the application oi sea surveys to test models, to investigate the stock- recruitment relationship and to measure spawning stock bio- mass, and (5) \he future. Systematics and Taxonomy A number of techniques have been developed to help in the identification and classification of fish larvae. Since the devel- opment of the skeleton and meristic characters are now so im- portant in identification, techniques of clearing and staining or x-radiography have become standard methods for examining the internal osteology of larvae (Ahlstrom and Moser, 1981). Morphometries and body pigmentation are also important and are used extensively by Russell (1976) in his monograph on fish eggs and larvae of the N.E. Atlantic. Rearing experiments have shown that the sequence of de- velopmental events may also be specific in character. For ex- ample the development of the acoustico-latcralis system and swimbladder in herring as shown by Allen, Blaxter and Denton (1976) is a long-drawn-out affair and quite different from that of the larval anchovy as described by O'Connell ( 1 98 1 a) or the menhaden or sprat. There are several larval features, such as the swimbladder and other internal organs, or features of the labyrinth, which would help in the separation of similar-looking species if only they were not obscured by fixation. Often the taxonomist (or fisheries biologist) resorts to count- ing menstic characters such as vertebrae, fin rays, scales or gill rakers. Yet many of these characters have been shown by ex- periment to be labile and to respond to environmental condi- tions during early development. The earlier work, mainly on freshwater species such as the sea trout, was summarised by Taning (1952). Since then a range of further studies by Fahy, Lindsey (e.g., see Fahy, 1982) and others have confirmed the earlier experiments, showing that temperature, salinity and oxy- gen level influence meristic counts and that there is a critical period when this influence operates. Little work has been done on marine species although Hempel and Blaxter (1961) showed that temperature and salinity both influence myotome and ver- tebral counts in herring (the species in which stock separation by meristic counts has been most widely applied). It seems likely that any environmental variable which influ- ences the relationship between differentiation and growth will affect the meristic count by determining the amount of embry- onic tissue which is present when the differentiation into skeletal units lakes place. The larval taxonomist needs to be cautious in interpreting small differences in meristic values, especially when they are related to clines or other types of geographical distribution. That is not to say, however, that there is no un- derlying genetic mechanism. The environment acts as a "fine- tuning" mechanism. Whether this fine-tuning is accidental or adaptive might well be worth discussion at the symposium. A warning also needs to be directed at morphometries. Rear- ing experiments in different-sized tanks by Theilacker ( 1 980b) show the influence of space on growth rates. Compansons of reared and wild fish larvae, especially of herring by Blaxter (1976), show that tank-reared fish are often shorter and fatter than their wild counterparts at the same developmental stage. There seems to be an interplay between diet and activity which is enhanced by the confinements of the rearing tank. This makes it difficult to extrapolate growth criteria from tanks, such as condition factor, to establish, for example, the nutritional status of larvae at sea (Fig. 1). A further and serious problem identified by the handling and use of live larvae is the shrinkage caused by capture and fixation. A number of workers such as Blaxter (1971), Schnack and Ro- senthal (1978), Theilacker (1980a) and Bailey (1982) have ad- dressed this problem but the most significant findings are those of Hay (1981) on Pacific herring. Feeding larvae from rearing experiments were released into the mouth of a plankton net at sea and then fixed by various techniques after capture. Shrinkage in body length ranged from a mere 5% to a massive 43% de- pending on the technique. Extensive voiding of gut contents also occurred. The implications of these results in morphometric or feeding studies will not be lost on the present audience. 1 ONTOGENY AND SYSTEM ATICS OF FISHES -AHLSTROM SYMPOSIUM •20 .16- (J •12 •08- HERRING WILD 12 16 LENGTH (mm) "20 Fig. 1 . Comparison between range of condition factors (C.F.) as dry weight/length^ of wild herring caught at sea by plankton net and reared herring larvae near starvation (from Blaxter, 1976). Finally, the ageing of larvae by daily ring formation in the otoliths should be mentioned. This technique was pioneered by Brothers et al. (1976) on anchovy larvae and California grunion following Pannella's suggestion that daily increments were being laid down in the sagittae of some temperate fish species. The findings were validated by rearing larvae in tanks and sampling the population at intervals of 1-7 days. Struhsaker and Uchi- yama (1976) supported these results from their work on the Hawaiian nehu and subsequently the technique was widely adopted in fisheries laboratories. Attempts by Geffen (1982) to manipulate ring formation in cod, herring, plaice, salmon and turbot larvae by varying the photoperiod, temperature and feed- ing regimes did not lead to any consistent result — the ring de- position was frequently not daily and the main determinant in herring and turbot seemed to be growth rate— the higher the growth rate, the higher the rate of ring deposition. Bailey (1982), however, found otolith rings deposited daily over a 10-day pe- riod in post yolk-sac Pacific hake larvae reared in tanks. Sea- caught larvae with more than about 30 increments were less satisfactory because of the appearance of different types of ring and it was not certain whether they were daily. Dale (1984) in a recent study of reared Atlantic cod otoliths using electromi- croscopy, found daily rings in a 12L/12D cycle but not in the dark. Daily ring deposition only continued, however, for a few days post-hatching. Although the ageing of anchovy and grunion from daily rings seems reliable, further validation experiments are required at sea. This is conceptually difficult on a wild stock of larvae of mixed age and it is notoriously difficult to remain over a single population of larvae for many days. Mass release of reared larvae into the sea remains an ambitious possibility. Perhaps best of all such a release should be into some large enclosure system initially free of a larval population. Validation experi- ments must also test the more unusual environmental condi- tions which apply in high latitudes where, for example, daylight prevails over the full 24 hours. Experimental Work The functional anatomy approach to taxonomy so elegantly described in a recent review by Moser (1981) shows the extent to which structure can be used to deduce function. The inter- action of this approach with that of the experimentalist has yielded much useful information. Since the 1950's increasing success in rearing marine fish larvae may have provided the taxonomists with help as well as some doubts as described in the last section. It has also led to a wide literature on the physiology, behaviour and physiological ecology of larvae (and the use of larvae in pollutant bioassay) as biologists seized the opportunity to exploit such new and valuable material. Perhaps the most credit should be given to Shelboume (1964) for his extensive and painstaking rearing ex- periments on plaice, and later sole, at Port Erin, Isle of Man. These experiments undoubtedly led to the present wide practice of marine finfish aquaculture with the expanding commercial use of turbot, sole, bass, bream and gilthead. Rearing may still be considered as something of an art and is often most successful in the hands of dedicated people with a "feel" for what is right or wrong. Undoubtedly a breakthrough was made in finding suitable food for larvae. It is significant that both plaice and sole can take Anemia nauplii from first feeding as can some races of herring. This resulted in another U.K. focus for rearing at Aberdeen, and later Oban, developed by Blaxter (1968) on the herring. Species with smaller larvae (with smaller mouths) were only successfully reared when Las- ker's group at La Jolla (Lasker et al., 1970; Theilacker and McMaster, 1971; Hunter, 1976) developed the use of the rotifer Brachionus plicatilis and the naked dinoflagellate Gymnodmium splendens as small food items for early-stage larvae of species like northern anchovy and jack mackerel. About the same time Howell (1973) also used Brachionus to rear turbot larvae at Port Erin. Subsequently a number of factors have been identified to add to our corpus of knowledge on rearing. These include the need for good water quality, with the interesting idea of "green water" culture of larvae in fairly high densitiesofC/j/oreZ/a which seems to damp out fluctuations in metabolites, and perhaps enhance oxygenation as well as providing secondary feeding for the larvae (e.g., Houde, 1977; Morita, 1984). Adequate light for visually- feeding larvae and the need to prevent excessive bunching of larvae or their prey are also important, as is the quality of the food. Success or failure may now depend on the fatty-acid profile of the Anemia nauplii which are still used by most workers in the later stages of rearing. Artificial diets of encapsulated or particulate food are also being developed but have yet to be introduced as a standard technique for early rearing. Before turning to the extrapolation and application of exper- imental data to modelling, mention must be made of Haydock's (1971) and Leong's (1971) work on the induction of spawning in the croaker and anchovy by pre-treatment with an appro- priate photoperiod followed by hormone injection. This has been applied subsequently to the menhaden by Hettler (1981), and to many other species, and has become a standard method for workers requiring eggs over long periods or at a specific time. We now have the widest knowledge of the development, be- haviour and physiology of both anchovy and herring larvae (see Fig. 2) but there are several species such as cod, jack mackerel, mackerel, plaice and turbot which run them a close second. BLAXTER: ONTOGENY, SYSTEMATICS, FISHERIES Lateral line Respiration Red muscle Reynolds number (Re) and hydrodynamic Viscous regimes Digestive tract Re200 Stomach forms .It- Expandable mouth 3.3 days 4 days T^ 1 1 1 — I 1 1 1 1 1 1 1 1 1 1 1 — 1- 15 20 25 30 Length (mm) Juvenile period Filler feeding 15 days * 1 — I 1 1 1 1 1 — 2.5 5 10 15 20 25 30 35 40 45 Days at 16° C — 1 1 1 1 1 1 1 — 50 55 60 65 70 75 80 Fig. 2. Events during development of the northern anchovy. RBC = red blood cells. Time to 50% starvation is number of days to starvation at which 50% of the fish died (from Hunter and Coyne. 1982). Much of this work is summarised by Theilacker and Dorsey (1980). Over the past few years the assembly of much basic data has allowed the current vogue for modelling to be applied to fish larvae. Modelling is an attempt to synthesise and simplify basic data usually in mathematical form. Mathematical models are often iterative and they have the value of being in a form suitable for computers. Laurence (1981) has recently reviewed modelling work on fish larvae and the complexity and type of interaction is shown in Fig. 3. The main problem addressed has been that of feeding. The earlier models of Blaxter (1966), Rosenthal and Hempel ( 1 970), Blaxter and Staines ( 1 97 1 ) and Hunter (1972) estimated the feeding efficiency of larvae, the volume of water searched in unit time and the density of food required to give good survival and growth. More sophisticated models have now been developed (e.g., Jones and Hall, 1 974; Beyer and Laurence, 1981) and Vlymen's (1977) model allows for the prey species being non-randomly distributed. The need for larvae and their prey to co-exist temporally was spelled out by Gushing ( 1 975) in his match-mismatch hypothesis. Thus the timing of reproduction appears to have evolved to synchronise the larval stages with the main phase of the annual production cycle. Spawning is probably controlled in most tem- perate fish species by photoperiod and temperature which are not the only determinants of plankton production. Hence a match or mismatch is possible between this production and the presence of fish larvae with a resulting influence on year class strength. An early paradox existed in that the density of the larger micro-zooplankton such as copepod nauplii required for good growth and survival in tanks was of the order of 1 organism/ ml. Such densities are rarely found in the sea as judged from normal plankton sampling. This led to the suggestion of micro- scale patchiness of food in the sea, which might occur at inter- faces such as steep thermoclines and at tide- and wind-induced fronts. The integrity of such microscale patchiness would not, of course, be obvious using nets sampling large volumes. This led Lasker (1975) to bioassay samples of water taken at different depths and places off the Califomian coast, using an- chovy larvae both hatched and tested on board ship. Chloro- phyll-rich layers with very high densities oi Gymnodinium were found near the thermocline. The bioassay showed good larval feeding in these water samples, suggesting that patchiness, in- deed, might be a valid concept. This was to some extent con- firmed by later findings that stable weather conditions (which maintained the thermocline) favoured good year classes of an- chovy larvae off the Califomian coast (Lasker, 1981). Owen ( 1 980) has subsequently shown from samples taken by plankton pumps and water bottles that patchiness of microzooplankton such as copepod nauplii and tintinnids and various protozoan species and phytoplankton (some of which are known to be the food of anchovy larvae) exist off the Peruvian and Califomian coasts on the scale of a few centimetres up to one metre (see Fig. 4). Only Houde and Schekter ( 1978) have attempted to rear larvae in simulated food patches and found that survival of sea bream was similar when they were exposed to 3 h of food per ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM PHYSICAL A CHEMICAL INFLUENCES CIRCULATION DIFFdSlON 80UNDARV EXCHANGE DISCONTINUITY EVENTS (STORMSj PRIMARY PRODUCTION & ORGANIC RECYCLING POLLL TION OR TOXICITY ABIOTIC FACTORS (TEMPERATURE SALINITY. OXYGEN) IMPORTANT I ARVAL FISH INTERACTIONS LARVAl PREDATORS MOSTLY UNIDENTIFIED RECRUITMENT ASSESSMENTS USHERY MODELS ECOSYSTEM MODELS MANAGEMENT STRATEGY LARVAL MORTALITY DETERMINED DIRECTLY IN LINKAGES A A B o o tl'RREMLY rWPHASlZED STL:nrFS STtlDIES FOR EMPHASIS SI I DthS ME IMMFniAlE I ESSEk IMPORT ASCE DIRECl HACnONAI IINKAGES I ISKAGES OE I ESSER IMPORTANCE Fig. 3. A generalised scheme for the main interactions between larval fish and their biotic and abiotic environment, providing a basis for modelling (from Laurence, 1981). day as when fed at the same food level continuously. Clearly, expeiiments need to be devised to test the effect of spatial rather than temporal food patchiness. The evidence is thus accumulating, but very slowly, that lai^al survival may depend on the extent and stability of microscale food patches or interfaces, at least in some areas. It may be that the rather high food densities required in small-scale tank rear- ing do indeed apply to conditions in the sea and that such densities are only found in patches. SCALING-UP Two major areas may be identified where rearing work has been extended into large-scale containers. The first of these are the large onshore enclosures and embayments used by the pres- ent generation of Norwegian biologists; the second are the deep- water plastic bags used by Scottish workers in Loch Ewe on the Scottish West Coast. The Norwegians have achieved remarkable growth and survival rates for herring and cod larvae, as high as 30-70% survival from hatching to metamorphosis, in shallow 4,000-60,000 m^ enclosures (Oieslad and Moksness, 1981; Kvenseth and Oiestad, 1984). The Loch Ewe bags, which are deep cylinders, of about 300 m\ have been used for rearing herring and cod, but with much less success than the Norwegians (Gamble et al., 1981; Gamble and Houde, 1984). Possibly volume itself is important, or more likely the ratio between volume and wall area. The interface between wall and sea water is not a natural one for fish larvae, feeding may be difficult al the interface, and food may aggregate there in an inaccessible form. Morita (1984) reports that Pacific herring larvae have recently been reared in 20 m' tanks with a 46% survival from hatching to a mean length of about 7 cm in 1 1 2 days. This spectacular result may have been partly a feature of a fairly large onshore tank but also the "green water" technique mentioned earlier. Hunter (1984) suggests that the high survival in some large tank or enclosure experiments is achieved by the elimination of predators. To the present author a combination of optimal feeding conditions and low predation seems to be the likely cause. The events have been described so far in a topsy-turvy way, in that sea surveys have always been the most widely-adopted approach to problems associated with the early life history of fish. The experimental and enclosure studies are the icing on the research cake, although both Norwegian and Japanese work- ers are seriously considering the possibility of restocking de- pleted inshore fisheries or topping-up poor year-classes of cod and herring by releasing reared late-stage larvae or O-group juveniles. Sea Surveys These are expensive in terms of ship-time and manpower. Originally designed to advance our knowledge of spawning grounds, larval drift, and horizontal and vertical distribution, they are often now linked to more practical aims. Nevertheless, superb time-series exist for areas like the California Current and North Sea as a result of the patience and foresight of earlier workers like Ahlstrom and later workers like Smith and Saville (see review by Smith and Richardson, 1977). Sea surveys have always been a rich ground for innovative science, in terms of sampling techniques, interpretation and usage. Experimenters and modellers have provided a great boost for this work, allow- ing new interpretations to be made and new hypotheses to be tested. No more mention will be made of the matrix-filling role of sea surveys— namely the completion of details of life history, which is still taking place and has been much aided by the vast improvement in egg and larval identification in the past two BLAXTER: ONTOGENY, SYSTEMATICS, FISHERIES CONCENTRATION (no/X) 20 D 1000 2000 3000 <■ ^ X J 20 4 "-^^ J e Prorocentrum — ^ ^ — ■C' ' _-^-^ ^ I ?0 8 J^^ \ — -Nit/schia 1— — "" " " ^ — \ a. - < J^ UJ Q 212 f^"^'^"^"'--- V L____ "jt 216 PHYTOPLANKTON 25 20 20 4 20 8 21 2 CONCENTRATION (no /i) 50 75 100 216- Fig. 4. Vanation in concentration of microplankton in samples from 20 cm depth intervals in the chlorophyll maximum layer over the coastal shelf of the Southern California Bight dunng March. 1976. Prorocentrum, tintinnids and copepod nauplii are all food items for larval anchovy (from Owen. 1980). - • •r- ) Tintinn r t { Nauplior s-^r copepods / - — Noctiluca MICRO-ZOOPLANKTON decades. Improvement in plankton nets and young fish trawls means that vertical profiling and quantitative sampling have finally come-of-age. This ability to sample quantitatively is the single most important advance in allowing larval populations to be assessed reliably and for allowing models to be tested. The outcome is two-fold. The door is open for biomass estimates of spawning stock from egg and larval surveys and for testing the possible factors in the stock-recruitment relationship. Each of these will be considered in the final part of this paper. /. Biomass estimation. — ¥or many years population dynami- cists lacked good information on the absolute size of the spawn- ing stock and regulation was largely achieved by minimum mesh and landing sizes. Of late, as a result of catastrophic declines in some species, whole fisheries have been closed or controlled by quotas and total allowable catch (TAC). The use of TAC's has been greatly aided by virtual population analysis and also by sonar-based fish counting surveys; these give an estimate of total stock size, the reliability of which depends on the extent of the survey, the ability to identify the species in question and the precision of the calibration of target strength. To supplement the results, estimates of spawning stock size have been made on an ad hoc basis by counting eggs and larvae and converting them into the parental spawning stock biomass by a knowledge of fecundity, age distribution and sex ratio. Some of the pioneering work was done by Sette and Ahlstrom (1948) on Califomian pilchard and Simpson (1959) on North Sea plaice. Saville, Baxter and McKay (1974) counted the demersal eggs of the herring on the small spawning ground of Ballantrae Bank in the Clyde. This was later extended by Saville and McKay (see Saville, 1981) to herring larval surveys in the North Sea and off the Scottish west coast. The biomass of Pacific hemng is now routinely assessed from the intertidal egg deposition along the coast of Canada and the USA as described in the recent Nanaimo Herring Symposium (Hay, 1984; Haegele and Schweigert, 1984). Similar, but ad hoc. data are available for the northern anchovy from the work of Smith (1972), Parker ( 1 980) and Picquelle and Hewitt (1983), for the Atlantic mack- erel from Lockwood, Nichols and Dawson (1981) and Berrien, Naplin and Pennington (1981) and for North Sea cod from Daan (1981). Some of these data give absolute measures, some relative ones from year-to-year, often related to biomass estimates by other means. This survey technique has notable disadvantages. It must be done at a limited time of year and is obviously easiest to interpret for one-off spawners. The survey must be done rapidly and as near the spawning season as possible to overcome any errors caused by mortality between spawning and sampling. Although it can be applied to a closed fishery, the age structure of the population is required to compute the aggregate fecundity, hence scientific sampling of the adults is required. 2. Stock-recruitment.— The relationship between the size of the spawning stock in any year and the number of recruits it supplies to the fishery subsequently is vital information for the regulation of fisheries. This is specially true where recruitment overfishing is prevalent as in the clupeoids. Over many years a stock-re- cruitment relationship may be obtained empirically in any fish- ery, but this is time-consuming and usually contains inexplicable features. While, as might be expected, low spawning stock leads to low recruitment, high spawning stocks may also give unex- pectedly low recruitment, as the result of density-dependent effects. Alternatively spawning stocks of a given size can yield enormously different brood strengths, of the order of 10-100 times, in a quite unpredictable way. It is not surprising that the underlying causes of the control of brood strength are of much interest to fisheries biologists and have received the attention of experimentalists and modellers. Most marine fish have a very high fecundity, of the orders of tens of thousands to a few million. From such a starting point mortality must be very high and it is surprising that brood strength variations are not even more variable than is actually the case. What then do we know of the mortality rate of eggs and larvae in the sea? Are there critical periods when it is es- pecially high? What are the causes of mortality? Hjort's original hypothesis, now some 70 years old, expressed the view that a critical period existed after yolk resorption as the larvae sought external food sources. This hypothesis was supported by earlier rearing experiments in which very high ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM mortalities occurred at first feeding. Measurements of mortality rates of eggs and larvae at sea tend to show a high but continuing mortality of perhaps 5-20% per day. The results of sea surveys are, however, often difficult to interpret because of the need to sample within a discrete larval population over a long time. May (1974), in his review of this subject, concluded that star- vation at the end of the yolk-sac stage may often have a major influence on brood strength but that mortality from fertilization to the O-group stage is the ultimate determinant. The results of modelling and the tests of the patchiness hy- pothesis which have already been discussed support the idea that first feeding is a critical time, although not having, neces- sarily, the dominant effect claimed by Hjort. Experimenters and modellers have also derived further concepts for testing. The major sources of mortality are identified as starvation and pre- dation. Starvation, of course, only operates from the end of the yolk-sac stage. Blaxter and Hempel (1963) used the expression "point-of-no-retum" to express the point at which larvae, as a result of starvation, are too weak to feed even if food becomes available. Sometimes called "ecological death" or "irreversible starvation" this is a useful concept for assessing the chances of larval survival under different conditions. For larvae in a good nutritional state the time to the point-of-no-retum may be only 1-2 days in a small newly feeding larva like the anchovy, but 2-3 weeks in a well grown flatfish larva like the plaice (see Theilacker and Dorsey, 1980). Implicit, also, in the concept is that larvae can live for some time after the point-of-no-retum. During this time they may be especially liable to capture by nets and, without adequate knowledge, a false impression might be obtained of the size or nutritional state of the larval population. The assessment of nutritional state of larvae has been of wide interest in recent years, in the hope of relating this to brood strength. Initially Blaxter ( 1 965) measured the condition factors of tank-reared herring larvae after varying periods of starvation and then later compared the results with the condition factors of sea-caught herring larvae (Blaxter, 1971). It was found that most sea-caught larvae had much lower condition factors than starving tank-reared larvae and it became apparent that the extrapolation of tank criteria to the sea was invalid because the tank larvae were short and fat compared with wild larvae (see Fig. 1). This means that condition factor comparisons of wild larvae are only valid on a relative basis from year-to-year or place-to-place (e.g., Chenoweth, 1970; Vilela and Zijlstra, 1971) and only then if one can be satisfied that shrinkage after capture is consistent. The problems of tank ; sea comparisons and shrinkage are unfortunately likely to be the most serious in long clupeoid larvae to which these experiments have been applied. No one has checked their validity in the more common type of larvae with a shorter body form. These problems led to work at Oban and La Jolla on histo- logical criteria for assessing starvation (Ehrlich et al., 1976; O'Connell, 1976; Theilacker, 1978). O'Connelfs work on an- chovy larvae deserves special mention. He found from screening the state of the body organs such as pancreas and gut that these showed increasing signs of degeneration as starvation pro- ceeded. On applying his criteria to sea-caught anchovy larvae O'Connell (1981b) found evidence for quite a high percentage of larvae suffering from advanced starvation and considerable differences in the incidence of starvation in closely adjacent areas. This method is now being applied by Theilacker on jack mackerel larvae from year-to-year and is likely to be adopted on a routine basis. The other cause of mortality, predation, has recently become fashionable following the work of Eraser, Lasker, Lillelund and Theilacker and subsequently Kuhlmann, von Westemhagen and Rosenthal, Bailey, Purcell and several other workers (See re- views of Hunter, 1981, 1984). Copepods, euphausiids, amphi- pods and chaetognaths are all implicated but perhaps medusae are the most voracious group of predators (Bailey and Batty, 1983), especially for inshore spawners like Pacific herring. Pre- dation, of course, operates from the moment of spawning and Hunter and Kimbrell( 1980) and MacCall (1980), in particular, have discussed the incidence of density-dependent cannibalism of spawning anchovies on their own eggs and larvae. It is gen- erally thought that strong selection pressure exists for fast growth which will take larvae speedily through the more vulnerable early stages. Larvae have been shown experimentally to be less vulnerable when they are larger, their escape speeds are higher and their recovery from a predator attack (for predators of a given size) more likely. As Hickey (1979, 1982) has shown, an efficient wound-healing mechanism exists, allowing larvae to recover from bites, stings and other forms of damage. The high survival rates of larvae reared in the absence of predators (Kven- seth and Oiestad, 1984; Morita, 1984) suggest strongly that predation is a major source of mortality in the sea. Although it is difficult to assess the relative importance of starvation and mortality in any larval population, it is also clear that the two must interact in the sense that starving larvae will be more susceptible to predation. The Future In this paper modelling has been only briefly discussed. The method is now widely used for setting up hypotheses about feeding, starvation, predation, cannibalism and other factors associated with the stock-recruitment relationship and biomass estimation. This approach is likely to continue as a basis for sea surveys. It seems uncertain whether biomass will be routinely estimated by egg and larval surveys except perhaps in Pacific herring and northern anchovy. The cost is too high and sonar surveys, if the problems can be ironed out, seem to be a better bet. Experimental data on predation still need to be collected and few correlations exist between predator populations and egg and larval mortality in the sea. In fact mortality studies on eggs and larvae in the sea in general need to be perfected since the prob- lems of following discrete populations and of ageing larvae are still not fully solved. At least one source of information is largely untapped and that is the explanation for the high survival rates of larvae in large enclosures. In particular the distribution of the larvae and their food in these enclosures is not known and may throw light on the validity of the patchiness hypothesis. Information on frontal systems, and interfaces as a result of tide, wind, upwelling and thermo— and halo— clines is now quickly being assembled by hydrographers and marine biologists. The larval biologists should be ready to exploit the results. It will be apparent to the audience how far research into the early life history of fish has advanced in the last 30 years. A major force has been the work off"the Califomian coast generated by Ahlstrom and his recruits at La Jolla. It is therefore very fitting that this symposium should be dedicated to his memory. Scottish Marine Biological Association, Dunstaffnage Marine Research Laboratory, P.O. Box 3, Oban, Argyll, Scotland. Ontogeny, Systematics, and Phylogeny D. M. Cohen THE work of Ahlie and his students and colleagues has brought to the fore great amounts of descriptive information about the early life history (ELH) stages of fishes gathered over many years. These data are of broad provenance, many being the results of original research by the Ahlstrom school, others being taken from the literature. Only a scientist with Ahlie's capabil- ities—an extensive knowledge of fishes and their ontogeny, a fine sense of order in nature, and a critical intellect— could per- ceive pattern in the bewildering diversity represented by the early life history stages of fishes. As would any good scientist, Ahlie questioned the meaning of these patterns, and it is chiefly to further this inquiry that this symposium was convened. Most students of comparative fish ontogeny know more about adult fishes than ichthyologists who study adults know about larval fishes; they have to. Ahlie stated in his lectures. "Larval taxonomy is just an adjunct to adult taxonomy and you have to start with the adults to know the larvae." Early on he dis- covered that data from early life history studies did not always confirm classifications based on adults alone. We all want to know which data sets most closely approximate phylogenetic relationships; how apparent conflicts best can be resolved; how the data of ontogeny can be integrated into the overall field of fish systematics? Answering these questions is not easy, espe- cially within the framework dictated by the widespread adoption of new methodologies in systematics, which claim to require more stringent evaluation of characters than has been heretofore customary. Many traditional character suites are being rejected for purposes of elucidating phylogenies, and new data are needed for testing. Our purposes m this volume are to state the bases for what has come to be called larval fish taxonomy and to consider the systematics of various groups of fishes in terms of the rich and virtually untapped store of data offered by the study of early life history stages. My own objectives in the present paper are several. First of all. I want to indicate the reasons, some obvious, some not, for the nearly exclusive use of adult fishes in systematics, which has prevailed until very recently. Secondly, I will briefly discuss the conceptual and methodological framework of classification within which early life history data is being used. Finally, I will comment on the possible importance of early life history data for the study of phylogeny with special reference to fishes. Why Has There Been So Little Use of ELH Stages in Fish Systematics? The fact that most fish classifications are based entirely or chiefly on the structure of adults was a source of concern to Ahlie and remains so to many of us, although this Symposium is an indication of positive change. I discuss below what may be some of the reasons for a long preoccupation with adults. In the first place, zoologists have been studying adults for a longer period of time than they have early life history stages. Although the dim beginnings of classification are often placed with Aristotle, it was the great naturalists Aldrovandi. Belon. Gesner. and Rondelet who in their cataloging of nature provided our earliest adult fish classifications. Several technological de- siderata would have prevented the study of early life history stages during the 1 6th century when these early scientists were at work. Even though lenses had been known for a long time, appropriate microscopes were not invented until the 1 7th and 18th centuries (Singer, 1959) when another requisite advance occurred, the use of alcohol and other fluids as a preservative for zoological specimens (Singer, 1950). Techniques for clearing flesh and staining bone and cartilage are modem acquisitions, as is the use of x-ray photographs (Ahlstrom and Moser. 1981). The invention of fine-mesh towing nets did not occur until 1 846 (Sverdrup. Johnson, and Fleming, 1942), deferring until rela- tively recent times the availability of suitable collections of early life history stages for scientific study. The rearing of early stages is another valuable component of the study of larval fish taxonomy, and although fish culture is an ancient art, the staging of fry and their preservation and microscopic study is technology-dependent and relatively re- cent. Lack of information on metamorphosis or of congruence of larval and adult stages has also delayed the adoption of early life history stages information into classification schemes. Of course not many kinds of fishes demonstrate an ontogenetic change as sudden and dramatic as do the eels, but the fact that this particular transformation was not described until 1897 (Grassi and Calandruccio) indicates the long advance start held by the use of adult stages. Even more recent have been discovery of the Anoplogaster-Caulolepis relationship (Grey, 1955a), the Gibberichthys-Kasidoron relationship (de Sylva and Eschmeyer, 1977), the Giganturidae-Rosauridae relationship (Johnson, this volume), and the as-yet-unpublished identity of larval forms such as Svetovidovia. These and other examples are described in this volume. And indeed, even when the study of the devel- opmental biology of vertebrates commenced, early emphasis in the mid- 18th century was on classical embryology, the describ- ing of processes and structures rather than on comparing them (Rostand, 1964). Not until the early years of the present century when fishery scientists began to use larval fishes in their inves- tigations of commercial species and required identifications were serious efforts made to compare data (Ahlstrom and Moser, 1981). Until Ahlie commenced his now famous courses on larval fishes, there were few places where a student could learn about them; hence, there are only rare instances of attention being paid to any potential value they might have in solving problems in systematics. By now, in contrast, there are courses and sem- inars available in a number of universities on the study of ELH stages of fishes. Another phenomenon that I believe has inhibited the use of early life history stages in fish systematics is what I call the curatorial mind set. Many curators of adult fish collections are wary of microscopic specimens stored in vials. Although these collections occupy small space, their maintenance and docu- mentation are labor-intensive and their use is foreign to most ichthyologists. There are many excellent collections of larval fishes, but they are mostly in fishery, environmental and marine biology laboratories— organizations that have no institutional commitment to long-term collection storage. Collections that document important publications or have potential value in systematics should ultimately be deposited in a museum that 8 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM has a mandate to guarantee long-term archival storage and easy access. Several such institutions that presently house larval fish- es or are willing to do so are the Zoological Museum of the University of Copenhagen, which maintains the extensive worldwide collections taken during the Dana Expeditions, as well as ones documenting the earlier classical studies on larval fishes by Johannes Schmidt and his students, the Harvard Mu- seum of Comparative Zoology, the Smithsonian Institution, and the Natural History Museum of Los Angeles County. If collec- tions of ELH stages are to realize their full potential in system- atics, then it is timely for ichthyoplankton specialists to offer good developmental series, especially illustrated ones, and for museum curators to accept them. Fossils have been studied for clues to the major classification of fishes since the days of Louis Agassiz (Patterson, 1981a) and to the extent that they were available have been widely consid- ered as important adjuncts or indeed prerequisites to compre- hending the phylogeny of particular groups. Although this view is now receiving heavy criticism (Patterson, 1981b), the fact remains that it did exist for many years and may have detracted from the potential contribution of the non-fossil suites of char- acters carried by early life history stages. Even so, students of fossils and of larvae share a preoccupation with the caudal fin skeleton, a structure that is often well preserved in fossils and can be studied in two dimensions and which, during the course of ontogeny, exposes a wealth of information of great value to the systematist. Because adult stages have been the chief source of characters used in fish systematics, a perception has arisen that these char- acters are in some way more useful or more indicative of a phylogenetic classification than are the characters of early life history stages. How did such a view arise? For many years, systematists tended to concentrate on the search for conserva- tive, "non-adaptive" characters (labeled the Darwin Principle by Mayr, 1969). They discarded not only ones that they believed were directly affected by the environment but also ones that appeared to smack of convergence. It seemed reasonable and proper, for example, to group together for phylogenetic purposes fishes with one spine and five soft rays in the pelvic fin because the character was apparently conservative, non-adaptive, and non-convergent. On the other hand, it seemed wrong to group together all fishes with canine teeth because the character was apparently non-conservative, adaptive, and surely convergent. With regard to larval fishes, Moser (1981) recently discussed the occurrence of a large number of apparently highly adaptive larval characters distributed across a broad taxonomic spec- trum. He states, "Marine teleost larvae have evolved an enor- mous array of morphological specializations, such that it seems to me we are looking at a distinct evolutionary domain quite separate from that of the adults. It is reasonable to assume that these remarkable structural specializations are adaptive and re- flect each species' solution to the challenge of survival in a complex and demanding environment." My point here is that if a systematist rejected adaptive characters (and many did), then he would have been unlikely to use ELH stages, and this may be another reason why they have not received sufficient attention. How Systematists Do Their Work Even if systematists agreed among themselves about their immediate goals and how best to achieve them, the task of this Symposium would be daunting. But contemporary systematists do not agree on either objectives or methodology. The concepts that purport to link systematics to phylogeny are being actively reassessed, and it is within the context of rapidly changing ideas in systematics that our presentations and discussions will occur. There are basically three conceptual methods now being used by systematists, and although the bare bones of these methods are easily comprehended, in practice they become more complex and their independence from each other less clear. The interested reader who is as yet unaware of the intense debate both between and within the several schools of systematic classification is referred to the pages of the journal Syslonatic Zoology for many articles and references as well as ones cited in this section. A recent description and comparison of the three methods is given by Mayr (1981), who lists many important references. Although 1 do not propose to use very much space here on a redundant treatment, 1 will briefly describe each method and comment on its strengths and weaknesses. The theoretically simplest method (or methods— there is more than one algorithm, and there is disagreement on which is best) is called phenetics or numerical taxonomy and is described in detail by Sokal and Sneath (1963) and Sneath and Sokal (1973). It is based on overall similarity. Many unweighted characters are used to generate clusters of OTUs (operational taxonomic units), which may be anything from individuals, populations, or species to orders, classes, or phyla. The hierarchically ar- ranged clusters, which lack a time dimension, are called phe- nograms. Neither homology nor the fossil record are considered in selecting characters. Each member of a cluster bears a closer resemblance, although not necessarily genealogical relationship, to other members of its cluster than it does to members of other clusters. Some pheneticists claim that if a sufficient number of characters is analyzed, any influence of convergence becomes dampened and the phenogram will express phylogenetic rela- tionships. Unfortunately, there seems to be no good way to ascertain how many characters are needed. Other pheneticists do not ascribe phylogenetic significance to their clusters and merely claim to be representing overall similarity. Replicability of results is the chief objective. Many classifications that purport to be based on the methods of cladistics or evolutionary clas- sification, upon close scrutiny appear to be basically phenetic. There are apparently few fish classifications using ELH char- acters, which are explicitly based on phenetic methods. One example is a paper on Northeast Pacific cottid genera (Rich- ardson, 1981a) which, according to the author, was not entirely satisfactory for phyletic purposes. Ichthyologists who restrict their data sources for a phenetic analysis to a single life history stage should consider a study by Michener (1977), who gener- ated four different phenetic classifications of a group of bees based on different life history stages or character suites. A second method is called cladistics or phylogenetic system- atics, and although it has been more or less on the scene for many years, it is only since the revision and translation into English of its original presentation (Hennig, 1950, 1966) that it has gained wide currency and is now used, either explicitly or implicitly, by many systematic ichthyologists all around the world but particularly in North America and western Europe. A recent guide to the method is a book by Wiley (1981), and the reader is advised to consult also Brundin ( 1 966) for a notably lucid interpretation. Cladistics requires a stringent evaluation of characters. Primitive or generalized ones (called plesiomor- COHEN: ONTOGENY, SYSTEMATICS, PHYLOGENY phic) for the group being analyzed are discarded for purposes of generating a phylogenetic classification; only derived char- acters (apomorphic) are of value, and monophyletic groups are defined by the degree to which they share such characters (syn- apomorphy). The distribution of derived character states among a monophyletic assemblage of taxa is analyzed and used to generate an hierarchically arranged chart called a cladogram, in which each node or branching point on the diagram gives rise to two branches that are interpreted as genealogical lineages and are called sister groups. In instances in which the data do not allow the unambiguous definition of two branches, more are often used. Each member of a monophyletic group is more closely related genealogically to other members of its group than it is to members of other groups. More than one cladogram can be generated with the same data set, and the most parsimonious, that is, the one requiring the fewest evolutionary steps, is taken as the most natural or best. According to Panchen ( 1 982), prob- lems in logic invalidate the use of parsimony in cladistics. Not all cladists agree about precisely what a cladogram represents, but some interpret it directly as a phylogenetic classification. One of the greatest problems in using cladistics is the difficulty in evaluating character states for primitiveness or degree of derivation. Two methods have been used; one involves onto- genetic stages and will be discussed later in this paper. A second method, called out-group comparison (Wiley, 1981, gives a good description), is the most subjective part of the entire cladistic procedure and to a certain degree may involve circular reason- ing. A practical problem that cladistics has not yet conquered is that of naming, for classifications must be used by many who have no interest in theory, and naming categories on a strictly genealogical basis raises many problems, as does the practice followed by some cladists of naming all branching points. Some attributes of ELH stages that might be considered unsuitable for use in evolutionary classification are available for use in cladistics. One example concerns character stages that are in- terpreted as being highly adaptive rather than conservative. If polarity can be ascertained, then so-called adaptive characters are available. Rates and sequences of ontogenetic change also constitute potentially valuable character suites. The third method, presently called evolutionary classification, is more difficult to define and discuss. It has a long history and an extensive literature (Mayr, 1981). The methods of evolu- tionary classification are eclectic and generally more subjective than those of phenetics and cladistics. They do not easily lend themselves to overall generalization. Characters are selected and weighted by paying particular attention to homology and con- vergence; to the extent that they are available, evidence from embryology and palaeontology are also used. Primitive char- acters are admitted to the system. Data are used from ecolog- ically oriented facets of evolution such as selection, competition, predation, and ecological biogeography. Historical biogeogra- phy, rate of evolution, and genetics are also considered. An hierarchical classification is derived, which has an inferred time axis and which may generally reflect genealogical relationships. However, degree of phenetic difference in selected characters, which is interpreted as reflecting degree of genetic difference, may be considered along with branching pattern in converting a strict genealogy into a classification. Patterson (1981b) has discussed and criticized such procedure. Whatever may be phy- letic relationships, the definition of taxa is essentially subjective, and each member of a group is not necessarily more closely related genealogically to other members of its group than it is to members of a different group. The test for goodness of a classification is pragmatic; if it has high predictive value it is good. (By prediction is meant the degree to which a classification encompasses additional data.) In commenting on evolutionary systematics Panchen (1982) writes that it, "has always been somewhat ad hoc in its procedure, yielding good results with competent taxonomists and bad with incompetent ones. The standard warks [sic] on procedure . . . are to some extent ra- tionalizations of a tradition that is too largely intuitive." As a summary, I have tried to compare in Table 1 some of the techniques, objectives, and assumptions of the three meth- ods. Phenetics requires the fewest assumptions but would seem to offer the systematist a classification with the least information value. Cladistics has the most constraints, so many and so strin- gent in fact, that they may limit its practical use, although the method is particularly valuable in indicating areas for which additional or more suitable data are required. Misuse of cla- distics may soon rival the long-time abuse by systematists of parametric statistics. Evolutionary classification tries to include the most information from the most sources, but the methods for doing so are not very well formalized. Cladists treat their method of classification as a general theory of biology (Nelson and Platnick, 1981), a forcing function among all evolutionary phenomena, which must therefore comply with a parsimonious model derived entirely from character state analysis. Evolu- tionary classification, on the other hand, incorporates infor- mation from a wide variety of biological phenomena and to that extent is forced, rather than forcing. Predictability, as a test of goodness for a classification, is more pragmatic and logically less satisfying than is parsimony. Perhaps an important question for theoretical systematists to consider is the formulation of comparable definitions for replicability, parsimony, and pre- dictability. Ontogeny and Fish Phylogeny Louis Agassiz, who fought the idea of organic evolution, pro- posed a "threefold parallelism" of arranging organisms in a series or classification. His three parallels were palaeontology, what we would now consider to be homology, and ontogeny. Even though he failed to interpret the parallels as evidence for evolution, his keen perception of the fact that they do exist in nature and are somehow interrelated has elicited extensive com- ment and reinterpretation (see especially Gould, 1977) and is a suitable point of departure for addressing the importance of ontogeny as a source of information about homology, the bio- genetic law, developmental stages as alternatives to outgroup comparisons in cladistics, paedomorphosis, and the application of life history stages to phylogenetic inquiry. If characters are the meat and muscle of classification, then homology surely shapes the skeleton on which phylogenetic clas- sifications are arranged. The worth of any allegedly phylogenetic classification is no better than the degree to which homology has been assessed, and how to do this is a major problem for the systematist. Like the weather, everyone talks about homol- ogy but does nothing about it— or almost nothing. The concept, which is so pervasive in the study of phylogeny and in evolution, has been with us since pre-Darwinian times, although not always in the way that we understand it today. The great comparative anatomist Owen defined it in 1866 as follows; "A 'homologue' 10 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Table 1 . Comparison of Three Methods Used in Biological Classification. Evolutionary' Character weighting Convergence Homology Fossil History Eco-evolutionary Data Rale of Evolution No. of Characters No. of Specimens Branches from a Node End Product Test of Goodness No Not Considered Not Considered Not Considered Not Considered Not Considered Many Few Two to Many Perhaps a Phylogeny Replicability Yes Important Important Not Important Not Important Not Important One to Medium Few to Many Two when Possible Phyiogenetic Classification Based on Genealogy Parsimony Yes Important Important Important Important Important One to Medium Few to Many Two to Many Phyiogenetic Classification Based on Genealogy and Degree of Difference Predictability is the same pail or organ in different animals under every variety of form and function." He goes on to note, however, that some earlier workers defined the concept as we now define analogy. But our problem remains identical with that of Owen— how to define same. In a recent discussion of homology, Patterson (1982) proposed similarity in ontogeny as part of a test of homology. But the use of similarity in development to help define Owen's "same" is tautological. Palaeontologists proceed in a basically circular fashion in their use of homology. They depend upon a time series to trace the history of transformed states of a presumably homologous char- acter along a sequence that is interpreted as a genealogy. But of course the characters are considered homologous because they are part of a genealogy. Whether they admit to it or not, most systematists use pure phenetics in the search for homology, and although this common sense, intuitive, non-scientific approach works much of the time, still, many systematists have misin- terpreted as homologues characters that are actually analogous and have filled the literature with many misdiagnosed conver- gences. In comparative vertebrate anatomy and systematics, the convention has grown up that certain organ systems are more conservative than others and therefore provide a better method for detecting homologies. The nervous system is generally con- sidered the best, the skeleton the next best, followed by viscera and muscles, with the integument the least good. In fishes, for example, Freihofer (1963, 1970) has used the patterns of the ramus lateralis accessorius and ramus canalis lateralis nerve systems relative to elements of the skeleton to propose groupings of fishes. But even here the possibility of convergence cannot be ignored (Gosline, 1968), and again the problem of circularity arises because many ichthyologists define osteological features on the basis of their topographic relation to elements of the nervous system. Another example relates to homologies of pho- tophore series in lantemfishes as determined by studies of their innervation (Ray, 1950). Here also, the conclusions based on this method appear to be equivocal (Moser and Ahlstrom, 1 972). A direct method for demonstrating the homology of structures would be to trace them back during development to their an- lagen. De Beer (1951) has commented on the apparent failure of experimental embryology to validate this approach. Even so, a survey of the development of bony structure during fish on- togeny presented by Dunn ( 1 983b) lists some observed instances of losses, gains, and modifications, chiefly in the caudal fin skel- eton, which interpret homologies in adult structure; unfortu- nately, these instances are too few. Ahlie had a long interest in the caudal fin skeleton, particularly of flatfishes, and the com- pletion of his work by colleagues hopefully will constitute an additional contribution to the use offish ontogeny in identifying homologous structures. The concepts of ontogeny and homology are intimately as- sociated in the idea that the study of early life history stages of an organism will reveal its adult ancestral stages— ontogeny re- capitulates phylogeny— as proposed by Ernst Haeckel in the latter half of the 19th century. Taken at its most extreme, the biogenetic law has been interpreted as meaning that an entire genealogy is encapsulated in an ontogenetic series. If adults of extant species of a group were to be matched up with their closest approximations in an ontogenetic series, homology would un- fold before our eyes. Of course its value to us in unraveling phylogeny would be redundant, because phylogeny would be there as well. It was soon evident however that the biogenetic model is far too crude to approximate nature. The embryologist von Baer had previously formulated four "laws" or general propositions about embryology that have been restated in var- ious forms by many authors and applied to the interpretation of phylogeny. The following are taken from De Beer ( 1 95 1 ): ( I ) In development from the egg the general characters appear be- fore the special characters. (2) From the more general characters the less general and finally the special characters are developed. (3) During its development, an animal departs more and more from the form of other animals. (4) The young stages in the development of an animal are not like the adult stages of other animals lower down on the scale, but are like the young stages of those animals. These propositions are useful generalizations and we can all think of obvious instances of fish , ontogeny that can be interpreted by one or more of them . Consider for example the bilaterally symmetrical larvae of flatfishes, the early presence and subsequent loss of a swimbladder in stromateoids (Horn, 1970a), the sequence of fusions during ontogeny in the caudal fin skeleton of myctophids (Ahlstrom and Moser, 1976), the ontogeny of the upper jaw bones and dentition in notosudids (Berry, 1964a), and the presence of a pectoral fin in larval Tac- tosloma and its loss in adults (Ahlstrom, lecture notes). On the other hand, a plethora of early life history stages of fishes man- ifests character states that represent morphological specializa- tions occurring early in development. Consider the egg stages of macrourids with their hexagonal patterns, atherinomorphs with their filaments, and argentinoids with their pustules. Other COHEN: ONTOGENY, SYSTEMATICS, PHYLOGENY 11 instances for which it is difficult to accept that ontogeny has recapitulated phylogeny include the leptocephalus of eels, the stalked eyes of assorted larval bathylagids, myctophids and Idi- acanthus. the elongated guts of larval melanostomiatids, the extensive armature of many spiny-rayed fishes during their lar- val stages, and the produced fin rays found in many kinds of larval fishes. Examples of all of these are illustrated and de- scribed in this volume. With regard to proposition three in particular, Ahlie often pointed out instances of fishes that were easily distinguished as larvae but became more similar in ap- pearance as adults; one example is Bathylagiis milleh and B. pacificus; Myctophum aurolaternalum and other myctophid species is another. Von Baer's propositions as applied to phy- logeny are tidy and appealing but are completely operative only under the rather special condition that major evolutionary changes (except for paedomorphosis) are restricted to the adult stage (Gould, 1977; Fink. 1982). For cladistic analysis, the polarization of characters through direct observation of their transformation during ontogeny has been discussed by Nelson (1978) and others as an alternative to the often unsatisfactory indirect method of outgroup com- parison. Such use of ontogeny, which depends on von Baer's first three propositions, has been analyzed by Henning (1966), who noted its uncertainty. As examples from fish ontogeny given above indicate, ontogeny could replace or corroborate outgroup comparison but only to the extent that the biogenetic law is valid for a particular situation. Patterson's (1982) statement, "that ontogeny is the decisive criterion in determining polarity," would seem to be based on limited acquaintance with ELH stages. Paedomorphosis refers to the presence in adults of larval char- acters (De Beer, 1951) and has been variously considered as insignificant to very important in evolution. For fishes at least, I think the latter is the case. As one example, small adult size could be considered a particularly widely distributed neotenic character. In his discussion of paedomorphosis and cladistics. Fink (1982) remarked that it is difficult to identify this phe- nomenon without paired taxa, but surely this is not always true. Although the relationships of the curious little fish Schindleria are unknown, it would be difficult to deny that it has many neotenic characters (Watson, Stevens and Matarese, this vol- ume). On a larger scale paedomorphosis may have been im- portant in establishing novel phyletic lines as well as isolated species or genera, and the study of ELH stages will be essential in detecting these divergences. I end this essay by noting that the most important use of all for information about fish ontogeny may be providing characters for charting fish phylogeny rather than theories about phylogeny. Distinguishing and identifying species for purposes of fish bi- ology and management has been the chief use for what is called larval fish taxonomy, and the large resulting literature is sum- marized in this volume. Many of the same descriptive data are of apparent value for purposes of grouping similar species or other taxa for phyletic purposes. Published examples of syn- thesis are far fewer than of descriptions, but accounts using each of the three methodologies previously described are available, either cited in this volume or presented here as original research. ELH characters can meet many methodological constraints and will be used increasingly by ichthyologists. To what advantage remains to be seen, but the prognosis is good. Life Sciences Division, Los Angeles County Museum of Natural History, 900 Exposition Boulevard, Los Angeles, California 90007. Early Life History Stages of Fishes and Their Characters A. W. Kendall, Jr., E. H. Ahlstrom and H. G. Moser Patterns of Teleost Early Life History IN discovering that Atlantic cod lay free-floating planktonic eggs which develop into pelagic larvae, G. O. Sars, in 1865 (see Hempel, 1979; Ahlstrom and Moser, 1981) had also come upon an example of the widespread life history pattern of marine fishes. Most marine fishes, regardless of systematic affinities, demersal or pelagic habits, coastal or oceanic distribution, trop- ical or boreal ranges, spawn pelagic eggs that are fertilized ex- ternally and float individually near the surface of the sea (Fig. 5). These eggs range from about 0.6 to 4.0 mm in diameter (mode about 1 mm) and generally are spherical. Within a species there is little variation in egg characters such as size, number and size of oil globules, and pigmentation and morphology of the developing embryo. Development time is highly tempera- ture dependent and also species-specific. The eggs hatch into relatively undeveloped yolk-sac larvae which swim feebly and rely on their yolk for nourishment while their sensory, circu- latory, muscular, and digestive systems develop to the point that they can feed on plankton. Even these yolk-sac larvae have characters (pigment patterns, body size and shape, myomere number) that reflect their heritage. After the yolk is utilized, they develop transient "larval" characters such as pigment pat- terns and, in some, specialized head spines and fin structures that are apparently adaptive for this phase of their life history. During this period more characteristics of the adult (e.g., me- ristic characters) gradually develop. At the end of the larval stage, they may go through an abrupt transformation to the juvenile stage, particularly if they move from a pelagic to de- mersal habitat, or the transformation may be gradual. In some fishes, there is a prolonged and specialized stage between the larval and juvenile stages. These pelagic (often neustonic) forms eventually transform into demersal juveniles. The juvenile stage is characterized by specimens having the appearance of small adults— all fin rays and scales are formed, the skeleton is almost EGGS YOLK SAC PRE FLEXION FLEXION POST FLEXION > < > m JUVENILE Fig. 5. Early life history stages of Trachurus symmelricus from Ahlstrom and Ball (1954), KENDALL ET AL.: ELH STAGES AND CHARACTERS 13 END POINT EVENTS TERMINOLOGY Primary developmental stages Transitional stages Subdivisions OTHER TERMINOLOGIES Hubbs, 1943,1958 Sette, 1943 Nikolsky, 1963 Hattori, 1970 Balon, 1975 (phases) Snyder, 1976,1981 (phases) \ E q q Larva Juvenile 1 ' ' 1 Yolk sac larva Transforma lion larva Early Middle Late Piftlexion Flexion PnMflexinn larva larva larva Pelagic or special juven ' 1 E m b y o Proiarva Post 1 a rv a Prejuvenile ' 1 1 Larva Post larva Embryo Prelarva Cleavage egg Embryo Eleulhero embryo Protoptery- qiolarva Pterygiolarva 1 1 1 1 1 1 1 1 Protolarva M e s 0- larva # M e t a 1 a V a Fig. 6. Terminology of early life history stages. completely ossified, the larval pigment pattern is overgrown or lost and replaced by dermal pigment similar to that of the adults, and the body shape approximates that of the adults. Although this is the most frequently observed life history pattern, there are many variations (see Breder and Rosen, 1 966) often related to increased parental investment in individual progeny with a concomitant decrease in fecundity and larval specializations. There is scant information on the young of many deep-sea fishes, and this may be due in part to life history strategies that do not include eggs and larvae that occur in the epipelagic zone (where most of the collecting is done). Marshall (1953) discussed life history adaptations of these fish such as the production of few, large yolky eggs that hatch into relatively advanced larvae. These young may remain far below the more productive surface layers, and thus not be susceptible to most sampling procedures. Markle and Wenner (1979) cite evidence for demersal spawning of two species of groups (Alepocephal- idae, Zoarcidae) that are seldom collected in the plankton as larvae. Many coastal marine and nearly all freshwater fishes lay de- mersal eggs which are generally larger than the I mm mode of pelagic eggs. In such fish development from hatching through juvenile stage is direct and the larvae gradually attain adult characters of shape, pigmentation, and meristic features. The demersal eggs frequently are adhesive and laid in some sort of nest. Parental care of the nest is observed in many species, and this care may extend to the larvae after hatching (e.g., mouth brooding in cichlids, ariids). Parental care takes another form in Sehastes. where development through the yolk-sac stage takes place in the ovary and first-feeding larvae are extruded. Vivi- parity, in which nourishment is supplied by maternal structures, has evolved many times (e.g., poeciliids, some zoarcids, em- biotocids), whereby the larval stage is bypassed and the fish are extruded ("bom") as juveniles (Wourms, 1981). Early Life History Stages Between spawning and recruitment into the adult population, most fishes undergo dramatic changes in morphology and hab- Table 2. Examples of Characters of Pelagic Eggs that May Be Useful for Systematic Studies of Certain Fishes. Character slates Systematic groups Egg size Egg shape Envelope sculpturing Oil globule position Embryonic characters < 1 mm->5 mm >3 mm->5 mm Round — oblong Varying distances between pores Varying length/density of filaments Anterior to posterior in yolk sac Slate of development of various organs/organ sys- tems at various develop- mental mileposts Pleuronectidae Anguilliformes Engraulidae Ostraciontidae Gadidae Atheriniformes (Exocoetidae) Perciformes Gadidae 14 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Fig. 7. Examples of features of yolk-sac larvae of teleosts. (A-C). Paracallionymus costatus. A. soon after hatching 0.98 mm NL; B. 1.8 mm NL; C. 1.9 mm NL. From Brownell (1979). Features demonstrated in; (A) include the small size of the larva, the lack of an oil globule, the segmented yolk, and the dorsally arranged melanophores; (B) demonstrates the migration of melanophores ventrally and the formation of the anus producing a preanal finfold; (C) demonstrates further ventral migration of melanophores, beginning of larval pectoral fin formation, the decrease in yolk-sac size, and beginning of pigment in the eye; (D) Diplodus sargus. 2.4 mm NL. From Brownell (1979). Single pigmented oil globule posterior in the unsegmented yolk and a short preanal finfold are demonstrated; (E) Trachurus I. capensis. 2.2 mm NL. From Brownell (1979). Single pigmented oil globule anterior in segmented yolk with moderately long preanal finfold demonstrated; (F) Cololabis saira. 5.1 mm SL. (original). Well-developed, heavily pigmented yolk-sac larva at hatching with notochord flexion beginning and some caudal rays formed; (G) Argentina silus. 1.1 mm. Redrawn from Schmidt (1906c). A large but poorly developed yolk-sac larva at hatching with a large oil globule; and (H) Hippoglossus slenolepis. 9.5 mm. From Pertseva-Ostroumova (1961). A large but poorly developed yolk-sac larva at hatching with no oil globule. its. As mentioned earlier, at hatching, particularly in marine fishes with pelagic eggs, the fish is in an extremely undeveloped state and then, as a free-living individual, it gradually develops the adult characters. This process is continuous, but there are morphological and ecological mileposts that are significant in the life of the fish and which allow us to subdivide this process so that we can communicate results of our studies and compare different fishes at the same moment in development. Fish early life history has been and continues to be studied from a number of different perspectives (Ahlstrom and Moser, 1976). Some studies deal directly with embryology and later ontogeny, others emphasize functional morphology of larval structures, apply larval features to taxonomic and systematic studies, investigate the ecology of eggs and larvae, or use these stages to address fishery-related problems such as assessment of spawning stock size and recruitment success. All of these studies have in common the need to subdivide early life history and communicate information based on processes and events occurring during these subdivisions. As with any communica- tion, it is vitally important to use terms that are clearly defined and this is particularly true with the diverse disciplines that are involved in larval fish studies. Historically, several disciplines have used different names for the same stage, or subdivided development differently [see Okiyama (1979a) and Fig. 6 in this paper]. This has led to confusion rather than communication. Several criteria seem appropriate for defining stages of de- velopment to be used by students of any discipline. The variety of developmental patterns should be recognized and the defi- nitions should apply to as many patterns as possible. Thus, stages should be based on very widespread, fundamental fea- tures of development. The stages should have some significance in the life history of the fish, both morphologically and func- KENDALL ET AL.: ELH STAGES AND CHARACTERS 15 From demersal eggs From pelagic eggs Clupea harengus harengus egg diameter = 1.2-1. 5mm NL at hatclning = 4.9mm Etrumeus teres egg diameter = 1.3mm NL at hatching = 4.8mm Krevanoski 1956 Mito 1961 O Mukhacheva and Zviagina 1960 Gadus macrocephalus egg diameter = 0.8-1. 4mm NL at hatching = 3.6mm Colton and IWarak 1961 Gadus morhua egg diameter = 1.1 -1.9mm NL at hatching = 3.6mm Lepidopsetta bilineata egg diameter = 1.02-1. 09mm NL at hatching = 3.9mm Isopsetta isolepis egg diameter = 0.90-0. 99mm NL at hatching = 2.9mm Pertseva-Ostroumova 1961 Richardson et al 1980 Fig. 8. Newly hatched yolk-sac larvae of related fishes with pelagic and demersal eggs of comparable sizes. tionally, such as a particular type of nourishment or locomotion. Also the endpoints for the stages should be easily observed and sharply defined. The most general scheme of terminology of early development of fishes includes (Fig. 5): The "egg stage" (spawning to hatching). The egg stage is used in preference to the embryonic stage because there are characters present during this stage other than just embryonic characters (e.g., those associated with the egg envelope). The "larval stage" (hatching to attainment of complete fin ray counts and beginning of squamation). One of the funda- mental events in development of most fishes is the flexion of the notochord that accompanies the hypochordal development of the homocercal caudal fin. It is convenient to divide the larval stage on the basis of this feature into "preflexion." "flexion," and "postflexion" stages. The flexion stage in many fishes is accompanied by rapid development of fin rays, change in body shape, change in locomotive ability, and feeding techniques. The "juvenile stage" (completion of fin ray counts and be- ginning of squamation until fish enters adult population or at- tains sexual maturity). Transitional stages can also be recognized: the "yolk-sac larval stage" (between hatching and yolk-sac absorption); and the "transformation stage" (between larva and juvenile). Meta- morphosis occurs during this stage and is considered complete when the fish assumes the general features of the juvenile. The life histories of some fishes include other specialized ontogenetic stages that have received various names. In some cases, these are the generic names under which these stages were described before they were recognized as larvae of other species (e.g., the leptocephalus stage of Anguilliformes, the scutatus stage of Anlennarius. the vexillifer stage of Carapidae. and the kasidoron stage of Gihhertchthys). In other cases, consistent fea- tures of development of a group permit useful subdivisions of stages (e.g.. in leptocephali the engyodontic and euryodontic stages). The Egg Stage Hempel (1979) reviewed the egg stage relative to fisheries investigations. Ahlstrom and Moser (1980) presented a concise review of the range of characters observed in pelagic fish eggs, particularly those useful in identifying eggs in plankton samples. Sandknop and Matarese in this volume also discuss this subject in detail. The characters that have proven useful for egg iden- tification include egg size and shape, size of perivitelline space, yolk diameter and character (homogeneous or segmented), num- ber and size of oil globules, texture of the egg envelope (smooth or with protrusions), pigment on the yolk and embryo, and characters of the developing embryo (relative rate of develop- ment of various parts, body shape, number of somites) (Table 2). The egg stage has been subdivided by a number of workers (e.g., Apstein, 1909). Fishery biologists need to determine the age of eggs at the time of collection for production, drift, and mortality estimates. Embryologists have designated stages to coincide with significant developmental features. While the stages of fishery biologists are designed to divide the embryonic stage into several easily recognized portions, embryologists are more 16 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Table 3. Examples of Use of Characters of Early Life History Stages in Taxonomic and Systematic Studies. X indicates range of stages and taxonomic levels at which characters vary. (X) indicates infrequent state. Developmental stage Character Lar\ac Taxonomic level Yolk- sac Pre- flexion Flexion Post- flexion Trans- forma- tion Rpfprenf**^ Spe- cies Genus Family Order IX \. IK 1 \,lkVV«^ Egg Keyed to Table 4 (X) X X (X) X 20 X X (X) (X) X 20,29 (X) X X X X X X (X) X 2,38 1, 2, 11, 19, 24, 27, 39 (X) X X X 11, 19,24,27,39 X X X X X 1,2,3.5, 11. 15, 17, 19, 20,25.27.28.33,34 X X (X) X X X (X) X 27,38 19 X X X X X X X 2, 3,4. 5. 10, 11, 13, 14, 19,20,23. 24,25, 26, 27, 28, 29, 31», 33, 37, 40 X X X X X 28,33,35,36,38 X X X X X X X X X X 1,2,3,4,8,9, 11, 13, 14, 15, 17, 19, 20,21,22, 25,27,28,29,33,36, 38, 39, 40 X X X X X X X 9, 11,23,24,25,27,36. 38,40 X X X X X 1,9, 14,23,27,29,33 X X X X X X X 14, 27, 29 27 X X X 8. 10, 14 X X X X X X X 36 8, 14, 15.33 X X X X X 20, 33, 38. 39 (X) X X X X X X X X X X X X X X X X X X 14,20,29,33,38 8, 10. 14,20,33 14,20,33 10, 14,20 29 X X X 6, 19,20,30,32 (X) X X X (X) X X X X X X X X X X 7, 16, 19,23.29,33,40 11,27 12, 14,21 X X X X X X X X 10, 11,22,23,29,30,39 (X) X X X X 13, 14,20,26,27,34 Meristic characters Fin spines/soft rays Principal caudal rays Pelvic fin Dorsal/anal fin Pectoral fin Vertebrae Branchiostegals Gill rakers Larval characters Body shape Snout shape Pigment patterns Head spines Fin ray elongation Fin ray ornamentation Fin ray serration Pinfold size/shape Preanal finfold Pectoral size shape Larval gut Shape Length Larval eye Shape Stalked Choroid tissue Migration Other characters Egg characters Osteological development Scale formation Photophore formation Size at developmental stage Fin development sequence • Emphasis on oil globule placement in yolk-sac larvae. interested in tracing the sequence of development. The em- bryologist's approach will probably provide more useful infor- mation for systematic investigations. Although excellent, early descriptive work was done on teleost embryology (e.g. Wilson, 1891), comparative research on de- velopment needs to be done to allow an evaluation of its value to syslematics, a subject that has proven so fruitful among in- vertebrates. It appears, from the characters that have been stud- ied in greatest detail, that convergence may overshadow phy- letically significant information. For instance, the egg envelope sculpturing on Pleuronichthys, a pleuronectiform, was found even on scanning electron microscope examination to be quite similar to that on Synodus, a myctophiform (Sumida et al., 1979). Phylogenetically diverse fishes often have round pelagic eggs, about 1 mm in diameter, with a single oil globule. Demersal eggs from equally diverse fishes are generally larger than I mm and develop a vitelline circulatory system. Yolk segmentation seems to be a character of more primitive fishes, but some carangids and other perciforms have yolks that are secondarily segmented in an evolutionary sense. Detailed studies are needed to sort out these and other features of the teleost egg and its embryonic development in a systematic context. KENDALL ET AL.: ELH STAGES AND CHARACTERS 17 Table 4. Some Contributions in Which Ontogenetic Characters have been used to Examine Systematic Relationships (Updated from Ahlstrom and Moser, 1981). References Dale Ciroup dealt with Egg Stages Ur- vac Juv ad Larval characters showing relationships No. Among spe- cies Among genera Among subfam- or Among families orders 1,3.5 Ege. V. 1930,53,57 Paralepididae — + + X X 2 Bertelsen. E. 1951 Ceratioidei — + + X X X 4 Bertelsen. E., and N. B. Marshall 1956 Minpinnati — + + X X X 6 Pertseva-Ostroumova. T. A. 1961 Pleuronectidae + + + X X 7 Berry, F. H. 1964a Mar. teleosts — + — X g Pertseva-Ostroumova, T. A. 1964 Myctophidae — + — X 9 Gutherz, E. J. 1970 Bothidae — + — X 10, 14 Moser. H. G., and E. H. Ahlstrom 1970, 74 Myctophidae — + + X X X 11 Mead. G. W. 1972 Bramidae — + + X X 12 Ahlstrom, E. H. 1974 Stemoptychidae — + + X 13 Johnson. R. K.. 1974b Scopelarchidae — + + X X 15 Okiyama. M. 1974a Myctophiformes — + — X X 16 Potthofr. T. 1974 Scombndae — + + X 17 Richards, W. J., and T. Potthoff 1974 Scombridae — + + X 18 Aboussouan. A. 1975 Carangidae — + — X 19 Ahlstrom, E. H.. J. L. Butler, and B. Y. Sumida 1976 Stromateoidei + + + X X X 20 Ahlstrom. E. H.. and H. G. Moser 1976 Mar. teleosts + + + X 21 Ahlstrom. E. H., H. G. Moser, and M. J. OToole 1976 Myctophidae — + + X 22 Bertelsen. E., G. Krefft, and N. B. Marshall 1976 Notosudidae — + ± X X 23 Futch. C. R. 1977 Bothidae — + — X X 24 Moser. H. G., E. H. Ahlstrom, and E. Sandknop 1977 Scorpaemdae — + ± X X X 25 Okiyama, M., and S. Ueyanagi 1978 Scombridae — + — X X 26 Powlcs. H.. and B. W. Stender 1978 Sciaenidae — + ± X 27 Kendall. A. W.. Jr. 1979 Serranidae — + + X X 28 Ueyanagi, S., and M. Okiyama 1979 Scombridae, Istiophoridae — + + X 29 Amaoka. K. 1979 Pleuronectiformes (in part) — + — X X 30 Dotsu. Y. 1979 Gobiidae + + — X 31 Suzuki. K.. and S. Hioki 1979a Percoidei + + — X X 32 Mito. S. 1979a. b Mar. teleosts + — — X X 33 Okiyama. M. 1979b Myctophoidei — + — X 34 Potthoff. T.. W. J. Richards, and S. Ueyanagi 1980 Scombrolabracidae — + + X X 35 Zahuranec, B. J. 1980 Myctophidae ( Na nnobrach lu m) — + + X X 36.37 Richardson. S. L. 1981a,c Cottidae — + + X 38 Washington, B. B. 1981 Cottidae — + X X 39 Johnson. R. K. 1982 Scopelarchidae Evermannellidae — + + X X X 40 Kendall, A. W., Jr., and B. Vinter 1984 Hexagrammidae - + + X X The Yolk-sac Larval Stage At hatching, larvae can be at various states of developmenl, dependent to a large degree on the size of the yolk (Fig. 7). Larvae from eggs with small yolks are less developed at hatching than those that hatch from eggs with larger yolks. Since the bulk of maiine fish spawn eggs that are about I mm in diameter and have a narrow perivitelline space, the yolk is only slightly less than I mm. Larvae from such eggs generally lack a functional mouth, eye pigment, and differentiated fins. They possess a large yolk sac relative to the size of the lai~va which supplies nour- ishment while the larvae develop to become self-feeding. Newly hatched larvae from demersal eggs are generally further ad- vanced in development than lai^ae from pelagic eggs of com- parable size (Fig. 8). In these and other fish with large eggs, hatching may be delayed until the yolk sac is absorbed and the larvae are ready to feed at hatching, having bypassed the yolk- sac larval stage. The delayed absorption of yolk reaches an ex- treme in fishes such as salmonines in which the yolk-sac larva transforms directly into a juvenile; Hubbs (1943) proposed the term "alevin" be applied to this yolk-sac larval stage. At hatching, locomotion and orientation of most yolk-sac larvae are aided by a continuous median finfold (dorsal, caudal, anal) and larval pectoral fins. During egg development, many fish embryos develop melanophores that originate in the neural 18 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM ,r ..'—v ^„.-«n.-)-'-T' ^ Fig. 9. Examples of teleost larvae illustrating extremes of some systematically useful larval characters. (A) Myctophum aurolaternatum. 26.0 mm (Moser and Ahlstrom, 1974). Note stalked oval eye with choroid tissue, trailing gut, and dorsal fin developing in finfold; (B) Epinephelus sp.. 8.4 mm (Kendall, 1979). Note elongate, serrate dorsal and pelvic spines; (C) Adioryx (Holocentrus) vexillarius. 8.5 mm (McKenney, 1959). Note head spines; and (D) Lopholatilus chamaeleonticeps, 6.0 mm (Fahay and Berrien, 1981). Note spines on head and body. crest and are generally aligned along the dorsal surface of the embryo. During the yolk-sac stage, these melanophores move laterally and ventrally to establish the beginning of the larval pigment pattern. Orton (1953a) describes these events in detail in Sardinops sagax. This realignment may begin during the late embryonic stages, before hatching. Some species hatch with few if any melanophores, and when they first appear, they are in ventral positions. Apparently, the pigment cells migrate before pigment formation occurs. The presence and position of oil globules in yolk-sac larvae vary and can be of diagnostic value. In fishes with single oil globules, it can be far forward (e.g., labrids, most carangids, muUids, and lethrinids), in the middle of the yolk sac (e.g.. some clupeids, serranids, and argentinids), or more usually near the rear of the yolk sac. The shape and relative size of the yolk sac itself are variable and provide additional taxonomic characters. In summary, although the yolk-sac stage starts at hatching and ends when the yolk is absorbed, fish are at different stages of development with regard to such features as pigmentation, eye development, and fin formation during this stage. The strik- ing pigment rearrangements that occur during this stage provide further emphasis that the yolk-sac stage is a transitional stage between the egg and larval stages. The Larval Stage During the larval stage many ontogenetic changes occur (Mos- er. 1981). Some of these relate directly to the development of the adult form while other changes and structures are specialized KENDALL ET AL.: ELH STAGES AND CHARACTERS 19 B Fig. 10. Apparent convergence in siphonophore-mimicking appendages on larval fish. (A) Loweina rara. 17.6 mm. Note lower pectoral fin ray (Moser and Ahlstrom, 1970); (B) Carapussp., 3.8 mm (Padoa, 1956j). Note elongate dorsal fin ray; (C) Exterilium larva, 64 mm. Note trailing gut (Moser, 1981); (D) Lopholus sp., 12.t mm. Note elongate dorsal and pelvic ray (Sanzo. 1940); and (E) Arnoglossus japonkus, 30.5 mm. Note elongate dorsal ray (Amaoka, 1973). and of presumed functional significance primarily for planktonic existence (Fig. 9). These latter features are of particular interest in systematic studies of larval fish ontogeny. They include pig- ment pattern, larval body shape, armature on head bones, and precocious (early forming), elongate, or serrate fin spines. The sequence and way of developing adult structures, such as the skeleton and fin rays, are also useful larval characters. All of the characters of the larvae— whether they are specialized larval characters or merely characters observable in the larvae— may have potential systematic value at some taxonomic level; how- ever, the usefulness of most of the characters has not been eval- uated (Tables 3 and 4). Among the most taxonomically useful larval characters, gen- erally at the specific or generic level, is the pigment pattern. Usually, each species has a distinct larval pigment pattern. In some the number and placement of individual melanophores are diagnostic, while in others the location, shape, and size of groups of melanophores are key characters. At a higher taxo- nomic level, in the myctophiforms for example, the peritoneal pigment blotches seem to indicate relationships on a suborder- family level. Problems associated with the usefulness of pigment patterns include 1 ) the widespread distribution of some patterns, and 2) the variable state of melanophore contraction on larvae of the same species. An example of the first problem is the frequent occurrence of a row of small melanophores along the ventral midline from just behind the anus to the tip of the tail. Another example is a pigmented area midlaterally on the caudal peduncle which occurs in numerous groups. A ventral spot at the junction of the cleithra is also quite common. These are just a few examples of widespread, presumably convergent pigment patterns that limit the usefulness of pigment in systematic stud- ies of larvae. The causes for the observed differences in degree 20 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Fig. 11. Liopropoma sp., 11.0 mm. Collected by G. R. Harbison, 16 May 1981, 6°31.8'S, 150°21.8'E. Note elongate dorsal spines. of contraction of melanophores are not well understood al- though they may be partially related to ambient light intensity. The relative size and placement of melanophores are genetically determined and therefore useful in a systematic context, while the degree of contraction seems to be physiologically deter- mined. In general, the body shape and size at various stages of de- velopment are characteristic of larvae at the generic or familial level, although subtle differences in body shape may be char- acteristic of species. Size at stage of development can be envi- ronmentally modified (e.g., by temperature or food) to some extent, but is primarily genetically determined. There appears to be some convergence in larval body shape, such as on a long tubular body in several divergent groups (e.g., Clupeiformes, Argentinidae, Blennioidea), just as there is on the "herring" morph of adults. A valuable and fairly widespread set of larval characters con- cerns the development of spines and armature on bones of the head and cleithral region. Such armature has provided diag- nostic larval characters as well as material for systematic infer- ence at levels from species to order. Larval head armature ap- pears to be a mark of the Acanthopterygii. Only a few scat- tered examples of such armature appear in fishes which have only soft rays as adults (e.g., Sudis). Within the spiny-rayed fishes, beryciforms are quite heavily armed with spines on many head bones. Perciforms usually do not have spines on the pa- rietals but the supraoccipital is armed in some. The Scorpaeni- formes are just the opposite: they tend to have head armature that includes spines on the parietals but do not have spines on the supraoccipital. Nowhere are larval specializations more evident or varied than in the fins. Elongation of particular spines or soft rays or enlargement of whole fins are frequently seen. Such elongations have been described for rays of the dorsal, pelvic, pectoral, and caudal fins; thus they occur with both spines and soft rays. In some, these long rays may bear pigmented "bulbs" or appear like flagellae. Such specialized rays are produced in the dorsal, pectoral, or pelvic fins of taxonomically diverse fishes. The ex- tended gut of "exlerilium" ophidioid larvae (Fraser and Smith, 1974) and the serial pigment pattern of some leptocephali (Smith, 1979) may give the same appearance to potential predators as these elongate rays. All of these structures may be mimicking siphonophores: a remarkable example of convergence (Fig. 10 and 1 1 ). Elongate fin spines are heavy and armed with serrations in some. Elongated rays are often precocious in development, with some even forming in the egg. These fin characters seem to vary at the family-species levels. Other characters associated with fin development include the sequence of formation and movement and loss of whole fins or some of the rays. Dorsal and anal fins move forward along the body during larval de- velopment in elopiform and clupeiform fishes. They develop in "streamers" in the finfold of argentinoids and attach to the body proper just before or during transformation. The shape of the finfold, presence or absence of a preanal finfold, and shape of the pectoral fin base provide additional characters at the family- genus level. Gut characters offish larvae include length and shape as well as the development of a protruding, trailing hindgut in some. In fishes with pholophores, their placement and sequence of development are excellent characters at the subfamily-species levels. The eye of a larva is specialized in a number of ways. Fig. 12. Examples of special juvenile stages. (A) Hexagrammos lagocephalus. 28.0 mm. A neustonic or epipelagic form of a species that is demersal as an adult (from Kendall and Vinter, 1984); (B) Forapiger longirosths. 17 mm. A spiny form that lives on tropical reefs as an adult (from Kendall and Goldsborough, 1 9 1 1 ); (C) Sehaslolobus altivetis, 26.8 mm. A barred pelagic form of a species that is demersal on the continental slope as an adult (from Moser et al., 1977); (D) Oncorhynchus kisulch. 37 mm. The freshwater alevin or parr stage of an andromous salmonid (from Auer, 1982); and (E) Kali macrodon. 45 mm. The juvenile of a bathypelagic species. Originally described as Gargaropteron pterodactylops (see Johnson and Cohen, 1974). KENDALL ET AL.: ELH STAGES AND CHARACTERS 21 22 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Its size and rate of development are useful, as well as whether it is round or oval. Some fish larvae have eyes borne on stalks that reach an extreme in Idiacanthus, while others develop an area of choroid tissue. Migration of the eye in flatfish larvae from a symmetrical position to one side of the head is well known. The sequence of development of ossified structures is proving to be a powerful tool in systematic studies offish larvae. The losses and fusions of bones, which are generally assumed based only on adult material, can and should be tested using developmental studies. The caudal fin skeleton has provided excellent developmental characters to be used for systematic inferences, mainly at the order-generic levels. The development of scales has been little studied but may prove valuable, espe- cially in fishes with precocious scales (e.g., some anthiins, hol- ocentrids). The Transformation Stage Between the larval and juvenile stages, there is a transitional stage which may be abrupt or prolonged and which, in many fish, is accompanied by a change from planktonic habits to demersal or schooling pelagic habits (Fig. 12). In some fishes migration to a "nursery" ground occurs during or just before this stage. Morphologically the transformation stage is charac- terized by a change from larval body form and characters to juvenile-adult body form and characters. At the end of this stage the fish generally looks similar to the adult, with major differ- ences only in pigmentation patterns. Two ontogenetic processes occur during this stage of transition between the larva and ju- venile: I ) loss of specialized larval characters, and 2) attainment of juvenile-adult characters. Changes that occur during this stage include pigment pattern, body shape, fin migration (e.g., in clu- peids and engraulids), photophore formation, loss of elongate fin rays and head spines (e.g., in epinepheline serranids and holocentrids), eye migration (pleuronectiforms), and scale for- mation. In several groups, where the transformation stage is pro- longed, the fish have developed specializations that are distinct from both the larvae and juveniles. This stage has been desig- nated the prejuvenile stage (Hubbs, 1943). The specializations generally involve body shape and pigmentation. In many, the morph resembles a herring-like fish and is apparently adapted for neustonic life. The dorsal aspect of the fish is dark green or blue and the lateral and ventral is silvery or white. The body tends to be herring shaped and the mouth terminal. Fins are generally unpigmented. Such a stage is present m Gadiformes (Urophycis), Beryciformes (Holocentrus), Perciformes (e.g., Po- malomus, MuUidae, Mugilidae) and Scorpaeni formes (e.g., Scorpaenichthys, Hexagrammos). In other fishes, such as some myctophiforms and carapids, the prolonged transformation stage may have distinctive body and fin shapes. Implications of Larval Fish Morphology When studying the appearance of larval fishes, one is im- mediately struck with their diversity and morphological dissim- ilarity to adults. This dissimilarity led early workers to establish names for several of these forms, not realizing that they were the young stages of known adults. After establishing the identity of many fish larvae in a variety of groups, we hypothesize that the larvae of all species are recognizably distinct. The use of diversity of larval form in vertebrate systematics was discussed some time ago by Orton (1953b, 1955c, 1957) and in this vol- ume we examine this use in detail in numerous groups of fishes. Why are the larvae so diverse?— Despite the tremendous mor- tality associated with living in the planktonic realm during the larval period, survival must be sufficient to maintain the species and provide a dispersal mechanism for it. To different degrees, various taxa apparently rely on survival and longevity of in- dividual larvae. The amount of reliance is presumably related to fecundity and importance of dispersal and colonization to the taxon. A number of structures have evolved that would be expected to enhance larval survival in the plankton. Practically no experimental work has been done to investigate the function of larval structures, but some structures probably assist flotation and feeding while others decrease predator mortality. Conver- gence on characters that are apparently functionally important to larval survival in the plankton is seen. These specializations develop in conjunction with the basic ontogeny of the taxon. In studying systematics using larval fishes, both the basic pattern of development and the specialized structures must be analyzed. Why are these larvae so morphologically unlike the adults?— Most larvae are adapted to survive in an ecological realm (gen- erally the plankton) that is far different from that of the adult. These are small organisms, compared to adults, and they live in the plankton, having to find and capture food there and avoid becoming food. They float and migrate vertically in a milieu that may be moving much faster than they are. During this larval period, these fish undergo extreme changes in morphology yet remain a functioning (eating, avoiding predators) organism and eventually end up in a suitable nursery area for the juvenile stage. How then can larval morphology help us understand the evolu- tion of these fishes?— Mler recognizing that each species has a morphologically distinctive larva, generally we see that species of the same genus are phenetically similar, and larvae of mem- bers of a family also share common features. Even larvae of suborders and orders share some larval characters. This would be expected since evolution operates on all stages in the life cycle, not just the adult. Evolutionary pressures on the larval stage seem to be particularly intense in those groups that rely on the larvae for widespread dispersal in the ocean. Here the larvae appear well adapted for life in the planktonic realm, and it can truly be said that the larva and the adult perform in "two quite separate evolutionary theaters" (Moser and Ahlstrom, 1974). In this volume we are focusing on what we know to date about larval evolution within various groups of fishes (Table 4). Northwest and Alaska Fisheries Center, 2725 Montlake Blvd. E., Seattle, Washington 98112 and Southwest Fisheries Center, P.O. Box 271, La Jolla, California 92038. TECHNIQUES AND APPROACHES Early Life History Descriptions E. M. Sandknop, B. Y. Sumida and H. G. Moser FISHERIES studies require accurate identification of subject species. Identification of the developmental stages of fishes is complicated by the small size of the specimens, their fragility, and the relatively great changes in their structure and pigmen- tation. Experience has shown that major changes can occur over very small growth increments and these can only be documented by a continuous growth series. Published descriptions of de- velopmental series vary in quality, perhaps more than do species descriptions of adults. Prior to Bertelsen (1951) and Ahlstrom and Ball (1954), most published descriptions were based on relatively few specimens, which were described individually. In their study of the early life history stages of the jack mackerel (Trachunts syinmetricus), Ahlstrom and Ball (1954) used over 500 eggs and a series of about 250 larvae, transforming speci- mens, and juveniles to describe development. Changes in struc- ture and pigmentation were thus described as a dynamic con- tinuum, with emphasis on variation, in contrast to the approach of most previous workers. Developmental osteology was con- sidered an integral part of the description as were seasonal and geographic distributions of eggs and larvae. This paper was fol- lowed by several others (Ahlstrom and Counts, 1955, 1958; Uchida et al., 1958; Kramer, 1960) and these became models for subsequent descriptive papers, including some which treated several species in various taxonomic groups (Moser and Ahl- strom, 1970; Ahlstrom, 1974; Ahlstrom et al., 1976; Moser et al., 1977; Kendall, 1979; Brownell, 1979; Richardson and Washington, 1980; Fahay, 1983; Leis and Rennis, 1983). The following is a brief account of the elements involved in preparing early life history accounts of teleosts. Sources The major source of material is plankton collections. Typical survey tows strain a column of water 200 m to the surface and sample eggs and subsequent larval stages of a major portion of the fish fauna (Smith and Richardson, 1 977). Fishes which have highly stratified vertical distributions are undersampled by oblique tows and require special gear or tow strategies. For example, surface dwellers can be sampled by neuston nets (Zait- sev, 1970; Nellen and Hempel, 1970; Hempel and Weikert, 1972; Nellen, 1973a; Ahlstrom and Stevens, 1976) and those species residing near the bottom may be sampled by epi-benthic plankton nets (Schlotterbeck and Connally, 1 982). Larger larvae and transforming stages are poorly sampled by typical survey tows principally because of accumulated mortality, increased avoidance capacity, and migration out of the sampling zone. These stages are more effectively sampled by trawls (Tranter, 1968), dip-netting with attractor lights (Klawe, 1 960), light traps (Faber, 1982), and fish predators (Haedrich and Nielsen, 1966). Recently, scuba divers have collected oceanic larvae with their delicate structures intact (Harbison et al., 1978; Govoni et al., 1 984). Developmental series may also be obtained by rearing larvae from eggs collected at sea or from captive brood stock (Houdeetal., 1970, 1974; Houde and Swanson, 1975; Richards etal., 1974; Houde and Potthoff, 1976; Moser and Butler, 1981). This method becomes essential when working with speciose faunas (e.g., Sebastes, warm water shorefishes), if only to de- termine which species cannot be identified. Use of Specimens The characters and techniques used in identifying develop- mental stages are discussed elsewhere in this volume (see Ken- dall et al.; Matarese and Sandknop; Powles and MarkJe). From the continuous developmental series two subseries are assem- bled and these form the basis for the description. The first series is used to describe morphology and pigmentation. Specimens in the second series are cleared and stained by a variety of techniques to describe the development of cartilaginous and osseus features (Potthoff, this volume). The number of specimens used to construct these series is dependent on several factors: 1) specimen availability, 2) length (duration) of the development period, and 3) complexity of developmental change. A guideline is that there should be enough specimens to demonstrate the beginning, progression and com- pletion of significant developmental changes in morphology and pigmentation. Usually more specimens are required for species which have extended larval periods; however, many fishes which transform at small sizes undergo great change over small length intervals. For example, lined sole {Achirus lineatus) hatch at 1 .6 mm, transform at about 4.0 mm, and complete a large suite of developmental changes over a 2.5 mm length interval (Houde et al., 1 970). The majority of marine teleosts transform between 10 and 30 mm and, for these, major developmental events can be documented by specimen length increments of 0.5-1.0 mm. Multiple samples representing 1 mm-intervals are required to study fine-scale character variation; however, such studies have rarely been done (Ahlstrom and Moser, 1981). A table of morphometric measurements constructed from the unstained series provides data on the size at important devel- opmental milestones (e.g., hatching, notochord flexion, fin for- mation, transformation) and provides a basis for analyzing structural change and allometric growth. These specimens can be used to construct character matrices of complex or diagnostic pigment changes. Illustration specimens chosen from the series provide an integrated view of major characters and also, if ac- curately executed, are themselves morphometric and meristic documents (Sumida et al., this volume). The stained series is used to construct a meristic table that forms the basis for following the development of fin rays and supporting elements, the axial skeleton and cranial bones (Dunn, this volume). Fine bony structures, such as cranial spines are also apparent in these preparations. Published descriptions employing these basic elements are 23 24 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM the basis for ontogenetic studies of fishes. These are essential for the identification of ichthyoplankton collections, and also present characters for systematic analysis. Data provided in these descriptions have proved useful in studies of the physi- ology, behavior and ecology of the early stages of fishes. National Marine Fisheries Service, Southwest Fisheries Center, P.O. Box 271, La Jolla, California 92038. Synopsis of Culture Methods for Marine Fish Larvae J. R. Hunter THE objective of this paper is to provide a synopsis of present technology for small-scale laboratory culture of marine fish larvae. The technology of marine fish culture is relevant to this book because it is one of the best ways to obtain a taxonomic series. "Ahlie" Ahlstrom was a strong proponent of this ap- proach and I lectured on the subject at his request for his courses on larval fish systematics. Marine fish culture has often been reviewed (May, 1970-, Houde, 1972a; Houde and Taniguchi, 1979; Shelboume, 1964; Kinne, 1977) and many additional references may be found in the previous reviews. The key feature of my review is that it is a condensed practical guide and key to the literature for beginners interested in small-scale laboratory culture of marine fish larvae; culture of freshwater fishes is not considered. Eggs Sources. — Pelagic fish eggs can be obtained from plankton tows, by catching ripe fish and fertilizing the eggs, and by induction of spawning of laboratory brood stock. Let eggs taken in plankton tows stand in quart bottles for 0.5 h, then remove plankton from bottom of jar and add fresh sea water (a second decanting may be required). Jars are stored on their sides in an insulated ice box with a refrigerant for 24 h or longer with the temperature kept within spawning range. Virtually all marine clupeoid fishes (Blaxter and Hunter, 1982) and probably most other pelagic marine fishes spawn at night, hence running ripe fish are more common at night or just before sunset (final egg maturation or hydration occurs just before spawning). After an egg is spawned in sea water its fertility decreases but the maximum time for it to become infertile is highly variable among species, varying from 6 minutes to over 3 hours (Ginzburg, 1972). Sperm in sea water may remain fertile for days (Ginzburg, 1972) although fertility periods as short as 30 seconds have been observed (Haydock, 1971). Owing to the great variation in the time eggs and sperm remain fertile it is preferable that sperm and eggs be mixed immediately after they are obtained. Storage of gametes may be helpful since mature males and females are not always available simultaneously and crosses between subpopulations may be desired. It is well known that sperm can be stored for extended periods ( 10 or more hours) if kept cool and maintained in the concentrated form and not activated by sea water (Ginzburg, 1972; Erdahl and Graham, 1980). Fertilization of Clupea harengus eggs may be obtained after 6-7 days dry storage at 4° C but a high hatching rate is expected only after periods less than 36 h (Blaxter and Holli- day, 1963). It is now possible to extend the life of fish sperm for much longer periods using cryopreservation techniques (- 196°C) (Erdahl and Graham, 1980). Various cryoprotective agents have been used to freeze sperm of marine fishes including glycerol (Blaxter and Holliday, 1963), glucose, NaCI, Ringer's solution and fish serum (Hara et al., 1982). The stress of capture causes female Katsiiwonus pelamis to ovulate and spawn within 24 h after capture but eggs are often not viable (Kaya et al., 1982), Maturing marine fish in the lab- oratory and spawning them by hormone injections has become routine in recent years and is preferable to stress techniques. Examples include Engraulis mordax (Leong, 1971), Scomber japonicus (Leong, 1977), Chanos chanos (Liao et al., 1979), Bairdiella icistia (Haydock, 1971), Paralichthys denial us and Pseudopleuronectes americanus (Smigielski, 1975a, b) and oth- ers (see review of Lam, 1982). Induction of spawning in the laboratory may require an open sea water system, large holding tanks (e.g., -3 m dia. or larger), temperature and light control. Handling and stocking.— To count eggs without damaging them we recommend a polished wide bore (~3 mm) pipette; count 30-50 late stage eggs at a time in a depression slide under a dissection microscope, and wash eggs off the slide by immersion of the entire slide in sea water. Counting eggs is critical because higher mortalities and slower growth result from excess stocking densities (Houde, 1975 and 1977). As a rule stocking densities in rearing tanks of 8 eggs/I or less seems preferable and most rearing successes have occurred when stocking did not exceed 20 eggs/1 (Houde, 1975). Similarly, the mortality of Mugil ceph- a/(« larvae seems to remain constant (2-3% loss/day) at stocking densities of 1-30 larvae/1 (Kraul, 1983). Apparatus Containers and lighting. — Larvae appear to grow faster and show fewer signs of starvation when reared in large containers (100 1) rather than in smaller ones (10 1) (Theilacker, 1980b). Opti- mum container size doubtless varies with species but 40 1 con- tainers are probably the minimum size that should be used and I prefer 100-400 1 containers. We use cylindrical black fiberglass containers although excellent results are obtained using ordinary rectangular glass aquaria (Houde, 1975). It is traditional to provide a daily cycle of illumination to HUNTER: CULTURE METHODS 25 larvae in rearing containers although constant illumination is occasionally used. Typically fluorescent lamps are used which provide 2,000-3,000 lux at the water surface (Houde, 1978; Hunter, 1976). Night light levels vary; we provide no light at night whereas Houde (1978) provides a dim light of 40-90 lux at night, which is substantially above the visual threshold for feeding for larval E. morda.x (6 mm larvae 50% feeding thresh- old = 6 lux, and 10-15 mm larvae 50% threshold = 0.6 lux, Bagarinao and Hunter, 1983). Clearly, longer periods for visual feeding will probably enhance growth if food is limited. Rearing at high light intensities such as natural sunlight may greatly increase production of algae and zooplankton in the culture tank and thereby increase larval survival (Kraul, 1983). On the other hand, solar UV radiation is clearly lethal to younger larvae (Hunter etal., 1 982) and use of deep tanks, or shaded or covered tanks (screen cloth, acrylic plastic, glass or mylar film) is rec- ommended for the first 1-2 weeks of larval life if tanks are to be exposed to solar radiation. Water qualily.—C\osed, non-circulating systems are typically used to rear marine fish larvae at least during the younger stages, because in an open system planktonic larvae and their foods are easily lost. Older (nektonic) larvae are able to resist the current and to consume a daily ration in a short period so a partially open system can be used. We fill our rearing containers with UV treated sea water that is passed through three, in line, cartridge filters (5, 3 and 1 ^m pore).' Although not a common practice in small scale rearing work, the addition to rearing tanks of antibiotics (sodium penicillin G at 50 i.u./ml plus strepto- mycin sulphate at 0.05 g/ml) slightly improved survival of Pleu- ronectes platessa eggs through hatching, but surprisingly this single treatment greatly improved survival of larvae through metamorphosis (Shelboume, 1975). Use of a closed system requires attention to water quality, a problem which may be intensified at higher rearing tempera- tures. In the most complete study of water quality in rearing tanks for marine fish larvae, Brownell (1980a, b) considered seven variables (pH, dissolved oxygen, carbon dioxide, am- monia, nitrite and nitrate), but only high pH, low dissolved oxygen and un-ionized ammonia had effects at levels likely to be encountered in rearing tanks. First feeding incidence declined by 50% in all species he studied when dissolved oxygen con- centrations were between 4 and 4.75 mg/1 (49-58% saturation). Dissolved oxygen in our rearing containers usually is not sat- urated after planktonic foods are added, and typically it is about 80% saturation even with aeration. Clearly water quality is im- proved by aeration and frequent water changes and lank clean- ing. Werner and Blaxler (1980) exchanged 20% of the water in Clupea harengus cultures (9° C) 3 times per week but at high temperatures greater replacement rates are required. For ex- ample Houde (1977) replaced 20% of the tank sea water on alternate days while culturing Anchoa mitchilli and Achirus lin- eatus at 26-28° C. Frequent tank cleaning is important as heavy mortalities may result from toxins produced by debris on the container bottom (Kraul, 1983). Aeration, unless very gentle, can cause heavy mortalities among delicate eggs and newly hatched larvae. In fact, Shelboume (1964) recommends no aer- ' Aqua-Pure model APIO. AMP Cuno Division, Inc., Meriden. Con- necticut USA. ation for Pleuronectes platessa larvae. I recommend very gentle aeration but not until a week or so beyond the first feeding stage. The mortality of cultured fish larvae often increases during the period of initial swim bladder inflation in physoclistous fishes (Doroshev et al., 1981; Kuhlmann et al., 1981) and this could be related to water quality. Symptoms include delay or complete failure of inflation or excessive inflation; in either case normal swimming patterns are disrupted and death frequently results. The causes of abnormal inflation are not clear; preven- tion of larvae from reaching the water surface prevented excess inflation in M. cephalus larvae (Nash et al., 1977), whereas the same treatment in Atractoscion nobilis larvae had no effect. In A. nobilis excess inflation was associated with abnormal devel- opment of gas secretory tissue suggesting a more complex etiol- ogy (SWFC. unpubl. data). Failure to inflate the swim bladder is a common problem in Morone saxatilus culture and turbulent aeration may reduce the incidence of this disease (Doroshev and Comacchia, 1979) but it now appears that reduction in salinity from 17 ppt to 4 ppt has a much greater eflect in reducing the incidence of swim bladder malfunction (S. Doroshev and J. Merritt, U. Cal. Davis, pers. comm.). Food The most critical aspect of rearing marine larvae is manage- ment of their food. Food must be the correct density, size, nutritionally adequate and must remain suspended in the water column which usually requires the use of living pelagic organ- isms. Food size.— Typ\c&\ pelagic fish larvae are 2.5-4.0 mm when they begin feeding and acceptable prey are 20-1 50 /um in breadth (Houde and Taniguchi, 1979). Some large larvae, e.g.. larval C. harengiis (B\di\\.QT. 1965). Pleuronectes platessa {Riley. 1966) or small larvae with large mouths, e.g., Merluccius productus {Sum- ida and Moser, 1980), can begin feeding on prey 300 Mm or larger in breadth. The optimal food size increases as larvae grow (Hunter, 1981), so any culture technique should provide a stead- ily increasing range of food sizes, because if the food is too small growth slows and mortality occurs (Hunter, 1981). Food size requirements can be expressed in terms of the ratio of prey width to mouth width. The 50% threshold for feeding on a prey of a particular width occurs when this ratio is about 0.75, although occasionally larvae consume prey as wide as the width of their mouth (ratio = 1) (Hunter, 1981). At the onset of first feeding a small prey of about 'A the mouth width seems to be preferable as capture success is low at this time but within a few days larvae are able to consume food of about V2 the mouth width. Wild zooplankton— V/i\d zooplankton, primarily the naupliar and copepodite stages of marine copepods but also mollusc veligers, tintinnids, cladocera, and appendicularia larvae, are the natural foods of most marine fish larvae and probably also the best source of food for rearing a larval taxonomic series. Wild zooplankton provide a wide range of sizes and types and are probably nutritionally superior to cultured rotifers and Ar- lemia nauplii (Kuhlmann et al., 1981). Collection of wild zoo- plankton may require less effort than production of cultured food except for brine shrimp nauplii (see below). Zooplankton is collected in nets of about 50 ^m, and is graded by size in the laboratory using various nylon nets (Houde, 1977, 1978), This eliminates the larger zooplankton which larvae would be unable 26 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM to consume and which may be larval predators. Fish larvae, particularly yolk-sac stages, are vulnerable to various carnivo- rous copepods, amphipods, euphausiids and chaetognaths (Hunter, 1981). Cultured foods.— T-wo cultured foods, the rotifer Brachiomts plicatilis, and nauplii of the brine shrimp, Arteinia. should be considered as potential foods for rearing marine fish larvae as many fish larvae can be reared on a combination of these two foods. These two foods may also be used as a supplement to diets of wild plankton. Groups of fishes that have been reared to metamorphosis on a combination oi Brachionus and Anemia or on Artemia alone include C. harengns, species of serranids, scombrids, atherinids, various flatfishes, sciaenids, and saganids (May, 1970; May etal., 1974; and unpubl. SWFC data). /lr?ew;a nauplii are recommended only for larvae with differentiated guts as they are quite resistant to digestion whereas copepods are not (Rosenthal, 1969). Methods for culturing rotifers using algae are given by Thei- lacker and McMaster (1971); culture methods employing for- mulated artificial diets or freeze dried algae (Gatesoupe and Robin, 1981; Gatesoupe and Luquet, 1981) and ones using brewers yeast also exist. Many of the essential facts given in these original papers will not be repeated here but I will point out a few practical points regarding rotifer culture using algae. Suitable algae species for rotifer culture include Dunaliella, Nannochloris, Tetraselmis, and Chlorella which may be grown using standard culture media (Guillard, 1975) or using liquid commercial plant fertilizers (dosage for fertilizer containing 8% total nitrogen = 0. 1 ml of fertilizer/1; dosage among brands is adjusted depending on total N content). We prefer commercial plant fertilizers that have an organic base such as liquid fish fertilizers and avoid those that have soil penetrants. A daily doubling rate can be expected in healthy rotifer cultures, and cultures can be maintained for weeks or even months by adding fresh algae or nutrients and sea water, although single batch harvesting after about 2 weeks gives more dependable results. Rotifers are harvested using gravity flow through a nylon filter (20-40 ^m mesh) as pumps may kill rotifers. Production ofArlemia nauplii is simple since all that is needed is to hatch the cysts ("Anemia eggs"). Cysts from a variety of strains of Anemia are commercially available. The strains differ considerably in average naupliar size (423-775 ^m length), in pesticide content (DDT, PCB, and chlordane) and in certain fatty acids (Klein-MacPhee et al., 1982). These authors show that very low survival (15%) of P. amehcanus larvae occurred when they were fed San Pablo Bay (San Francisco) nauplii whereas survival of larvae fed other strains varied from 60- 80%. Beck et al. ( 1 980) gave similar results for Menidia menidia larvae. Of all the strains tested in these papers the Australian and Brazilian strains seem the most suitable for rearing larvae and the San Pablo Bay (USA) the least. - Anemia hatcheries vary from a jar to complex automated systems. The J. D. Riley Anemia hatching box has been used with slight modification in many laboratories for over 20 years. It is a sea water filled box separated in half by a sliding partition; Anemia cysts are added to one side (I g/l) and they hatch 1-2 ^ Exotic Anemia cysts are available from: Artemia Inc., P.O. Box 2891, Castro Valley, California 94546 USA and Biomarine Research. 4643 W. Rosecrans, Hawthorne, California 90250 USA. days later depending on the temperature selected (23-30° C). The tank is then illuminated, the partition raised slightly off the bottom, and the nauplii, attracted by the light, swim beneath the partition leaving behind the hatching debris and unhatched cysts (Shelboume, 1964). A semiautomatic version of this sys- tem is described by Nash (1973), and various other improve- ments in aeration, illumination, temperature, and other factors have increased yields to lO' nauplii per 4.8 g of cysts (San Francisco Bay Brand) (Dye, 1 980). In recent years decapsulation of Anemia cysts using hypochlorite bleach has become popular because it increases yields, increases the dry weight of the nau- plius (Bruggeman et al., 1 980) and eliminates contamination of larval fish rearing tanks with unhatched cysts. It should also be noted that freshly hatched Anemia nauplii are clearly more nutritious than older starving individuals and consequently new batches should be frequently produced. In general, prey with full stomachs are probably nutritionally pref- erable to ones with empty stomachs. Similarly, more Dicen- trarchits labrax larvae seem to survive when rotifers are nutri- tionally enhanced by 30 min immersion in a solution containing vitamins and soluble proteins (Gatesoupe and Luquet, 1981). Mass culture of marine copepods is difficult and laborious and therefore not recommended when a taxonomic series is the sole objective. Nevertheless, culture of marine copepods may be the only way some fish larvae can be reared if wild zooplank- ton is not readily available and larvae die when fed Anemia nauplii (rarely are more than a single strain of Anemia tested, however). Harpacticoid copepods (Tignopus sp., Tishe sp., and Euterpina sp.) are the most frequently used copepods because of ease of culture; for culture techniques see Kahan et al. (1982) and Hunter (1976). Euterpina may be preferable to Tignopus or Tishe because the nauplii and copepodites of Euterpina are pelagic and therefore available to the larvae whereas nauplii and copepodites of Tigriopus and Tishe tend to remain on surfaces and are therefore less available (Kraul, 1983). See Nassogne (1970) and Zurlini et al. (1978) for laboratory culture of Euter- pina. Eood density. —The optimal food density for fish larvae depends upon the size of the food organism and size or age of the larvae. Densities of 1-3 organisms/ml have been routinely used for larvae fed wild zooplankton (largely copepod nauplii) during the first 1-2 weeks of feeding (Houde and Taniguchi, 1979). The same density range is used when cultured .Anemia nauplii are the food. A higher density range (IO-20/ml) is used for cultured B. plicatilis which are about 1/10 of the weight of an .irtemia nauplius (Theilacker and McMaster, 1971). A very small food particle, the dinoflagellate Gymnodinium splendens (40 nm dia), is used for the first 2 days of feeding in northern anchovy larvae (Lasker et al., 1970; Hunter, 1976) at a high density of about lOO/ml. In very active species such as S. ja- ponicus or the siganid Siganus canaliculatus high food densities can cause heavy mortality because of overfeeding since most larval fishes seem to lack a satiation mechanism (May et al., 1974; Hunter, 1981). Overfeeding seems to occur only when such easily captured prey as .irtemia nauplii are used as food. Piscivorous fish /arvac — Piscivorous fish larvae such as the scombroids, Sphyraena and others pose special problems in culture. Fish larvae are an ideal food for such larvae; in fact, our only success in rearing Katsuwonus pelamis larvae to meta- morphosis was probably related to an abundant supply of yolk- HUNTER: CULTURE METHODS 27 sac fish larvae as food. Zooplankton is the initial food until piscivorous feeding habits develop (Houde, 1972b; Mayo, 1973; Hunter and Kimbrell, 1980). Piscivorous larvae manipulate their larval prey and consequently are less dependent on mouth size when consuming larval fish. Sibling cannibalism is common under reanng conditions in such fishes. Increasing the food den- sity may increase survival as may elevating the temperature, thereby accelerating growth through the most cannibalistic sizes; at least in scombroids sibling cannibalism declines at meta- morphosis (Mayo. 1973; Hunter and Kimbrell, 1980). Sorting by size and isolating the larger larvae is probably the only certain method for controlling losses due to cannibalism, however. Phytoplankton Phytoplankton blooms are often maintained in larval culture tanks to reduce the detrimental effects of metabolic by-products which accumulate in static rearing tanks (Houde, 1974) and to provide food for larval food organisms. In many cases dense blooms of phytoplankton enhance larval growth and survival and I recommend the practice but the mechanism is obscure. The phytoplankters used are various, easily grown, small species such as Chlorella. Anacystis, Nannochloris, Tetraselmis. Dun- aliella. Isochrysis. Phaeodactylum and others.' They are main- tained at high densities (10,000 or more cells/ml) in the rearing tanks. At high cell densities larvae ingest these small phyto- plankters, perhaps inadvertently (Moffatt, 1981) but they appear not to be able to exist on them as a sole food source (Houde, 1974; Scura and Jerde, 1977). They may supplement the food ' For a nominal fee starter cultures of manne phytoplankton can be obtained from R. R. L. Guiliard. Bigelow Laboratory for Ocean Sciences. McKown Point, West Boothbay Harbor, Maine 04575 USA; culture methods are discussed by Guiliard (1975). ration either directly or indirectly through the ingestion of prey having guts full of algal cells (Moffatt, 1981). Evidence now exists that enhancement of growth and survival of larval Scoph- ihalmus maximiis by blooms of Isochrysis and Phaeodactylum is due to the inclusion in the diet of certain polyunsaturated fatty acids not occurring in the normal laboratory rotifer diet (Scott and Middleton, 1979). It is interesting in this regard that Dunaliella which lacks the fatty acids did not enhance S. max- imiis larval growth or survival. Effects of Culture Extrapolation from cultured larvae to natural populations must be done with caution because culture may affect the morphology, behavior and biochemistry of larvae (Blaxter, 1976). The mor- phological characteristics most susceptible to modification in tanks are those partially controlled by environmental conditions such as vertebrae and fin ray counts. Reared larvae also may be more heavily pigmented than sea caught specimens (Watson, 1982). This appears to be related to the expanded nature of the melanophores, not to added numbers of pigment cells. In ad- dition, pigmentation events may occur at smaller sizes in reared material (S. Richardson, Gulf Coast Research Laboratory, Ocean Springs, Mississippi, pers. comm.). Laboratory reared larvae are often heavier and have deeper bodies than their wild counter- parts, making some morphometric measurements on laboratory specimens useless (Blaxter, 1975). The differences in preserva- tion and handling between laboratory and sea-caught larvae also make direct size-specific comparisons difficult. Shrinkage in length may vary greatly depending on the duration larvae re- main in plankton nets and shrinkage differences between reared and wild specimens can be misinterpreted as morphological differences (Theilacker, 1980a). National Marine Fisheries Service, Southwest Fisheries Center, P.O. Box 271, La Jolla, California 92038. Identification of Fish Eggs A. C. Matarese and E. M. Sandknop A wide variety of egg types exists among teleost fishes in both freshwater and marine environments. Eggs may be pelagic and nonadhesive or demersal and either adhesive or not. They may possess a variety of specialized structures aiding in flotation or attachment. Depending on egg type and associated repro- ductive ecology, many characters are useful in identification. These characters have been reviewed for pelagic marine eggs by Rass(1973), Robertson (1975a), Russell (1976), and Ahlstrom and Moser ( 1 980); we have liberally and extensively drawn from the latter. Important characters for other egg types have been discussed in part by Balon (1975a, 1981a), Hardy (1978a, b), Jones et al. (1978), and Snyder (1981). Characters such as size and possession of oil globules are important for all types; how- ever, perivitelline space and chorion sculpturing are more im- portant in pelagic eggs, while in demersal eggs special coatings. chorion thickness, or nature of egg deposition may be more useful. A wealth of potential characters useful in egg identification exists; however, it is still difficult to identify eggs of most species with certainty. Except for late stages, few may be recognized at the species level. Some characters are useful at a family level, but presently it is not productive to speculate on the systematic significance of any characters (see Kendall et al., this volume). Presently, the main goal of taxonomy with respect to fish eggs is identification. Regardless of egg type or reproductive ecology, a summary of identification characters useful to an egg taxonomist is pre- sented. Additionally, we recommend using available literature for reference and encourage the building of local fish egg col- lections. We follow Ahlstrom and Ball (1954) in subdividing 28 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM 1.34x0.66 Engraulis mordax B 1.0x1.06 Ophidion scrippsae Unidentified 0.58-0.74 Vinciguerria lucetia 1.9 Glyptocephalus zachirus 0.80 Symphurus atricauda H Prionotus stephanophrys 2.92 Icosteus aenigmaticus 1.35 Etrumeus teres Fig. 13. Fish eggs. Captions under each illustration indicate the species and the diameter or dimensions of the egg in millimeters. A. Engrauli mordax. original; B. Ophidwn scrippsae. onginal; C. Unidentified, original; D. Vincigiierna tucetia. from Ahlstrom and Counts (1938); E Glyptocephalus zachirus. from Ahlstrom and Moser (1980); F. Symphurus atricauda. original; G. Prionotus stephanophrys. onginal; H. Icostei. aenigmaticus, original; and I. Etrumeus teres, original. 'is E. •osleus MATARESE AND SANDKNOP: EGG IDENTIFICATION 29 egg development as follows: Early— from fertilization to closure of blastopore. Middle— from closure of blastopore to tail bud lifting off yolk, and Late — from tail bud lifting off yolk to time of hatching. Identification Characters Shape.— The vast majority of all egg types are spherical. Ex- ceptions include ellipsoidal eggs as found in anchovies, En- graulis and Anchoa. and slightly flattened or ovoid eggs as seen in members of the families Gobiidae, Scaridae, and Ophidiidae (Fig. 13A. B). A number of demersal eggs have somewhat ir- regular shapes, especially those associated with large egg masses. The perciform family Congrogadidae has cruciform shaped eggs (Herwig and Dewey, 1982). An unidentified, star-shaped egg is encountered infrequently in the Alaska region (Fig. 13C). Size.—T\\t average marine and freshwater fish egg size is about 1.0 mm. According to Ahlstrom and Moser (1980), pelagic fish eggs range from 0.5 mm [Mncigiicnia (Fig. 13D)] to about 5.5 mm (Muraenidae). Demersal eggs may range higher in size (up to 7.0-8.0 mm), e.g., members of the families Salmonidae, An- arhichadidae, and Zoarcidae. Mouth brooders, e.g., in the catfish family Ariidae, have among the largest eggs with sizes from 1 4 mm to 26 mm. Oil globules.— The oil globule provides useful characters in fish egg identification; these include presence or absence, number, size, position, color, and pigmentation. Among both pelagic and demersal eggs, the most common form contains a single oil globule. Eggs may lack an oil globule as in most gadines and pleuronectids (Glyplocephaliis). contain only one (Icosteiis), or have multiple oil globules as in the cynoglossids and triglids (Symphums and Prionotus) (Fig. 13E, F, G, and H). In pelagic eggs with a single oil globule, the size ranges from <0.10 mm to > 1.0 mm (Ahlstrom and Moser, 1980). The position of the oil globule within the yolk sac is usually posterior, but several groups contain species that have an anterior placement (e.g., labrids and carangids) and others have an intermediate place- ment (argentinids). In some fishes, oil globules migrate during embryonic development. Some members of the family Bathy- lagidae initially possess multiple oil globules that eventually coalesce into a single globule (Ahlstrom, 1969). Although not a totally reliable character, the oil globule color can be useful, especially in the identification of freshly taken demersal eggs. Lastly, many species have oil globules with melanistic pigment, Icosteus (Fig. 13H) and Icichthys. Yolk.— The degree of yolk segmentation is an important iden- tification character. Yolk is usually segmented in primitive forms, e.g., Etruineus (Fig. 131), and homogeneous in higher forms (Rass, 1973; Ahlstrom and Moser, 1980). The opaqueness of yolk found in catfishes, salmonids, and gars can be diagnostic' Pigment, which may also be diagnostic, can be present dunng various developmental stages from middle to late. Yolk color is often important especially in demersal eggs. Among demersal eggs vitelline circulation patterns within the yolk sac are useful in identification.' ' P. Douglas Martin, Chesapeake Biological Laboratory, P.O. Box 38, Solomons, Maryland 20688. Personal communication, October 1982. Chorion. — A. number of characteristics associated with the cho- rion or egg envelope can be useful in identifying fish eggs and have been shown to be highly adapted to the environmental conditions under which an embryo develops (Ivankov and Kur- dyayeva, 1973; Stehr and Hawkes, 1979; Laale, 1980; Stehr, 1982). The most important character of the chorion is whether it is smooth, as is in most fishes, or sculptured. Among fish eggs with patterns, the size and texture (e.g., raised hexagons, pus- tules) of the design are diagnostic. Raised polygonal surfaces are found in several unrelated species (Stehr, 1982), e.g., Synodus and Pleuronichthys (Sumida et al., 1979), and pustules occur among some bathylagids and argentinids. Mugil cephalus eggs (Fig. 14A), previously considered to have a smooth chorion, have a raised patterned surface visible by scanning electron microscope (Boehlert, this volume). In many groups of fishes, the chorion has various degrees of ornamentation consisting of projections, threads, filaments, or stalks which may aid in flo- tation (pelagic) or attachment (demersal). In some scombere- socids, e.g., Cololahis (Fig. 14B). some exocoetids and ather- inids, pelagic eggs are attached to each other or to a substrate by filaments. Spines are found in some myctophiforms and exocoetids, and stalks occur in some demersal egg groups, e.g., blenniids and Osmerus mordax. In ostraciid eggs, a patch of pustules is present near the micropyle (Fig. 14C). Recently, thickness of the chorion has been of diagnostic value (Ivankov and Kurdyayeva, 1973; Boehlert, this volume). Stehr and Hawkes (1979), using scanning electron microscopy, found that most marine teleosts with pelagic eggs have thin chorions in relation to egg diameter whereas demersal eggs tend to de- velop much thicker chorions. Color of the chorion is an im- portant diagnostic character, especially for freshly taken de- mersal eggs in the marine intertidal environment (Matarese and Marliave, 1982). A number of freshwater demersal fishes have eggs that possess a special coating associated with the chorion which can be either gelatinous or adhesive, e.g., Perca. Icialurus, and Notropis (Snyder, 1981). Penvilelline space. — Most fish eggs have a narrow- to medium- width perivitelline space, but wide spaces are common in some groups, especially among the more primitive fishes that have a segmented yolk, e.g., Clupeiformes (Sardinops. Fig. 14D), An- guilliformes, and Salmoniformes (Chauliodus. Fig. 14E) (Ahl- strom and Moser, 1980). Large perivitelline spaces are also found among some unrelated higher forms, such as cypnnids (Nolro- pi.s). percichthyids (Morone saxatill.s). or pleuronectids (Hip- poglossoides). Embryonic characters.— CharacXers associated with the devel- oping embryo are extremely useful in egg identification, partic- ularly in the middle and late stages of development. Many eggs not identifiable in the early stages are easily recognizable using embryonic characters such as pigment on embryo or finfold and morphology. In some fishes, embryonic pigment in the late stages has already undergone sufficient migration and rearrangement to the point where it resembles the yolk-sac larva; this is com- mon in several groups including gadiformes, e.g., Merluccius (Fig. 14F), Gadus. and Theragra. and heavily pigmented flat- fishes like Pleuronichthys and Hypsopsetta. Characteristic late- stage pigment bands appear in Glyptocephalus (Fig. 13E). In most freshwater species, pigment is not present prior to pigment cell migration but appears sometime after the cells have mi- 30 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM 0.76-0.80 Mugil cephalus B 1.7x1.9 Cololabis saira 1.54x1.68 Ostraciidae 1.35-2.05 Sardinops sagax 2.93 Chauliodus macouni 1.07-1.18 Merluccius productus H 2.0 Eumicrotremus orbis 2.65-2.90 Trachipterus altivelus 0.88 Stomias atri venter Fig. 14. Fish eggs. Captions under each illustration indicate the species and the diameter or dimensions of the egg in millimeters. A. Mugil cephalus. original; B. Cololabis saira. original; C. Ostraciidae, original; D. Sardinops sagax. original; E. Chauliodus macouni. original; F. Merluccius productus. from Ahlstrom and Counts ( 1 955); G. Eumicrotremus orbis. from Matarese and Borton unpubl. MS; H. Trachipterus altivelus. original; and I. Stomias aim-enter, original. grated lo their actual destinations (Snyder, 1981). As seen in the cyclopterid, Eumicrotremus. most late-stage demersal em- bryos resemble the newly hatched larva with respect to all char- acters (Fig. 1 4G). The morphology of the head, gut, and postanal body as well as the number of myomeres is used for identifi- cation within all tish egg groups. A number of specialized char- acters associated with the embryo are essential for identification when present, e.g., elongated fin rays— J'rachiplerus (Fig. 14H), MATARESE AND SANDKNOP: EGG IDENTIFICATION 31 precocious fin development (caudal— exocoetids and Tricho- don\ pelvic— Trachi mis), and pelvic disc development in some cyclopterids (Eumicrotremus) (Fig. 14G). Miscellaneous characters. —The presence of a secondary mem- brane inside the chorion occurs in some groups, although it is lacking in most fishes. Sloniias alnvcnter eggs have a double membrane (Fig. 141). These membranes occur in some of the more primitive fishes including members of the Anguilliformes, Clupeiformes, and Salmoniformes. In some species, like the freshwater cyprinid Abbottina rivularis (Nakamura, 1969), the secondary membrane is thick and gelatinous. The presence and size of the micropyle are diagnostic in other fishes, particularly freshwater demersal eggs (Laale, 1980; Riehl, 1980). Among freshwater fishes, the cleavage pattern is important for egg iden- tification. In the more primitive families (Acipenseridae, Poly- odontidae, Lepisosteidae, and Amiidae), cleavage pattern is typ- ically semiholoblastic as opposed to the meroblastic pattern seen in the higher teleosts. Genetic studies have shown differences in LDH A zymograms to be a useful, diagnostic tool for the identification of Gadus morhua and Melanogrammus aeglefinus eggs (Mork et al., 1983). Ecological and behavioral considerations.- \ number of con- siderations related to mode of reproduction and collection rather than the characters of the eggs themselves are essential when identifying any type offish egg. In identifying demersal eggs one must consider where they were collected — on rocks, on plants, in masses, and if parental care is involved. Nest type, nature of egg deposition, and the presence of guarding parents can all be essential clues to proper identification. Also, for any egg type one must note spawning time (season), location depth, and gear used for collection. In addition, the rearing of unknown eggs to an identifiable larval stage is useful in species determination as shown by Stevens and Moser (1982) for the blenny, Hypso- blennius. Of course, a necessary prerequisite to accurate iden- tification of eggs is a thorough knowledge of the species present in any given area and their breeding seasonality. Summary of Characters Characters most useful in identification of fish eggs are the following: ( I ) egg shape— spherical, ellipsoidal, irregular, or oth- erwise; (2) egg size— fish eggs range in size from 0.5 to 26.0 mm; (3) oil globules— presence or absence, number, size, color, po- sition, and pigmentation; (4) yolk — segmented or homogeneous, nature of segmentation, color, pigmentation, and circulation pattern; (5) chorion— smooth or ornamented, type of ornamen- tation, thickness, color, and coatings; (6) perivitelline space- width; (7) embryonic characters— morphological features, pig- ment patterns, and special structures; (8) miscellaneous char- acters—inner or secondary membrane (presence or absence, lo- cation), cleavage pattern, micropyle (size), and biochemical analysis; and (9) ecological and behavioral considerations— col- lection (gear, location, season, etc.), and mode of reproduction (nests, parental care, etc.). (A. CM.) National Marine Fisheries Service, Northwest AND Alaska Fisheries Center, 2725 Montlake Boule- vard East, Seattle, Washington 98112; (E.M.S.) Southwest Fisheries Center, P.O. Box 271, La Jolla, California 92038. Identification of Larvae H. POWLES AND D. F. Markle MINOR errors in identification of larval fishes can lead to major misinterpretations of ecological and taxonomic phenomena. Fish identification and taxonomy are largely based on adult characteristics and since these develop during the larval period, new characters must be discovered and validated in order to identify larval fishes. Usually larvae possess fewer char- acters than adults and are more fragile. Identification can, there- fore, be difficult and, frequently, must be based on a combi- nation of character states. Since larval anatomy is by its nature dynamic (a given spec- imen being a snapshot of the process linking embryos to adults), developmental series are essential to identification. Three dif- ferent approaches are used to identify larvae, the first two of which arc based on developmental series: I) to raise eggs and larvae from fertilized eggs of known parents; 2) to work back- wards from the adult utilizing characters common to succes- sively earlier ontogenetic stages; and 3) to extrapolate from pre- vious results obtained by (1) or (2) to synthesize generic or familial diagnoses and identify by process of elimination or limited corroboration (Ahlstrom in Berry and Richards, 1973; Leiby, 1981). There are pitfalls in all approaches. Laboratory-reared larvae are frequently more heavily pigmented than wild-caught spec- imens and may show greater meristic variation (Lau and Shaf- land, 1982). Laboratory rearing may be financially and logis- tically difficult or impossible for fishes of interest. Ontogenetic transformations arc based on associations of adult diagnostic characters with characters that persist in progressively earlier ontogenetic stages. This method requires careful attention to methodology, as well as good ontogenetic series which are not always available. Purely descriptive accounts of larval series (laboratory-reared or reconstructed) may not be useful for iden- tification purposes if no diagnostic characters that will distin- guish sympatric congeners and/or similar-looking forms are pre- 32 ONTOGENY AND SYSTEMATICS OF FISHES -AHLSTROM SYMPOSIUM sented. Novel sorts of characters or ways of manipulating data are sometimes needed to identify larvae and the data required may not be retrievable from "standard" descriptive accounts. Synthesis and elimination is the normal procedure used by tax- onomists to identify adult fishes. It has been called the "look- alike" system when applied to larval fishes (Leiby, 1981). It is basically a simple procedure but the pitfalls are numerous and subtle. As with some early adult fish taxonomy, premature syn- thesis may often be based on the wrong characters (e.g. con- vergent characters) and lead to spurious identifications. General references on larval fish identification include Berry and Richards (1973), Ahlstrom and Moser (1976) and Moser (1981). Some recent works which provide exposure to a wide range of larval forms and literature are Ahlstrom and Moser (1981) and Fahay (1983) for marine taxa, and Auer (1982) and Balon (1975a, 1981a) for freshwater taxa. The purpose of the following is to describe the tools— pref- erably sharpened, polished and comfortable to use— which should be at hand when the ichthyologist sits down to identify larval fishes. Our emphasis is on three main factors: 1 ) the larval fish — its anatomy, ontogeny, and phyletic relationships; 2) the study area— its ecology and zoogeography and 3) the investigator— his experience, knowledge and ingenuity. Systematics, Ontogeny and Anatomy Perhaps the most important type of character for identifica- tion of larvae is meristic, as counts usually do not increase or decrease once established. All meristic characters can be im- portant, but vertebra/myomere counts and fin element counts are of particular value. Meristic variables are useful at different taxonomic levels, e.g., principal caudal fin ray and pelvic fin element counts at the family or order level, median fin elements at the genus/species level, pectoral fin ray counts at the species level. Frequency distributions of meristic counts are extremely important (particularly when it is uncertain whether develop- ment of a character is complete) but often are not given in published literature. Some important characters may not be included in published studies (e.g., pectoral fin rays, procurrent caudal rays). Differences in methodology and variable attention to detail may also affect the quality of published meristic data. Thus, published studies must be treated with caution and one must be prepared to collect and compile one's own information when opportunities arise. Despite potential problems with pub- lished works, these are the obvious place to start with compi- lations. Few "regional" meristic publications as exemplified by Miller and Jorgensen (1973) exist, but many publications on larval fishes include extensive tabulations of meristic infor- mation. Various ways exist for facilitating use of meristic compila- tions. A simple taxonomic listing (e.g.. Miller and Jorgensen, 1973) can be time-consuming to use, while a "gazetteer" format, with species arrayed in order of counts (e.g., Fahay, 1983) may be more practical. X-Y plots of two meristic variables (e.g.. Berry, 1959b) can include frequency distributions and be very useful for separating closely-related forms. A second suite of characters of broad use is specialized larval characters which may characterize whole groups. These include but are not limited to: characteristic shapes (e.g., Anguilli- formes/Elopiformes, Pleuronectiformes), spination (Acanthur- idae, Holocentridae), fin development patterns (argentinoids), fin element development (Pleuronectiformes, epinepheline Ser- ranidae), fin placement (pelvic fin placement in Pleuronecti- formes), eye shape (myctophid subfamilies, salmoniform groups), and phoiophore development pattern (Gonostomati- dae). The elucidation of such characters is a focus of this volume, and reference should be made to specific chapters for further detail. The important point is that a broad knowledge of larval fishes is frequently necessary for accurate, efficient identification of larvae. Finally, identification of larvae depends on a suite of dynamic characters (pigmentation, body form, spination, fin develop- ment pattern, etc.), which may change rapidly and differentially over a small size range. Generally, a combination of such char- acters is required for accurate identification; this is particularly true in early stages. These characters can vary extensively, even within a species, due to regional differences; method, time or area of collection; preservation method or duration. Develop- mental changes can be extremely rapid (e.g., changes in mela- nophore distribution from some yolk-sac to post-yolk-sac lar- vae). Again, no extensive treatment of these characters is possible here, but the important point is that detailed, disciplined ob- servations of larvae are essential for accurate identification. The importance of osteological characters for larval identi- fication is increasingly recognized (Dunn, this volume). Use of these depends on clearing and staining techniques (PotthofT, this volume) or X-ray techniques (Tucker and Laroche, this vol- ume). As with meristics, osteological characters may be useful at different taxonomic levels. Caudal osteology has been widely used because of its early development and relative simplicity, but cranial osteology and pterygiophore patterns are also useful. Recent application of cartilage-staining techniques has permit- ted use of cartilaginous structures in identifying larvae (e.g., Fritzsche and Johnson, 1980). Other internal characters such as gut shape (Ahlstrom and Moser, 1976; Govoni, 1980) may also be useful. Keys have not generally been used in larval fish identification because of the dynamic nature of characters (a separate key would be required for each size class or development stage) and because of "incompleteness" of information (i.e., it has usually been impossible to completely cover a defined region or sys- tematic group with a key). Generally, much more information is required to identify a larva than an adult, and summarizing this in a key has been impractical (the information-organizing capacity of computers may eventually help to permit this). Ex- ceptions, such as Bertelsen's (1951) key to larval Ceratioidea, Johnson's ( 1 974b) key to genera of larval scopelarchids, and the key of Bertelsen et al. (1976) to notosudids do exist. Because of the complexity of identification of larvae, a wide ichthyological background is important. A good knowledge of fish anatomy is essential, particularly when (as often occurs) damaged specimens must be identified. Published descnptions exist, for example, which interpret broken branchiostegal rays as jugular pelvic fin rays. A general knowledge of suspected phylogenies and inter-relationships (e.g.. Greenwood et al., 1966; Nelson, 1976) is essential if attempting to identify by synthesis or elimination. This should at least cover those groups to be expected in a given area, but wider knowledge is desirable, par- ticularly in the marine environment where exotic larvae may be transported great distances (e.g., Markle et al., 1 980). Finally, thorough familiarity with the ontogenetic continuum is neces- sary to place unknown specimens in perspective. Absorption of the yolk sac, flexion of the notochord in the caudal region, development of median fins, and transformation from larval to POWLES AND MARKLE: LARVAL IDENTIFICATION 33 juvenile stages (as defined by completion of fin element devel- opment, development of scales, etc.) are major events in fish development which have been used by various authors to define stages (e.g., Ahlstrom, 1968; Snyder, 1976). Ecological Considerations There are two basic ecological or zoogeographic consider- ations when identifying larvae: the expected composition of the larval ichthyofauna of the study area and the potential for influx from "upstream" areas. Thorough knowledge of the adult ichthyofauna of the study area is essential in order to know what larvae may occur; thus, the most complete possible list of adult species is required. Literature may be incomplete or erroneous, so this list should be based on unpublished or personal observations as well as on standard faunal works or other literature. For ease of use, the list should be organized by systematic groups (e.g.. Greenwood et al., 1966; Nelson, 1976). In addition to knowledge of the adult ichthyofauna, knowl- edge of spawning seasons is central to prediction of the larval fish composition. As with meristic or anatomical information, published information may be incomplete so that personal col- lections and unpublished information may be important. Al- though capture location and season can be important in elim- inating some species from consideration, caution is essential here as with other "elimination" methods. Since most marine fishes have planktonic eggs and/or larvae and many have a prolonged planktonic life the basic hydrog- raphy of a study area must be understood. A "downstream" study area is potentially vulnerable to an influx of larvae from "upstream" spawning. In addition, the direction of "streams" can differ at different depths of the water column so the influx may come from more than one direction. On the shelf oR"Nova Scotia the general circulation is from the northeast but there is a strong influence from the Gulf Stream, both from eddies and mixing which produces Slope Water. Thus, for some species, the "downstream" effect comes from the northeast while for tropical and oceanic species it comes from the southeast. Knowledge of an area's fish communities may help in inferrmg which larvae may occur together— for example, an unknown specimen taken together with larvae from a coastal community is probably not a mesopelagic species. Again, however, such inferences should be considered critically. One sort of ecological observation may be misleading— al- though spawnmg biomass may be calculated from egg and larval abundance for some species, the relative apparent abundance of adults is not always in proportion to the relative abundance of planktonic larvae. Cryptic species may appear rare in collec- tions of adults but larvae may be extremely abundant (e.g., Gobiidae in tropical and subtropical waters) while species which appear extremely abundant as adults may be rare as planktonic larvae (e.g., the clupeid Jenkmsia lamprotaenia in the Carib- bean, Powles, 1977). Some General Considerations Like larval development, identification of larvae is a dynamic process— the cumulative knowledge of the student is the key to accurate identification. The complexity of larval identification requires that a wealth of information be applied to the task, and for this reason some degree of specialization in identification of larvae is required for all but the simplest identification prob- lems. There are many examples of superficially similar but sys- tematically very different larvae, and most students, including the authors, have experienced embarrassment at an uncritical identification. Identification of larvae is frequently comparative, by elimination, so that wide knowledge of larval fishes as well as caution are necessary. The student must have information of the kinds identified above. Organization and ingenuity are required in order to keep this information usable — card files, looseleaf binders, drawings and sketches, and well-curated reference series should be de- veloped or readily available. Finally, although many beginning students are hesitant to draw, sketching and drawing (freehand, on squared paper, or with camera lucida) is one of the best ways to "see" and un- derstand larval anatomy. The process is painstaking and often frustrating in the early stages, but will pay off in the long term with increased understanding. (H.P.) Fisheries and Oceans, P.O. Box 15500, Quebec GIK 7Y7, Canada; (D.F.M.) Huntsman Marine Laboratory, Brandy Cove, St. Andrews, New Brunswick, EGG 2X0 Canada. Illustrating Fish Eggs and Larvae B. Y. SuMiDA, B. B. Washington and W. A. Laroche SCIENTIFIC illustrations of fish eggs and larvae are an in- dispensible component of any descriptive work, providing a visual reference of form and structure which is not possible to express by written descnptions and measurements alone. Illustrations facilitate identification by emphasizing distinctive but often subtle morphological characters and allow for com- panson of features at difl^erent developmental stages and with morphologically similar taxa. These qualities make illustrations the preferred and most frequently used aid for taxonomic iden- tification of fish eggs and larvae. The broad range of morphological diversity found among larval fishes requires flexibility in technique and style to produce eflTective illustrations, but the criteria of accuracy, clarity, and consistency of style should be met. The basic concept behind illustrating a fish larva involves accurately representing a three- dimensional, somewhat transparent organism on a two-dimen- 34 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM sional sheet while emphasizing characters which are most useful in identifying the actual larva from the drawing. Such characters include the fins, pigmentation patterns, and details of the head such as the jaws, spines and eyes. Internal structures such as myomeres, the gut, cleithrum, and posterior end of the noto- chord may also be emphasized but without masking important external characters. Details of other internal structures as well as shading or stippling for contrast are best excluded or de- emphasized to maintain clarity. Pigmentation is important in identification of most larvae and should be depicted clearly. External melanophores can be drawn with a fine-tipped pen as realistically as possible. Internal pigmentation can be effectively represented by using light stippling with a smaller sized pen- point. Care must be taken to avoid confusion of internal struc- tures with pigmentation. Specimens selected for illustration should ideally be those of the best condition available and representative of the particular developmental stage in both pigmentation pattern and mor- phology. The number of specimens to be illustrated is deter- mined by the nature and objective of the publication, the amount of material available in various size groups, and the degree of morphological and pigmentation change undergone by the par- ticular species during ontogeny. Specimens from described series should be archived in a museum collection for proper care and future reference after completion of the illustrations, and catalog numbers should be published. The detailed drawing begins with an accurate body outline showing the proper body proportions and position of fins and critical pigment spots. This is most easily achieved by drawing in light or blue pencil from a camera lucida-equipped micro- scope. Other methods include drawing from a projection of a slide transparency of the specimen or tracing a photograph. By convention the lateral view of the larva is drawn, with the head to the left. The exception to this is made with right-eyed pleu- ronectiforms. In some instances a dorsal or ventral view is also necessary to clarify a pigment pattern or laterally projecting morphological structures. If sketching through a camera lucida, it is helpful to use a magnification which allows the entire spec- imen to be in the field of vision as long as important details remain visible. Any resulting distortions at the periphery of the field can be compensated for by differentially focusing the mi- croscope on the particular region involved while carefully pen- cilling along the image, then reconstructing a smooth line where disjointed lines meet. Problems involving specimens that are too large or too small can often be overcome by using lens adapters or eyepieces of lower or higher magnification. Large specimens may require being drawn in sections which are later pieced together. This original sketch should be made large enough to clearly indicate fine details such as the full complement of fin rays, but not excessively so with the result of producing lines which bleed in the final reduction for publication. Related to this is the use of appropriate sizes of pen points which produce lines fine enough to draw minute details yet not be lost in re- production. Therefore, in determining the original size of each drawing, thought should be given to the desired reduction ratio as well as the number of illustrations comprising each plate. An opaque projector is most useful for obtaining a specific size for the final drawing from the initial sketch, but photocopy reduc- tions also work well. With this final pencilled sketch, the illus- trator can work with the larva under a microscope as a reference to complete details of the drawing before attempting to ink it. A light table can be helpful when tracing or inking over a rough pencilled sketch. The illustrator should always have a set of meristics of the specimen being drawn and an understanding of the important characters to be emphasized. A thorough inspec- tion for accuracy is essential to insure agreement between il- lustrations and descriptive text, especially concerning pigmen- tation and meristic elements with size and stage of development. Ideally exact counts and measurements can be obtained directly from the illustration, allowing easy identification of the larva. Illustrations are often designed for comparison of features at different stages of development or for comparison of similar features which occur among different taxa. Special care should be taken to represent similar features in a consistent style from illustration to illustration. For example, a partially ossified fin ray element, an ossified fin ray, and a fin spine may each be depicted in a consistent but slightly different manner so that the illustration not only shows the number and position of fin ele- ments but also the type of element and its relative stage of development. Literature dealing with larval fishes contains a broad array of illustrative styles, techniques, and quality. Many of these are of limited use since they fail to meet the criteria discussed above. Photographs frequently yield unsatisfactory results due to dif- ficulties in focusing on small, transparent organisms so that all body parts appear equally sharp, and they preclude emphasizing inconspicuous but important features for identification. Color illustrations in a variety of media, although potentially valuable, particularly for xanthophores, are limited due to prohibitive publication costs, poor reproducibility, and the absence of a long-lasting color preservative. Half-tone illustrations (see Ahl- strom, 1965) are effective but difficult to reproduce. These latter two techniques may become more practical with advances in photocopy technology. The preferred technique in widespread use consists of pen and ink drawings done in black India ink. Various styles of illustrations of diverse groups of larvae are represented in Moser (1981) and in this volume which serves as a useful overview. Poul Winther, George Mattson, and other artists (Ahlstrom and Ball, 1954; Ahlstrom and Counts, 1955; Bertelsen and Marshall, 1956; Ege, 1953, 1957, and 1958; Grey, 1955b; Moser, Ahlstrom and Sandknop, 1977; Moser and Ahl- strom, 1970; Tuning, 1961; Richardson and Washington, 1980) have been instrumental in establishing a fine style of pen and ink drawings which we emulate and have found most effective in its applicability to larval fish identification. We maintain a degree of flexibility in technique and style which varies with the taxonomic group under consideration but falls within the gen- eral framework discussed above. Illustrating a fish egg poses a more difficult problem than illustrating a fish larva and will be limited to a brief discussion. Encapsulation by the chorion necessitates representing the three- dimensional quality of the egg in the drawing while showing important morphological and pigmentation characters of inter- nal structures (Ahlstrom and Moser, 1980; Matarese and Sand- knop, this volume) with as much clarity as possible. Difficulties arise due to the superimposing of these characters from a two- dimensional perspective, particularly when the chorion is or- namented, when an oil globule(s) is present, and when the de- veloping embryo is fully coiled. In spite of the more complex structural representation re- quired, the same criteria of accuracy, clarity and consistency of style apply to egg illustrations. The relative proportions of the egg size to the size of the embryo, oil globule(s), and width of perivitelline space, the number of myomeres, and length of gut SUMIDA ET AL.: ILLUSTRATING 35 need to be accurately drawn. An effective balance between show- ing important characters for identification and three-dimen- sional reahsm of the egg is required to maintain clarity. Several illustrations of the egg at different stages of development and from different perspectives are helpful in demonstrating key characters such as embryonic pigmentation, myomeres, and po- sition of the oil globule(s) in the yolksac. Adherence to a con- sistent illustrative style is primarily critical for a developmental series of eggs. As with fish larvae, pen and ink drawings provide the most practical technique for illustrating fish eggs, but the specific style of illustrating and details shown depend upon the character of the egg and its stage of development. Many kinds of illustrative styles and techniques are found in the literature (see Ahlstrom and Moser, 1980 and references cited therein) and examination of these is most helpful in effectively illus- trating a particular type of fish egg. (B.Y.S.) National Marine Fisheries Service, 8604 La Jolla Shores Drive, La Jolla, California 92038; (B.W.) Gulf Coast Research Laboratory, East Beach Drive, Ocean Springs, Mississippi 39564; (W.L.) Department of Fisheries, Humboldt State University, Arcata, Cal- ifornia 95521. Clearing and Staining Techniques T. POTTHOFF THE clearing of tissues and the staining of cartilage and bone are indispensable in the study of larval and juvenile fishes. At the National Marine Fisheries Service Miami Laboratory modifications of the clearing and differential cartilage-bone staining technique proposed by Simons and Van Horn (1971) and Dingerkus and Uhler (1977) are used. The modifications are in part based upon an unpublished manuscript by W. R. Taylor and G. C. Van Dyke from the National Museum of Natural History, Washington, D.C. A wide size range of fish from 3 mm NL to larger than 500 mm SL can be cleared and stained. The technique works well for all sizes, but adjustments in the various solution soaking times are made dependent on fish size (Table 5). Method F/.Ya/ZoA!. —Specimens are fixed in 1 0-15% marble chip buffered formalin. Samples previously fixed in formalin of lower than 10-15% concentration and specimens presently in alcohol or fixed in alcohol should be refixed in 10-15% formalin for best results. Eighty to 90% of all larvae of different perciform families fixed in alcohol totally disarticulated during clearing and staining. In juvenile and adult fish > 100 mm SL the flesh is routinely removed from the left side before or after fixation. Dehydration— This is an important step, because even small amounts of water interfere with the staining of cartilage. Place specimen from the formalin into solution of 50 parts of 95% cthanol and 50 parts distilled water. Do not wash or soak spec- imens with water during transfer from formalin to alcohol. After one day for larvae < 20 mm SL and two days for specimens 20-80 mm SL and three to five days for specimens >80 mm SL transfer from 50% ethanol into absolute ( 100% or 200 prooO ethyl alcohol. If absolute ethanol is not available, 190 proof or 95% ethanol can be substituted for the absolute, although stain- ing of cartilage will not be as intense. A second change of ab- solute alcohol is desirable in larger than 20 mm SL specimens. Leave larvae <20 mm SL for one day in the absolute alcohol and juveniles 20-80 mm SL for 2 days. Adult and juvenile fish 80-200 mm SL should be kept in absolute ethanol for 3 days and fish >200 mm SL should be soaked for one week. An intermediate absolute alcohol change should be given to all specimens with longer than one day soaking time. Cartilage staining. — This is accomplished by placing specimens in an acidified alcohol solution of the alcian blue stain. For best results 70 parts of absolute alcohol should be mixed with 30 parts of acetic acid 99% glacial. To every 100 ml of acidified alcohol 20 mg of alcian blue powder should be added. The above solution should be used on larvae and juveniles from 3 mm NL to 80 mm SL. For larger fish, a staining solution of 60 parts absolute alcohol and 40 parts of acid with 30 mg of alcian blue for every 100 ml of acidified alcohol should be used. Fish larvae and juveniles <80 mm SL should be left in the alcian staining solution no longer than 24 hours. Larger juveniles and adults should be stained no longer than 36 hours. Specimens >500 mm SL can remain 48 hours in the alcian staining solution. After the specified time in the alcian solution the stain is per- manently fixed in the cartilage and cannot be removed with any chemicals used in the clearing and staining process. Staining solution can be used twice for staining larvae but should be discarded after staining a juvenile or adult fish. Neutralization. — This process raises the pH within the specimen thus allowing proper subsequent bleaching. The higher pH pre- vents further calcium loss from the bones for better alizarin red stain. To neutralize the specimen remove it directly from the alcian staining solution and place it in a saturated sodium borate solution for 12 hours for specimens <80 mm SL and for 48 hours for larger specimens. For the juveniles and adults that soak for 48 hours, change the sodium borate solution once. Bleaching (an optional .s/cpA — Larvae with little pigment on their body (e.g., Scombridae) should not be bleached. Larvae covered with pigment (e.g., Istiophoridae) and all juveniles and adults must be bleached. Prepare bleaching solution by mixing 36 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Table 5. Method of Clearing and Staining Cartilage and Bone in Larvae, Juvenile and Adult Fish. Length in mm, NL or SL Steps 10 20 .10 40 50 60 70 80 90 100 200 .100 400 500 >500 Fixation: 10-15% formalin marble chip buffered. --►h - — 5 days, flesh removed— on left side ...-► Dehydration: 1. 50% distilled H,0, 50%of95%ethanol. 2. Absolute ethanol (95% ethanol may be substituted). -1 day ►[- -1 day ►(- h 2 days • 2 days >h- 3 days ►h 3 days -one intermediate change -►h • -5 days- -7 days- -► ■-► Staining cartilage: 100 ml solution: A. 70 ml absolute ethanol, 30 ml acetic acid, 20 mg alcian blue. 100 ml solution: B. 60 ml absolute ethanol, 40 ml acetic acid, 30 mg alcian blue. — I day -Solution A- -►h I'/idays ►h"2 days->- -►H Solution B ► Neutralization: saturated sodium borate solution. '/2 day - -►I-- I- 2 days -one intermediate change - -► -► Bleaching: pigmented specimens only. 100 ml solution: 15 ml 3% H,0„ 85 ml 1% KOH. -20 min. -►!-■ -40 min. -► h - 1 hour ► I 1 1/2 hours - Trypsin digestion: 100 ml solution: 35 ml saturated sodium borate. 65 ml distilled H,0. trypsin powder. -Keep in solution until 60% clear, change to fresh solution every 10 days- Staining bone: 1% KOH solution with alizarin red stain. I day - -►[-■ 2 days • -►h -4 days Destaining: 100 ml solution: 35 ml saturated sodium borate, 65 ml distilled HjO, trypsin powder. -2 days — ►!- Change to fresh solution every 10 days until solution remains - unstained and specimen is clear Preservation: 30% glycerin and 70% of 1% KOH. 60% of glycerin and 40% of I % KOH. 1 00% glycerin with thymol as final preservative*. 1 week — ►!- - -2 weeks- — ►!- 4 weeks- * Direct sunlight and 100% glyceiine help to clear and destain difficult specimens. 15 parts of 3% hydrogen peroxide solution with 85 parts of 1% potassium hydroxide solution. Bleach larvae and small ju- veniles up to 80 mm SL for 20 to 40 minutes depending on size. Larger juvenile fish and adults may be bleached 1 to 1 Vi hours. Trypsin digestion and alizarin red staining. — The clearing and alizarin staining process has been well described by Taylor ( 1 967) and need not be repeated here. Simply continue after bleaching with the Trypsin digestion, which are Taylor's steps 4 and 5. We saw no need in modifying Taylor's method. Removal of semitransparent tissue. ~^\\ex\ studying cleared and stained material of large fish, the structures studied (caudal com- plex, pectoral fin supports, pterygiophores, vertebral column, etc.) may have to be dissected out and adhering tissue removed. This can be accomplished by time consuming picking with tweezers or by placing the material in a two-phase phenol so- POTTHOFF: CLEARING AND STAINING 37 lution with the addition of heat (Miller and Van Landingham, 1969). With this method the bones are not disarticulated, but some bone distortion was experienced. Variables affecting results.— The results of the clearing and staining procedure are not always satisfactory because of known and unknown variables. Results can never be predicted with certainty. The known variables are: ( 1 ) Time and ambient tem- perature the organism is subjected to between death and fixation. The longer an organism remains unpreserved after death and the higher the temperature, the less the tissues will clear. For best results, specimens should be killed in the fixative, or if that is not possible, they should be kept cool or frozen before fixation. (2) Effect of fixative and preservative. Marble chip buffered formalin is a good fixative for larval fish if specimens are re- moved from it after 24 hours. Buffered formalin as a preser- vative destroys first the stain uptake in cartilage. Bone decalcifies as buffered formalin becomes acid over a longer time period and decalcified bone will not stain. Therefore, it is best to fix specimens in 10% formalin and then to preserve them in 70- 95% ethanol. Specimens fixed and preserved in ethanol should be re-fixed in formalin before clearing and staining. (3) Time in a preservative. The longer a specimen has been preserved, the less predictable the clearing and staining outcome will be. Some fish larvae from the Dana collection in the 1920's were cleared and stained. The results were startling for both Formalin and alcohol preserved material because some specimens cleared and stained well, but most were unfit for study. Other vanables which affect the results of clearing and staining exist, but are not understood. No matter how carefully one adheres to the procedures, the clearing and staining results are not predictable. Interpretation of results. — Frequently specimens will remain opaque and overstain with alcian or alizarin for unknown rea- sons. This makes viewing of cartilage and bone structure diflicult or impossible. Such specimens can be used for study of fin ray development and for fin ray counts. Cartilage or bone does not always stain but can be made visible in cleared preparations by changing light conditions at the microscope and manipulating the substage mirror. Cartilage appears reticulated in structure whereas bone is structurally clear and hyaline. Erroneous conclusions can be made if one solely relies on color to determine cartilage and bone. In general, cartilage will appear blue and bone red, but often alcian blue is taken up by bones and rarely alizarin red by cartilage. For instance, devel- oping fin rays often appear blue. Generally larger developed cartilage structures will stain bet- ter than small developing ones. Thus, in the same specimen one may find brightly blue stained cartilage, pale blue cartilage, and cartilage with no stain at all. Therefore, special care is indicated when viewing newly developed cartilage. The ossification onset in cartilage is difficult to determine. A thin layer of bone forming all around the cartilage can be de- tected by examining the outer edges of the cartilage structure: a shiny hyaline line forms there, probably only a cell layer thick. Investigators are often discouraged by clearing and staining results, particularly when their sample is small. In a larval de- velopmental series I usually clear and stain 200 to 400 speci- mens, and I am able to study each aspect and area of devel- opment that I wish to examine because of the large sample size at hand. For example, in a specimen in which the pectoral fin support area is unclear and stained poorly the caudal area may be clear and stained well. Thus, this specimen is utilized only for caudal development, whereas in another specimen the pec- toral area may be clearer and better stained. Thus, with a large sample size, the uncertainties and vagaries of the clearing and staining procedure are overcome. Application of clearing and staining.— Cleanng and staining is helpful in identification offish larvae when external characters are inadequate. It also aids systematic and phylogenetic studies of larvae to adult fishes. This subject has been discussed in detail by Dunn (1983b). National Marine Fisheries Service, Southeast Fisheries Center, Miami Laboratory, 75 Virginia Beach Drive, Miami, Florida 33149. Radiographic Techniques in Studies of Young Fishes J. W. Tucker, Jr. and J. L. Laroche RADIOGRAPHY is useful for obtaining skeletal informa- tion in studies of fish taxonomy and morphology. Al- though clearing and staining provides more detail, radiography has other advantages. It produces an easily stored, long-term record of the skeleton and does not permanently alter the con- dition of the specimen. In many cases, counts can be obtained more accurately from radiographs than from the specimens themselves. If an x-ray unit and darkroom are available, ra- diography is usually faster and easier than clearing and staining. The time saved may be of value in studies of population vari- ation, in which many specimens must be examined. Radiog- raphy has also been used to monitor decalcification of larvae stored in formalin (Tucker and Chester, in press), and has been suggested for use in toxicological studies to check large numbers 38 ONTOGENY AND SYSTEM ATICS OF FISHES -AHLSTROM SYMPOSIUM of larvae for skeletal deformities. The consensus among ichthy- ologists who have used both techniques is that, although clearing and staining methods provide the detail necessary for describing developmental osteology, radiography is a simple and quick way of obtaining counts from large numbers of specimens. Hard (shortwave) x-rays have been used to form shadow pic- tures, or radiographs, of large, well-ossified fish for almost four decades (Goshne, 1948; Bartlett and Haedrich, 1966), but the use of soft (longwave) x-rays for small specimens is relatively new. Although first suggested by Bonham and Baylifr( 1953) and used by Watson and Mather (1961 unpubl. manusc), useful techniques for larval radiography have only recently been de- scribed (Miller and Tucker, 1979). Potential larval fish radiog- raphers should consult Miller and Tucker's paper for method- ological details and Quinn and Sigl ( 1 980) for basic radiographic principles. Although specimen fragility determines the mini- mum size of larvae that can be x-rayed, sensitivity of the tech- nique, which depends to a large degree on spectral characteristics of the radiation, determines the amount of detail present in the finished radiograph. This section, therefore, reviews the prin- ciples and current methods useful for maximizing detail in ra- diographs of fish larvae. Radiographic sensitivity refers to the clarity of details in the radiographic image and depends on a combination of two fac- tors, definition and radiographic contrast. Definition is sharp- ness of the image. Radiographic contrast refers to the density (darkness) range of the image and depends on two factors, sub- ject contrast and film contrast. Subject contrast refers to the ratio of radiation intensities that pass through different parts of the specimen. Film contrast refers to the ratio of densities in parts of the film that have received different degrees of exposure. In larval fish work, radiographic sensitivity can be improved by several means. Definition can be improved by using the longest possible radiation wavelengths, by using the finest grained film available, and by minimizing geometric production of over- lapping shadows at tissue discontinuities in the specimen. Ab- sorption by x-rays of a given wavelength depends mostly on the atomic numbers of components in the x-rayed material, and to a lesser degree on thickness and density of the material. Larval skeletons, which are thin, poorly calcified, and of relatively uni- form composition and thickness, do not contrast radiographi- cally with the rest of the body as much as in older fish. High contrast techniques should, therefore, be employed. Subject con- trast can be increased by increasing wavelengths and by de- creasing the thickness of non-skeletal tissue by dehydrating the specimen. Film contrast can be increased by using a high con- trast film and by increasing development time; however, over- development will also increase graininess and reduce definition, and probably should be avoided. The longwave (soft) end of the x-ray spectrum is the portion most useful for x-raying small fish, because this low energy radiation does not pass through materials as easily as that at the shortwave (hard) end. Decreasing the tube voltage (kv) caus- es a shift of the emitted spectrum toward longer wavelengths. Resultant elimination of some of the hard radiation contributes to better subject contrast and improves definition by reducing clumping of silver grains in the film emulsion (graininess). The x-ray unit should be equipped with a thin beryllium window, which allows passage of soft rays. A 25 mil (0.63 mm) window allows work at a kv of 20; a 10 mil (0.25 mm) window extends capabilities to about 8 kv (Joseph Fowler, Hewlett Packard, pers. comm.). However, the lower practical limit for fish larvae may be governed by restrictions on exposure time, rather than kv limitations. Another relevant factor is the source-to-specimen distance, to which image definition is directly related. Increasing the source- to-specimen distance improves definition by minimizing en- largement and distortion. Practical limits are set by air atten- uation, loss of radiation intensity (roughly as the square of the ratio of the distances), and dimensions of the x-ray unit. Geo- metric unsharpness is the maximum width of the zone of over- lapping shadows that are caused by a non-point source. This factor can be calculated to determine the minimum source to specimen distance that can be tolerated. Use of the minimum distance will permit the shortest possible exposure time and reduce relative attenuation of soft rays, thus contributing to subject contrast. The formula for geometric unsharpness, Ug (Quinn and Sigl, 1980) is: U„ D, in which F is the radiation source size. Do is the source-to- specimen distance, and t is the specimen to film distance (max- imum specimen thickness). For F = 0.5 mm, D,, = 460 mm, and t = 1 mm, U^ is 0.00 1 mm. This level of unsharpness would not be visible without magnification and could be tolerated at moderate magnification depending on the requirements of the investigator. To ensure that geometric unsharpness is not large enough to affect quality of radiographs, it should be calculated for the set of factors relevant to each operation, keeping in mind the level of magnification to be used. With most modem x-ray units, a distance of 46 cm or less can be used. Because air attenuates soft rays more than hard, elimination of air between the x-ray source and specimen allows a greater proportion of soft radiation to reach the specimen. Decreasing the source to specimen distance helps some, but also increases geometric unsharpness, unless the source is very small. A vac- uum would be ideal but is impractical. Replacement of the air in a cabinet unit with helium allows the use of lower kv with reasonably short exposure times and provides an increase in subject contrast. Helium can be conserved and reused if it is placed in a small volume plastic cylinder that has its ends sealed with dry-cleaning plastic. Before a specimen is x-rayed it should be dehydrated as much as can be tolerated to increase the signal (skeleton) to noise (non-skeleton) ratio. For best results, the specimen should be placed in 50-75% ethyl alcohol for a short period, maybe 30- 60 min, depending on size. Then the specimen should be placed on the film holder, blotted to remove surface liquid and bubbles, and quickly x-rayed and returned to a container of liquid before desiccation damage occurs. The specimen should be placed as close as possible to the film emulsion. This can be accomplished without wetting the film by sandwiching it between two thin sheets of black polyethylene. Details for construction of a convenient film holder (cassette) are presented in Miller and Tucker (1 979). Polyethylene is trans- parent to soft x-rays and is good cassette material. Vinyl, as well as wood, paper, and any metal are relatively opaque to soft x-rays, and vinyl or metal make good labels. Single coated Type R (now Type XAR) film has provided the best quality radiographs of larvae. High resolution plates give better resolution but are too slow. Type R film is slow relative to other films but within practical limits. It has ultra-fine grain TUCKER AND LAROCHE: RADIOGRAPHY 39 Fig. 15. Positive image of radiograph of a southern flounder (Paralichthys tethosligma) larva, 9.7 mm SL, stored m 7% borax buffered seawater formalin for seven years. Radiographic exposure data: Faxitron Model 43805N; Kodak Type R film; source to film distance. 46 cm; 9 kv; 600 mAs; under helium. Intemegative processing data: radiograph was projected onto 4 in x 5 in professional copy film (Kodak 4125) with an Omega (4 in X 5 in) Pro Lab Enlarger; exposure was 1 s at f S'/j; film was developed in Kodak HCl 10 (dilution E) for 5 min at 23 C. Print processing data: a positive pnnt was made on Kodak Polycontrast Rapid 11 RCF paper using a polycontrast no. 3 filter in the Omega enlarger; exposure was 5 s at f 5.6; print was developed in Kodak Ektaflo diluted to simulate Dektol 1:1, at 23 C. (The intemegative and printing procedure was devised and performed by Tom Smoyer of Harbor Branch Foundation.) and high contrast. The single emulsion is necessary for avoiding two images (on both sides of the film). Coarser grained and lower contrast films will produce inferior radiographs. Exposures should not be longer than about 5 min, and for many specimens 5 min is too long. Larvae will quickly desiccate, and even if not damaged, may shrink and cause blurred images. Specimen damage or image blurring will determine the mini- mum size of larvae that can be x-rayed. Specimens can be pro- tected by an overlying sheet of dry-cleaning plastic if care is taken to remove bubbles. During exposure, unneeded portions of the film can be protected for later use with lead vinyl masks. The manufacturers' instructions for mixing chemicals and processing films should be followed as closely as possible. Fre- quent agitation of the film while it is developing, rinsing, and fixing is important to ensure uniformity of chemical reactions. Both undeveloped and developed films should be stored away from light, heat, humidity, and chemical fumes (particularly formalin, alcohol, and hydrogen peroxide). Radiographs are best observed directly, emulsion side up, with a dissecting or phase contrast microscope. Printing of radiographs is best done via an intemegative (Fig. 15). This compresses the tonal range so that finer detail can be preserved in the print. The major limitation of the technique is probably inadequate radiation intensity at low kv. This limit may have been reached with x-ray units equipped with 10 mil beryllium windows. Sat- isfactory radiographs of 4-1 5 mm larvae have been made at 8- 10 kv and 300-800 mAs (milliamperes x seconds). Some im- provement can be expected if the air is replaced with helium; however, exposure time will eventually become prohibitively long. Because machine and specimen characteristics vary, a stan- dard formula for producing high-quality radiographs cannot be provided. At least initially, the larval fish radiographer must proceed by trial and error with the machine and specimens at hand. As familiarity develops, the results will improve signifi- cantly. We stress that an accurate and detailed logbook con- taining specimen and exposure data should be kept, and that procedures should be standardized. (J.W.T.) Harbor Branch In.stitiition, Inc., RR l,Box 196-A, Fort Pierce, Florida 33450; (J.L.L.) Gulf Coast Re- search Laboratory, East Beach Drive, Ocean Springs, Mississippi 39564. Histology J. J. GOVONI WHILE contemporary systematists rely upon a broad scope of biological features to infer relationships among taxa, the definition and comparison of morphological characters re- mains one of their most useful tools. The small size and often altricial development of fish larvae, however, make it difficult to resolve the morphology of structures other than skeletal ele- ments. By clarifying tissue composition and by enhancing mor- phological resolution, histological techniques may aid the sys- tematist in defining characters at the tissue as well as at the microanatomical level, thereby providing additional character states to be examined for synapomorphies and perhaps onto- genetic precedence. Because of their small size, sections of whole larvae can be prepared (Fig. 16) and structural relationships of organ systems examined. Insofar as there is no clear separation between gross and micro-anatomy beyond the limits of human visual resolution, histological techniques may otfer yet another tool useful in phylogenetic analysis. Techniques Flvi2;/o«. — Inasmuch as autolysis is rapid in larval tissue (Thei- lacker, 1978), fixation is difficult (Richards and Dove, 1971). Specimens reared in the laboratory or specimens taken from brief plankton tows (O'Connell. 1980) are the most suitable for histological preparation and study; specimens sorted from field collections fixed in formalin and seawater will usually yield poor quality preparations. Neutral buffered (phosphate buffi;rs) for- malin (see Humason, 1979) enhanced with <4% acrolein (van der Veer, 1 982) is recommended for rapid and thorough fixation. Glutaraldehyde (2.5%) is also a useful fixative (Hulet, 1978). Difference in the osmolality of tissues and ambient water may distort cells and tissues, especially of marine larvae. Such arti- facts have not been observed in preparations of clupeiform and perciform larvae, but may be of concern in the preparation of anguilliform leptocephali (Hulet, 1978). Forsterand Hong (1958) and Hulet (1978) provided applicable saline solutions that may eliminate distortion and enhance staining. Sectioning and staining. — Sxandsivd animal tissue techniques (e.g., Humason, 1979)— dehydration, paraffin embedding, and sectioning— have been used to trace the development of organ systems (O'Connell, 1981a), as well as to assess the pathology of starvation in fish larvae (Umeda and Ochiai, 1975; O'Con- Fig. 16. Sagiual section ot a Leiostomus xanlhurus larva, 4.4 mm notochord length (glycol methacrylate section stained with alkali blue 6B- neutral red). Fig. 17. Example comparisons of larval fish tissue and microanatomy. Abbreviations: AM, axial musculature; CS, collagenous supporting shafts; EP, epidermal cells; M, midgut; MC, mucous cell; NF, nerve fiber. (A) The integumentary epithelium of a Brevoortia patronus larva showing hyaline plates (arrow), a tissue characteristic of some clupeiform larvae. Note that erosion of the outer layer of epithelium is evident. (Scale bar = 20 /jm; glycol methacrylate section stained with acid fuchsin — toluidine blue.) (B) The integumentary epithelium of a Leiostomus xanthurus larva showing lack of hyaline plates in epithelial cells. (Scale bar = 10 iim; glycol methacrylate section stained with alkali blue 6B — neutral red.) (C) Axial musculature of a Brevoortia patronus larva showing two opposing layers of muscle fibers, a tissue characteristic of clupeiform larvae. (Scale bar = 50 livn, glycol methacrylate section stained with acid fuchsin — loluidine blue.) (D) Axial musculature of a Leiostomus xanthurus larva showing muscle fiber layers in parallel alignment, a tissue characteristic of perciform larvae. (Scale bar = 50 iim\ glycol methacrylate section stained with alkali blue 6B — neutral red.) (E) Cross section of the elongate dorsal ray of an Echiodon dawsoni larva. (Scale bar = 20 ixm: glycol methacrylate section from Govoni et al., 1984.) (F) Cross section of the elongate dorsal ray of a Bregmaceros atianticus larva. (Scale bar = 15 Min; glycol methacrylate section stained with acid fuchsin — toluidine blue.) 40 GOVONI: HISTOLOGY 41 .:•»' /■ B .f \ MC tiSNHBC^- EP AM '^ .- i^. t w •tr\ ►'i'^ . ,^ " f >i ••' D V- *,'" AM ^ "« '-« . \ ~<^ .» ■^, •- *k- \. cs 42 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM nell, 1976; Theilacker, 1978). These techniques will suffice for the examination of soft tissue morphology given adequately fixed specimens. To avoid their loss, small specimens may be prestained with borax-carmine before embedding and section- ing; this stain can be washed out before subsequent histological staining (Engen, 1968). Plastic embedding (Bennett et al., 1976) is advantageous for examination of small delicate structures, for precise records of specimen orientation and section plane, and for the resolution of fine cellular detail. Glycol methacrylate (Bennett etal., 1976), epoxy resins (Humason, 1979), and other low viscosity plastics (Hulet, 1978; L. R. White resin, London Resin Company Lim- ited) are useful embedding media. Small specimens that can become indistinguishable or even lost in paraffin blocks can be easily observed in the plastic block during sectioning. As whole mounts, specimens can be examined, measured, and meristic characters counted before sectioning (Hulet, 1978). Techniques developed by Ruddell (in press) reduce swelling of tissues, an artifact sometimes encountered with glycol methacrylate embedding. While the spectrum of histological and histochem- ical stains applicable to plastic sections is somewhat limited, toluidine blue counter stained with acid fuchsin has staining reactions analogous to the more commonly used hematoxylin and eosin. Other stain combinations also are applicable to larval tissue embedded in glycol methacrylate (for examples see Go- voni, 1980; Govoni et al., 1984): alkali blue 68 counter stained with neutral red reveals fine cellular structure; VanGiesen's picric acid counter stained with acid fuchsin reveals collagenous fibers, the anlagen of actinotrichia; periodic acid-Schiff reagent reacts strongly with acid mucopolysaccharides, including chon- dromucin, and can be used to reveal cartilaginous precursors of cartilage (endochondral) bone; alizarin red S reacts with Ca + + ions and can reveal both calcified cartilage and bone. Examples of Application Histological preparations may serve the systematist in two ways: by clarifying tissue composition and by resolving struc- ture, thereby allowing for the determination of ontogenetic pres- ence or absence of tissues and by offering comparisons of tissue organization among taxa. An example of the first use is in the identification of cartilage and bone. The literature is replete with errors that result from the naive interpretation of alcian blue and alizarin red S reac- tions with cartilage and bone tissue in whole mounts. Alcian blue reacts histochemically with the sulfate and carboxyl groups of mucopolysaccharides (Pearse, 1968) including chondromu- cin, the ground substance of cartilage, but it may also react with developing bone matrices, which are rich in mucopolysaccha- rides as well (Belanger, 1973). An alcian blue reaction, therefore, may indicate cartilage when developing membrane (dermal) bone is present. The reaction of alizarin red S with calcium ions (Pearse, 1968) may indicate calcified cartilage as well as true bone. While the clearing and staining of skeletal elements re- mains a powerful tool (Potthoff, this volume), histological prep- arations can clarify the identity of cartilage and bone tissue in extremely small specimens wherein their identity may not be clear in whole mounts. To date, comparisons of larval fish characters revealed by histological techniques have not been extensive and examples of application are few. Comparative histological sections of elo- pomorph and clupeomorph larvae illustrate the unique char- acter of the elopomorph leptocephalus (Smith, this volume). The unique configuration of organs and tissues is apparently inclusive of anguilliform, elopiform, and notocanthiform lep- tocephali. Inasmuch as Hulet (1978) also found peculiarities in the kidney structure of the eel leptocephalus that may be unique among vertebrates, the kidney structure of anguilliform lepto- cephali should be compared with that of other elopomorph leptocephali. Transient, hyaline plates occur in the basal end of the outer integumentary epithelium of some clupeiform larvae (Jones et al., 1966; Lasker and Threadgold, 1968; O'Connell, 1981a; Fig. 17 A), but this feature was not mentioned in the integumentary descriptions of anguilliforms (Hulet, 1978) and pleuronectiforms (Wellings and Brown, 1969; Roberts et al., 1973), nor is it apparent in the perciform Leiostomus xanthunis (Fig. 1 7B). These plates presumably function as osmotic barriers (O'Connell, 1981a), but their systematic presence or absence is not completely established and remains unexplained. The or- ganization of axial musculature is another histological difference among higher taxa. The two-layered musculature of clupeiform larvae is aligned in opposing directions within myotomal seg- ments (Blaxter, 1969b; O'Connell, 1981a; Fig. 17C), whereas in perciform larvae the orientation of axial muscle fibers is closely parallel (O'Connell, 1981a; Fig. 1 7D); this difference may have a functional basis related to gross body form and swimming postures (O'Connell, 1981a). An example of the use of histological preparations to compare microanatomical characters is the differences exhibited in elon- gate dorsal fin rays. Elongate dorsal fin rays are features of many unrelated taxa offish larvae (Moser, 1981), but the microana- tomical structure of these homologous derivatives differs among taxa (Govoni et al., 1984). A major difference is the bilateral, paired, collagenous supporting elements of the carapid elongate ray, as in Echiodon dawsoni (Fig. 1 7E), and the singular supports of elongate rays of the bregmacerotid Bregmaceros atlanticus (Fig. 17F) and the serranid Liopropoma (Kotthaus, 1970). Monophyly in carapids has been inferred, in part, from the distinctiveness of this synapomorphy, the elongate first dorsal ray of their highly specialized larvae (OIney and Markle, 1979; Markle and OIney, 1980; Gordon et al., this volume). The often remarkable similiarity of cells and tissues, even among phyla (Andrew, 1959), and the development of tissues from the undifferentiated to the complex, may limit the use of a histological approach to systematics. Yet, the unusual diversity that characterizes ontogenetic patterns of fishes (Wourms and Whitt, 1981), and some apparent contrasts in tissue organiza- tion and composition that correlate with current supraordinal classification, make histological comparisons tenable. The pre- ceding examples of tissue and microanatomical dissimilarities may serve to illustrate the kinds of comparisons that may prove useful in inferring relationships as more information becomes available. Histological techniques may provide a potentially useful tool to the systematist; more comparative work is clearly warranted. National Marine Fisheries Service, Southeast Fisheries Center, Beaufort Laboratory, Beaufort. North Carolina 28516. Scanning Electron Microscopy G. W. BOEHLERT SCANNING electron microscopy is an ideal tool for descrip- tion of microstructure in taxonomic studies. The scanning electron microscope (SEM) provides a surface image character- ized by high resolution and depth of field and a three-dimen- sional quality unavailable with other techniques. In many cases this allows one to objectively describe microstructure where only subjective descriptions were available in the past. It is the pur- pose of this contribution to describe the techniques and use of scanning electron microscopy and its application to systematic investigations of fish eggs and larvae. The SEM has been used in a wide variety of systematic and evolutionary investigations. With available magnifications from 10 to greater than 100,000 times, the SEM covers the range from dissecting and compound light microscopy to transmission electron microscopes. It has thus been immensely important to progress in classification in the study of micropaleontology, bot- any, insects and mites, and a wide variety of microorganisms, among other taxa (Heywood, 1971; Kormandy, 1975). Taxo- nomic applications of the SEM to fishes have been more limited. Several studies have used the SEM for studies of morphology, including epidermis, gill tissue, optic capsules, eggs, sperm, and embryosof fishes (Dobbs, 1974, 1975). Microstructural analysis of otoliths of fishes with the SEM is now common (Pannella, 1 980). For early life history stages, the most frequent use in identification and classification has been with the egg stage. The chorion, or external membrane, of many species is variously ornamented with filaments, spines, patterns of ridges, loops, blebs, and pustules ( Ahlstrom and Moser, 1 980; Robertson, 1981; Matarese and Sandknop, this volume). These ornamentations and the ultrastructure of the chorion are species- specific (I vankov and Kurdyayeva, l973;Lonning, 1972). While many of these structures may be easily visualized with light microscopy (Hubbs and Kampa, 1946; Kovalevskaya, 1982), the SEM often provides the best means of adequately describing structures which are very small or transparent under the light microscope. The egg chorion of Maurolicus muelleri, for ex- ample, was described as "drawn up into hexagonally arranged points," by Robertson (1976) based upon light microscopy but as "drawn up into hexagonal ridges . . . and slightly raised at the point of intersection" under the SEM (Robertson, 1981). Similarly, Boyd and Simmonds ( 1 974), among others, suggested that the chorion of southern populations of Fundulus fietero- clitus lacked fibrils using light microscopy, whereas the SEM showed the presence of numerous short and thin fibrils (Brum- mett and Dumont, 1981). Thus for purposes of classification, the SEM allows visualization of surface structures that are dif- ficult to describe with light microscopy. Methodology Preparation of biological material for examination under the SEM is concerned with preservation, dehydration, and coating with a conductive material. Fixation of labile biological speci- mens is necessary because removal of water during the stages of dehydration may result in collapse of cells and other artifacts. Depending upon the method of fixation and dehydration, the artifacts can range from shrinkage to collapse or fracture of the structures to be observed. It is preferable to begin with fresh, live material. For eggs this requires either laboratory spawning or abundant eggs from the field which can be reliably collected. For larvae at different stages, it is diflicult without laboratory rearing facilities. Results with formalin-fixed material from plankton collections will generally be satisfactory for lower mag- nification analysis of surface morphology, but may not reflect the quality of freshly prepared material. Fresh material should be fixed for electron microscopy. Larval stages may first be relaxed in anesthetant solution (such as MS- 222). Initial fixatives for both eggs and larvae are generally based upon glutaraldehyde, with concentrations ranging from 0.5 to 4.0%; lower concentrations are typically followed by post-fix- ation. A fixative which I have found acceptable is that from Dobbs (1974) as follows: 70% glutaraldehyde-2.0 ml, flounder saline— 34 ml, and distilled water— 34 ml. The flounder saline follows Forster and Hong (1958) and contains the following (in grams per liter): NaCl, 7.890; KCl, 0.186; CaCK, 0.167; MgCK- 6H,0, 0.203; NaH,FO,H_,0, 0.069; NaHCO,, 0.84. The fix- ative has a final osmolarity of 380 mOsm/l. Fixation should be for 24 hours. Other authors provide several other fixatives. One suggested by Stehr and Hawkes (1979), while more difficult to prepare, is also useful should transmission electron microscopy be desired for the same material. Post-fixation in osmium te- troxide is recommended by several authors as a means of hard- ening particularly soft tissues. Generally, 1-2% osmium tetrox- ide in buffered saline is used. I have found this unnecessary with fish eggs and larvae, as suggested by Dobbs (1974) and Stehr and Hawkes (1979). It may be considered, however, if collapse is a problem. Lonning and Hagstrom (1975) suggested that egg chorions not post-fixed would rupture under the electron beam; I have not noticed this. It is the process of dehydration where the greatest artifacts are likely to occur. With larvae, shrinkage of tissue may occur, while eggs may suffer complete collapse. On larger eggs, punc- turing the chorion with a sharpened dissecting needle may fa- cilitate transfer of fluids and prevent this collapse (Stehr and Hawkes, 1979). Removal of water from the tissues is prerequisite to coating and observation, which are both conducted under high vacuum. Two methods are available, freeze drying and critical point drying. For freeze drying, unfixed fresh material may be used. Fixed material should first be rinsed with distilled water to remove salts, and then plunged with little adhering water into liquid nitrogen. Damage here may result from formation of ice crystals if freezing rate is too slow, but this is typically not a problem with small eggs and larvae in liquid nitrogen. Boyde and Wood (1969) recommend using 20 ml chloroform per liter of distilled water to increase nucleation rates and decrease ice crystal formation. After freezing, the material is immediately 43 44 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM introduced into the freeze dryer, where water subUmes, leaving the specimen dry and intact. Critical point drying, on the other hand, requires dehydration through a graded series of alcohols (20% for 24 h. then 10-20 min each in 50%, 70%, 80%, 90%, 95%, and two changes of absolute ethanol). The ethanol is then replaced with either freon or acetone depending on whether freon or carbon dioxide critical point dryers are used. The steps of dehydration and transfer can be done in small specimen holders to minimize handling and possible surface damage. Af- ter dehydration, specimens must be mounted on SEM studs with any of several available adhesives and tapes. The dried specimens are particularly delicate and should be handled with a small camel-hair brush to avoid damage to the surface. They are then oriented onto the stud under a dissecting microscope. Before coating, no further preparation is necessary with larvae, but eggs have only a small area of electrical contact with the stud. It is therefore advisable to use a conductive adhesive (such as silver paint) to make a more complete electrical connection and prevent charging, which decreases image quality. This paint should be allowed to become tacky prior to positioning the eggs, or it may cover portions of the egg itself Finally, specimens are coated with a thin conductive layer, typically of gold or gold- palladium, by either vacuum evaporation or ion sputtering, prior to viewing on the SEM. At most facilities, trained SEM tech- nicians are available; their advice and assistance are invaluable and should be sought. Results and Discussion Shrinkage and other artifacts will vary depending upon the type of material, preservation, and method of dehydration. For fresh material preserved in a mixture of formalin, glutaralde- hyde, and acrolein, Stehr and Hawkes ( 1979) observed a shrink- age of approximately 10% in the eggs of Platichthys stellatus and Oncorhynchus gorbuscha; the latter had been punctured prior to dehydration. In the present study, eggs of Maurolicus muellen initially preserved in 5% buffered formalin showed varying degrees of shrinkage and collapse depending upon sub- sequent treatment. The least shrinkage (12%, Fig. 18B) was noted in material which was freeze dried, whereas post-fixation and dehydration through freon 1 1 3 associated with critical point drying resulted in shrinkage of up to 67% of the original diameter (Fig. 18D). Eggs of this species show a hexagonal sculpturing; under the light microscope the sculpturing is hyaline and difficult to interpret (Fig. 18A). Eggs prepared by freeze drying clearly show the surface sculpturing; note particularly the ridges, which are more clearly defined (Fig. 188). For comparison, an egg which had partially collapsed during dehydration is shown (Fig. 18D). The obvious differences in shrinkage point out the im- portance of specifying method, initial size, and shrinkage values, particularly for comparative or taxonomic studies. Eggs from other species are shown to give an idea of the range of chorion structures which may be observed. The hexagonal pattern on M. muellen overlies a highly porous surface structure Fig. 18. (A) Egg of Maurolicus muellen from off South Africa taken under the compound light microscope with transmitted, polarized light. Note the emphasis of the points on the hyaUne chorion, which represent the intersections of ridges. Bar = 100 ^m. (B) Egg of A/, muellen under the scanning electron microscope. Note the areas between what one would interpret as points on Figure 18A. which are now seen as polygonal facets or ridges. Bar = 500 nm. (C) Individual facet of the egg of At. muellen. Note the porous and diaphanous nature of the egg surface. Bar = 50 Mm. (D) Egg of A/, muelleri post-fixed in osmium tetroxide and critical point dried. The shrinkage of this specimen is approximately 65%. Note the differences in morphology of the ridges and surface of the egg. Bar = 100 /jm. (E) E^of Pleuronichlhys coenosus. The facets are relatively small by comparison with M. muellen and the pattern units are more regularly hexagonal. Bar = 100 Mm. (F) Detail of two hexagons from the egg of P. coenosus. Note the morphological differences between both the ridges and chorion surface as compared to M. muellen. Bar = 10 Mm. Fig. 19. (A) Egg of Alherinopsis californiensis. The filaments are single, terminate in loose ends, and are distributed over the entire egg surface. Bar = 1 ,000 Mm. (B) Egg of .-itherinops affiiUs. The egg of this species is characterized by filaments which are looped, with no free ends (Curless, 1979). This differentiates it from the egg of ,-1. californiensis, as do filament length, abundance, and basal morphology. Closed-loop filaments have also been noted in .Aniennanus caudimaculatus eggs by Pietsch and Grobecker ( 1 980). Bar = 1 ,000 Mm. (C) Chorion of Paracaltionymus costatus collected off South Africa. The surface features are irregular and cover the entire egg surface. This differs from species of Callionymus. which have hexagonal patterns. Bar = 10 Mm. (D) Chorion surface of Mugil cephalus. These structures are irregular and cover the entire egg surface. Note the superficial similarity to Paracallionymus. Bar = 10 Mm. (E) Chorion surface of an advanced ovarian egg of Coryphaenoides filifer. Note that the surface "blebs" are arranged in hexagonal patterns and may be the precursors of a hexagonal pattern typical on eggs in this family. The pelagic egg of this species has not been described. Bar = 10 Mm. (F) Chorion surface of an advanced ovarian egg of Coryphaenoides acrolepis. The hexagonal ridges are better developed than in Fig. I9E. There are holes under the ndges between the intersections, which might indicate that this species, whose egg is also undescribed, may have the hexagonal network supported on "stills" as described for eggs of Coelorhynchus spp. (Robertson. 1981; Sanzo, 1933a). Bar = 10 Mm. Fig. 20. (A) Spines on the chorion surface o( Oxyporhamphus microplerus. These are distributed over the entire surface of the egg. Bar = 100 Mm. (B) Chorion surface from Scomhereso.x saurus collected off South Africa. The tufts are characterized by a relatively complex basal morphology and depending upon method of fixation, may resemble small bundles of hairs or, as here, simply coalesced tufts. Bar = 10 Mm. (C) Micropyle and associated pores of the egg of Laclona diaphana from the Eastern Tropical Pacific. The pores shown here are restricted to this region around the micropyle and appear to penetrate the outer layer of the chorion. Bar = 50 Mm. (D) Secondary, smaller pit structures on the remainder of the egg of Laclona diaphana. I refer to these depressions as "pits" because closer examination does not reveal penetration through any layer of the chorion, as opposed to the pores surrounding the micropyle in 20C. Bar = 1 Mm. (E) Head region of a larval Sebasles melanops shortly after parturition. Polygonal epidermal cells may be noted on some parts of the body. Bar = 100 Mm. (F) Epidermis on the dorsal surface, just posterior to the head, on an embryonic S. melanops approximately 28 days post fertilization. Note the distinct microndges and cell borders characteristic of developing teleost epidermis. Bar = 10 Mm. BOEHLERT: SCANNING ELECTRON MICROSCOPY 45 •/N .\!^^V-U 46 ONTOGENY AND SYSTEMATICS OF FISHES -AHLSTROM SYMPOSIUM K BOEHLERT: SCANNING ELECTRON MICROSCOPY 47 48 ONTOGENY AND SYSTEMATICS OF FISHES -AHLSTROM SYMPOSIUM (Fig. 18C) as compared to that oi Pleuronichthys coenosus (Fig. 18E, F). Here, the hexagons are not only smaller, but the area within the facets does not appear porous. SEM was used for this species and its congeners for egg description by Sumida et al. (1979). It is interesting to note that these authors discussed the similarity in chorion structure of Plenronichthys spp. with that oi Synodus lucioceps. While there were slight differences in sizes of the polygons, the superficial similarity of chorion structure on these phylogenetically distant genera supports a functional role (Robertson, 1981) and independent derivation. In this in- stance, however, SEM was valuable for understanding and in- terpreting the differences between species and genera subse- quently observed under the light microscope (Sumida et al., 1979). Similarly, Keevin et al. (1980) used chorion ornamen- tation to distinguish among genera of killifishes. Other ornamentations include more random ridges (Para- callionymus costatus. Fig. 19C, and Mugil cephalus. Fig. 19D), filaments of varied length, diameter, and base morphology (Ath- erinopsis califormensis and Athehnops affinis. Fig. 19A, B; see also Hubbs and Kampa, 1946), tufts (Scomberesox saurus. Fig. 20B), spines (Oxyporhamphus microptents. Fig. 20A), and pits and pores (Lactoria diaphana. Fig. 20C, D). In thecallionymids, the small eggs of species of Callionynms have hexagonal sculp- turing similar to that oi Pleuronichthys (Fig. 18E). In Paracal- lionymus costatus (Fig. 19C), however, random ridges similar to those in Mugil cephalus are apparent. Since chorion microstruclure is formed by follicle cells during oogenesis (Sponaugle and Wourms, 1979; Stehr, 1979), patterns may also be discerned in ovarian eggs. The pelagic eggs of mac- rourids are poorly known but have been described for selected species by Sanzo ( 1 933a), Robertson ( 1 98 1 ), and Grigor'ev and Serebryakov (1981). For Pacific species of Coryphaenoides. pe- lagic eggs remain poorly known but apparently have hexagonal patterns as in other members of the genus; this, is clearly shown in ovarian eggs near the maximum size observed by Stein and Pearcy (1982; Fig. 19E, F). Thus SEM of developing ovarian eggs may be used to discern differences which then aid in iden- tification of eggs from plankton samples. For larval stages, SEM has been used for the description of development of several surface structures, such as the olfactory organ (Elston et al., 1981) and lateral line neuromasts (Dobbs, 1974). For taxonomic studies, differentiation of fine-scale mor- phological differences, such as dentition or fine-scale spine ser- ration, may be useful. Its most valuable use may therefore be for later larval development, since pigmentation and other char- acteristics in early larvae are better seen with conventional methods (Fig. 20E, F). To conclude, SEM may serve as an adj uct to traditional meth- ods in the description of fine structure in fish eggs and larvae. For high magnification, high resolution visualization of surface morphology, it remains the most effective tool available. Under lower magnifications, it may allow one to clearly visualize struc- tures which are difficult to interpret using standard microscop- ical methods (Fig. 1 8A, B). Oregon State University , Marine Science Center, Newport, Oregon 97365. Developmental Osteology J. R. Dunn ONE legacy left by Elbert H. Ahlstrom was an appreciation of the value of developmental osteology of teleosts as a taxonomic aid and as an indicator of phylogenetic affinities. Although numerous studies have been made on the growth of various bones in teleosts, such descriptions have not been widely used in assessing relationships of fishes. I have recently re- viewed, in some depth, the application of developmental os- teology in taxonomic and systematic studies of teleosl larvae (Dunn, 1983b). Here I present a brief overview of some skeletal structures in teleosts whose ontogeny offers potential utility in inferring phylogenetic affinities. It is hoped that this precis will encourage ichthyologists to examine the development of bones in the course of their systematic studies. Ontogenetic Changes in Skeletal Structures Cranial and associated bones— CTaniaX osteology has, of course, been the foundation of systematic studies of adult fishes, but the development of cranial bones has been little used in phy- logenetic studies. Numerous descriptions of the ontogeny of cranial bones exist in the literature (e.g., Bhargava, 1958; Bert- mar, 1959; Kadam, 1961; Weisel, 1967; Moser and Ahlstrom, l970;Mook, l977;Leiby, 1979b; Yuschak, 1982). Additionally, the sequence of ossification of head bones has been described for a variety of taxa (e.g., Moser, 1972; Aprieto, 1974; Leiby, 1979a; Dunn, 1983a; Kendall and Vinter, 1984). The devel- opment of certain cranial structures has also been shown to be of taxonomic value (Fritzsche and Johnson, 1980), yet com- parative studies of the developmental osteology of the skull of related groups of teleosts seem rare (e.g., Norman. 1926b; De Beer, 1937). Available evidence suggests that the sequence of ossification of the skull of teleosts is a conservative (i.e., relatively constant among different phyletic groups) process (De Beer, 1 937; Mook, 1977). Among the bones which ossify first are those in areas of high stress, such as feeding (jaw bones) and respiration (bran- chial region), as noted by De Beer ( 1 937), Weisel ( 1 967), Moser and Ahlstrom (1970), Mook (1977), Yuschak (1982). Examples of ontogenetic changes in skull bones which suggest that these structures might offer insight into phylogenetic affin- DUNN: DEVELOPMENTAL OSTEOLOGY 49 ities include upper jaw bones (Berry, 1 964a), head spines (Ken- dall, 1979; Washington, 1981; Yuschak, 1982; Washington and Richardson, MS), gill arches (Leiby, 1979b; Yuschak, 1982; PotthofTet al., 1984), and lateral skull bones (Leiby, 1979b). Patterns of chondrification may also be of value in inferring phylogenetic relationships. Washington and Richardson (MS) noted that while chondrification of skeletal bones in most scor- paeniform fishes is a relatively brief process, occurring in pre- flexion and early flexion larvae, chondrification was prolonged (occurring through most larval development) in hexagrammids and in three genera of cottids. These authors also considered a unique pattern of ossification of cartilaginous rings in the regions of the parietal and frontal spines as a synapomorphic character uniting three genera of cottids. Vertebral column and associated bones. — Vertebral centra, neural and haemal spines, apophyses, and ribs all undergo variable changes in configuration with growth. A number of workers have documented the development of the vertebral column and as- sociated bones in a variety of taxa, but attempts have not been made to analyze the phylogenetic significance of the ontogeny of these structures. The sequence and direction of ossification of vertebral centra is known to vary among taxa (e.g., Moser and Ahlstrom, 1970; Mook, 1977; Potthoff" et al., 1984), but this character has yet to be analyzed among groups of fishes. Among those elements of the vertebral column which have been studied in various taxa, Potthoff"and Kelley (1982) noted that the neural and haemal arches in Xiphias first develop dis- tally opened, whereas in other perciforms studied, split arches were observed in small larvae on the anterior two centra only. Washington and Richardson (MS), in their study of cottid larvae and scorpaeniform outgroups, noted in various taxa the reduc- tion or absence of the first neural spine, presence or absence of autogenous neural arches on centrum one, shape of anterior neural arches, and whether or not the first neural arch was distally fused or open. Potthoff" and Kelley (1982) cited the unique position and development of ribs in Xiphias compared to other perciforms studied, and Washington and Richardson (MS) examined the location, number, and position of ribs in cottids and perciform outgroups. Fins and their supports— Y>OTsaX and anal fins— The sequence of formation of dorsal and anal fins as well as the order of development of their constituent spines and/or rays varies among taxa (Dunn, 1983b). This succession of formation may be rel- atively constant among related groups or it may vary, but the phylogenetic significance of these events, if any, has yet to be analyzed. Additionally, numerous taxa of larvae possess tran- sient, often bizzare, structures, such as elongate dorsal spines or rays or anal rays (e.g., Kendall, 1979; Moser, 1981). These structures are of taxonomic value and may contain phylogenetic information, but the homologies of these structures, if any, are not known (Govoni, this volume). PotthoflTet al. (1984) indicated that the second dorsal and anal fins are the first to develop in most perciform fishes. How- ever, in generally more advanced species, dorsal fin rays (or spines) develop first anteriorly and second dorsal and anal fin ray development starts after the first dorsal fin is either partially or fully developed. Fahay and Markle (this volume) described the sequence of fin formation in gadiform fishes. Usually the vertical fins ossify at nearly the same time, but two or more centers of ossification are present in those genera (e.g., Molva. Merluccius) with a single long dorsal fin (or a short first dorsal fin preceding a longer second dorsal fin). The ontogeny of pterygiophores has received considerable attention from Potthofl"and colleagues (e.g., PotthofT. 1975, 1980; Potthoff'et al., 1980, 1984). The developmental pattern of fin pterygiophores may suggest phylogenetic relationships. PotthofT and Kelley (1982) noted that the first dorsal pterygiophore in Xiphias arose from either one or two pieces of cartilage, as is the case in Morone (Fritzsche and Johnson, 1 980), but not in scombrids. Washington and Richardson (MS) observed the on- togenetic migration of dorsal fin pterygiophores, relative to neu- ral arch position, in three cottid genera. Proximal and distal radials may fuse during ontogeny (Yuschak, 1982) and the pres- ence or absence of medial radials may characterize certain groups of fishes (PotthofT and Kelley, 1982). Pectoral and pelvic fins and their supports.— 'Wilh some excep- tions, pectoral fins develop rays later in the larval period than median fins (Dunn, 1983b). Transient, elongate spines and rays also develop in the pectoral fins of some taxa (Moser and Ahlstrom, 1974; Moser, 1981); such structures may be of taxo- nomic value, but their phylogenetic significance, if any, and their homologies are not known. Relatively few descriptions have been published on the development of the pectoral fin (e.g., Houdeand PotthofT, 1976; Potthoff", 1980; Potthoff"and Kelley, 1982; Yuschak, 1982; Potthofl["et al., 1984), and few systematic inferences have been drawn. PotthofTet al. (1984) noted, in Anisotremus virginicus. the ontogenetic fusion of the supratem- poral-intertemporal, the elongation of the anterior coraco-scap- ular cartilage, and the reduction in length of the posterior pro- cess. Washington and Richardson (MS) examined the orientation of the cleithrum, as well as its outer lip, the length of the scapula- coracoid complex, the base of the cleithrum, and the cleithral extension over the pelvic bone (among other characters of the pectoral girdle) in their analyses of cottids and their allies. The ontogeny of the pelvic fin and its supporting structures also has been little investigated (PotthofT, 1980; PotthoflTet al., 1980; Fritzsche and Johnson, 1980) and infrequently used in systematic studies. Dunn and Matarese (this volume) indicated that in gadid larvae the length of the posterior-lateral process of the basipterygia differed among subfamilies and tended to be reduced or wanting in those genera presently considered ad- vanced. Caudal fin.— The development of the caudal fin in teleosts, a subject Dr. Ahlstrom was extremely interested in (e.g., Ahlstrom and Moser, 1976), seems to have received more study than other bony structures. However, few workers have attempted to in- terpret the phylogenetic significance of the development of this fin (Dunn, 1983b). The fusion of bones, reduction in size of structures, or'loss of elements by absorption can frequently be observed in the development of the caudal fin in some fishes. Additionally, based on ontogenetic evidence, the structure of this fin may differ from that commonly accepted based on adult specimens (Dunn, 1983b). Ontogenetic changes in the caudal fin and associated bones which have been used to infer phylogenetic relationships include the reduction through fusion of ural centra (Moser and Ahl- strom, 1 970; and others), discreet or fused hypural bones (Wash- ington and Richardson, MS; Dunn and Matarese, this volume), absence of the parhypural in certain taxa which normally possess one (Washington and Richardson, MS), characteristics (e.g.. 50 ONTOGENY AND SYSTEMATICS OF FISHES -AHLSTROM SYMPOSIUM shape, modification, autogenous or fused to the centra) of neural and haemal spines on preural centra associated with the caudal fin (Washington and Richardson, MS; Dunn and Matarese, this volume), and number of vertebral centra supporting the caudal fin (Washington and Richardson, MS; Fahay and Markle, this volume). Attention has recently been directed toward the presence of radial cartilages (their position and shape during development) in the caudal fin of certain teleosts (Kendall'; PotthofT et al., 1984). These structures may contain information of value in assessing phylogenetic relationships. Squamation.—The development of scales in teleosts has been described for a variety of taxa (e.g.. Berry, 1960; Burdak, 1969; Fujita, 1971; White, 1977; Potthofl'and Kelley, 1982). The se- quence of development of scales and their origin on the fish differs among taxa, and scales undergo changes with ontogeny (e.g.. White, 1977; Potthoffand Kelley, 1982). The acquisition ' Kendall, A. W., Jr. 1981. Ventral caudal radials— oft overlooked structures. (Paper presented at annual meeting Amer. Soc. Ichthyol. Herpetol., Corvallis, OR, June 1981; Abstract in Copeia 1981:935). of scales on fish usually occurs during their transformation to the juvenile stage; however, a number of groups (e.g., Zaniolepis. serranids, holocentrids, and xiphiids) acquire scales during the larval period. Such developmental changes have apparently not been analyzed among diverse groups of fishes. Perspective Developmental osteology of teleosts appears to be an under- exploited approach of potential value in increasing our under- standing of the relationships of fishes. Studies of developmental osteology of teleosts may contribute much to our understanding of homology, the central concept of all biological comparisons (Inglis, 1966; Bock, 1969; Wake, 1979) in our search for prim- itive and derived character states. A number of investigators present at this symposium are actively engaged in evaluating ontogenetic changes in ossified structures in their studies of various taxa of larval fishes. An appraisal of this method may well be in the future, but evidence provided during the course of this meeting will contribute to such an evaluation. Northwest and Alaska Fisheries Center, National Marine Fisheries Service, 2725 Montlake Boulevard East, Seattle, Washington 981 12. Otolith Studies E. B. Brothers ALTHOUGH the value of otolith studies in systematic ich- thyology is well established, essentially all studies to date deal with the otoliths of adults, or only incidentally juveniles, and are usually limited to the external morphology of the typ- ically largest otolith, the sagitta (see reviews of Weiler, 1968; Casteel, 1974; Hecht, 1978; Huygebaert and Nolf, 1979). Oto- liths of larvae, which are of recent interest in terms of age, growth, mortality, and life history studies (Brothers et al., 1976; Struhsaker and Uchiyama, 1976; Methot and Kramer, 1979; Townsend and Graham, 1981; Kendall and Gordon, 1981; La- roche et al., 1982; Lough et al., 1982; Bailey, 1982; Brothers et al., 1983) have been ignored from a taxonomic point of view. This is perhaps not surprising due to their very small size and generally simpler form, with an apparent lack of obvious dis- tinguishing external features. Although the internal structure of larval otoliths appears to be more variable than the external form, no comparative taxonomic studies have been attempted to date. In addition, relatively little has been done on compar- isons of these features of adult otoliths, noting that in a real sense, the internal anatomy of the adult otolith is just the cu- mulative historical record of ontogenetic changes in external structure and growth patterns. Comparative studies on features other than external appearance have tended to be at the crys- tallographic, mineraiogical and chemical level. Carlstrom's ( 1 963) research on the crystallographic structure of fish otoliths and otoconia was a pioneering attempt to apply structural and com- positional information to understanding the broad outlines of vertebrate evolution. A few studies have followed this line of investigation (Lowenstam, 1980, 1981; Lowenstam and Fitch, 1978, 1981), however the discrimination ability of crystallo- graphic techniques is certain to be limited by the relatively few crystalline varieties known to exist in ear stones. Analysis of the amino acid composition of the major organic fraction of otoliths (Degens et al., 1969) offers another possibility for taxo- nomic information, however it is unlikely to be useful for spe- cific identification of individuals. Finally, trace element analysis of otoliths (Gauldie et al., 1980; Papadopoulou et al., 1978, 1980) may allow for stock and perhaps species discrimination, but again the small sample sizes offered by larval otoliths impose severe or impossible methodological problems unless x-ray mi- croprobes or ion microscopes are employed. New analytic tools for chemical studies could offer unique insights into fish sys- tematics. Recently renewed interest in fish otoliths, due primarily to the recognition of daily growth increments (Pannella, 1971, 1980). has resulted in an expanding effort toward collecting, examining and cataloging the otoliths of larval fishes. As we begin to study the external and internal structure of this material for systematically useful characters, we should begin to develop a new set of morphological criteria for species identification, taxonomic relationships, and perhaps phylogenetic reconstruc- tion. BROTHERS: OTOLITH STUDIES 51 Fig. 2 1 Abrupt changes in external and internal morphology of the sagitta associated with the end of the larval stage. (A) Scanning electron micrograph of the medial face of the left sagitta (9 mm SL) of a french grunt {Haemuton flavohtwatum). (B) 12 mm SL, showing development of "secondary growth centers." (C) Enlargement of area in previous specimen. (D) 44 mm SL. Scale omitted: 12 mm = 500 ixm. (E) SEM of ground and etched hake (Merluccius sp.) sagitta. showing growth centers around the larval otolith. (F) Photomicrograph of ground sagitta of a largemouth bass, Micropterus salmoides. The larval portion of the otolith is in the lower right comer. 52 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Fig. 22. Photomicrographs of otoconia in teleosts. (A) Bonefish, Albula vulpes. free otoconia. (B) Bonefish, otoconia embedded m the sagitta. General Methodology The otoliths (sagittae, lapilli, and asterisci) of larval fish are usually the first calcified structures to appear in the development of an individual. At least some of the otoliths are frequently evident before hatching. Over the larval life, they vary in size from a few to several hundred micra for different taxa and ages. Because of their composition and small size (high surface to volume ratio), larval otoliths are very sensitive to degradation, decalcification, and dissolution in acidic solutions (McMahon and Tash, 1979), and great care must be exercised in preserving larval fish and otoliths. Improper handling results in rapid and irreversible damage. Fresh larvae are best stored for later otolith extraction in three ways: 1 ) frozen, 2) fixed and maintained in strong ethanol solutions (preferably 95%), 3) dried (e.g., on glass slides). The last technique is least preferred due to increased difficulties in otolith removal and general damage to the larvae. Removal from embryos and larvae involves microscopic dis- section with fine needles. The use of crossed polarized filters is sometimes helpful in locating the otoliths, although they are generally clearly visible in the otocysts or otic capsules with standard transmitted illumination. Dissection is best carried out in water, and opaque larva can be cleared by brief exposure to a weak KOH (1%) solution. Air dried otoliths should be trans- ferred on the tips of oil wetted (immersion) needles, and for light microscopy may be stored in oil on slides or permanently mounted under coverslips with a neutral medium (non-acidic). In the latter case, care must be taken to prevent the otoliths from being cracked or crushed as the mounting medium shrinks and pulls down the coverslip. In most cases larval otoliths are small and thin enough to preclude a need for grinding. Light microscopy is best applied to studies of internal structures, al- though some external features can be viewed with either surface microscopy or transmitted light and wide openings of the con- denser diaphragm. Compound microscopes should have high quality oil immersion optics (preferably to at least 1 ,000 x ) and polarizing filters. For the latter, a single, rotatable field polarizer helps in resolving internal structures, while an analyzing polar- izer can be employed to locate the very small, but highly bire- fringent otoliths on slides. A moderately high resolution (at least 500 lines) black and white video system is an additional, but invaluable accessory. Such a system reduces eye fatigue, sim- plifies group viewing, measurement and photography, and most importantly can substantially enhance image quality by elec- tronic adjustment. It is also a necessary component in a variety of automatic and semi-automatic image analysis systems. Scanning electron microscopy is most useful for high reso- lution views of external structures, for examination of fine (< 1 fim) internal features, and for confirmation of suspected optical artifacts. However the technique is also more expensive and time consuming and may necessitate critical preparation. Whole, cleaned and air-dried otoliths can be mounted and coated by standard techniques. Internal views require embedding, grind- ing, polishing and etching before stub mounting and coating. The most recent important development in SEM preparation is the use of etching solutions other than the initially preferred HCl. Haake et al. (in press) summarize a technique for SEM preparation of larval otoliths. Otolith Morphology and Early Ontogeny There are a number of papers which deal with the general structure and composition (Hickling, 1931; Degens et al., 1969: Blackler, 1974: Pannella, 1980), mechanism of growth (Irie, 1960: Dunkelburgeretal., 1980; Campana, 1983), and functions of the otoliths and otolith organs (Popper and Coombs, 1 980a, b; Piatt and Popper, 1981). This work has not specifically dealt with larvae, however the gross morphology and processes should be comparable with older fishes. The otic capsule or otocyst forms very early in the ontogeny of fishes and is an obvious landmark in the head of newly hatched larvae. The earliest evidence of the otoliths is one to several small (usually less than 10 ixm) optically dense bodies, referred to here as primordia. From their physical appearance and etching properties, the primordia are assumed to be sub- stantially composed of organic matrix (probably the fibroprotein otolin), and are soon calcified and surrounded by an accreted layer of calcium carbonate and matrix. There are distinct dif- ferences between certain taxa, usually at the supraspecific level, with regards to the morphology of the primordia. Distinctions also exist between the sagitta, lapillus, and asteriscus, so com- parative studies must be careful to properly identify the otoliths examined. Variation in primordial form involves the size, shape, and number per otolith. Surrounding the primordium (partic- BROTHERS: OTOLITH STUDIES 53 Fig. 23. Otolith primordia and cores. (A) SEM of single primordium and core in a french grunt (Haemulon flavolineatum) lapillus. (B) Photomicrograph of single primordium and core in a mimic blenny {Labrisomus guppyi) sagitta. (C) Multiple primordia in the lapillus of a white sucker {Caloslomus commersoni). (D) Multiple primordia in the sagitta of a seahorse (Hippocampus sp.). (E) Multiple primordia and cores in the lapillus of a banded killifish (Fiindulus diaphamis). (F) SEM of multiple primordia and cores in the sagitta of a rainbow trout {Salmo gairdneri). ularly in the sagitta and lapillus) is a discrete, relatively ho- mogeneous zone of calcified material usually delimited by a distinct, thin, optically dense (matrix-rich) layer. This layer de- fines the boundary of the core. In some cases, careful exami- nation of the core may reveal diffuse, very faint, or extremely fine growth increments, however, they are easily distinguished from the more distinct incremental growth pattern distal to the core. Taxonomically related differences in core size, shape and number generally parallel differences in the primordia. The external morphology of larval fish otoliths is much less 54 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Table 6. Occurrence of Multiple Primordia in Fish Otoliths (see Text for Explanation). Order Mormyriformes Mormyridae Order Salmoniformes Esocidae Umbridae Salmonidae (including Coregoninae) Osmeridae Order Cypriniformes Characidae Cyprinidae Catostomidae Order Siluriformes Ictaluridae Bagridae Order Atheriniformes Exocoetidae Oryziatidae Cyprinodontidae Belonidae Anablepidae Poeciliidae Atherinidae Order Syngnathiformes Gasterosteidae Syngnathidae Order Scorpaeniformes Cyclopteridae (Cyclopterinae and Liparinae) Order Gobiesociformes Gobiesocidae Order Perci formes Istiophoridae Stichaeidae Percichthyidae Order Pleuronectiformes Pleuronectidae variable than seen for adults. Similarity between taxa is greatest in the youngest and smallest individuals, in which the otoliths, particularly the sagittae and lapilli, tend to resemble flattened spheroids or hemispheres. Landmark features used in char- acterizing adult otoliths such as the form of the sulcus, rostral projections, cristae, colliculi, ostia etc. are initially not evident or weakly developed in most fishes. Exceptions to this gener- alization may prove to be useful taxonomic characters (e.g., in various istiophorids, the sulcus acousticus is clearly developed in larvae only 6 mm SL). Exaggerated or distinctive morpho- logical features of adult otoliths of some taxa may also begin to develop in the early larval stages. For example, if a species has a markedly elongate sagitta, such as found in some callionymids or fistulariids, then the larval otolith may show a tendency for greater growth along the anterio-posterior axis. Unfortunately, such early evidence for adult otolith characters is often not present, particularly for the many species which show an abrupt change in otolith growth patterns at the end of the larval phase. Nevertheless, there are other unique or distinctive larval otolith features in many taxa, and they are potentially valuable for systematic studies. Aside from shape, there are at least two other "external" otolith characters which may be used for taxonomic work; these involve the relative sizes and times of formation of the different otoliths; the sagittae, lapilli, and asterisci. In certain taxa, such as the Ostariophysi, the sagitta is highly modified from the typical teleost condition, being smaller and very elongate; and the asteriscus is relatively enlarged. In clupeids, the lapillus is unusually small and distinctively shaped. Differences of this sort exist to a lesser degree at lower taxonomic levels and may be used in larvae for distinguishing groups. The time of appearance of the otoliths in development is also a variable feature offish ontogeny. Many or perhaps most species have sagittae and lapilli at hatching, the former usually noticeably larger even at this stage. There is a general positive relationship between egg size, time to hatching and state of otolith development at hatching. Fishes with very large eggs and corresponding hatching size may also have the asterisci present at this early stage, however for the majority of fishes, these otoliths appear later, and are sometimes not apparent until the end of the larval stage. The asterisci are distinctive in other respects as well; all species I have looked at have a poorly defined core with multiple primordia; the calcium carbonate is deposited as vaterite (Lowenstam and Fitch, 1981) rather than the aragonite of the sagittae and lapilli; and there are qualitative differences in the appearance of growth incre- ments. Internal structures other than the primordium and core may also have direct or indirect systematic applications. It is well documented that otoliths grow by the addition of layers which are deposited on a diel cycle (see earlier references on larvae, plus review by Pannella, 1 980; also Barkman, 1 978; Wilson and Larkin, 1980; Steffensen, 1980; Victor, 1982; Victor and Broth- ers, 1982). These daily growth increments are usually simple bipartite structures composed of one protein-rich and one pro- tein-poor calcareous layer. In certain situations (especially fast growth and large otoliths) subdaily increments (formed over shorter time intervals) of similar structure may also be present. The timing of the production of the defining boundary of the core, which also corresponds to the onset of incremental growth around the core, is another "internal" character that varies be- tween taxa. Some groups start incremental growth before hatch- ing, others at hatching, and still others at about the time of yolk absorption and the onset of exogenous feeding (Brothers et al., 1976; Radtke and Waiwood, 1980; Radlke and Dean, 1982; Radtke, 1984). There appear to be clear taxonomic trends in these characters which are also related to other trends in egg size and developmental rate and pattern. Some Examples of Taxonomically Related Trends in Larval Otolith Form: External Morphology The development of the general form of the adult sagitta is a gradual process in many species, whereas in others there may be one or more relatively abrupt changes in growth form, par- ticularly around the time of transformation from larva to ju- venile. This change involves the development of "secondary growth centers" which first appear externally as angular to rounded protuberances on the sagitta surface (Fig. 21; internal structure is discussed below). The result of the expanding growth around these centers is the eventual surrounding of a discrete larval otolith and the stronger development of form and surface characters of the adult sagitta. In examining the otoliths of over BROTHERS: OTOLITH STUDIES 55 IOh™ 10h"> B Fig. 24. Pnmordia and cores of goby otoliths. (A) Sagilta of adult sirajo goby {Sicydiuni plumieri). (B) Sagitta from an unidentified goby larva. 100 families of fishes, this soil of sagittal growth pattern appears to be characteristic in a number of higher level taxa (e.g., many, but not all, perciform families; some myctophids; certain but not all anguilloid families, pleuronectiform, gadiform and scor- paeniform fishes; Percopsis, and others). It is not certain whether the presence of this character is consistent enough to be used as a diagnostic feature, and it also occurs too late in development to be of use in larval identification. Lapilli and asterisci tend to show more gradual changes in shape and growth (Brothers and McFarland, 1981) and I have not observed the discontinuous pattern described above. Lapilli undergo transitions in incre- mental patterns at about the same time that the sagitta changes in growth form (Brothers and McFarland, 1981; Brothers, un- published), however these are not obviously evidenced in ex- ternal morphology of the former. An unusual and surprising character has been found in a preliminary survey of several of the "lower" teleosts. This fea- ture, the presence of otoconia in the sacculus and/or utriculus in addition to the otoliths, has only been noted for non-teleos- tean bony fishes, i.e., holosteans, chondrosteans, brachiopte- rygians, dipnoans (Carlstrom, 1963) and probably Latimena (Brothers, unpublished). Osteichthyan otoconia or statoconia are numerous (hundreds to thousands), small (from a few to 1 00 ^m) calcareous bodies (vateritic, sometimes aragonitic) which are found in close association with the otolith (Fig. 22). They generally have a very characteristic lens shape, although some may tend towards an hexagonal outline. Internal features are variously developed; a primordium-like body is usually present and incremental growth is seen in some. Unexpectedly, otoconia were found in representatives of the following teleost families: Albulidae, Congridae, Anguillidae, Muraenidae. Moringuidae, Notopteridae, Osteoglossidae and Pantodontidae. The character appears to be an example of a synplesiomorphy shared between non-teleostean osteichthyans and two teleostean superorders, and Osteoglossomorpha and the Elopomorpha. Not all species and possibly families in the latter two groups show the character, so apparently it has been lost independently more than once. The presence of otoconia is usually not apparent until the early juvenile stage, they are not seen in the few larvae I've had available, however, their taxonomic interest warrants mention here. Internal Morphology There are a number of taxonomically related trends in the size and shape of the primordium and core of sagittae and lapilli. Table 6 lists all the families (of 1 13 sampled) found to have representatives with multiple or clustered primordia (inclusion in the table does not necessarily indicate that all family members have the character). In some, particularly the salmonids and related families, the primordia are clearly separated and may each be surrounded by discrete multiple cores, whereas in others, such as the Atheriniformes and Gasterosteiformes, the multiple primordia are more lightly grouped and are usually surrounded by a single core (Fig. 23). Two other primordium and core characters have been found to be unique to certain taxa. In the gobies and related families ( 1 5 genera; Gobiidae, Microdesmidae, Eleotridae, and Gobioid- idae) all species invariably have an elongate primordium in the sagittae and lapilli (usually with a slight central constnction. Fig. 24) which has not been seen in any other group. Since this feature is present at hatching, it allows for rapid and certain identifi- cation of these speciose families. The parrotfishes (Scaridae, 4 genera examined) appear to have a family-specific early growth pattern in the sagitta which also allows for the identification of very young larvae. The nearly spherical primordium and core grow asymmetrically for about the first 5 days, adding new increments in a restricted area on the distal face before the growth pattern changes to one producing a hemispherical larval otolith. The result of this pattern (Fig. 25) is that the core is clearly on a different focal plane from a section normal to the majority of larval growth increments. The core is therefore asymmetrically placed nearer to the proximal or internal face of the sagitta. This feature is easily observed in whole larval otoliths and has not been found in related families such as the labrids, although these families share other larval otolith char- acters. A second class of internal features has obvious external man- ifestations described above, although they may be distinguished externally for only a discrete period in development. "Secondary growth centers" appear in optical sections or SEM views as foci for increment formation removed from the core (Fig. 2 1 ). Sp)ecies in which otoconia occur are also found to have these bodies 56 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM IOh"" lOt"" Fig. 25. Primordia and cores of parrotfish sagittae. (A) Unidentified scarid larva, medial face up, core in focus. The dark crescent is a portion of the crista on the surface. (B) Same as previous, but with increments in focus. (C) Suspected scarid larva, core in focus. (D) Same as (C), increments in focus. incorporated into the otoliths. The mechanism appears to be that the otoconia adhering to the otolith surface are surrounded by new material accreting on the otolith, and eventually these "included" otoconia are found deep within the otoliths of larger fish. In some species, such as Anguilla rostrata otoconia are found in dense bands corresponding to annual zones. "Includ- ed" otoconia have only been observed in juveniles or older individuals. Transitions in otolith microstructure involving changes in the width and optical density of growth increments (Fig. 26) may be related to a variety of morphological and eco-behavioral changes in the early life history offish (Pannella, 1 980; Brothers, 1981; Brothers and McFarland, 1981; and numerous other pa- pers; also related works by Postuma, 1974, and McKem et al., 1974). Hatching, yolk absorption, changes in feeding and hab- itat, postlarval transformation, and settlement can all poten- tially influence the deposition pattern of daily and subdaily growth increments. To the extent that life history patterns consistently diflfer between taxa, we may expect to find microstructural evi- dence of events in the early life history which are of systematic value. Difierences between taxa will then be expressed as dif- ferences in the timing of marks (e.g., hatching) and otolith tran- sitions and in their intensity and duration. Thus we may use otoliths to record ecological information which may then be applied to systematic studies. An even simpler approach might just be a quantitative comparison of growth rates as determined from daily growth records (once validated, and the fish growth- otolith growth relationships are known), however care should be taken to avoid problems due to high intraspecific variability in growth rate (e.g., Methot, 1981; Bailey, 1982; Brothers et al., 1984). Another possibility is the use oflarval life duration as a taxonoinic character. There is evidence to both species speci- ficity and very limited variability in some taxa, as well as vari- ability or flexibility in others (Brothers et al., 1983; Thresher and Brothers, in press; Brothers and Thresher. MS.; Brothers and Erdman, unpublished), so caution must be exercised in using this character as a taxonomic tool. A final ecologically related application is the determination of spawning time (and perhaps place, by correction for current drift) by age determination of larvae, with correction for the lag between fertilization and increment initiation (Townsend and Graham. 1981; McFarland et al., unpublished). When difl^er- ences in spawning times are suspected or known to exist for taxa, then larval age may be used to help in assigning identifi- BROTHERS: OTOLITH STUDIES 57 IOh" B Fig. 26. Transitions in otolith microstructure associated with settlement and transformation from the larval to juvenile stage. (A) Striped parrotfish (Scarus iserti) sagitta. (B) Queen angelfish (Holacanthus ciliaris) sagitta. cation. Under the best of circumstances, when spawning is rel- atively discrete in time, differences of only a few days could potentially be resolved. The last area in which otolith studies might be of value in systematic studies is in the presentation of descriptive papers on fish development. Until now all illustrations and descriptions of development of wild caught larvae were related to body size since we had no information on the age of these specimens. We suspect, and in some cases have direct knowledge (cited earlier) that growth rates of larvae are moderately to highly variable, yet we have no data on the relationship between age and growth rate and the appearance and form of standard characters such as pigment, ossification, meristics, and morphometries. Perhaps some of the variability seen in size specific descriptive accounts is the result of the effects of different growth rates on the char- acters. I urge that we should make an extra effort to determine the age of wild-caught larvae, used in descriptive studies so we may be able to establish age and/or growth rate specific accounts as well as size specific ones. Of course another problem with size is the highly variable shrinkage rates caused by handling and preservation. Alternately we should perform laboratory ex- periments to examine the relationship between growth rate and developmental rate. In this way we may be able to understand some of the underlying causes for intraspecific variation in larval fish characters. Section of EcoLOCiv and Systematics, Cornell University, Ithaca, New York 14853. Present Address: 3 Sunset West, Ithaca, New York 14850. Preservation and Curation R. J. Lavenberg, G. E. McGowen and R. E. Woodsum THOSE processes by which we fix or kill living tissues without significantly altering their gross anatomy, and preserve or maintain these tissues on a long-term basis have routinely re- quired the use of formalin solutions (Fink et al., MS; Markle, 1984). This certainly is the case for fish eggs and larvae. The protocols for use of formalin as a fixative and preservative for ichthyoplankton have been reviewed and standardized in sev- eral techniques manuals (Ahlstrom, 1976; Castle, 1976; Smith and Richardson, 1977). These protocols are well established and it is not our intention to repeat them here. Rather we wish to elaborate on some of the problems associated with preservation and curation, and to propose recommendations to resolve those areas of real or potential conflict. There are two areas of special concern to us that dictated how our investigations proceeded. First, we wish to ensure that em- bryonic pigment is retained in both the egg and larval stages in both the fixation and long-term preservation procedures. Sec- ond, for ontogenetic stages of larvae we were guided by a concern for protection of mineralized structures, guarding particularly against their loss. Specimens that are well-fixed and properly preserved are im- portant not only to ichthyoplanktologists but to a broad spec- trum of biologists, fish systematists, and museum curators. Among fixatives, bufters and preservatives there is no unani- mous agreement on the most appropriate ones. The problems that plague our understanding of the processes associated with 58 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Fig. 27. A proposed method to archive the early life history stages of fishes. In the left foreground is a series of three vials, the first contains the specimens and preservation fluid and is capped with a polyseal closure. This first vial is placed into the second with the documentation. The third vial is a complete unit. As evaporation occurs the outer vial pops free of its plastic closure, indicating that the vial requires curatorial attention. The vials can be placed together in commercially available paper trays, which can be arranged in commercially available wooden trays much like entomological collections are maintained. these chemicals and prevent us from standardizing a protocol are not biological ones but rather those of chemistry. Fixatives. — Formahn generally is accepted as the most appro- priate fixative. However, it must be used in a specific concen- tration, polymerizes with age and with contact with metals, and is a poison. Tucker and Chester (in press) found that formalin used with salt water causes significant shrinkage, whereas an unbuffered 4% solution of formalin mixed with freshwater caused the least amount of shrinkage and distortion during fixation. They found that pigment preserves best in a solution of un- buffered freshwater formalin. Although the pigment holds up well in this solution, the skeleton decalcifies and reduces or may even prevent staining for either bone or cartilage using the meth- ods of Dingerkus and Uhler (1977). In the absence of a suitable, inexpensive substitute we recommend that formalin be used for fixing zooplankton samples, using the standard ichthyoplankton protocols described by Smith and Richardson (1977). This pro- tocol could be modified so as to use freshwater rather than seawater in preserving the sample (Smith and Richardson, 1977: 16-section 2.1.3.1) so as to reduce shrinkage. Buffers— The problems associated with buffers are more diffi- cult to unravel. Buffers have been used in an attempt to control fluctuating pH during fixation and preservation. Buffers are needed to prevent a reduced pH in either the fixative or pres- ervation solution to avoid excessive acidity in formalin that may decalcify bone (Taylor, 1 977). However, tissues clear when the buffer makes the solution alkaline. Taylor's (1977) data indicate that pH can fluctuate only in a narrow range without LAVENBERG ET AL.: PRESERVATION AND CURATION 59 causing some degree of specimen damage. A pH of less than 6.4 begins the process of decalcification, mineral loss in bone, whereas a pH in excess of 7.0 initiates the clearing process that results in translucency. Tucker and Chester (in press) recommend that sodium borate not be used as a buffer on the basis that it results in high pH, i.e., loss of pigment may occur. Calcium carbonate also is not recommended because it tends to precipitate out of solution and onto the larvae. Hexamine should not be used at all because it tends to clear specimens independent of pH, and to damage them (Steedman, 1976) Markle (1984) summarized five years of data for phosphate buffered formalin solutions used as a preservative. He used the standard ichthyoplankton protocol for fixation of his samples. He gives compelling reasons for using a phosphate buffer to control pH of formalin solutions used as a preservative for fish larvae, on the basis that the amount of the buffer can be adjusted to control pH. A review of the ichthyoplankton protocols indicates that so- dium borate (borax) and calcium carbonate (marble chips) are the preferred buffers, although Tucker and Chester (in press) recommend sodium acetate. We wish to stress that our knowl- edge is inadequate, particularly in understanding the chemistry of these processes. Clearly, a study of the chemistry of fixation and preservation must occur before a recommendation of an acceptable buffer can be made. However, we agree with Markle (1984) that phosphate buffers offer the best alternative to borax and marble chips for long-term preservation on the basis of their versatility in adjusting pH. Presenarives. — Afler the fixation process is completed, the zoo- plankton collections are processed to obtain data on plankton volumes. Then the samples are sorted to remove the ichthyo- plankton component, the eggs and larvae of fishes. After the identification, enumeration, and measurements offish eggs and larvae, they are ready for long-term archival preservation. Through this process the collections are usually maintained in a buffered formalin solution. However, Ahlstrom (1976) indi- cated that if an investigator was sensitive to formalin then eth- anol or a similar preservative was acceptable. For final long-term archival preservation Ahlstrom (1976) indicated that fish eggs and larvae were separately vialed, and placed in fresh preservative. This fresh preservative was a one percent buffered formalin solution made with freshwater. Ac- cording to Ahlstrom (1976) the larvae remained in excellent condition for a period of 15-20 years. Tucker and Chester (in press) recommend a long-term preservative consisting of a 4% formalin solution made from distilled water with sodium acetate used as a buffer. Whenever formalin is used as the basis for a long-term preservation fluid for fish eggs and larvae there will be problems of pH. Phosphate buffers apparently control pH best as they are capable of maintaining pH within a narrow range between 6.4 and 7.0. Unfortunately the use of formalin as a final preservative has the potential to incur considerable curatorial expenses just to monitor pH levels. We recommend that 70% ethanol be used as the final pres- ervation fluid on the basis that it renders the pH problem moot, eliminates working with the fumes of formalin, and eliminates problems associated with the staining process. In recommending ethanol we wish to reduce or eliminate the bufliering problems and their associated pH problems in formalin solutions. After fixation, the concentration of formalin can be reduced to a 1% solution, then this fluid can be drained off during the volume determination process and replaced with ethanol. It is important to transfer the collections directly from the I % formalin solution into ethanol without washing them through a water bath. Thus a small concentration of formalin fixative will be retained in the ethanol preservative. Also, the transfer should be a staged one through a series of ethanol solutions, from 1% formalin to 20% ethanol to 45% ethanol to 70% ethanol, rather than a direct transfer. Zooplankton collections should be stored in the dark, specifically avoiding light. Also, the storage facility should be as cold as possible, and it should avoid fluctuating temperatures. In summary, we recommend that formalin be the fixative of record until a suitable alternative can be established. Buffers should be investigated to determine how they affect long-term effects of fixation and preservation. Phosphate buflfered formalin is recommended as the most suitable one to control pH within a narrow range to prevent melanistic pigment loss and deminer- alization. We recommend that ethanol replace formalin as a preservative fluid. Finally, the chemistry of fixation and pres- ervation should be addressed by a chemist to establish a suitable protocol for processing zooplankton samples. Curation.— The chief problems with storage and curation of larval fish collections are to prevent fluid loss, stabilize collec- tions, and to allow for retrieval availability. Fluid losses through evaporation in small containers, such as vials, can be disastrous. There are means to reduce evaporation. We propose that a double vialing procedure be established (Fig. 27). First, evaporation may be significantly reduced, and second, a double vialing system provides a mechanism to eliminate abrasion and damage to fish eggs and larvae. The procedure calls for an inner vial containing the specimens and preservation fluid sealed with a poly-seal closure. This vial is inserted into another glass vial, which leaves sufficient space for labels and specimen documentation. The second vial is sealed with a plas- tic closure. The outer vial is placed upside down over the inner one. The procedure here is to allow gravity to work on vapor evaporating from the inner vial in such a manner that it must be compressed before escaping from the outer vial. Essentially an equilibrium would be achieved that would act to prevent further evaporation. In addition, a means for specimen docu- mentation can be achieved that allows for maximizing these data for curation without causing abrasion or damage to the delicate specimens. Another important aspect of this curation technique would be its contribution to retrieval availability. The vials can be integrated into an existing ichthyological system so as to make them immediately available to researchers while offering to maximize long-term archival preservation protection. We would like to thank all of our colleagues who provided us with information relative to the fixation, preservation and curation of the early life history stages of fishes. On behalf of the steering committee of the Ahlstrom Sym- posium we would like to recommend that the National Museum of Natural History in Washington, D.C., the Museum of Com- parative Zoology (Harvard University), and the Natural History Museum of Los Angeles County in Los Angeles be considered for the deposition of the early life history stages of fishes for long-term archival care. Section of Fishes, Natural History Museum of Los Ange- les County, 900 Exposition Boulevard, Los Angeles, California 90007. DEVELOPMENT AND RELATIONSHIPS Elopiformes: Development W. J. Richards THE Elopiformes comprises four genera of recent fishes and each of these genera is composed of at least two species. The species are found in tropical waters of the Atlantic, Indian and Pacific oceans. Elops, a cosmopolitan genus, is composed of several species and Megalops is composed of two species. M. atlantica Valenciennes is found in both the eastern and western Atlantic and M. cyprinoides (Broussonet) is found in the Indian and western Pacific Oceans. Alhula has two recognized species. A. vulpes is cosmopolitan and A. nemoptera is found on the Atlantic and Pacific coasts of the Americas. Recent electropho- retic work indicates that there may be additional species (Shak- lee and Tamaru, 1981). Pterothnssus has one species along the coast of West Africa, P. helloci Cadenat, and one off Japan, P. gissu Hilgendorf Larval stages of elopiform fishes have attracted great interest among ichthyologists because of their unusual leptocephalus development, a stage found in no other group but the Anguil- liformes and Notacanthiformes. Consequently most recent clas- sifications have combined all fish with leptocephalus larvae into the Elopomorpha (Patterson and Rosen, 1977). Forked tails of the elopiform leptocephali provide an easy means of sepa- rating them from other leptocephali which have reduced or no tails at all. The non-fork tailed leptocephali are treated sepa- rately in the three subsequent papers in this volume. Recent classifications have altered our classical view of elo- piform fishes by suggesting a much closer relationship with eels. Greenwood et al. (1966) included all fishes with leptocephalus larvae in the superorder (Elopomorpha). This superorder con- tained: Elopiformes with two suborders, the Elopoidei (Elopidae and Megalopidae) and the Albuloidei ( Albulidae including Pter- othrissidae); Anguilliformes with two suborders, the Anguil- loidei and Saccopharyngoidei; and Notacanthiformes with two families (Notacanthidae and Halosauridae). A number of papers have discussed this proposed classification and a majority has sustained the opinion that the Elopomorpha is a monophyletic assemblage. Forey (1973a) discussed the intragroup relation- ships and made some interesting observations on leptocephali in a second paper (1973b). Two significant classifications ap- peared in 1977, one by Greenwood and one by Patterson and Rosen. Both classifications concluded that Elopomorpha is a natural, monophyletic group and that Albula and Pterothrissus are related to the Halosauridae and Notacanthidae. Greenwood (1977) presented a concept of Elopomorpha as a Cohort Tae- niopaedia with two superorders: Elopomorpha comprised of Elops and Megalops in the Order Elopiformes (Suborder Elo- poidei) and Anguillomorpha comprised of two orders, the Al- buliformes with two suborders (Albuloidei and Halosauroidei) and the Anguilliformes. Patterson and Rosen (1977) defined a cohort Elopomorpha of three orders: Elopiformes, Megalopi- formes and Anguilliformes, the latter with two suborders— the Anguilloidei and Albuloidei. Patterson and Rosen (1977) con- cluded that the interrelationships of the Elopidae, Megalopidae and Anguilliformes are best represented by an unresolved tri- chotomy. However, it would seem that those with forked tails would be monophyletic and the reduced or tailless leptocephali would be derived from those with tails. The trichotomy scheme results in paraphyletic forked tailed forms. With the exception of the species of Pterothrissus. the species of the remaining genera are coastal with some stages entering hyposaline environments. Pterothrissus helloci occurs benthi- cally from 70 to 500 m, most abundantly from 120 to 250 m, off the coast of West Africa from 9°N latitude to 20°S latitude (Poll, 1953). All elopiforms are presumed to have pelagic eggs although the eggs of all are undescribed. According to Smith and Potthoff (1975) the eggs and early larvae of Harengula jaguana were erroneously attributed to Megalops atlanticus by Breder (1944), Mansueti and Hardy (1967), and Mercado and Ciardelh (1972). The larval stages have been well described for all genera and are unique (Fig. 28). The larval stage is represented by the lep- tocephalus which has been defined by Hulet (1978) and Smith (1979). The leptocephalus is compressed, transparent and leaf- like with a mucinous pouch which distinguishes it from all other fish larvae. It grows to large size compared to other fish larvae, it has fang-like teeth at the early stages which are subsequently lost (possibly reabsorbed), its viscera is confined to a narrow strand along the ventral midline, its musculature forms a thin layer outside of the mucmous pouch and the remainder of the pouch consists of a mass of acellular material composed of mucoproteins and polysaccharides enclosed by a continuous layer of epithelial cells. Its gut is in two sections, an esophagus and an intestine which are separated by a gastric region com- posed of the stomach, liver and gallbladder. The kidney, of various lengths, lies over the gut beginning near the gastric region and contmuing posteriorly. Ventral blood vessels conspicuously appear between the aorta and the kidney and gut. In elopiform leptocephali dorsal, anal, pectoral and pelvic fins are present and the caudal fin is large and forked. Genera of elopiform leptocephali are easily identified except at small sizes prior to caudal development when myomeres are difficult to count. The number of myomeres for elopiforms ranges from 51 to 92 whereas most anguilliform leptocephali have more than 95. Leptocephali of the Cyemidae have 80 myomeres. Smith ( 1 979) provides a key, characterizations and illustrations of the genera. Many other workers have described complete series or individual stages. Complete series of Elops have been described by Gehringer (1959a), Megalops by Wade (1962), Alhula by Alexander (1961), and Pterothrissus by Matsubara (1942). Among other papers which describe and illustrate var- ious stages are: oi Megalops by Delsman (1926b), Mercado and Ciardelli ( 1 972), Gehringer ( 1 959b), Eldred ( 1 967b, 1 972) and Richards (1969); of Pterothrissus by Smith (1966b) and Rich- 60 RICHARDS: ELOPIFORMES 61 rv' ve- \vN\v V V -^ z^-^^^.. Fig. 28. Elopiform leptocephali. Top to bottom: Elops sp., 33.8 mm SL, Luanda, Angola (redrawn from Richards, 1 969); Megalops allanticus. 22.8 mm SL. Luanda, Angola (redrawn from Richards, 1969): Plerolhnssus belloci. 123.9 mm SL, off Angola (redrawn from Richards, 1969); and Albula vulpes, 64.2 mm (redrawn from Alexander, 1961). ards (1969); of Elops by Hildebrand (1963a), Eldred and Lyons (1966), Gomez Caspar (1981), Richards (1969); and of Albula by Eldred ( 1 967a), Poll (1953), Gomez Gaspar ( 1 98 1 ) and Hil- debrand (1963b). The Albula leptocephali heads illustrated by Meyer-Rochow (1974) may be incorrect. The characters used for distinguishing the families and genera (following Smith, 1979) are as follows: Albula and Pterothhssus leptocephali have the origin of the anal fin well behind the dorsal fin by a distance exceeding the length of the anal fin base whereas Elops and Megalops have the origin under the dorsal fin or close Table 7. Meristic Characters for Selected Elopiform Leptoc ephali. Taxon Source Dorsal rays Number of anal rays Myomeres Elops saurus spp. Gehringer (1959a) Richards (1969) 21-26 usually 22-24 20 12-15 usually 13-14 15-17 78-82 usually 79-80 70-73 Megalops allantica cyprinoides Wade (1962) Wade (1962) 9-13 usually 12 10-17 usually 12-17 16-22 usually 19-21 18-25 usually 23-25 51-57 59-68 usually 62-67 Alhula vulpes nemoptera Alexander (1961) Rivas(1967) 16 7 65-70 usually 67-68 69-74 Pterolhrissus belloci Richards (1969) 51-56 10-13 85-92 62 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM behind it, by a distance not exceeding the length of the anal fin base. Flops and Mega/ops leptocephali have lateral pigment but Albula and Pterothrissus leptocephali do not have lateral pig- ment. Elops is distinguished from Mega/ops by having a de- pressed head, more dorsal than anal rays and the origin of the anal fin is under the posterior end of the dorsal fin or slightly behind it. Megalops does not have a depressed head, has fevk'er dorsal rays than anal rays and the origin of the anal fin is under the middle of the dorsal fin. Albula leptocephali are separable from Pterothrissus leptocephali by the distance between the pos- terior edge of the dorsal fin and the origin of the anal fin. In Albula this distance is about 2.5 times the length of the dorsal fin base and in Pterothrissus this distance is about 6-7 times the length of the dorsal fin base. Also the snout is short in Albula and prolonged in Pterothrissus. Within genera, meristic char- acters are useful in identification of the species (Table 7). The interrelationships of the elopiform fishes are discussed by Smith in a subsequent paper in this volume. National Marine Fisheries Service, Southeast Fisheries Center, 75 Virginia Beach Drive, Miami, Florida 33149. Notacanthiformes and Anguilliformes: Development P. H. J. Castle THE Notacanthiformes (spiny eels) and Anguilliformes (true eels) were united with the Elopiformes (tenpounders, tar- pons, bonefishes) by Greenwood et al. (1966) as the superorder Elopomorpha. These authors noted that members of the three orders share osteological similarities, swim bladder not con- nected with ear (except for Megalops), and a distinctive larval phase (leptocephalus). More recent authors (Nelson, 1973; Fo- rey, 1973b; Patterson and Rosen, 1977) recognised this rela- tionship, though not precisely in this form. There seems little doubt that they are indeed closely related, but in being exclu- sively elongate fishes the notacanths and eels are readily distin- guished externally from the short-bodied, herring-like Elopi- formes. NOTACANTI form ES McDowell (1973) reviewed the notacanths, a morphologically discrete group of fishes, found on or near the bottom on the deeper continental slope into the deep sea, recognising 2 sub- orders, 3 families, 6 genera and 22 extant species (Table 8). He chose to give subordinal distinction to the Halosauridae on the one hand, and the Notacanthidae and Lipogenyidae jointly on the other, although Marshall (1962) had already demonstrated major structural similarities between these families. The Notacanthiformes have in common with the Anguilli- formes a leptocephalus phase, an elongate body form, the as- sociated lengthening of the anal fin, and a reduced caudal fin. Members of the two orders are otherwise dissimilar. Notacanths have well developed pelvic fins; a compact, dorsal fin with spines in some species; scales present and prominent in some; and a large gill opening and opercular flap. Eels lack pelvic fins; the dorsal, unless secondarily reduced or lost, is always long and is supported by delicate rays; scales, if present, are greatly reduced; and the gill opening and its supporting structures are also re- duced. Furthermore, notacanth leptocephali are as distinctive from those of the true eels as are their adults (Fig. 29). They are greatly elongate (up to 180 cm), having a thin post-caudal fil- ament in place of a normal caudal fin; dorsal and pelvic fins are represented by compact, short-based structures present at some stage of larval growth; there is a minute pectoral, straight gut, subterminal anus and the myomeres are V-shaped, not W-shaped; pigment occurs in a ventral series and (rarely) below the mid- lateral level. Several quite different notacanth leptocephali of this type are known, some almost certainly halosaurids ( Tiluropsis. Lepto- cephalus attcnuatus), some possibly notacanthids (Tilurus) and others of unknown identity (Leptocephalus giganteus). Eggs and early larvae have not yet been identified and information on vertebral numbers is mostly lacking for the group. Until con- firmed identifications have been made and more information is forthcoming from leptocephali, ontogeny is unlikely to con- tribute further to the little that is known of relationships in this order. Anguilliformes The Anguilliformes make up a much larger and more diverse assemblage. I recognize 21 families. 153 genera and 720 species for the group (Table 9). Within the Anguilliformes itself Bohike (1966) reviewed the Table 8. Composition, Distribution and Habitat of the Nota- canthiformes. + = All or most species; ( + ) = some species only. Halo- saundae Nola- canlhidae Lipo- genyidae Taxonomic components: Known genera (adults) Known genera (larvae) Known species (adults) 3 ?1 13 2 ?l 8 1 1 Distribution: Atlantic: Genera Species 3 7 2 3 1 1 E. Pacific: Genera Species 1 2 1 1 I.-W. Pacific: Genera Species 2 5 2 4 Habitat (species): Shelf Slope Abyssal (+) (+) ( + ) ( + ) ( + ) + CASTLE: NOTACANTHIFORMES, ANGUILLIFORMES 63 Leptocephalus giganteus 390mm TL 'Tilurus" "Tiluropsis' Fig. 29. The three major forms of notacanth leptocephali showing in upper two the elongate snout, distinct dorsal (arrow), and ventral melanophore series; in lower left the myoseptal pigment; and in lower right the oval eye. superfamily Saccopharyngoidea (gulpers), a small group of 3 families, 4 genera and 8 species of highly modified mid-water, oceanic eels, unmistakeable in body form and possessing a lep- tocephalus of distinctive type. Although they are currently ac- cepted to be true eels, they are so highly aberrant in form and osteology that a case could be made for their retention in a separate suborder, as indeed was proposed by Greenwood et al. (1966). Other eel families have been studied in some detail, notably the Congridae (Smith. 1971), Synaphobranchoidea (Robins and Robins, 1976), Ophichthidae (McCosker, 1977), Nemichthyidae (Nielsen and Smith, 1978) and others, but there are several major gaps and the order has never been compre- hensively reviewed. With some exceptions, the families and genera of eels occur worldwide (Table 9) while eel species have a more restricted distribution in one or other of the major oceans. Some meso- pelagic, slope/abyssal species and just a few shelf species are known from both Indo-west Pacific and Atlantic. As for many other teleosts. the Indo-west Pacific is richest in genera and species, despite relatively limited collecting there, and infor- mation is scattered (Alcock. 1889 e/.yf(7!/.: Fowler, 1934;Asano, 1962; Karrer, 1982). The eel fauna of the Atlantic is rather better known (Blache, 1977; Bohlke, 1978) but by comparison the group is rather poorly represented in the East Pacific. Characters.— The families and genera of Anguilliformes are dis- tinguished principally by external characters, including mor- phometries (Table 10) but the limits are not yet firmly estab- lished for all families in the order. Osteological characters, which mostly reflect these external modifications are also of value at family and generic levels (Table 1 1 ) but are inadequately known, especially in the Congridae and related families, and the Mu- raenidae. Too few genera have been identified in their larval form for ontogenetic characters to have been used extensively 64 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Table 9. Composition, Distribution and Habitat of ihe Anguilliformes. All or most species; ( + ) = some species only. Synapho- Ophich- Netla- Dench- branchi- Dysom- Simcn- thi- Con- Muraenc- stomali- Colo- thyi- Semvo- Anguil- dae matidae chelyidae dae gndae socidae dae congndae dae mendae lidae Helcr- Monn- enchelyi- guidae dae Taxonomic components: Known genera (adults) Known genera (larvae) Known species (adults) Distribution: Atlantic: Genera Species E. Pacific: Genera Species I.-W. Pacific: Genera Species Habitat (species): Freshwater Shelf: Tropical Temperate Slope/abyssal Pelagic 3 9 1 55 28 9 6 1 2 3 1 2 2 1 2 25 15 4 5 1 2 3 1 2 2 7 16 1 250 131 16 32 4 3 12 15 13 8 3 5 1 29 17 2 6 1 2 2 1 2 2 7 6 1 73 32 5 13 2 3 6 2 2 7 1 17 10 2 3 1 1 1 1 2 39 12 2 3 1 1 1 1 4 6 1 35 24 7 6 1 2 2 1 1 8 8 1 137 ( + ) + ( + ) 63 + ( + ) 9 + 8 ( + ) ( + ) 2 3 6 13 + 10 ( + ) + (+) + -1- -1- + ( + ) ( + ) ( + ) ( + ) -1- -1- ( + ) ( + ) -H in determining relationships. Eel species are principally distin- guished externally, by teeth and cephalic pore patterns and by meristics, especially the number of vertebrae. The latter reflects the number of myomeres in the leplocephali. Many of the adult characters by which the families and genera differ from one another appear to be correlated with the extent to which the rather sedentary mode of life associated with bur- rowing, crevice-dwelling or pelagic habits has been elaborated throughout the group. In most families of eels there are species in which the body is very slender, with vertebrae numbering 180 or more (Table 10). The pectoral fins are reduced or lost variously in families (Muraenidae, Heterenchelyidae), genera (Ophichthidae, Xenocongridae), or even within the life span of individuals {Moringita). The median fins may also be reduced to vestiges either in height or in length by a posteriorwards shift of their origin, or they may be entirely lost, though pterygio- phores can be retained. Scales occur only in some of the syna- phobranchoids and in the Anguillidae. Other characters are not so clearly associated with the adop- tion of fossorial, cryptic or pelagic habits. These include the ventral displacement of the gill openings (the extreme devel- opment being in some Synaphobranchidae and a few Ophich- thidae where they are confluent ventrally); the ventral displace- ment of the posterior nostril (most Ophichthidae, Xenocongri- dae, to some extent the Synaphobranchidae) so that it may even open within the mouth; or its dorsal displacement (Muraenidae), Table 10. Some Morphological Characters of the Anguilliformes. + = All or most species; ( + ) = some species only; * = presumed primitive condition. Synapho- Dysom- Simen- Ophichlhi- Con- Muraenc- Nella- Colocon- Dcr- branchidae matidae chelyidae dat- gndae socidac stomatidae gndac ichthyidae Vertebrae: Min.* 126 107 121 110 105 120 186 148 126 Max. 172 204 125 270 225 261 290 163 159 Scales: Present* -t- ( + ) -h Absent ( + ) -t- -1- -1- -1- -1- -1- + Pectoral: Present* + + + ( + ) -t- -1- Reduced { + ) { + ) (+) ( + ) ( + ) -1- -1- Absent ( + ) (+) ( + ) ( + ) + Caudal: Present* -1- + + -1- + + + H- Reduced (+) (+) Absent -1- Dorsal origin: Over pectoral/gill opening* + + H- + -1- + + + Between pectoral and anus (+) ■f Over or behind anus (+) Gill openings: Lateral* + + + + -1- Displaced ventrally -1- + + (+) + Posterior nostril: Before eye* + + + + + + Displaced dorsally (+) (+) Displaced ventrally + + -1- -1- (+) (+) Lateral line: Complete* + + + + + -1- + + Incomplete + CASTLE: NOTACANTHIFORMES, ANGUILLIFORMES 65 Table 9. Extended. Murae- nidae Myro- congn- dae Xeno- congn- dac Nem- ichlhyi- dae Cyema- Iidae Sacco- pharyngi- dae Eury- pharyngi- dac Mono- gnathi- dae 13 1 8 3 2 1 1 2 7 6 2 1 1 1 1 170 1 15 8 2 3 1 6 8 1 6 6 2 1 1 2 30 1 7 7 2 3 1 2 6 2 3 1 1 1 12 2 6 1 1 2 13 7 3 1 1 2 120 8 7 1 1 3 (+) + + + ( + ) + + + + + or both (the genera of Nettastomatidae). There may be a re- duction of the lateral line (Muraenidae. Xenocongridae) or, con- versely, its great elaboration (the congrid Scalanago). In some eels there is an enlargement of the mouth and teeth-bearing surfaces, either by a forward prolongation of the premaxillary- ethmovomer and dentary (Nemichthyidae, Cyematidae and others), or by the turning backwards of the suspensorium with a coincident reduction or loss of the palatopterygoid arch (Ophichthidae, Muraenidae). In all eels the branchial region is elongate, the pectoral girdle is separated from the skull and the posttemporal is lost. This lengthening is accompanied by a reduction of the opercular series, narrowing of the gill opening and increased importance of branchial pump respiration. The branchial series is displaced backwards with enlargement of the 4th arch as pharyngeal jaws, especially in the Muraenidae. The long branchial wall is sup- ported by an increased number of branchiostegal rays which curve up around the branchial region and expand distally. In the ophichthids the throat is further supported by numerous accessory branchiostegal rays (Parr's "jugostegalia") which are not attached to the hyoid arch and overlap in the ventral mid- line. Overall, there is clearly a strong functional correlation be- tween the lengthening, narrowing and smoothing out of the body outline, the increase in body flexibility and modifications in nostrils, jaws, gill openings and lateral line with the mode of life which is a feature of the eels as a group. Eggs.—T\\e best known stages in the early life history of the Anguilliformes (less so in the Notacanthiformes) are undoubt- edly their highly distinctive leptocephali. Eggs and earliest larvae are very poorly known. Those of the saccopharyngoids and no- tacanths have not been identified. Grassi (1913), Schmidt (1913), D'Ancona(1931b)and Sparta (1937 e^^e^M.) described eggs and developmental stages of several Mediterranean eel species, mostly from reared material. The basis for identification of eel eggs was thus reliably established. Some errors have been made: Eigen- mann's (1902) eggs of Conger oceanicus were apparently those of Ophichlhus cruenlifer {Nap\m and Obenchain, 1980); Fish's (1928) Angiulla rostrala eggs were those of the muraenid An- archias yoshiae (Eldred, 1968). Little further information has been added recently, although Naplin and Obenchain's (1980) detailed account of Ophichlhiis cruent ifer demonsUalcs the use- fulness of matching planktonic, newly hatched larvae with late stage embryos. Yamamoto et al. (1975a, b) described live eggs and early larvae of Angnilla japonica spawned from a ripe fe- male that had been artificially matured, but there have been few //; v/vo studies. There is no comprehensive information available for the identification and comparison of eel eggs, principally Tabi E 10. Extended. Serrivo- Anguil- Monn- Heteren- Muraeni- Myrocon- Xenocon- Nemich- Cyemati- Sacco- Eury- Mono- meridae hdae guidae chelyidae dac gndae gridae ihyidae dae pharyngidae pharyngidae gnalhidae 137 100 98 108 107 131 97 170 74 138 97 88 170 119 + 180 227 216 131 156 400 + 108 250 125 95 + + + + + + + + + + + -1- + + (+) + + + (+) (+) + + (+) (+) + + + + -1- -1- -1- + + + -1- -t- + + + ( + ) + + -1- -1- -1- -1- ( + ) + (+) + ( + ) + + ( + ) -1- + + + + + + -1- -1- + + + -1- -1- + + + -1- -1- -1- -1- + + + -1- + + -1- -1- + + + -1- -1- + -1- + 66 ONTOGENY AND SYSTEMATICS OF FISHES- AHLSTROM SYMPOSIUM Table 1 1. Some Osteological Characters of the Anguilliformes. + = All or most species; ( + ) = some species; * = presumed primitive condition. Synapho- Dysom- Simen- Ophichthi- Muraene- Netlasto- Colocon- Dench- branchidae matidae chelyidae dac Congridae socidae matidac gndae thyidae Frontals: Separate* Fused + + + + + + + + + Pterygoid: Present* + + + + Reduced + (+) + + ( + ) Absent (+) Hyomandibula: Forward* + + + + + + Vertical ( + ) (+) ( + ) Backward + + + Lateral line ossifications: Present + + + + Absent* + + + + + Gill arches: More or less complete* + + ( + ) + + + ( + ) Variously reduced + ( + ) ( + ) because only a few species have been studied from just six families. Major characters of eggs of these families are collated in Table 12, which also includes selected references. Eggs and earliest larvae of Ophichthus cruentifer are illustrated as an ex- ample in Fig. 30. Eel eggs are large; the chorion is thin and clear, but may have minute chromatophores; the perivitelline space is wide; the yolk makes up about one half of the egg diameter and is segmented, with or without chromatophores. Oil globules are usually pres- ent (absent in Muraenidae and Nettastomatidae) but the number and size may vary during development. Development takes around 4 days at about 20 C in Gnathophis mystax (Thomo- poulos, 1956) and in O. tTM£'n//7er(NaplinandObenchain, 1980) but may be several days longer. The yolk reduces in size and the embryo reaches a hatching length of about 4.5-5.5 mm, coiling once or more around the yolk. While the late embryo may possess conspicuous melanophores and segmentation, the definitive number of myomeres and the characteristic pigmen- tation of the lai~vae, if any, are not usually fully established until after hatching. Leptocephali.—The yolk-sac larva ("preleptocephalus" or en- gyodontic stage) which is liberated from the egg is characteris- tically elongate, with a tear-drop shaped to elongate yolk. It Table 12. Characters of Anguilliform Eggs. Family 1 2 Ophichthus Ophichthus Dalnphi . ipterichtus Ophisurus Echehis Ophichthid Facciolella Character cruentifer remicaudus imberbis caecus .serpens mvnis (unident ) oxyrhvncha Diameter of chorion: Min. 1.62 2.10 2.20 3.00 3,04 3.04 3.40 2.96 Max. 2.89 2.40 2,40 3.60 4.00 3.80 3.68 3.24 Diameter of yolk: Min. 1.32 1.32 1.68 2.10 1.60 1.32 1.48 Max. 1.60 1.60 1.60 1.92 2.20 1.85 1.80 1.84 Oil globule(s): Absent + Present -1- -1- -1- + -1- + + Number Min. 1 6 1 3 11 1 Max. 1 22 4 40 28 1 11 Size Min. 0.26 0.08 0.32 Max. 0.65 0.16 0.36 0.36 Pigment of embryo: Present on caudal -t- -t- -1- + -1- Present on gut -1- -1- + + + Present on spmal cord Chorion smooth: -1- -1- + + + + + + Yolk segmented: -1- -1- + + + + + + Reference a b b c d e f g Families represented: References: a- -Naplin and Obencham 1980 h — Sparta, 1942a 1 Ophichthidae b- -Sparta, 1937 i— Sparta, 1939d 2 Nettastomatidae c- -Sparta, 1938a j— Sparta, 1939b 3 Xenocongridae d- -Sparta, 1939c k — Sparta, 1938b 4 Congridae e- -Sparta, 1940a 1— Castle and Roberison, 1974 5 Muraenidae f- -Sparta, 1940b m — Mannaro, 1971 6 Anguillidae g- -Sanzo, 1938a n-Eldred, 1969 0— Yevseyenko, 1974 CASTLE: NOTACANTHIFORMES. ANGUILLIFORMES 67 Table 11, Extended. Scmvo- mendae Anguil- hdac Monn- guidae + + + + + + + + + Heleren- Myrocon- Xenocon- chelyidae Muraenidae gndae gridae Eury- Ncmich- Sacco- pharyngj- Mono- thyidae Cyematidae pharyngidae dae gnathidae +? +? +? (+) + +? + + +? somewhat resembles later stages but the development of larval characters is progressive. There may be substantial differences in pigmentation between this stage and the fully grown lepto- cephalus (e.g.. the congrid Ariosoma, Table 17 E,, Mj, O and Fig. 37); typically the pigmentation pattern is much less com- plex. The engyodontic stage has few, needle-like teeth, lower jaws equal to, or longer than upper, an unformed nasal capsule, and undifferentiated median fin-folds and hypurals. At about 20 mm TL the leptocephalus then enters the eury- odontic stage which lasts until metamorphosis. It begins with shedding of the engyodontic teeth and their replacement by 3 series (usually) of shorter, broad-based teeth, the lower jaw shortens relative to the upper, the head decreases in relative length, and the fins and hypurals differentiate. At this stage leptocephali are highly distinctive and well-known forms amongst fish larvae. At full growth they are typically around 50-80 mm but may attain 300-400 mm (Nemichthyidae) or 1,800 mm (Notacanthiformes). They are almost transparent except for eye and other pigmentation and the blood lacks erythrocytes and haemoglobin. The body is greatly compressed and leaf-shaped or filamentous, typically with a small head, prominent, for- wardly-directed larval teeth and a posteriorly placed anus. The electrolyte make-up of their body fluids differs markedly from that of postmetamorphic forms (Hulet, 1978). Table 12. Extended. Family 2 3 4 5 6 Neltastoma nielanurum C'h/opsis hicolor Conger conger'' Ariosoma baleancum (jnalhophis Gnathophis sp, mystax Muraena helena (ivmnothorax unicolor ( i nigro- marginanis Angiulla iinguiUa? 2.40 3.00 1.44 1.48 + 2.72 3.04 1.40 1.48 -I- 13 0.04 0.08 2.60 1.7 -I- 1 0.40 1.80 1.92 1.00 1.04 -I- 1 5 0.30 2.93 3.43 1.25 1.50 -I- 1 9 0.03 0.10 2.50 3.00 1.50 1.85 -t- 5.0 5.5 2.3 3.4 3.3 4.0 1.5 2.0 -I- 2.3 2.9 1.3 1.6 1 2 0.31 0.42 -I- + J + + + + 1 + + m + -I- m -I- -I- -I- n 68 ONTOGENY AND SYSTEM ATICS OF FISHES -AHLSTROM SYMPOSIUM B C mm D OPHICHTHUS CRUENTIFER Fig. 30. Embryonic and early engyodontic stages of Ophichthus cruenltfer (adapted from Naplin and Obenchain, 1980). I CASTLE: NOTACANTHIFORMES, ANGUILLIFORMES 69 Metamorphosis follows the euryodontic stage. It is relatively abrupt and involves the replacement of many of the character- istic leptocephalus features by those of the juvenile. The body rounds up in section, tissue transparency is lost, the postorbital portion of the head lengthens, the larval teeth are lost and the definitive teeth are gradually substituted. The anus and median fin origins move forwards, though not in all species. Pectoral and caudal fins are lost late in metamorphosis in those species which lack the fin in the juvenile and adult. There may be a substantial reduction in body length, extremely so in the No- tacanthiformes. The principal characters which are retained are the definitive number of myomeres/vertebrae which is estab- lished very early in larval life, the number of dorsal and anal fin-rays which is attained rather late in development, and for some species the larval pigment. The maintenance of larval pigment through metamorphosis is of prime importance in iden- tification at the generic level. However, metamorphic larvae are relatively rare in collections, possibly because they are in any case a transient stage; metamorphics are also benthic and hence less accessible to collection. Information on these important stages is therefore sparse. Identification Leptocephali are thus readily recognisable amongst other fish larvae, apparently abundant in the warmer ocean, and accessible near the surface. Large collections of leptocephali have accu- mulated, for some families and genera there being many more specimens available than of the adults (e.g., the moringuid, Neo- conger. Smith and Castle, 1972; the Nettastomatidae, Smith and Castle, 1982). The availability of such collections and the need for identification of leptocephali have resulted in the recent rapid advance of larval studies (Castle. 1969; Blache. 1977; Smith. 1979; Fahay. 1 983). These studies have, understandably, emphasized identification rather than inter-relationships based on larval characters. Larvae of all but the monotypic families Simenchelyidae and Myrocongridae and those of about half (82) of the genera are known. Several distinctive larval forms, possibly of undescribed genera rather than families, are also known (e.g., the congrid- like Leptocephalus thorianus Schmidt, Smith, 1979). Family identification, largely by morphological and pigment characters, may be arrived at from Table 13, which incorporates infor- mation set out in key form by Smith (1979) and Fahay (1983). This "look-alike" approach to identifying leptocephali largely suffices at the family level but is less satisfactory in identifying genera, especially of the Ophichthidae and Congridae (Leiby, 1981). More detailed information may be necessary, especially for species identification, but this will be slow to accumulate. Some attempt to collate available data for identification pur- poses is made in Tables 14-23, with their complementary figures (Figs. 34 to 43). More than 500 different leptocephali have been described, 200 as nominal species of the invalid genus Leptocephalus Gron- ovius, 1763. The procedure of formally naming eel larvae in this way has been both opposed (Bohike and Smith, 1968) and advocated (Castle, 1969). However, nomenclatural problems associated with naming larval forms will not be readily over- come by ignoring the priority of larval names or attempting to apply a blanket restriction on their use. Some alternative ref- erence scheme, or at least an agreed descriptive procedure, does seem appropriate (Fahay and Obenchain, 1978) to accommo- date the large number of distinctive ontogenetic stages of eels. Fig. 31. Anterior region of leptocephalus of an unidentified ?net- tastomatid (DANA St. 4181 II, 34<'23'N, 25°53'W, 9 June 1931), show- ing tab-like extensions of the intestine. Few complete growth series have been described and illus- trated, and developmental osteology is known only for Anguilla anguilla (Norman, 1926b), Serrivoiner spp. (Bauchot. 1959). Ariosorna baleancum (Hulet. 1977). Ophichthus gomesi (Leiby, 1979a), and Atyrophispunctatus {Leiby, 1979b). At least in Oph- ichthus gomesi ossification of the head skeleton does not occur for most elements until metamorphosis, although the jaws, sus- pensorium and branchial skeleton are present as cartilage during the pre-metamorphic stage. Leiby's recent papers (1979b, 1981) contain detailed information on the sequence of development of the skeleton and emphasize the relevance of a more thorough evaluation of developmental osteology in identification of lep- tocephali. In overall body form leptocephali range from the greatly elon- gate notacanths (Castle, 1973, for references; Smith, 1979; Fig. 29), Nemichthys (Nielsen and Smith, 1978; Smith, 1979; Table 19) and some Nettastomatidae (Smith and Castle, 1982) to the short, deep larvae of Thalassenchelys (Castle and Raju, 1975; Table 22 and Fig. 42). the Xenocongridac (Smith. 1969; Table 22 and Fig. 42) and Cyema atrum (Smith, 1979; Table 23 and Fig. 43). The snout is typically rather sharp, especially so in some Notacanthiformes (Fig. 29), Dysommatidae (Table 14 and Fig. 70 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM 20 3 mm ENGYODONTIC 37'» mm Gnathophis 856 mm EURYODONTIC Fig. 32. Development of teeth-series in the congrid Gnathophis. 34), Nettastomatidae (Table 19 and Fig. 39) and Cyematidae (Table 23 and Fig. 43), but characteristically short and rounded in the Heterenchelyidae (Table 18 and Fig. 38) and Muraenidae (Table 21 and Fig. 41), especially near metamorphosis. In some Dysommatidae (Table 14 and Fig. 34) it is produced forwards as a conspicuous, narrow, ethmoid rostrum bearing at its tip a pair of "premaxillary" teeth and, in some also, fleshy tabs or tentacles along its length. The rostrum itself is lost at meta- morphosis so that the snouts of post-metamorphic dysomma- tids, apart from their characteristic papillae and plicae, are sim- ilar to those of other eels. In full-grown leptocephali the anus lies just in advance of the midpoint (some Nettastomatidae, Table 19 and Fig. 39; some Muraenidae, Table 21 and Fig. 41; some Xenocongridae, Table 22 and Fig. 42), well behind the midpoint (most genera), or is subterminal (the congrid group Ariosoma-Bathymyrus. Table 1 7 and Fig. 37). For those in which it is subterminal, it advances during metamorphosis, taking with it the anal fin origin and the developing pterygiophores and actinotrichia. Its position in these species is thus a very rough measure of the stage of metamor- phosis. Broadly speaking, the amount of forward movement of the anus is correlated with the length of larval life, generally long in Notacanthiformes, Anguillidae (1-3 years) and Congri- dae (10 months for species of Gnathophis, Castle, 1968; Castle and Robertson, 1974) but much shorter in Moringuidae (3'/2 months for Moringua edwardsi. Castle, 1979) and probably also for Muraenidae, Xenocongridae and many Ophichthidae. How- ever, little is known of the duration of larval life in most eels. A special feature of some Ariosoma-Bathymyrus larvae is an exterilium or external intestine (Mochioka et a!., 1982; Table 17Q and Fig. 37) and in the unidentified larva illustrated by Weber (1913) and Smith (1979), there are tab-like extensions of the intestine, of unknown significance (Fig. 31). The olfactory organ is a round to oval sac immediately in front of the eye. As growth proceeds its single aperture pro- gressively becomes vertically subdivided by flaps growing from the upper and lower margins. After separation of the two nos- trils, the olfactory sac lengthens in many leptocephali, except the Cyematidae, Nemichthyidae and Serrivomeridae, so that the anterior nostril moves forwards to near the tip of the snout. There it becomes subtubular and often turns downwards; late in metamorphosis the posterior nostril may move dorsally or ventrally to adopt its final position above or behind the eye or ventrally on or through the upper lip. The eye is usually round, but in the notacanthiform larvae referred to the larval genus Tiluropsis, and in Leptocephaliis attemiatus, it is characteristically oval, with the long axis ver- tical. In all Synaphobranchoidea, probably also including the Simenchelyidae, the eye assumes a so-called "telescopic" or "tubular" shape (Table 14 and Fig. 34) and the body of the eye faces anterodorsally and is elongate, with a very deep retina. Teeth develop shortly after hatching. These engyodontic teeth (Fig. 32) are few, needle-like, forwardly directed, each one pro- gressively shorter along the rami of the jaws; typically there is a pair of larger teeth anteriorly. The engyodontic teeth are shed at the beginning of the euryodontic growth stage and are pro- gressively replaced with the 3 series of shorter, broad-based teeth in upper and lower jaws; the upper teeth are preceded by an anteriormost pair, slightly smaller than the first maxillary pair, which are very large in the supposed xenocongrid Thalassenche- lys (Table 22 and Fig. 42). As growth proceeds teeth are added progressively, to reach 40-50 at metamorphosis. They are blade- like and slightly recurved in Paraconger, bicuspid in Coloconger (Table 1 8 and Fig. 38), or needle-like and distinctly spaced in the Heterenchelyidae (Table 18 and Fig. 38). Leiby (1979b) notes that the splanchnocranium is so weakly developed in the engyodontic stage of the ophichthid Myrophis pimctatus that the first series of larval teeth cannot be used in feeding. I CASTLE: NOTACANTHIFORMES, ANGUILLIFORMES 71 150 UO 130- 120- nO' CO O O ^ 100 c/) O I CO 5 90 70' 60 50 40 MYROPHINAE 1 Myrophis punctalus 2 M. plumbeus 3 Ahlia egmontis u Pseudomyrophis nimius OPHICHTHINI 5 Aplatophis chauliodus 6 Ophichthus rex 7 O. ocellalus 8 O. gomesi 9 O. cruentifer 10 O. melanoporus 11 Echiophis mordax 12 Myrichthys oculatus 13 M. acuminatus SPHAGEBRANCHINI u Ichthyapus ophloneus 15 Apterichtus ansp 16 ^. kendalh 17 Stictorhinus potamius CALLECHELYINI 18 Letharchus velifer 19 Callechelys muraena 20 C, springer! 21 C. perryae BASCANICHTHYINI 22 Carolophia loxochila 23 Bascanichthys scuticans 2'- 6. bascanium 25 Gordiichthys irrelitus 26 Phaenomonas longissimus /» /• 25 -y^' i^^ A^^ 20 .^^^ ^^,^^' -\^^ ^^ '23 ^i' 019 022 04 017 150 7- 1= / 016 •3 1 •^ 13 09 ^^ oio 02 cTu "-r~ 110 I 120 I 130 I uo — I — 150 — I — 180 100 I 160 1 170 190 — I — 200 — I — 210 — I — 220 MEAN TOTAL VERTEBRAEIADULTS) MYOMERESO-ARVAE) Fig. 33. Position of kidney in adults and larvae of 26 species of Western Atlantic Ophichthidae; black circles adults, open circles larvae. Adults of not all species shown. The gill opening is anteroventral to the pectoral base and any movement to take up an adult ventral position (Synaphobran- choidea, Ophichthidae) does not occur until very late in meta- morphosis. Pectoral fins are present as fleshy tabs in all very early lep- tocephali. If absent or much reduced in the post-metamorphic stage, the loss does not occur until late in larval life or at meta- morphosis (Muraenidae, Ophichthidae, the muraenesocid Gav- laliccps). Actinotrichia do not develop until late in the eury- odontic stage and lepidotrichia not until metamorphosis. The range is 8-22 among the species of eels. Median fins are first visible as undtflferentiated folds of tissue and remain so until the beginning of the euryodontic stage. The dorsal and anal fin skeletons begin to develop posteriorly first, and then progressively forwards, the anal more rapidly than the dorsal. Pterygiophores and associated muscle blocks appear be- fore the actinotrichia but lepidotrichia do not complete devel- opment until metamorphosis is complete. The anal fin supports are usually closely packed before the anus moves forwards dur- ing metamorphosis. The dorsal origin is less easy to define until late in the euryodontic stage and may not take up its final po- sition until well into metamorphosis. In the muraenids Anar- 72 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Table 13. Major Morphological and Pigment Characters of Anguilliform Leptocephali (Families). + = All or most species; ( + ) = some species only. Synapho- Dysom- Simen- Ophichthi- Muraeneso- Nettasto- Colocon- Derich- branchidae matidac chelyidae dac Congndae cidae matidae gridae thyidae Eye: Tubular + + ?+ Normal + + + + + + Hyomandibula: Backwardly oblique Normal + + ? + + + + + + + Gut: A simple, straight tube + + + + + with swellings or loops 1 Swelling 2 Swellings (+) 3 Or more + + ( + ) (+) Body depth: a50%TL Much <50%TL + + + + + + + + Tail tip: Broad, rounded Narrow + + + + + + + + Gut length: sHalfTL ( + ) ( + ) >HalfTL + + + + + ( + ) + + Head: Elongate ( + ) ( + ) Short + + + ( + ) + ( + ) + + Snout: Rounded Acute + + + + + + + + Pigment: Entirely absent At least some present + + + + + + + + None on gut + + Present on gut + + + + + + Present dorsally in orbit ( + ) Absent from orbit + + + + + + + ( + ) Present on spinal cord ( + ) Absent from spinal cord + + + + + + + + Patch below iris + Absent below iris + + + ( + ) + + + + chias. Uropterygius and to a lesser extent Channotnuraena the dorsal and anal fins are much restricted and distinctive as such early in the euryodontic stage (Table 21 and Fig. 41). At least in the Ophichthidae (Leiby, 1982), even in those species which lack a dorsal fin in the adult, pterygiophores and actinotrichia develop in the larvae. There is also a marked correlation between position of dorsal fin origin in larvae and adults. In the congrid Ariosoma and related genera, the anus is subterminal and the dorsal and anal are also restricted but develop progressively forwards during late larval growth (Table 17 and Fig. 37). Dorsal fin-rays range in number from 1 10 in Neocyema erythrosoma to 600-700 in some ophichthids, anal rays usually being some- what fewer. The large number and apparent considerable vari- ability of median fin rays in most eels has resulted in this meristic character being neglected, but it may be of considerable use in larval identification (Leiby, 1981). The caudal fin develops at least as early as the anal, its sup- porting structure being 3 hypurals, the first two joined distally, enclosing a foramen. Typically hypurals 1 and 2 support 4 rays, hypural 3 supports 5 rays, but the hypurals are much broader in the Synaphobranchoidea, supporting about 16 rays. The fin is resorbed, the rays shorten, and finally become embedded in the tail tip of heterocongrin and many ophichthid larvae shortly before metamorphosis. Myomeres differentiate during embryonic development but because of their relatively high number and small size it is not known for any species whether the definitive number of the adult is established then, or after hatching. However, differen- tiation of the most posterior myomeres, as evidenced visually, appears to occur during the engyodontic stage, even for species with very high total numbers of myomeres. Total counts for species with more than about 180 are difficult to make accu- rately, even in fully grown leptocephali. Myomeres are less readily counted as body transparency is lost at metamorphosis. The range in myomere number across the Anguilliformes is 74-78 in the short-bodied Cyema atrum to more than 400 in the greatly elongate Neinichthys scolopaceus (Table 10) with ranges for species of about 10 myomeres at the lower end (e.g.. for Anguilla, Jespersen, 1942) to about 30 in the range 200-300 (e.g., for Nettastomatidae, Smith and Castle, 1982). Vertebrae first begin to differentiate posteriorly just before metamorphosis with the constriction of the terminal portion of the notochord proceeding anteriorly. The value of vertebral counts in defining eel species has be- come firmly established in eel studies (Bohlke, 1978). The cor- relation of vertebral number with number of myomeres in larvae was demonstrated by Jespersen (1942) for Angidlla and taken upextensively in recent years (Blache, 1977; Smith, 1979; Smith and Castle, 1982). In utilizing this agreement between larvae and adults, associated phenomena need to be further explored and assessed, e.g.. pleomerism (the correlation in related species of vertebral number and maximum body length attained: Lind- sey, 1975), "Jordan's Rule" (the tendency for fishes in polar or cool waters to have more vertebrae or other meristic parts than have related forms in tropical warm waters, Jordan. 1892), and sexual dimorphism in vertebral number (as occurs in Aforingua edwardsi. Castle and Bohlke. 1976). The existence of latitudinal dines in vertebral number in eels has been proposed, but not convincingly demonstrated, except possibly for the muraenid Gymnothorax panamensis which CASTLE: NOTACANTHIFORMES. ANGUILLIFORMES Table 13. Extended. 73 Scrrivo- mcndae Anguil- lidac Monn- guidac Heicrcn- Myrocon- chelyidae Muraenidae gndac Xenocon- gndae Nemich- Sacco- Eury- Mono- ihyidae Cyematidae pharyngidae pharyngidae gnathidae + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + {+) + + + + + + + + + + + + + + + + + + + + + + + + (+) + + + + + + + + + + + + + + + + (+) (+) + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + Randall and McCosker (1975) show to have a mean vertebral range of 1 43 in Chile and 1 25 in the Gulf of California. Variation across longitude is apparently not usual but may be consider- able; for example, McCosker (1977, 1979) demonstrates that the ophichthid Myrichlhys maculatus has a mean vertebral count of 153 in the East Pacific to 195 in the Red Sea. Two other problems arise in using vertebral/myomere char- acters in matching leptocephali with their adult species. These are the prevalence of damaged tails in adults of some species, especially those that are slender-tailed (Nettastomatidae, some Congridae and Muraenesocidae) and hence the unavailability of vertebral counts; and the overlap or near concordance of vertebral numbers within species groups. For example, in the western Indian Ocean there are 15-20 species of the muraenid genus (iymnolhorax which have vertebral numbers within the range 130-145. Unless other characters (e.g., fin-ray numbers) can be shown to differ significantly between these species, it is likely that their leptocephali, all having rather similar pigmen- tation, will prove difficult, if not impossible, to identify. However, there is a reliable correlation between the segmental position of the larval kidney and that of the adult. The larval nephros (opisthonephros) is typically an elongate sac lying above the gut approximately in the middle of the body, i.e., near the anus in those larvae with a relatively short gut (Xenocongridae, Nettastomatidae, Ophichthidae) or some distance in front of it in those having a long gut (Congridae). The segmental position of the kidney changes little, ifat all, during larval life and through metamorphosis into the juvenile. Its position then very ap- proximately agrees with the end of the body cavity and the first caudal vertebra. The correlation in the nephros position has been successfully employed as an identification character for the Muraenidae and other families (Blache, 1977) and for some Ophichthidae (Leiby , 1981) but its value has not yet been com- prehensively explored across the Anguilliformes as a whole. Further evidence for the stability of nephros position from larva to adult, at least in the Ophichthidae, is provided in Fig. 33. The figure expresses the mean segmental positions of the end of the nephros in the larvae and adults of various western At- lantic ophichthids of the subfamily Myrophinae and the four tribes of the subfamily Ophichthinae. There is close agreement in position of the nephros between larvae and adults of all species. Furthermore, the position of the kidney (and first caudal vertebra) is conspicuously further back along the body in the tribes Callechelyini and Bascanichthyini. These are readily re- cognisable short-tailed ophichthids whose larvae can be im- mediately identified as such by the posterior position of the nephros. There is considerable overlap in this character between the Myrophinae, Sphagebranchini and Ophichthinae although individually the species are distinct. The larval nephros is typically supplied and drained by two prominent blood vessels passing vertically between the lateral muscles to the aorta and cardinal veins below the vertebral column. The segmental position of the last of these vessels in the leptocephalus and its correlation with the position of the first caudal vertebra in the adult has been emphasised in larval identification. However, it seems simpler to use nephros posi- tion instead. In those groups of larvae in which the anus does not move forwards during metamorphosis, there is some agreement be- tween number of preanal myomeres and preanal vertebrae. 74 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Table 14. Pigment and Morphological Characters of the Synaphobrachoidea. to Fig. 34. + = All or most species; ( + ) = some species only. Refer atei al pigment A. A large midlateral patch at about level of anus B. On caudal only C. A midlateral row of compact or dendritic spots 1 . Row complete 2. Postanal row only D. A dorsolateral row E. A ventrolateral row F. A ventral row G. A postanal row Gut pigment H. Absent 1. An irregular series of dendritic melanophores along its length Morphological J. Posterior flexures of myomeres rounded An opaque midlateral area of myomeres along length of body Posterior flexures of myomeres angular Rostrum absent Rostrum present Gut straight Gut swollen or lightly arched at points along its length Posterior end of gut markedly flexed downwards K. M. N. O. P. Taxa Synapho- bronchus Nettodarus Dvsommma Type Characters A B C D (+) + + + + + + + + + (+) + + + + (+) (+) (+) (+) (+) (+) + + + (+) (+) (+) (+) (+) (+) (+) + + + + However, this character is not generally applicable in larval identification because of forward movement of the anus during metamorphosis in some species. The gut is most often a narrow straight tube, flexed down- wards under the pectoral fin and following the ventral margin to the posteriorly placed anus. The stomach is usually visible as a finger-like sac at about segment 10. The most frequent modifications of the gut tube are loops or swellings at intervals along its length, each usually accompanied by groups of mela- nophores (Ophichthidae, Tables 15-16 and Figs. 35, 36; Ac- romycter. Table 18 and Fig. 38; some Nettastomatidae, Table 19 and Fig. 39). The number and state of development (low, moderate or conspicuous) of the swellings may be diagnostic at family, genus or species level but is not always so (Leiby, 1981). The liver, with associated gall bladder, fills much of the space anteriorly between the gut and the ventral margin of the lateral muscles. It has two or three lobes in the Ophichthidae (Table 15 and Fig. 35), the gall bladder on the second or third lobe, and the lobes may be distinct or connected by a thin band of liver tissue. Larval pigment is present in larvae of all families except the Anguillidae and may be highly elaborated to form complex and distinctive patterns. The pigmentation, if present, is usually much simpler in the engyodontic stage than later stages. Melanophores may begin to appear in the embryo (in some Ophichthidae as several pigment patches on the gut similar to those in the larvae; in some Muraenidae on the spinal cord) but typically do not do so until the early engyodontic stage. Pigmentation sometimes reaches its full expression by the beginning of the euryodontic stage but typically the complex patterns characteristic of the Ophichthidae and other families are not complete until full larval growth. Subsequently pigment may be lost during meta- morphosis (the congrid Ahosoma), but may serve as a highly important character in matching larvae with adults. Individually, melanophores may be dendritic (Dysommati- dae. Table 14 Ci-C, and Fig. 34). ocellate (Congridae, Table 18B and Fig. 38B), compact (Muraenidae, Table 21 D and Fig. 4 1 ) or rather diffuse (Moringuidae, Table 23 C, and Fig. 43). They may be isolated, grouped in clusters to form conspicuous pig- ment patches (the congrid Bathymynis. Table 1 7G and Fig. 37), or they may form well defined lines, series or patterns. In most families they occur on the lateral body surface, including the caudal fin, on the myosepta (Ariosoma, Table 17E and Fig. 37; Bathymyrus. Table 1 7E and Fig. 37; many Ophichthidae, Table 16 and Fig. 36), or on the ventral body wall (Dysommatidae, Table 141 and Fig. 34; Congridae, Table 18Land Fig. 38). They may occur deeper in the tissues, either on the gut, liver, kidney, suspended in the mucinous space between the lateral muscles, associated with the spinal cord or vertebral column or, fre- quently, on the bases of the caudal, anal and dorsal fin-rays. Although Blache (1977) and Fahay and Obenchain (1978) have attempted to summarise pigment patterns in some groups CASTLE: NOTACANTH I FORMES, ANGUILLIFORMES 75 Fig. 34. Illustrations accompanying Table 14. 76 ONTOGENY AND SYSTEM ATICS OF FISHES- AHLSTROM SYMPOSIUM Table 15. Morphological Characters of Ophichthidae (Myrophinae and Ophichthinae). + = All or most species; ( + ) = some species only. Refer to Fig. 35. Myrophinae Ophichthinae Characters Murae- Neen- Pseudo- Ophich- Sphage- Bascanich- Calle- 4h!ia mchlhys Myrophis chetys myrophts thini branchini thyini chelyini A. Body depth (euryodontic stage) 1. >10%TL 2. <10%TL B. Gut loops or swellings 1. Low 2. Moderate to pronounced C. End of nephros 1. Above or just before anus 2. 4-14 myomeres before anus D. Liver lobes and oesophageal swellings 1. Two 2. Three E. Caudal fin at metamorphosis 1. Present, normal 2. Absent (or much reduced) F. Dorsal pterygiophores and rays before metamorphosis 1 . Well developed; dorsal origin migrates forwards 4-6 myomeres 2. Weakly developed; origin migrates forwards 5-50 myomeres (or resorbed) + (+) + (+) + (+) + (+) + (+) + (+) (+) + (+) (+) of lai-vae, the significance of these has not yet been comprehen- sively reviewed across the Anguiliiformes. Furthermore, the ex- tent of intraspecific variability of pigment patterns has also not been assessed. Any present discussion as to the significance or otherwise of similarities and differences in larval pigmentation must therefore be preliminary. The range of pigmentation in genera for which larvae have been identified, and for some other forms, is summarized in Tables 14-23, family by family. These tables, with their accom- panying figures and morphological information, may be used as a guide to generic identification, and also as a synopsis of pigment patterns. Because these are both complex and diverse in some families, they cannot always be simply displayed in keys. In the Ophichthidae also, and other families, further pig- ment patterns are known, probably representing other genera. This is particularly so of Indo-Pacific Anguiliiformes which have not been extensively studied. These tables and figures highlight common features of pig- mentation: (1) on the gut or its adjacent body wall, often as a regular, spaced series from throat to anus (Notacanthiformes, Congrinae, Heterocongrinae. Heterenchelyidae, Colocongri- dae), or as an interrupted series (Nettastomatidae. Muraene- socidae. Dysommatidae. Ophichthidae) or in some other form (Bathymyrinae, Heterocongrinae, Muraenidae, Nemichthyidae, Xenocongridae); (2) on the lateral body surface (Dysommatidae, Congrinae, Nettastomatidae, Xenocongridae). often associated in some way with the myosepta (Ophichthidae, Bathymyrinae, Heterocongrinae, Serrivomeridae. Derichthyidae); (3) on the spinal cord (Nemichthyidae. Muraenidae); or (4) on the bases of the dorsal, anal and caudal fins. The broad perspective on the ontogeny of the Anguiliiformes and Notacanthiformes given by the preceding deserves com- ment. As adults, eels have adopted a somewhat conformist body plan notable for reduction and loss of external features, though the component families of the group are more or less discrete osteologically. In contrast, through elaboration of the leaflike body form and pigment patterns their larvae display a diversity which matches that of any other group of teleosts. This diversity involves some distinctive larval characters (morphological and pigmentary) which allow leptocephali to be identified at the family level. These characters have not been comprehensively assessed; further definitive identification of larval forms will aid any future analysis. Within families, larvae are generally similar in body form and pigmentation but there are several remarkable exceptions. There are some discernible character gradients in larvae (e.g., the complexity of gut swellings or loops in Ophich- thidae; pigmentation of Congridae). but these may or may not be matched by adult character gradients. Detailed meristic in- formation, as forthcoming throughout larval development, is the only satisfactory medium for species identification, espe- cially in the larger eel families. Zoology Department, Victoria University of Wellington, Wellington, New Zealand. CASTLE: NOTACANTHIFORMES. ANGUILLIFORMES 77 OPHICHTHINAE Fig. 35. Illustrations accompanying Table 15. 78 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Table 16. Pigment Characters of Ophkhthidae (Mvrophinae and Ophichthinae). to Fig. 36. + = All or most species; ( + ) = some species only. Refer Characters Myrophinae Ophichlhinae Bas- Aturae- Myr- Seen- Pseudo- Ophich- Sphagc- canich- Calle- Ahiia mchthys ophis chelys myrophis thini branchini Ihyini chelyini B. C. D. E. F. Lateral pigment A. Absent A single spot mid-laterally on nearly every myomere An oblique row (or streak) of compact spots below midlateral level 1. On all or most myosepta 2. On only a few myosepta, often associated with deep axial pigment Round groups of spots scattered over body Extra spots on dorsal and ventral myosepta A group of spots midway along body Axial pigment G. Several deep postanal pigment clusters below vertebral column (sometimes preanal also; may be associated with myomere pigment) Gut pigment H. Scattered spots along gut, usually prominent groups above upward loops, below downward loops Irregular along length, mostly between nephric duct and crest of each gut loop Loop pigment associated with spots on body wall Conspicuous pigment patch at crest of each gut loop I. J. K. ( + ) ( + ) (+) + ( + ) ( + ) (+) (+) ( + ) ( + ) ( + ) + (+) {+) (+) + (+) (+) + Head pigment L. Spots along upper jaw near bases of teeth and often on lower jaw M. On postorbital region, pectoral base or oesophagus Other pigment N. On bases of anal rays O. On body wall above anal base P. On bases of dorsal rays Q. On body wall below dorsal base, or before it R. On caudal base + (+) (+) + + + + + (+) + (+) + + + + (+) + + + (+) CASTLE: NOTACANTHIFORMES, ANGUILLIFORMES 79 OPHICHTHIDAE Fig. 36. Illustrations accompanying Table 16. 80 ONTOGENY AND SYSTEMATICS OF FISHES- AHLSTROM SYMPOSIUM Table 1 7. Pigment and Morphological Characters of Congridae (Bathvm'i rinae, Heterocongrinae) and Miiraenesocidae. + = All or most species; ( + ) = some species only. Refer to Fig. 37. I^ara- Allo- Ario- Uniden- Balhv- Paru- Goi- fivlern- Con- (iuvi- Miirae- xcnonjy- .\cnn- conger soma tified niyriis conger gasta conger gresox aliccps ncso.x s/av mv^lax Lateral pigment A. Absent B. A midlateral row of single spots, often with extra spots below C. A row of few large spots between midlateral and ventral levels D. A large group of dendritic spots at about myomere 80 E. Oblique rows of compact spots on myosepta below midlateral level 1 . Spots very close together + + + + 2. Spots scattered F. Additional oblique rows present 1. Above midlateral level + 2. Below midlateral level + G. A large midlateral patch of minute spots at one third of body length + H. Scattered minute spots above and below midlateral level + Head pigment I. small spots on throat J. Small spots elsewhere on head Gut pigment K. Small spots ventrally before stomach and dorsally behind stomach + + + L. Small spots ventrally behind stomach (+) + M. Series from throat to anus 1. Approx. one spot every 1-2 segments 2. Spots widely spaced (in young only) + 3. 6-9 groups of spots Other pigment N. Small spots on anal and dorsal bases + + + O. A series of spots before dorsal fin; few, large (young); many, small (full grown) + + Morphological: P. Posterior teeth bladelike Q. An "exterilium" intestine ( + ) + + CASTLE: NOTACANTHIFORMES, ANGUILLIFORMES 81 CONG Rl DAE El 7^ z;! MURAENESOCIDAE . < ; «t■ C £ o ■6 3 w £ 3 m ■^ t o :^ o e V b ■a 3 c o c at CT i 3 tl W 3 3 c i 3 E a o o E o o o « w • 3 u a. u a> (D Z 3 <3 s VI o tn O O a. u UJ ^ -* o = •) w ^ tt -o ^ •» ly Callechelyin Ancestor Bosconichf hy s-like Ancestor Moderately Specolized Optiictittiin-like Ancestor oderotely Speciolized Ophictttttin-like Ancestor Moltfolioptiis Or E vips- like Ancestor Ouossiremus-like Ancestor Ance strol Myroph Ancestrol Optlictlttlin = Tribe BenThenchelyin Congrid-like Ancestor Fig. 55. Hypothesized relationships of the subfamilies and genera of the eel family Ophichthidae. tin having instead a hardened tail tip with, at most, a few ru- dimentary caudal rays embedded in the flesh of the tail. The monotypic genus Leptenche/ys. known only from the 1 1 5 mm type specimen, has caudal-fin rays, but they are weakly devel- oped compared to those of a myrophin (McCosker, 1977). Since all ophichthid larvae have a well-developed caudal fin until the onset of metamorphosis, the presence of weakly developed rays in the only known specimen of Leplenchelys may be an anomaly resulting from incomplete resorption during metamorphosis. The well developed caudal fin of Echelus has prompted most earlier authors to place it in the family Echelidae (=Ophichthi- dae, in part) or to ally it with the subfamily Myrophinae (e.g.. Dean, 1972; Blache, 1977); however, the osteology of the genus (McCosker, 1977) and its larval morphology (Blache, 1977: Figs. 72 and 74) clearly place Echelus in the subfamily Ophichthinae and ally it with the tribe Ophichthini. Adult Myrophinae have four to seven branchiostegal rays attached to the epihyal and ceratohyal and 1 3-45 free (unat- tached) branchiostegal rays which originate posterior to the tips of the epihyals. Most adult Ophichthinae have the majority of their branchiostegal rays attached to the epihyal and ceratohyal. The free branchiostegal rays of all Ophichthinae originate an- terior to the tips of the epihyals. The ceratohyal, epihyal and hypohyal of both the Myrophinae and the Ophichthinae originate from a single block of cartilage with the first center of ossification being a thin strip along the lateral face of the cartilage (Leiby, 1979a, b; 1981). When de- velopment is complete, the ceratohyal of the Myrophinae is a simple bone which terminates about midpoint along the lateral face ofthe epihyal (Dean, 1972; McCosker, 1977; Leiby, 1979b). The ceratohyal ofthe Ophichthinae has a slender, elongate distal portion which terminates about midpoint along the lateral face of the epihyal and a medial portion which is attached to the proximal end ofthe epihyal by a cartilage (McCosker, 1977; Leiby, 1981). The urohyal ofthe Myrophinae and Ophichthinae ossifies in a bifurcated medial ligament which is attached to the developing hypohyals. In the Myrophinae, the urohyal is generally limited 104 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM L L- Fig. 56. (Upper.) Anterior portion of Myrophis punclalus larva depicting typical myrophin gut morphology. Abbreviations: LL|_j, liver lobes 1-3; GB, gall bladder. (Lower.) Anterior portion of Neenchelvs microlrelus larva depicting gut morphology. Abbreviations: LL,.,, liver lobes 1- 2; GB. gall bladder. to a basal plate which ossifies from the hypohyal to the bifur- cation of the ligament. The urohyal of the Ophichthinae gen- erally ossifies to include a spike which extends well posterior to the area of the bifurcation. The gill openings of the Myrophinae are midlateral and con- stricted. Ophichthine gill openings are variable in position, their major axis ranging from midlateral to ventral, but always un- constricted. Leptocephali belonging to five of the nine myrophin genera have been identified. Larvae of four of these five genera have three unconnected liver lobes with the gall bladder on the third lobe (Fig. 56-upper). Larvae of the fifth genus, Neenchelvs. which differ trenchantly from all other ophichthid larvae, have two unconnected liver lobes with the gall bladder on the second lobe (Fig. 56-lower). Leptocephali belonging to twenty of the forty- four ophichthin genera have been identified. All twenty of these genera have two connected liver lobes with the gall bladder on the second lobe (Fig. 57-upper). LEIBY: OPHICHTHIDAE 105 5mm Fig. 57. (Upper.) Anterior portion of Ophichthus gomesi larva depicting typical ophichthin gut morphology. Abbreviations: LL|_2. liver lobes 1-2; GB, gall bladder. (Lower.) Middle portion of Ophichthus gomesi larva depicting position of nephros relative to anus in some members of the Ophichthus lineage of the tribe Ophichthini. Abbreviations: N, nephros; A, anus. The dorsal fin of known myrophin lai^ae has well-developed pterygiophores and fin rays prior to the onset of metamorphosis and migrates only a few myomeres anteriorly (4-6) during meta- morphosis to reach its adult position. The dorsal fin of known ophichthin larvae, which is weakly developed having only pte- rygiophores and rudimentary rays in its anterior portion prior to metamorphosis, must migrate 5-20 myomeres anteriorly dur- ing metamorphosis in species having the dorsal fin antenor to the branchial aperture as adults, and 20-50 myomeres in species having the dorsal fin posterior to the branchial aperture as adults, and is resorbed m species which are finless as adults. The subfamily Myrophinae contains two tribes (sensu McCosker, 1977), the Myrophini and the Benthenchelyini. Os- teological examination of adults in the tribe Myrophini indi- cated the presence of three lineages consisting of Pseudomyro- phis and Neenchelys; Myrophis, Ahlia. and a currently undescribed genus; and Muraemchlhys and its allies. The My- rophis and Muraemchlhys lineages share a common ancestor (Fig. 55). Larval morphology oi Myrophis, Ahlia and Muraen- ichthys is very similar and supports the determination of a close relationship for the two lineages. Larvae of these three genera have three unconnected liver lobes, similar gut and opistho- nephros morphology, and similar body length to depth ratios (Fahay and Obenchain, 1978; Leiby, 1979b; Ochiai and No- zawa, 1980). Pseudomyrophis larvae have three unconnected liver lobes and a body length to depth ratio which is similar to that of the Myrophis and Miiraenichthys lineages, but gut and opisthonephros morphology is significantly different from that seen in the Myrophis and Muraemchthys lineages and supports the conclusion drawn from adult data that the Pseudomyrophis lineage is distinct from the Myrophis and Muraenichthys lin- eages. Nelson ( 1 966a) suggested that Pseudomyrophis micro- pinna, the type of the genus, was congeneric with Neenchelys hiutendijki, but that P. nimius, while belonging to the same lineage, was separable at the generic level from either of the other two species. Dean (1972) also felt that the differences 106 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM between P. micropinna and P. nimiiis warranted a separate ge- nus for P. nimius. However, McCosker (1977, 1982) demon- strated that Pseudomyrophis and Neenchelys are both valid gen- era and that P. micropinna, P. nimius, P. atlanlicus and an undescribed Pseudomyrophis from the eastern Pacific are con- generic. Dean (1972) indicated that Myrophis frio properly be- longs in the Pseudomyrophis lineage. Evidence from larval mor- phology supports McCosker's (1977, 1982) recognition of Pseudomyrophis and Neenchelys as valid genera, and supports the recognition of P. micropinna, P. nimius, P. atlanlicus, the undescribed Pseudomyrophis from the eastern Pacific, two un- described Pseudomyrophis known only from their larvae in the western Atlantic, one undescribed Pseudomyrophis from the eastern Atlantic known only from its larva and erroneously identified as P. nimius (Blache, 1977), and Myrophis frio as congeneric. Pseudomyrophis larvae are readily distinguishable from all other ophichthid larvae by a combination of the fol- lowing characters: three unconnected liver lobes, undulating gut and nephros, characteristic head shape, and pigmentation (Blache, 1977; Leiby, in press a). Neenchelys larvae differ tren- chantly from Pseudomyrophis larvae in having two, rather than three, unconnected liver lobes, a gut lacking the marked un- dulations seen in Pseudomyrophis larvae, and a much deeper body than any other known ophichthid (Castle, 1 980; this paper. Fig. 56-lower). Studies of adult Pseudomyrophis and Neenchelys have clearly demonstrated that the two genera are more closely related to each other than either is to any other genus ( McCosker, 1977, 1982). In the light of this information, the most parsi- monious interpretation of the data on the larval morphology of the two genera is that Neenchelys was derived from Pseudo- myrophis or a Pseudomyrophis-hke ancestor. Pseudomyrophis and all other known myrophin larvae except Neenchelys have three unconnected liver lobes and similar body length to depth ratios. It seems likely, therefore, that larvae of the ancestral myrophin also had three unconnected liver lobes and a similar body length to depth ratio. Neenchelys larval morphology can be easily derived from this proposed ancestral larval morphol- ogy by significantly deepening the body and foreshortening the gut so that one liver lobe is lost. Derivation of Pseudomyrophis larval morphology from a Neenchelys-Wkc ancestor requires a change from the ancestral larval morphology body plan to the Neenchelys larval body plan and a later re-emergence of the ancestral larval myrophin body plan in Pseudomyrophis. Benthenchelys cartieri. a highly specialized pelagic eel (Castle, 1972) is the sole member of the tribe Benthenchelyini. The larvae of this species have not yet been described, but based on the hypothesized evolutionary history of the Ophichthidae (Fig. 55), it seems likely that the larvae of 5. cartieri will have three unconnected liver lobes, a well-developed dorsal fin which mi- grates little during metamorphosis, and a body length to depth ratio that is typical of the Ophichthidae. Discovery of these larvae should help clarify relationships within the Myrophinae. The subfamily Ophichthinae contains four tribes (sensu McCosker, 1977); the Ophichthini, Sphagebranchini, Bascan- ichthyini and Callechelyini. The tribe Ophichthini lies at the evolutionary base of the subfamily Ophichthinae, and contains the most primitive, least specialized members of the subfamily. The ancestral ophichthin was probably Ophichthus-hke. The tribe Ophichthini, which contains two lineages, and the tribe Sphagebranchini can be easily derived from the generalized ophichthin character states which are represented in the genus Ophichihus (sensu McCosker, 1977). One lineage in the tribe Ophichthini appears to be directly derived from the generalized Ophichthus condition. The genus Echelus has been represented as belonging to its own unique lineage in the Ophichthinae and has been considered the most primitive member of the tribe Ophichthini because in addition to having all the primitive characters of its closest relative Ophichthus. it possesses a well- developed caudal fin. A re-examination of adult Echelus char- acters in conjunction with the larval characters oi Echelus sug- gests, however, that Echelus belongs to the Ophichthus lineage and that the caudal fin of Echelus is either a case of character reversal or paedomorphosis which resulted in Echelus retaining the larval caudal fin rather than losing it, as is apparently the case in all other members of the Ophichthinae. In addition to the generalized genera Echelus. Ophichthus, and Ophisurus, the Ophichthus lineage contains two groups of specialized genera which are closely tied to Ophichthus by a nearly continuous character series. The Pisodonophis-Myrichthys-Cirrhimuraena group differ from the basic Ophichthus body plan by having an increased number of branchiostegals, multiserial dentition, and individual sp)ecializations found in each genus. The second group, containing Mystriophis and seven allied genera, are specialized for the capture of large active prey by having a strengthened suspensorium and enlarged dentition. The close relationship of this group to Ophichthus is emphasized by similar adaptations in some species of Ophichthus (McCosker, 1977). The close relationship of the Ophichthus lineage is further emphasized by the unique positioning of the nephros relative to the anus found in many members of this lineage. Larvae from seven of the fourteen genera in the Ophichthus lineage have been identified. While there is considerable inter- and intrageneric variability in the general morphology of these larvae, five of the seven genera (Echelus. Ophichthus, Ophisurus, Echiophis, and Apla- tophis) are generally characterized by having larvae with a neph- ros which terminates 4-14 myomeres anterior to the anus on the next to last gut loop or between the last and next to last gut loop (Fig. 57-lower). This condition has not been observed in any genera of the Ophichthinae outside of the Ophichthus lin- eage of the Ophichthini. The larvae of Myrichthys, one of the specialized genera in the Ophichthus lineage, has a nephros which terminates above or just anterior to the anus (Leiby, in press a). Blache (1977) identified a series of larvae as Brachysomophis atlanlicus. This series of larvae differs from the larvae of the closely related genus Aplalophis in having the nephros termi- nating above or just anterior to the anus. Larvae of the western Pacific species of Brachysomophis have not yet been identified. Consequently, it is unknown whether this nephric position is a secondarily derived character of the genus Brachysomophis or whether it is limited to the eastern Atlantic species B. atlanlicus. The other lineage to arise from the generalized Ophichthus- hke ancestor contains eight genera including Quassiremus and Malvoliophis (Fig. 55), which are characterized by various re- ductions and modifications of the generalized Ophichthus-Vike condition such as reduced gill arches, cephalic lateralis systems, and pectoral fins. This lineage probably gave rise to the Sphag- ebranchini and subsequent lineages by continued modification, reduction, and specialization of the ophichthin condition (McCosker, 1977). The larvae of the Quassiremus- Malvoliophis lineage are virtually unknown. Leiby (in press) tentatively identified three larvae as Quassiremus produclus, but no other larvae from this lineage have been identified. There is a natural LEIBY: OPHICHTHIDAE 107 progression in larval morphology from some Ophichthus spp. through Quassiremus morphology to sphagebranchin mor- phology which tends to support McCosker's (1977) hypothesis that the other ophichthin lineages arose through modification, reduction, and specialization of the ancestral Ophichthus-like condition. Quassiremus larvae look much like the larvae of some Ophichthus spp., but differ in having the nephros termi- nate over or just anterior to the anus, and in having reduced gill arches. The tribe Sphagebranchini is distinguished from the other tribes of the Ophichthinae by a combination of the following adult characters: the pectoral girdle is reduced; the pectoral fin is absent; the gill openings are low to entirely ventral; the neu- rocranium is elongate (neurocranium depth going 4 or more times into its length), generally depressed, and truncate poste- riorly; the gill arches are generally much reduced; the body is equal to or shorter than the tail; the tail tip is sharply pointed; and, the cephalic lateralis system is generally better developed than in other tribes (McCosker, 1977). Larval characters which distinguish this tribe from other tribes in the Ophichthinae or which distinguish lineages within the tribe, are reflections of the adult characters (e.g., reduced gill arches, short gut, dorsal fin origin) (Leiby, 1982). As yet, there are no independent larval characters which confirm the monophyletic origin of this tribe or which confirm the proposed lineages within the tribe, al- though the larval morphology is similar to, and sometimes dif- ficult to distinguish from, the larval morphology of some Oph- ichthini and is consistent with the hypothesis of modification, reduction, and specialization of the ancestral ophichthin con- dition which has been proposed based on adult data. The tribe Bascanichthyini, apparently derived from a mod- erately specialized ophichthin-like ancestor (McCosker, 1977), is distinguishable from the other tribes of the Ophichthinae by a combination of the following adult characters: the body is equal to, or longer than the tail; the gill openings are low lateral and crescentic, never entirely ventral; dorsal-fin origin is on the head in most genera; the pectoral fin is reduced or absent; the cephalic lateralis system is reduced; and, the gill arches are generally much reduced (McCosker, 1977). The genus Dalophis is provisionally placed in the Bascanichthyini despite its pos- session of a gill arch skeleton and a body length which are more ophichthin than bascanichthyin, due to its reductions, general cephalic appearance and several osteological characters (Mc- Cosker, 1977). If this placement oi Dalophis is correct, it seems likely that the ancestral bascanichthyin was similar in appear- ance to Dalophis. Larval characters which distinguish this tribe from other tribes in the Ophichthinae are reflections of adult characters (e.g., reduced gill arches, relatively long gut and opis- thonephros, and dorsal-fin origin). Larvae have been identified from each of the three proposed bascanichthyin lineages [e.g., Dalophis (Blache, 1 977; Palomera and Fortuno, 1981), Bascan- ichth\'s(B\?Lc\\e, \971\ Leiby, 1981), Gordiichthys (Leiby. in press), Caralophia (Leiby, in press)], but there are currently no clear larval characters which are useful for elucidating relationships within the Bascanichthyini. With one exception, all of the larvae assigned to the Bascanichthyini are characterized by extremely low to moderately developed gut loops and, except for gut length, nephros length and dorsal-fin origin, look much like larvae of the Sphagebranchini. One larval form which cannot yet be as- signed to a genus, has tentatively been placed in the Bascani- chthyini based on gill arch and caudal osteology although its gut morphology is more like some Callechelyini than Bascani- chthyini (Leiby, in press). Discovery of the adults of this species may help clarify relationships within the Bascanichthyini. The tribe Callechelyini is apparently derived from a bascan- ichthyin-like ancestor. Adults of this tribe are distinguished by a short neurocranium (neurocranium depth > 33% of its length); the dorsal-fin origin on the head or nape; the body longer than the tail; absence of a pectoral fin; low lateral to entirely ventral anteriorly convergent gill openings; reduced gill arches; reduced cephalic lateralis system; laterally compressed body; small eyes; and, a stout hyoid (McCosker, 1977). Larvae of three of the five known Callechelyin genera have been identified (Leiby, 1984) and are readily distinguishable from larvae of the other ophich- thin tribes. Callechelyin larvae are characterized by moderate to pronounced gut loops; variable but distinctive pigmentation (see Leiby, in press b, for full descriptions); anterior dorsal-fin origin; nephric myomeres more than 56% of total myomeres; a distinct fourth hypobranchial which may be separate from or united with a reduced fifth ceratobranchial (a remnant of the fourth hypobranchial united with a reduced fifth ceratobranchial may occasionally be found in gill arches of larval Sphagebran- chini and Bascanichthyini; a distinct fourth hypobranchial is found in some larval Ophichthini, but, when present, is united with a well developed fifth ceratobranchial); and usually two hypurals rather than the three seen in other ophichthids. McCosker and Rosenblatt (1972) and McCosker (1977) recog- nized the presence of subgeneric lines in the genus Callechelys. Evidence from larval morphology confirms the presence of two subgeneric lineages in Callechelys (Leiby, 1984). Adults of one subgenus have a split urohyal and two rod-shaped elements in the pectoral girdle. The larvae of this subgenus have pronounced gut loops; the fourth hypobranchial free from the fifth cerato- branchial; most or all of the myosepta without pigment; most or all of the anal pterygiophores without pigment; no pigment on the esophagus; pigment on the dorsal surface of each gut loop but no pigment between gut loops; pronounced, round pigment patches in the body wall lateral to each gut loop; and, three to five pronounced, circular postanal pigment patches which consist of subcutaneous and body-wall pigment. Adults of the second subgenus have a simple urohyal and one or two rod- shaped elements in the pectoral girdle. The larvae of this sub- genus have moderate gut loops; the fourth hypobranchial united with the fifth ceratobranchial; dark pigment every third to elev- enth myoseptum, or light pigment on every myoseptum; round or saddle-shaped patches of pigment in the body wall on the ventral margin of the tail extending onto the anal pterygio- phores, or pigment on every anal pterygiophore but none in the ventral body wall; pigment on the esophagus, on the dorsal surface of each gut loop, and between each gut loop; occasionally some body-wall pigment lateral to each gut loop; four to seven irregular, subcutaneous pigment patches on the tail, usually not flanked by body-wall pigment. Relationships to other taxa The family Ophichthidae is generally considered to be a co- hesive group which is the sole member of the superfamily Oph- ichthoidea. The unique nature of ophichthid larvae supports this allocation. Most workers (e.g., Gosline. 1951; Nelson, 1966b; McCosker, 1 977) consider the Ophichthidae to be a specialized offshoot of the Congridae, although Dean (1972) decried the 108 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM value of the characters used to associate the Ophichthoidea with the Congroidea and implied that the Ophichthidae could just as easily be a specialized offshoot of the Anguilloidea. While the only known larvae which could be confused with the Ophichthidae are members of the family Congridae (e.g., Ac- romycter larvae have pronounced gut loops. Nystactichthys lar- vae have a gut which expands abruptly between the esophagus and intestine), there are no known larval characters which un- equivocally establish the evolutionary relationships of the Ophichthidae. Careful osteological studies of ontogenetic series of eel larvae from the various families may eventually clear the currently clouded picture. Department of Natural Resources, Marine Resources Laboratory, 100 Eighth Avenue Southeast, Saint Pe- tersburg, Florida 33701. Clupeiformes: Development and Relationships M. F. McGowAN AND F. H. Berry THE order Clupeiformes contains four families of fishes: the herrings, Clupeidae; the anchovies. Engraulidae; the wolf- herrings, Chirocentridae; and the denticle herring, Denticipiti- dae (Nelson, 1976). Denticeps clupeoides. the monotypic den- ticipitid, occurs in freshwater in southwest Nigeria (Clausen, 1959). Two species of Chirocentrus occur in marine waters of the Indo-Pacific region from the Red Sea to the western Pacific (Whitehead, 1972). They are unusual among the Clupeiformes in that they are piscivorous. The herrings and anchovies are, in general, small schooling planktivores of marine coastal waters. The Indo-Pacific shad, Tenualosa reevesii. reaches 509 mm standard length; the West African riverine species, Thrattidion noctivagns and Sierrathrissa leonensis. are mature at 18 mm (Wongratana, 1980). There are 192 species of clupeids in 62 genera and 122 species of engraulids in 16 genera (Table 24) based on our review of the literature. Herrings and anchovies are most speciose in the tropics, and individual species are most abundant in cold temperate regions and eastern boundary cur- rents (Longhurst, 1971). Some are found in fresh or brackish water; some are anadromous. They support major fisheries worldwide. Their biology has been reviewed most recently by Blaxter and Hunter (1982). Development The eggs and the larvae of Chirocentrus are known (Delsman, 1923, 1930b); the egg and larva oi Denticeps are unknown; and the eggs or larvae of at least one species in a genus have been described for approximately one-half the genera of herrings and anchovies but for only one-third of all species. Ontogenetic stages of herrings and anchovies are best known for species of commercial interest or potential commercial interest in regions with low clupeoid diversity such as the northeast Atlantic (e.g., Chtpea, Sprattus. Sardina. Engraulis) and the California current (e.g., Sardinops, Etrumeus. Engraulis). The ontogeny of mor- phology and behavior, and the requirements for growth and survival of the herring, Cliipea harengus, and the anchovy, En- graulis mordax, are well known (Blaxter and Hunter, 1982). Very little detailed information exists for clupeids from species- rich areas, especially western African freshwaters and the New World tropics. Descriptive taxonomy is still needed in these areas. Table 25 lists the clupeiform fishes for which we found some information about eggs and larvae. Published descriptions of clupeoid eggs and larvae may not be adequate for systematic studies for a variety of reasons. When there are few species in an area with which to confuse the de- scribed species, only the key identifying features are described. When eggs are hatched but the larvae are not reared to meta- morphosis, usually an atypical starving early larva is described. When a well-described series of field-caught larvae is compared with a laboratory-reared series there may be differences in pig- mentation and size at a particular stage of development due to the rearing environment. Future descnptions should describe the eggs and yolk-sac larvae thoroughly because these stages have characters other than those such as meristics which, be- cause they are shared with the adults, are redundant for system- atic purposes. Future descriptions should also try to describe the development of characters which are of phylogenetic im- portance in adult-based classifications because the ontogenetic transformation of a character provides information about the polarity of states of that character (Nelson, 1978). Because the eggs and larvae of so many clupeiform genera are undescribed and because existing descriptions vary in com- pleteness, it is premature to attempt a phylogenetic classification of the Clupeiformes based on early life history stages. However, because many species' eggs and larvae have been described it is possible to identify and describe characters of taxonomic and phylogenetic value, to discuss their distribution among the Clu- peiformes, and to point out some similarities and conflicts be- tween the distribution of egg and larval characters and current hypotheses of clupeiform phylogeny. Taxonomic characters of eggs and larvae The taxonomic characters of clupeoid eggs include size, shape, chorion thickness and sculpturing, width of perivitelline space, degree of yolk segmentation, number and size of oil globules if present, whether they are pelagic or demersal, whether they are adhesive or not, and whether they are spawned in fresh, brackish or full seawater. The egg of Chirocentrus is 1.60-1.65 mm in diameter, has a very small perivitelline space, is pelagic, spherical, and is abun- dant near shore, especially around river mouths (Delsman, 1930b). The egg of Chirocentus nudus has a chorion with fine hexagonal sculpturing (unique among clupeiforms) and up to 9 small oil globules, while the egg of C. dorab has a smooth cho- rion and may have a single oil globule (Delsman, 1923, 1930b). The eggs of clupeids are all globular and they range in size McGOWAN AND BERRY: CLUPEIFORMES 109 Table 24. Families, Subfamilies, Genera, and Species of Clupeiformes with Selected Meristics. Classification follows Whitehead (1972) and Nelson (1976) for subfamilies; Wongratana (1980, 1983) and Nelson (1983) where pertinent for genera and species; otherwise the nomenclature is that of the author cited in the table. Data compiled by F. H. Berry for species presumed valid. A; Atlantic; P: Pacific; c: central; e: east; n: north; s: south; w: west; FW: Freshwater; IcP: Indo-central Pacific; IwP: Indo-west Pacific; 1: India; Aust; Australia; Philipp: Philippines; US: United States of America; Braz: Brazil, Venz: Venezuela; Arg; Argentina. Localion Dorsal Anal P2 Gillrakers Vertebrae Upper Lower Reference DENTICIPITIDAE Denticeps clupeoides Nigeria 9 26-27 5 10 41 Clausen, 1959 CHIROCENTRIDAE Chirocenlrus dorab IcP-Aust 72- -74 Delsman, 1923: White- head, 1973 nudus IwP CLUPEIDAE Clupeinae Sardinelta longiceps I 17-19 14-18 9 117-241 150-253 Wongratana, 1980 neglecta se Africa 17-19 16-18 9 108-166 143-188 Wongratana, 1983 lemuru China-Aust 17-19 15-19 9 51-153 77-188 Wongratana, 1980 Jussieui China-Aust 19-20 19-21 8 52-61 88-101 Wongratana, 1980 sindensis I 17-20 17-21 8 16-46 38-77 Wongratana, 1980 gibbosa IwP 17-20 17-22 8 16-36 38-66 Wongratana, 1980 fimbriata IwP 18-20 19-22 8 27-47 54-82 Wongratana, 1980 albella IwP 18-20 18-23 8 20-36 41-68 Wongratana, 1980 dayi 1 18-19 19-20 8 51-103 87-134 Wongratana, 1980 fijiense N. Guinea 17-18 18-19 8 33-40 61-74 Wongratana, 1980 la Wilis Philipp 18-19 1-22 Wongratana, 1980 hauliensis Taiwan 18-20 19-22 8 Wongratana, 1980 brachysoma 1-Aust 17-20 18-22 8 25-39 48-67 Wongratana, 1980 richardsoni China 18-19 18-22 8 36-42 63-74 Wongratana, 1983 zunasi China-Japan 17-19 17-21 8 21-23 42-58 Wongratana, 1980 marquesensis Marquesas 16-18 17-21 7-8 15-58 27-85 42- -44 Wongratana, 1980 melanura IcP 16-18 17-20 8 20-41 38-74 Wongratana, 1980 alncauda se Asia 18-19 17-18 8 20-26 39-43 Wongratana, 1980 aurita wAeA 17-20 16-18 9 56-81 95-132 45- -47 Wongratana, 1980 hrasiliensis wA 17-18 18-20 9 >150 46 Hildebrand, 1963d; Whitehead, 1973; Berry inaderensis eA 8 >70 Whitehead, 1981 rouxi ecA 8 34-40 Whitehead, 1981 Amblygasler sirm IwP 18-20 17-22 14-18 36-43 Wongratana, 1980 clupeoides wP 18-19 17-19 12-14 26-31 Wongratana, 1980 leiogaster IwP ,19 17-20 13-16 31-33 Wongratana, 1980 Herk/olsichlhys quadrimaculalus IwP-Aust 18-20 16-21 13-17 30-37 Wongratana, 1980 konigsbergeh wP-Aust 18-19 19-22 15-17 30-34 Wongratana, 1980 caslelnaui wP-AusI 17-20 17-22 18-22 39-52 Wongratana, 1980 gotoi N. Guinea 19 17 16 34 Wongratana, 1983 lossei Persian G. 18-19 15-18 12-15 29-35 Wongratana, 1983 spilura I 17-19 15-18 12-15 29-34 Wongratana, 1980 punclatus Red Sea 17-20 13-18 12-17 31-39 Wongratana, 1980 dispilonotus se Asia 17-20 16-19 14-17 34-38 Wongratana, 1980 Escualosa elongala Thailand 16 19 26 41 Wongratana, 1983 thoracata IwP-Aust 15-17 17-21 16-25 29-40 Wongratana, 1980 Opisthonema bidleri eP 18-21 20-23 8-9 35-47 65-83 46- -48 Berry and Barrett, 1963 medirasire eP 17-20 19-23 8-9 70-99 110-156 45- -48 Berry and Barrett, 1963 herlangai Galapagos 19-20 19-22 8-9 75-117 133-171 46- -48 Berry and Barrett, 1963 liherlale eP 17-20 19-22 8-9 1-149 161-224 44-48 Berry and Barrett, 1963 oglinum wA 18-22 22-25 8-9 43-60 72-107 45- -49 Berry and Barrett, 1963 captivai Colombia A 19-20 18-21 8 (c25-28) 49 Rivas, 1972; Berry 110 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Table 24. Continued. Location Dorsal Anal P2 Gillraker Vertebrae Upper Lower Reference Harengula humeralis wA 18 16 8 13-15 26-29 40-41 Whitehead, 1973; Berry clupeota wA 18 18 8 14-16 27-31 41-42 Whitehead, 1973; Berry jaguana wA 17-18 17-18 7-8 16-20 31-35 41-43 Whitehead, 1973; Berry peruana esP 18-19 15-17 8 15-19 31-51 40-42 Berry thrissina enP 16-20 14-17 8-9 9-18 24-33 40-43 Hildebrand, 1946; Berry; Miller and Lea, 1972 Ramnogaster arcuata Arg 7 Whitehead, 1973, 1965 melanostoma Arg Whitehead, 1965 pallida Arg Whitehead, 1965 Platanichthys platana Braz 14 16 7 13 25 Whitehead, 1973 Sardinops sagax sagax esP 17-20 17-20 8 49-54 Ahlstrom sagax caerulea enP 17-20 17-20 8 21-23 44-45 48-54 Berry; Miller and Lea, 1972 neopilchardus Aust 18-20 17-21 58-93 50-52 Berry melanosticta e Asia ocellata s Africa 8 Whitehead, 1981 Sardina pilchardus enA 17-18 17-18 8 44-106 50-53 Whitehead, 1981 Rhinosardinia amazomca Guyanas 13-16 15-19 8 ca. 20 33-43 Hildebrand, 1963d; Whitehead, 1973; Berry bahiensis Braz 17 18 Hildebrand, 1963d Lile piqmtinga wcA 15-17 17-19 7-8 12-17 30-36 38-41 Whitehead, 1973; Berry stolifera eP 17-18 17-23 8 13-18 32-36 42-44 Hildebrand, 1946 Clupea harengus nA 16-20 16-20 37-52 53-60 Hildebrand, 1963d; Wheeler, 1969 pallasi nP 13-21 14-20 20 45 46-55 Berry, 1964b, Ahlstrom; Miller and Lea, 1972 bentincki Chile Whitehead, 1965 Sprallus spraltus enA 16-19 18-20 7-8 46-49 Whitehead, 1965; Wheeler. 1969 antipodum Aust 8 Whitehead, 1965 muelten Aust 8 Whitehead, 1965 hassensis Aust 8 46 Whitehead, 1965 fuegensis Chile 8 49-51 Whitehead. 1965 Clupeonella cultiventris Whitehead, 1965 grimmi Whitehead, 1965 engraulifonnis Whitehead. 1965 abrau Whitehead. 1965 Dussumieriinae Eirumeus teres Cosmop. 18-22 10-19 8-9 12-15 28-35 48-50 Wongratana. 1980; Miller etal.. 1979; Miller and Lea, 1972 whiteheadi S. Africa 18-20 12-13 8 16-18 36-39 54-56 Wongratana, 1983 Dussumieria elopsoides IcP 18-23 14-18 8 11-16 21-32 54-55 Wongratana, 1980; Delsman, 1925 acuta 1-China 19-22 14-18 8 11-15 19-26 54-55 Wongratana, 1980; Delsman, 1925 McGOWAN AND BERRY: CLUPEIFORMES Table 24. Continued. Ill Location Dorsal Anal P2 Gillrakcrs Venebrae Upper Lower Reference Spratelloidinae Spralelloides gracilis IwP Aust 11-14 11-14 S 10-12 28-37 Wongratana, 1980 lewisi N. Guinea 11-13 10-13 8 9-11 28-32 Wongratana, 1983 delkatulus IwP Aust 10-14 9-11 8 9-12 26-32 44- -45 Wongratana, 1980; Miller etal., 1979 robustus Aust 12-13 10-11 8 9-11 28-35 Wongratana, 1980 Jenkinsia lamprolaenia wcA 12-13 13-16 8 19-24 39- -40 Whitehead. 1973; Berry; Cervigon and Velazquez, 1978 stolifera wcA 9-12 13-16 18-25 Whitehead, 1973 majua wcA 11-13 21-28 Whitehead, 1973 parvula Venz 10-13 12-16 20-24 38- -39 Cervigon and Velaz- quez, 1978 Dorosomatinae Clupanodon ihrissa wP 16 21-26 8 (190-480) (200-420) Wongratana, 1980 Konosirus punctatus China 16-19 21-25 8 (145-270) (160-250) Wongratana, 1980 Nematalosa erebi Aust 14-16 19-22 8 (155-370) (145-370) Wongratana, 1980 chanpole IwP 15-17 22-26 8 (250-315) (255-355) Wongratana, 1980 arabica I 17-19 18-20 8 (145-335) (180-390) Wongratana, 1980 come I-Aust 17-18 20-24 8 (175-245) (170-250) Wongratana, 1980 nasus I-wP 15-19 20-26 8 (155-310) (165-315) Wongratana, 1980 japonica wP 16-18 19-22 8 149-205 156-193 Wongratana, 1980 vlaminghi Aust 16-17 19-25 8 216-300 239-328 Wongratana, 1980 paubuensis N. Guinea 14-16 22-27 8 72-342 82-309 Wongratana, 1980 flyensis N. Guinea 14-16 21-26 8 152-553 195-508 Wongratana, 1983 Gonialosa whitcheadi Burma 15 27 8 (92) 90-93 Wongratana, 1983 mammmna I 14-16 22-27 8 87-160 96-166 Wongratana, 1980 modesta Burma 15-17 24-28 8 (125-170) (140-185) Wongratana, 1980 Anodontostoma chacunda IwP 17-21 17-22 8 52-98 54-96 Wongratana, 1980 selangkat wP 18-20 17-21 8 129-186 100-166 Wongratana, 1980 ihailandiae IwP 17-20 18-23 8 43-125 46-140 Wongratana, 1983 Dorosoma cepedianum wnP 10-13 25-36 7-8 (ca. 300- 400) 48- -51 Miller, 1960; Berry petenense wnA 11-14 17-27 7-8 (ca. 300- 400) 40- -45 Miller 1960; Berry anale eMexico 29-38 Miller, 1960 chavesi eNicaragua 12-14 (22-31) Miller, 1960 smithi wMexico 9-13 (22-31) 43- -46 Hildebrand, 1963d; Miller, 1960 Berry Congothnssinae Congothrissa gossei Congo 14-16 15-17 7-8 ca. 40 Poll, 1964 Alosinae Hilsa kelee IwP 16-19 17-22 8 (45-105) (70-180) Wongratana, 1980 Tenualosa toli IwP 17-18 15-21 8 (38-55) (60-95) Wongratana, 1980 macrura Java 19 21-22 8 (46-52) (63-74) Wongratana, 1980 reevesii wP 17-19 16-20 8 53-131 80-248 Wongratana, 1980 ilisha wP 17-20 18-23 8 46-196 62-272 Wongratana, 1980 thibaudeaui Thailand 16-18 19-23 8 (170-248) (205-320) Wongratana, 1980 112 ONTOGENY AND SYSTEMATICS OF FISHES- AHLSTROM SYMPOSIUM Table 24, Continued. Upper Gadusia chapra variegata Alosa sapidissima pseudoharengus mediocris chn^sochloris alabamae aestivalis fallax alosa Pakistan Burma wnA-enP eUS-Canada eUS eUS eUS eUS, Canada enA enA 14-18 16-17 17-20 15-19 15-20 16-21 16-20 15-20 18-21 18-21 21-25 25-27 20-23 17-21 19-23 18-21 19-22 16-21 19-23 20-26 (160-235) (170-255) (250-270) (252-270) Wongratana. 1980 Wongratana, 1980 59-73 54-59 Hildebrand, 1963d; Berry 38-43 46-50 Hildebrand, 1963d; Berry 18-23 54-55 Hildebrand, 1963d; Berry 20-24 53-55 Hildebrand, 1963d; Berry 42-48 55 Hildebrand. 1963d; Berry 41-51 49-53 Hildebrand, 1963d; Berry 20-40 55-59 Whitehead, 1981; Wheeler, 1969 55-85 57-58 Whitehead, 1981; Wheeler, 1969 Ethmalosa fimbriata eA 18 22 8 53 136 44 Whitehead, 1981; Berry Brevoortia aurea Braz gunteri Gulf Mexico 17-20 20-25 7 144 42- -44 Hildebrand, 1963d patronus Gulf Mexico 17-21 20-23 7 138-142 42- -48 Hildebrand, 1963d; Berry smilhi eUS 18-20 22-23 7 151 45- -47 Hildebrand, 1963d; Berry tyrannus eUS, Canada 18-22 18-24 7 137-145 45- -50 Hildebrand, 1963d; Berry Ethmidium chilcae Chile-Peru 18-23 15-18 7-8 123-129 147-159 48- -50 Hildebrand, 1946; Berry Pellonulinae Ehirava fluvial His I 14-16 12-18 8 12-14 24-30 Wongratana, 1980 madagascarensis Nelson, 1970 Dayella malabanca I 14 17 8 10-11 24-27 Wongratana, 1980 Clupeoides borneensis Borneo 15-18 15-19 8 9-12 18-24 Wongratana, 1980 hypselosoma Borneo 14-15 16-18 8 10 12-19 Wongratana, 1980 paupensis Borneo 13-16 17-22 8 9-11 15-19 Wongratana, 1980 venulosus N. Guinea 13-15 20-22 8 Corica laciniata Borneo 15-17 13-16 + 2 8 10-13 23-27 Wongratana, 1980 soborna I 15-16 14-15 + 2 8 9-11 19-21 Wongratana, 1980 Pellonulinae Laevisculella dekimpet Nelson, 1970 Odaxothrissa losera Nelson, 1970 Potamothrissa aculiroslris Nelson, 1970 Spratellomorpha bianalis Nelson, 1970 Pristigasterinae llisha sirishai I 17-18 39-43 8-12 22-26 Wongratana, 1980 novacula Burma 16 43-45 10-12 21-23 Wongratana, 1980 megaloplera 1 16-19 38-53 8-11 19-23 47- -48 Wongratana, 1980; Berry elongala 1-China 16-20 43-53 9-13 21-25 Wongratana. 1980 filigera I 17-21 46-52 9-12 19-23 50- -52 Wongratana, 1 980; Berry macrogaster I 18-19 49 11-12 23-25 Wongratana, 1980 pristigaslroides Java 17-18 47-48 9-10 17 Wongratana, 1980 kampeni 1 16-18 38-46 9-12 20-24 Wongratana, 1983 striatula 1 15-18 40-48 10-13 21-24 Wongratana, 1980 melastoma IwP 15-18 35-48 10-13 21-25 Wongratana, 1983 McGOWAN AND BERRY: CLUPEIFORMES 113 Table 24. Continued. Location Dorsal Anal P2 Gillrakcrs Venebrae Upper Lower Reference obfuscala I 16 39-42 7 12-13 27-28 Wongratana, 1980 afncana ecA 15 47 Whitehead, 1981 amazonica Braz 20 34 6 15 29 Hildebrand, 1963d furlhii ecP 15-17 46-50 11-12 20-25 50-52 Peterson, 1956; Hildebrand, 1946; Meek and Hildebrand, Neoopisthoplerus 1923 cubanus Cuba 12-15 39-43 10 17-19 47 Hildebrand, 1963d, Berry tropicus 15 43-48 8 20 45-47 Peterson, 1956; Hildebrand, 1946 Pellonulinae Clupeichthys hleekeh Borneo 14-15 16-18 + 2 8 8-10 16-18 Wongratana, 1980 aesarnensis Thailand 13-15 14-16 + 2 8 8-10 17-19 Wongratana, 1983 goniognathus Thailand 14-15 15-17 + 2 8 8 15-16 Wongratana, 1980 perakensis Malaya 13-15 14-17 + 2 7 5-9 13-15 Wongratana, 1980 Pellonula leonensis ecA 8 20-30 Whitehead, 1981 vorax ecA Whitehead, 1981 Microthrissa royauxi Nelson, 1970 Poecilothrissa congica Nelson, 1970 Hyperlophus villala Nelson, 1970 Cynolhnssa ansorgii Whitehead, 1981 memo Potamalosa richmondia Wongratana, 1980 Gitchnstella aestuarius Wongratana, 1980 Limtwlhrissa mtodon Wongratana, 1980 Stolothrissa tanganicae Wongratana, 1980 Pristigasterinae Prist igaster cayana Brazil 13-16 44-55 10 20-23 43-44 Hildebrand, 1963d; Berry Opisthoplerus valenaermesi China 16-18 54-65 7 9-12 23-25 Wongratana, 1980 lardoore I 14-17 51-63 7 8-12 22-28 50-52 Wongratana, 1980; Berry dovii ecP 12-13 53-62 17-18 51-52 Meek and Hildebrand, 1923; Ahlstrom equalorialis esP 11-12 59-62 10 25 46-47 Hildebrand, 1946; Ahlstrom Raconda russehana I 81-92 8-11 23-27 62 Wongratana, 1980; Berry Pellona ditchela I-Aust 16-19 34-41 7 10-14 22-27 42 Wongratana, 1980; Berry day! I 17-18 35-42 7 9-11 20-21 Wongratana, 1983 altamazonica Braz 18 37-40 6-7 9 12-14 Hildebrand, 1963d; Berry castelnacana Braz-Venz 18-20 34-42 6-7 13-14 24-25 45-46 Hildebrand. 1963d; Whitehead, 1973; Berry flavipinnis Braz-Arg 17-21 38-47 7 14-15 28-31 43 Hildebrand, 1963d; Whitehead, 1973; Berry harroweri wcA 14-17 36-42 5-6 12-13 24-28 38-40 Hildebrand, 1963d; Whitehead, 1973; Berry 114 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Table 24. Continued. Location Dorsal Anal P2 Gillrakers Vertebrae Upper Lxjwer Reference Odontognathus mucronatus wsA 10-12 74- -85 7-9 22-26 53- -54 Hildebrand, 1963d; Whitehead, 1973; Berry compressus wcA 10-14 52- -62 9 18-23 46- -47 Hildebrand, 1963d; Whitehead, 1973; Berry, Meek and Hildebrand, 1923 panamensis ecP 11-12 61-68 ca. 21 51- -53 Peterson, 1956; Meek and Hildebrand, Chirocentrodon 1923 bleekenanus wcA 14-16 38- -45 6-7 4-6 15-17 44- -45 Hildebrand. 1963d; Whitehead, 1973; Berry Pliosteostoma lutipinnis ecP 49-51 50-51 Peterson, 1956; Berry macrops CLUPEIDAE Status not verified Alosinae Caspialosa maeolica Nelson, 1970 Clupeinae Clupeonella delicalula Nelson, 1970 Dorosomatinae Nematatosa horm Nelson, 1970 Thratlidion noctivagus Sierrathrissa leonensis ENGRAULIDAE Coilinae Coilia ramcarati I 14-16 9-10 21-23 29-30 Wongratana, 1980 borneensis Borneo 14-15 7 21-23 32 Wongratana, 1980 reynaldi I 13-14 7 20-27 28-36 Wongratana, 1980 coomansi Borneo 14 7 21-24 31-33 Wongratana, 1980 rebentischii Borneo 14-15 7 15-19 22-27 Wongratana, 1980 neglecta I 13-15 7 17-19 21-27 Wongratana, 1980 dussumieri I 13-15 7 17-20 23-26 Wongratana, 1980 rendahli China 13-15 7 grayii I-China 13-14 7 21-23 28-31 Wongratana, 1980 lindmam Thailand 12-15 7 18-25 29-34 Wongratana, 1980 macrognalhos Borneo 14-15 7 15-16 22-24 Wongratana, 1980 mystus China 13-15 79- -89 6-7 17-22 25-29 Wongratana, 1980 nasus China-Japan 13-15 87- -117 7 16-20 23-26 Wongratana, 1980 Engraulinae Engraulis japonicus IwP 14-17 14-22 22-34 26-39 Wongratana, 1980 (=australis) eA (=encrasicolus) eA Wongratana, 1980 (=capensis) sAfrica Wongratana, 1980 anchoita swA Whitehead, 1973 euryslole nwA 15-16 16- -19 7 28-31 43- -45 Whitehead. 1973 ringens seP 15-18 19- -24 35-43 38-48 46- -49 Hildebrand, 1946; Berry mordax neP 14-19 19- -26 28-41 37-45 43- -47 Miller and Lea. 1972 "juruensis" Amazon Whitehead, 1973 A nchovia clupeoides swA 14 31 7 105 41 Whitehead, 1973 rastralis eP 12-14 26- -30 ca. 50 Meek and Hildebrand, 1923; Whitehead, 1973 tnnilatis cubana parva lamprotaenia hepselus filfera lyok'pis ginsburgi tricolor choerosloma januaria mitchilli pecloralis cayorum argenteus argentivitlala ischana McGOWAN AND BERRY: CLUPEIFORMES Table 24. Continued. 115 Location Dorsal Anal P2 GUItakei^ Vertebrae Upper Lower Reference surinamensis macrolepidota magdalenae cwA eP neP 13-15 12-14 25-28 27-33 7 47-62 ca. 95 40-42 Whitehead, 1973 Meek and Hildebrand, 1923; Whitehead, 1973; Peterson, 1956 A nchoa spinifer wcA-ecP 15-17 30-40 7 12-16 12-18 19-21 ± Hildebrand, 1963c; Venz wA wcA wA wA wcA wA Venz pananwnsis ecP compressa mundeoloides walkeri anatis curta ecP delicatissima P helleri P slarksi ecP clarki eigenmanma P ecP scofteldi P lucida ecP 13-15 14-15 26-32 14-16 20-24 15-16 21-25 13-16 19-27 13-16 18-24 13-15 19-23 12-16 19-27 18-22 wsA 14-16 18-22 Bermuda 13-15 22-24 wsA 14-15 21-24 wnA 14-16 24-30 Braz 14-16 25-27 wA 13-15 25-29 Venz 16 32 ecP 18-20 enP 18-21 ecP 12 32-26 14-19 19-21 16-18 23-25 13-15 23-26 18-21 23-26 20-23 32-40 20-22 21-22 41 7 7 17-23 17-20 23-33 23-28 42-43 38-41 7 13-18 16-22 39-42 7 15-21 19-25 40-44 7 17-19 20-26 39-40 7 16-23 20-27 41-43 44-45 Whitehead, 1973; Peterson, 1956; Cervigon, 1966; Nelson, 1983 Whitehead. 1973; Cervigon. 1966; Hildebrand, 1963c Whitehead, 1973 Whitehead, 1973; Hildebrand, 1964 Whitehead. 1973; Hildebrand, 1964 Whitehead, 1973; Hildebrand, 1964 Whitehead, 1973; Hildebrand, 1964 Whitehead, 1973; Cervigon, 1966; Hildebrand, 1963c Cervigon, 1966; Hildebrand, 1963c 25-28 18-22 24-28 40- -42 Hildebrand. 1963c 17-20 23-26 41- -42 Hildebrand, 1963c 20-23 23-26 41- -42 Hildebrand, 1963c 15-19 20-26 38- -44 Hildebrand, 1963c 13-14 17-19 4 2 Hildebrand, 1963c 13-15 15-17 43 Hildebrand, 1963c 14 19 Hildebrand, 1963c 17-21 24-25 + 19-22 Peterson, 1956; Nelson, 1983 19-21 22-24 + 19-21 Peterson, 1956; Nelson, 1983 22-24 18-20 + 21-24 18-19 + 20-22 18-20 + 21-23 18-20 + 21-24 17-19 + 20-23 Peterson, 1956; Nelson. 1983; Hildebrand, 1946 Nelson, 1983 Nelson, 1983 Nelson, 1983 Nelson, 1983 22-25 19-22 + 19-22 Peterson, 1956; Nelson. 1983 23-26 19-21 + 19-21 20-23 + 18-21 Nelson. 1983; Miller and Lea, 1972 Nelson, 1983; Miller and Lea, 1972 22-26 20-22 + 19-21 21 + 21 Peterson, 1956; Nelson, 1983 Nelson, 1983 12-13 17-21 + 20-25 20-22 + 21-23 Peterson, 1956; Nelson, 1983 Nelson, 1983 19-22 17-20 + 19-22 Peterson, 1956; Nelson, 1983 116 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Table 24. Continued. Location Dorsal Anal Gil I rakers Vertebrae Refere '2 Upper Lower ice naso ecP 14-16 23-27 21 -24 23-27 19-21 + 19-22 Peterson, 1956; Nelson, 1983; Hildebrand, 1946 chamensis eP 21 + 22 Nelson, 1983 nasus ecP 15-16 21-27 21 -25 24-28 20-21 + 20-22 Nelson, 1983; Hildebrand, 1946 exigua ecP 17-22 23-25 43-45 Peterson, 1956; Nelson, 1983 Anchovielta leptdentostole wsA 14-16 22-25 7 17-18 19-23 Whitehead, 1973; Cervigon, 1966 urevirostris wsA 16-18 18-20 7 24-27 Whitehead, 1973 guianensis wcsA 14-15 18-20 7 23-26 40 Whitehead, 1973 cayennensis wcA 13 16 7 30 Whitehead, 1973 nattereri Braz 12 25-29 Whitehead, 1973; Cervigon, 1966 perfasciata wnA 12-15 15-19 18-23 25-28 42-44 Cervigon, 1966 elongata Panama A 13-14 22-24 17-18 22-24 39 Cervigon, 1966 blackbumi Venz 13-15 25-27 10-12 15-17 43 Cervigon, 1966 jainesi Braz 12-13 19-21 12-13 20-21 40 Cervigon, 1966 vaillanti 23 19 Whitehead, 1973 carrikeri 17-18 14-15 Whitehead. 1973 Slolephorus indicus IwP 14-17 17-22 16-20 20-28 20-23+ Wongratana, 19-21 980 commersonii IwP 15-17 20-23 12-27 21-35 Wongratana, 980 brachycephalus Papua 16-17 22-25 15-17 20-22 Wongratana, 983 chinensis China 16-18 20-23 18-19 26-27 Wongratana, 980 wailei 1-Aust 15-17 19-24 14-17 1-4 Wongratana, 980 holodon seAfr 15-18 20-23 17-22 24-29 Wongratana, 980 andhraensis el-Papua 15-17 19-23 14-15 20-21 Wongratana, 980 lysoni Papua 15-17 21-25 15-18 21-25 Wongratana. 983 insulahs I-China 14-17 19-23 16-20 22-28 Wongratana, 980 dubwsus I 14-16 19-24 19-24 25-31 Wongratana, 980 baganensis I 14-16 20-23 16-19 20-24 Wongratana, 980 iri Thailand 14-15 19-22 15-17 19-22 Wongratana, 980 oligobranchus Philipp 14-16 18 7 13-14 17-18 Wongratana, 983 Thryssa baelama IwP 15 29-34 14-20 19-26 Wongratana, 980 chefuensis China 14 29-34 23-28 27-30 Wongratana, 980 rastrosa N. Guinea 14-15 32-35 39-44 55-61 Wongratana, 980 scratchteyi N. G.-Aust 14 33-36 15-18 18-20 Wongratana, 980 aesluaha N. G.-Aust 13-15 32-36 22-25 27-29 Wongratana, 980 kammalcnsis Thailand 14-15 32-37 23-27 28-32 Wongratana, 980 kammalensoides I 14 34-35 18 24-25 Wongratana, 983 vilrirostris e Africa 13-15 34-43 14-17 20-23 Wongratana, 980 adetae China 13-14 38-44 13-16 20-22 Wongratana, 980 dussumieri I-Taiwan 12-15 34-38 13-16 17-19 Wongratana, 980 mysto-x I-China 13-15 35-39 9-11 13-16 Wongratana, 980 polybranchialis I 13-15 38-42 18-21 25-27 Wongratana, 983 gualamiensis I 13-15 36-40 11-13 17-19 Wongratana, 980 malabarka I 13-15 37-41 14-16 17-19 Wongratana, 980 hamiltonii IwP 13-15 35-41 7-10 11-15 Wongratana, 980 whiteheadi Pers. G. 12-14 42-46 13-15 18-20 Wongratana, 983 purava I 12-14 42-47 14-16 18-19 Wongratana. 980 stenosoma I 12-14 43-48 13-15 17-19 Wongratana, 983 dayi I 13-14 44-49 10-13 14-18 Wongratana, 983 spinidens I-Thai 12-14 44-48 9-11 13-15 Wongratana, 980 setirostris I-China 13-15 32-39 5-6 10-12 Wongratana, 980 Encrasicholina purpurea Hawaii 12-15 14-18 7 1 5-22 23-29 41-44 Miller etal., 1 Wongratana Nelson, 198 979; , 1980; 3 McGOWAN AND BERRY: CLUPEIFORMES Table 24. Continued. 117 Localion Dorsal Anal P2 Gillrakers Vertebrae Upper Lower Reference punclifer IwP 12-16 14-17 7 15-22 23-29 24-25 + 17-20 Miller etal., 1979; Wongratana, 1980; Nelson. 1983 heterolobus IwP 13-15 15-19 20-25 23-29 22-24 + 19-21 Miller etal., 1979; Wongratana, 1980; Nelson, 1983 devisi I-Aust 13-16 17-21 17-18 20-22 21-23 + 19-21 Miller etal.. 1979; Wongratana, 1980; Nelson, 1983 ronquilloi Philipp 15-17 19-22 20-21 28-30 Wongratana, 1980 Pterengraulis alherinoides wcA 12-14 29-35 7 10-12 12-15 43-45 Lycengraulis hatesii wcA 14-16 27-30 7 9-13 12-15 47 Whitehead, 1973; Cervigon, 1966 grossidens wcA 14-16 24-28 7 13-19 17-23 41-48 Whitehead, 1973; Cervigon, 1966 poeyi eP 13-15 22-27 14-18 18-23 43 Whitehead, 1973; Peterson, 1956; Meek and Hildebrand, 1923 Cetengraulis edenlulus wcsA 13-16 21-27 7 45-53 Whitehead, 1973; Meek and Hildebrand, 1923 mysticelus ecP 13-17 18-26 40-58 43-60 39-43 Peterson, 1956; Hildebrandichthys seliger Venz 12 25 Papuengraulis micropinna N. Guinea 5-6 54-56 Lycolhrissa crocodilus China 10-13 47-51 Setipinna tenuifilis papuensis melanochir taty wheeleri phasa brevifilis I-China N. G.-Aust China I-China Burma I I 14-16 14-15 13-15 13-15 14 13-15 13-15 49-59 54-57 48-53 48-58 72-77 69-82 68-75 Heterothrissa breviceps I-China 17-18 59-64 Status not verified: Thrissa grayi Lycengraulis barboun olidus Cetengraulis juruensis Amazon-FW 20-22 Anchoa arenicola Anchoviella hubbsi pallida balboae llisha indica ca. 23 ca. 33 6-7 15-16 25-27 6-7 8-10 10-12 13-17 11 15 7-10 9-12 13-17 18-20 16-18 21-22 15-16 18-19 14-15 17 7-8 11-12 Hildebrand, 1946; Meek and Hildebrand, 1923; Miller and Lea, 1972 Cervigon, 1966; Schultz, 1949 Wongratana, 1980 Wongratana, 1980 Wongratana, 1980 Wongratana, 1980 Wongratana, 1980 Wongratana, 1980 Wongratana, 1983 Wongratana, 1980 Wongratana, 1980 Wongratana, 1980 Nelson, 1970 Nelson, 1970 Nelson, 1970 20 + 20 Nelson, 1984 Nelson, 1970 Nelson, 1970 Nelson, 1970 Nelson, 1970 Nelson, 1970 118 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM from 0.59-0.75 mm in Sardinella jussiem (Bensam, 1970) to 2.5-3.8 mm in Alosa sapidissima (Jones et al., 1978). Most clupeid eggs are 1-2 mm in diameter. All have a segmented yolk. The chorion is not ornamented or sculptured. The peri- vitelline space varies in thickness among species. It may be as large as 45% of the egg diameter (Sardinella zunasi) or as small as 5-10% (Anodontostoma. Opisthoplerus). The egg yolk may shrink relative to the egg diameter when preserved (Bensam, 1967) and the yolk decreases in size during the development of the embryo. Oil globules are present in the eggs of most clupeids. One is often present (e.g.. Sardinella. Harengula, Sardinops); Escualosa thoracata has nine (Delsman, 1932a, described as Clupeoides Hie). The eggs of clupeids which lay demersal adhe- sive eggs (Clupea. Dorosoma, Spratelloides) have a gelatinous covering around the egg. The pelagic egg of Tenualosa ilisha is also covered by a gelatinous sheath. In Dorosoma petenense the adhesive layer is composed of transformed ovarian follicular epithelium, an unusual feature among teleosts (Shelton, 1978). Eggs of anchovies, family Engraididae. range in size from 0.7 mm (Lycengraidis) to 1.75 mm (Slolephorus, long axis). Their shape varies from globular to extremely elliptical. The ratio of the long axis of the ellipse to the short axis has been used to identify anchovy eggs (Peterson, 1956; Phonlor, 1978). Some Slolephorus species have a distinct knob on one end of the egg surrounding the micropyle. A perivitelline space is present but smaller and less noticeable than in clupeid eggs because of the elliptical shape. Oil globules are absent except in the genera Coilia and Setipinna, which have spherical eggs like clupeids, and the Indo-Pacific species of Slolephorus. Fig. 58 illustrates representative eggs of clupeiforms. Yolk-sac larvae are characterized by their size at hatching (2- 5 mm), which is related to yolk size; whether the yolk-sac is rounded or pointed posteriorly, the number and position of oil globules, number of myomeres and pigmentation. Larvae from demersal adhesive eggs may hatch with pigmented eyes (Clupea harengus); those from pelagic eggs hatch with unpigmented eyes. Oil globules may be present in the anterior, ventral, or posterior part of the yolk sac. Multiple oil globules in early embryos coalesce into a single large one before hatching in Seiipinna phasa (John, 1 95 1 a). A spherical yolk sac usually remains spher- ical although shrinking in size during development (Sardinella zunasi), while a yolk sac which is pointed posteriorly may be- come more rounded as yolk is utilized (Coilia sp.). Larval clu- peiforms are slender and elongate with long straight guts. Series of melanophores are variously arranged above and below the gut and along the ventral body wall. Subtle differences in pig- mentation are very useful for identifying co-occuring larval clu- peoids prior to fin development. Larvae of Engraulis mordax. Sardinops sagax. and Etrumeus leres are illustrated for com- parison in Moser (1981). Median dorsal melanophores in clu- peid embryos migrate, reaching their characteristic ventral po- sitions soon after hatching (Orion, 1 953a). In engraulids, pigment cells are presumed to migrate similarly but they don't become pigmented until after hatching. Melanophores are commonly present ventrally just anterior to the pectoral symphysis in small larvae, (e.g., Opislhonema. Harengula, Engraulis, Sardinops, Etrumeus). During development external rows of melanophores become dark streaks and internal melanophores may increase in size and number at first but disappear or become occluded at transformation. A thorough description of pigment devel- opment of laboratory-reared Opislhonema oglinum larvae com- plete with dorsal, lateral, and ventral illustrations is given by Richards et al. (1974). Preanal myomere number is taxonom- ically useful but it does not correspond exactly with precaudal vertebral count in the adult because of changes during trans- formation. Pectoral fin buds and a continuous dorsal-caudal- anal finfold are present at hatching. Fin rays first appear in the caudal fin then in the dorsal, then the anal, next the pelvic, and last the pectoral fin. Ossification of fin rays proceeds in the same order. A full complement of fin rays is not attained until trans- formation, which occurs at approximately 20 mm standard length (e.g., Harengula jaguana. Houde et al., 1974; Opislhonema og- linum Richards et al., 1974). Figs. 59 and 60 illustrate yolk sac larvae of herrings and anchovies. The most useful single character for identifying larval clu- peiforms is total myomere or vertebral number. Pigment pat- terns are useful when vertebral counts overlap. The relative positions of dorsal and anal fins and the length of the gut can be used to separate clupeids from engraulids: clupeids have a longer gut relative to body length and there is a gap between the posterior margin of the dorsal fin and the anterior margin of the anal fin; engraulids have a shorter gut and tend to have the posterior margin of the dorsal over the anterior insertion of the anal fin. The number of myomeres between dorsal and anal fins has been used as a taxonomic character in larvae of certain size classes (Houde and Fore, 1973) and in clupeid adults (Sve- tovidov, 1963). During metamorphosis the position of the gut and the dorsal and anal fins shift forward relative to myomere number. The dorsal insertion moves 10 myomeres forward in Sardinops sagax (Ahlstrom, 1968); it moves eight myomeres in Harengula jaguana (Houde et al., 1974). The migration of the fin takes place at approximately the time when the fin ray number stabilizes. The pelvic fin migrates posleriad in Clupea harengus (Lebour, 1921). Because of these dramatic changes in morphology during development different characters must often be used at different stages to separate species. However some morphometric characters show a small but consistent difference between species at all sizes as between .4losa pseudoharengus and .4. aestivalis (Chambers et al., 1976). Additional care must be taken when using information from laboratory-reared spec- imens to identify field samples. Fin development began at 4 mm in laboratory-reared Opislhonema oglinum. but was not observed in wild-caught larvae less than 7 mm long (Richards et al., 1 974). Shrinkage due to preservation and handling (Thei- lacker, 1980a) also presents problems when comparing devel- opment of larvae based on length. Meristic characters in Clupea Fig. 58. Eggs of Clupeiformes illustrating taxonomic characters: number and size of oil globules, width of perivitelline space, degree of yolk segmentation, shape, size. (A) Chirocemrus nudus. 1.56 mm. Delsman, 1923; (B) Etrumeus leres. 1.35 mm, Ahlstrom and Moser. 1980; (C) Opisthoplerus tardoore, 0.85 mm, Bensam, 1967; (D) Dussumiena. 1.5 mm, Delsman, 1925; (E) .Anodontostoma chacunda. 0.92 mm, Delsman, 1926c; (F) Sardinops melanosticta. 1.60 mm, Mito, 1961; (G) Coilia. 1.04 mm, Delsman. 1932b; (H) Setipinna phasa. 1.10 mm, Jones and Menon, 1950; (I) Anchoa mitchilli. 0.84 x 0.65, Kuntz, 1914b; (J) Engraulis mordax. 1.40 x 0.74, Bolin, 1936a; (K) Slolephorus msulans. 1.92 X 0.69, Delsman, 1931; (L) Slolephorus indicus or commersonii. 1.15 x 0.81, Delsman, 1931. All redrawn by J. Javech. McGOWAN AND BERRY: CLUPEIFORMES ,19 H K 120 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Fig. 59. Yolk-sac larvae of Clupeidae and Chirocentrus illustrating taxonomic characters: number, size, and position of oil globules; shape of yolk sac; degree of segmentation of yolk; preanal myomeres. (A) Sardinella zunasi. 2.1 \ mm, Takita, 1966; (B) Sardmelta :unasi, 4.79 mm, Takita, \9(>(>.(C) Elrumeus teres. AM mm. Mao, \9(i\:(D) llisha elongata. 5.59 mm, Sha and Ruan, \9i\:{E) Dussumieria. 3.17 mm, Delsman, 1925; (F) Chirocentrus mtdus. 3.79 mm, Delsman, 1923. All redrawn by J. Javech. McGOWAN AND BERRY: CLUPEIFORMES 121 Fig. 60. Yolk-sac larvae of Engraulidae illustrating taxonomic characters: oil globules, shape of yolk sac, yolk segmentation, preanal myomeres. (A) EngrauUs japomcus. 3.02 mm, Mito, 1961; (B) Coilia. 2.83 mm, Takita, 1967; (C) Coilia. 2.46 mm, Delsman, 1932b; (D) Slolephorus msularis. 2.19 mm. Delsman, 1931; (E) Thryssa hamiltomi. 2.42 mm, Delsman, 1929a; (F) Cetengraulis mysticetus, 1.99 mm, Simpson, 1959. All redrawn by J. Javech. 122 ONTOGENY AND SYSTEM ATICS OF FISHES- AHLSTROM SYMPOSIUM harengus larvae were shown to be affected by temperature and salinity (Hempel and Blaxter, 1961); morphometric characters in Gikhristella aestuarius adults were found to differ between estuaries with different types of prey items (Blaber et al., 1981). There are several characters which may be useful in system- atics when they are described for more clupeiform species. The melanophores on the caudal fin dorsal and/or ventral to the notochord tip in small larvae have been described for a few species. Harengida jaguana has dorsal melanophores only at first, then both dorsal and ventral (Houde et al., 1974). Opis- thonema oglinum has ventral ones (Richards et al., 1974). Sar- dinella brasiliensis. S. maderensis and S. zunasi have just ven- tral melanophores but Sardine/la rouxi has both. Slight differences in pigmentation over the brain and on the mid-dorsal and mid-ventral postanal body midline have been used to iden- tify scombrid larvae. Small scombrid larvae are otherwise very similar to each other as are clupeoid larvae. The development of free neuromasts and the lateral line has been described for a few species (Blaxter et al., 1983). Development of the swim- bladder and its unique connection with the inner ear should be useful (Hoss and Blaxter, 1982). Ephemeral basihyal teeth were observed on Opisthonema oglinum and Harengula jaguana lar- vae (Richards et al., 1974; Houde et al., 1974). Two patterns of nasal epithelium cells have been observed with scanning elec- tron microscopy (Yamamoto and Ueda, 1 978). Harengula, Sar- dinops and Konosirus had one pattern while Etrumeus (a clu- peid) had the same pattern as Engraulis, an engraulid. Although the eggs and yolk-sac larvae of clupeiforms have many characters of potential systematic importance, the taxo- nomic characters of the older larvae (meristics, fin position, and pigmentation) will tend to be redundant with the same adult characters. However, clupeoids are easily reared in the labo- ratory so direct experimental evaluation of the polarity of adult character states by comparative developmental studies is pos- sible. Relationships The clupeiform fishes are considered a well-defined mono- phyletic group based on their unique otophysic connection, the caudal skeleton, and other characters (Greenwood et al., 1966). The distribution of species within genera, genera within subfam- ilies, and number and taxonomic rank of categories within the group are not agreed upon (Gosline, 1971, 1980; Miller, 1969; Nelson, 1967, 1970, 1973; Whitehead, 1972, 1973). J. S. Nelson (1976) lists the families Chirocentridae, Denticipitidae, Clu- peidae, and Engraulidae. He gives seven subfamilies of herrings (Dussumieriinae, Clupeinae, Pellonulinae, Alosinae, Doroso- matinae, Pristigasterinae, and Congothrissinae) and two subfamilies of anchovies (Engraulinae and Coilinae). Spralel- loides is separated from the Dussumieriinae and given subfamily rank by Whitehead (1972. 1973). Jenkinsia is the western At- lantic spratelloidin. Based on the gill arches Nelson (1967) concluded that the Dussumieriinae (including Spratelloides and Jenkinsia) were the most primitive clupeid family; the Pristigasterinae were also primitive but with distinctive specializations; the Clupeinae were more advanced, but linked to the Dussumieriinae by Clupea and Sprattus; the Alosinae and Dorosomatinae were closely related and perhaps both derived from the Clupeinae; and the Pellonulinae, lacking the specializations of the Alosinae and Dorosomatinae, most resembled the Clupeinae. Expanding his study of gill arches in the Clupeidae to the hyobranchial ap- paratus in the Clupeiformes, Nelson (1970) divided the order into the superfamilies Chirocentroidae, Engrauloidae. Pristi- gasteroidae, and Clupeoidae. The Clupeoidae were suggested to consist of two families: the Clupeidae composed of the Dus- sumieriinae, Pellonulinae, and Alosinae in part; and the Do- rosomatinae composed of the Dorosomatinae plus Hilsa from the Alosinae and Harengida and Sardinella from the Clupeinae. Sardina and Alosa were aligned with Clupea, Polamalosa, and Etrumeus in his tree depicting relationships of representative genera (Nelson, 1970: Fig. 1 1). Whitehead (1972, 1973) acknowledged that radical changes in clupeid classification could be expected but retained the subfamilies Dussumieriinae, Spratelloidinae, Clupeinae, Pel- lonulinae, Alosinae, Dorosomatinae, and Pristigasterinae in his works which were chiefly concerned with the identification of genera and species. The most recent comprehensive work is that of Wongratana (1980) on the Clupeidae and Engraulidae of the Indo-Pacific. He examined over 14,000 specimens and considered many me- ristic and morphological characters including gill rakers, epi- branchial organs, predorsal bones, caudal osteology, circumor- bital bones, gut form, the gas bladder, scale striae, and the patterns of scale distribution on the body. No numerical, cladistic, or phenetic analyses were done. Taxonomic characters were dis- cussed with respect to apparent evolutionary trends and relative importance. Wongratana retained the subfamilies of Whitehead (1972). The Spratelloidinae were diagnosed by a bony process on the 6th and 1 2th principal caudal rays. Spratelloides is also unique among Indo-Pacific clupeids in having a single epural. Jenkinsia, the spratelloidin in the Western Atlantic, also has a single epural (Hollister, 1936). The Alosinae and Dorosomatin- ae were kept separate and the Pristigasterinae were accorded subfamily status although considered quite distinct from the other clupeids. The Dussumieriinae and Pellonulinae were con- sidered the most primitive groups, the Alosinae and Doroso- matinae the most advanced, and the Spratelloidinae and Clu- peinae were considered intermediate. Within the anchovies, the Coiliinae have one epural while the Engraulinae have two {En- graulis) or three (Papuengraulis). The Coiliinae were considered primitive relative to the Engraulinae although specialized in many respects. Wongratana ( 1 980) found that the number of predorsal bones varies from one to thirty in the clupeids and engraulids (Chi- rocenlrus has none). Some engraulids and pellonulins have a gap between the posterior predorsal bone and the first dorsal pterygiophore which he interpreted as evidence that the dorsal fin has migrated posteriad during evolution. It would be inter- esting to compare the patterns of dorsal bones and the anteriad migration of the dorsal fin during larval metamorphosis. The "dorsal scutes" of Clupanodon ihrlssa were found to be the exposed tips of predorsal bones (Wongratana, 1980). The only double-armored herrings known now are Polamalosa and Hy- perlophus in the Pellonulinae, and Elhmidium in the Alosinae. Dorsal scutes are interesting because they occurred in herring- like fossils (Diplomystus, Knightia, and Gasteroclupea) which resemble pristigasterins (Nelson 1970). Because he examined so many species from such a wide area Wongratana (1980) was able to clear up many nomenclatural questions and to correct previous misidentifications which had been based on limited material. He also described 24 new species McGOWAN AND BERRY: CLUPEIFORMES 123 (Wongratana, 1 983) and provided keys to all Indo-Pacific species (Wongratana, 1980). However no direct comparison between his classification and that of Nelson (1967, 1970, 1973) is pos- sible because he only examined Indo-Pacific material while Nel- son included West African and New World material. Evidence from eggs and larvae There are two major problems with using characters of eggs and larvae to criticize classifications based on adult characters. First, the planktonic stages of fishes are exposed to different selective pressures than the adults so they may show patterns of specializations for planktonic life which are not congruent with the distribution of adult character states. Second, relatively few genera of clupeiform fishes have had the eggs or larvae described for even one species in the genus. The first problem must be dealt with the same as any character complex in a group with more than one character complex. More knowledge of the ecology of the larvae in the sea would indentify species with different funtional requirements for their larvae. The second problem may be resolved by using the available evidence in a parsimonious fashion. Eggs and young larvae are similar within genera. Seven species of Sardine/la (Table 25) all have moderately sized clupeid-type eggs with a wide perivitelline space and a single oil globule. The egg described by Takita (1966) and Chang et al. (1981) as that oi Harengida ziinasi is similar. Wongratana ( 1 980) places zunasi in Sardinclla. Within subfamilies there is little apparent consistency in egg morphology among genera. Etruineus has no oil droplet but Dussumieria does. Brevoortia has eggs 1.3 mm or larger with a single oil globule; HHsa kelee has 1.00-1.07 mm eggs with sev- eral small oil droplets. Clupea has demersal adhesive eggs while Sprattus has pelagic eggs with a small perivitelline space. The Indo-Pacific pristigasterin species of Ilisha have large eggs with adhesive coatings and a single large oil globule but Opislhopterus tardoore and the eastern Pacific O. dovii have small eggs with small perivitelline spaces and no oil droplets. The functional significance of egg characters is unknown. Sep- arate lineages within the group which have radiated into several habitats could show parallel adaptations such as oil droplets for buoyancy or nutrition, adhesive coating for retention nearshore or demersally. and egg size as a trade-off between broadcasting and parental investment. Alternatively, different types of eggs within taxonomic categories could also support splitting the category. The anchovy genus Stolephorus contains species with eggs which range from oval with no oil globule to varying degrees of eccentricity with an oil droplet, to unusually shaped eggs with knobs on one end (Delsman, 1931). Nelson (1983) separated Stolephorus into two groups, a Stolephorus group with 1 3 species and an Encrasicholina (new usage) group of 5 species which he considered more closely related to New World anchovies than to the 1 3 Stolephorus species. The three Encrasicholina species whose eggs are known have an oval egg without a knob. One of the three, E. hetcrolobus. was reported by Delsman (1931) to have a small oil droplet and to be relatively more abundant near shore than Stolephorus zolingeri. The other two, E. pur- purcus and E. punctifer (^buccanceri, Strasburg, 1960; =zolin- geri. Delsman, 1931), occur in Hawaii and neither has an egg with an oil droplet. New World anchovies don't have eggs with knobs or oil droplets; therefore, the evidence from eggs supports Nelson's revision and in addition provides some basis for zoo- geographic speculation. Whether the pristigasterins should be given equal rank with the clupeids and engraulids cannot be answered with the avail- able ontogenetic information. There are two very different egg types in the group, small with small perivitelline space and large with gelatinous coating, both of which could be considered spe- cializations. Etrumeus. Jenkmsia. Spratelloides, Clupea. Sprat- tus, and Potamalosa were linked based on a foramen in the fourth epibranchial (Nelson, 1970). Eggs of Spratelloides and Clupea are both demersal and adhesive. The planktonic eggs of Etrumeus and Sprattus both have narrow perivitelline spaces and lack oil globules. Eggs of Potamalosa and Jenkinsia are unknown. Jenkinsia is related to Spratelloides and has demersal larvae (Powles, 1977) so it may have demersal eggs. The de- velopmental osteology of these genera could be studied to de- termine if the shared foramen is phylogenetically homologous. The egg of Anodontostoma, Dorosominae, is similar to eggs of the Alosinae in that it has multiple small oil droplets. Otherwise both the Alosinae and Dorosomatinae contain species with de- mersal adhesive eggs and species with buoyant planktonic eggs. Other suggestions of Nelson (1970) that Sardinclla. Opistho- nema. Harengula. and Herklotsichthys should be placed with the Dorosomatinae and Sardina and Sardinops with the Alo- sinae and then that the Alosinae and Dorosomatinae should be combined leaving just Clupeinae and Dorosomatinae cannot be critically evaluated with existing ontogenetic data. These hy- potheses could be tested by comparing the osteological devel- opment of the characters used by Nelson, augmented by other early life history characters. Relationships of the Clupeiformes Greenwood et al., (1966) placed the Clupeomorpha and Elo- pomorpha together in their Division One but gave serious con- sideration to the possibility that the Clupeomorpha should be recognized as a separate division. Using information on the gut and lower jaw. Nelson (1973) proposed that the Clupeomorpha were distinct from the Elopomorpha but perhaps related to the non-osteoglossomorph teleosts. Gosline (1980) concluded that the clupeiform fishes should be grouped with the elopiform, the salmoniform, gonorynchiform, and ostariophysine fishes; sep- arated on one side from the osteoglossiform fishes and from the iniomous— acanthopterygian teleosts on the other. His conclu- sions were based on five morphological character complexes: the caudal skeleton, the swim bladder-ear connection, the post- cleithrum, the structures associated with pectoral fin movement, and the various types of premaxillary movements and jaw pro- trusion (Gosline, 1980). Gosline (1980) considered the elopomorphs to be an early offshoot from a basal lower teleostean group. He considered the gonorynchiforms and ostariophysines to be more closely related to each other than to the clupeiforms. A clupeiform— osteo- glossiform link has also been mentioned (Greenwood, 1973). J. S. Nelson (1976), who put the superorders Clupeomorpha (Clu- peiformes) and Elopormorpha (Elopiformes, Albuliformes, An- guilliformes) into Division Taeniopaedia, slated succinctly that "the relation of superorders recognized here is poorly known and they are essentially "loose ends." " Lauder and Liem (1983) provisionally follow Nelson (1970) for most groups within the Clupeomorpha but represent the interrelationships of clupeoid lineages as an unresolved polychotomy. Lauder and Liem (1983) 124 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Table 25. Sources of Early Life History Information for Clupeiformes. Reviews and readily available works with superior illustrations are cited rather than original descriptions in some cases. Genus species Eggs Lar- vae Ju- ven- Mor- Mens- lies phology tics Pig- menta- Oste- Fins lion ology Keys or com- pan- sons Wild- Fe- Spawn- Spawn- with Reared caught cun- ing ing others speci- speei- dity region season species mens mens References X X X X X X X X X X X X X X X X X X X X X X X Chirocentnis dorab X X Chirocentrus nudus X X Sardinella zunasi X X Sardinella jussieui XXX Sardinella aurila XXX Sardinella albella X X Sardinella fimbriata X X Sardinella brachysoma X X Sardinella brasiliensis X X Sardinella longiceps X X Sardinella maderensis X Sardinella rouxi X Clupea harengus XXX Clupea pallasi XXX Clupea bentincki X X Spratlus sprattus XXX Sprattus antipodurn X Elrumeus teres X X X X Elrumeus whiteheadi XX XX Dussumieria sp. XX XX Spratelloides delicatulus X X X X X Jenkinsia lamprolaenia X X X X Konosirus punctatus XX XX Anodontostoma chacunda X X X X X Dorosoma pelenense X X X X X Amblygaster leiogasler XX XX Amblygaster sirm X X Escualosa thoracata XX X Opisthonema lihenate X Opisthonema oglinum X X X X X Harengula jaguana X X X X X Harengula peruana X Sardinops sagax caerulea X X Sardinops sagax musica X X Sardinops melanosticta XX X Sardinops ocellata X X X X X Sardina pilchardus X X X X X Lile stolifera X Dorosoma cepedianum X X X X X Hilsakelee XX XX Tenualosa itisha XX XX Alosa sapidissima X X X X X Alosa pseudoharengus X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X XXX XXX X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X XXX XXX X X X X X X X X X X X X Delsman, 1923, 1930b Delsman, 1923, 1930b Takita, 1966; Chang et al., 1981 Bensam, 1970 Jones et al., 1978; Houde and Fore, 1973 Delsman, 1933b Delsman, 1926 Delsman, 1926 Matsuura. 1975 Nair, 1960 Conand, 1978; Conand and Fagetti, 1971 Conand, 1978 Jones et al.. 1978; Fahay, 1983 Wang, 1981 Orcllana and Balbontin, 1983 Saville, 1964 Russell, 1976; Robert- son, 1975a Mito, 1961a Brownell, 1979; OToole and King, 1974 Delsman, 1925 Uchidaet al., 1958; Miller et al., 1979 Powles, 1977 Mito, 1961a Delsman, 1933a Shelton and Stephens, 1980; Jones etal., 1978 Delsman, 1926b John, 1951a Delsman, 1926c, 1934a Peterson, 1956 Richards et al.. 1974; Jones et al., 1978 Houde etal.. 1974; Gorbunova and Zvyagina 1975; Houde and Fore, 1973 Peterson, 1956 Ahlstrom, 1943; Miller, 1952 Santander and de Castillo, 1977; Orellanaand Balbontin, 1983 Mito, 1961a Brownell, 1979; Louw and OToole, 1977 SaviUe, 1964; Russell, 1976 Peterson, 1956 Shelton and Stephens, 1980; Jones etal., 1978; Cooper, 1978 Rao, 1973 Kulkami, 1950 Bainbridge, 1962; Jones etal., 1978 Jones etal., 1978; Chambers et al., 1976 McGOWAN AND BERRY: CLUPEIFORMES 125 Table 25. Continued. Genus species Eggs Ur- vae ven- Mor- Mens- iles phology tics Fins Pig- Fe- menta- Oste- cun- tion ology dity Keys or corn- pan - sons Wild- Spawn- Spawn- with Reared caught ing ing others speci- speci- region season species mens mens References Alosa mediae ns Alosa aestivalis Caspialosa sp. Elhmalosa fimbriala Brevoortia aurea Brevoortia patronus Ethmidium macutata Gilchristella aesluanus Laevisculella dekimpei PeUonula vorax Ilisha elongata Ilisha melasloma IHsha afncana Ilisha furthi Neoopislhopterus tropicus Opisthopterus tardoore Opisthopterus do\i Opisthopterus equatorialis Odontognathus panamensis Anchoa ischana Anchoa panamensis A nchoa curta Anchoa tucida Anchoa naso Anchoa exigua A nchoa arenicola Anchoa marinii Anchoa hepsetus Anchoa mitchilli Anchovia macrolepidota Engraulis japo nicus Engraulis eur\'slole Engraulis anchoita Engraulis inordax Engraulis encrasicolus Engraulis ringens Slolephorus purpureus Stolephorus buccaneeri Stolephorus heterolobus Slolephorus tri Thryssa hamiltonii Thry'ssa sp. Lycengraulis poeyi Lycengraulis gross idens Celengraulis mysticetus Setipinna melanochir Selipmna taty Setipinna phasa Heterothrissa breviceps Coilia sp. Coilia sp. X X X X X X X X X X X X X X XXX XXX X X X X X X XXX X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X X XX X XX X XX X XX X XXX X XX X X X X X X X X X X X X X X X X X X X X X X X X X Jones etal., 1978; Chambers et al., 1976 X X Jones etal., 1978 Pertseva, 1936 X X Bainbndge, 1961 X Conand 1978; de Ciechomski. 1968 X Houde and Fore, 1973 X Orellana and Balbontin, 1983 X Brownell, 1979 X Conand, 1978 Bainbndge. 1962; Conand, 1978 X X Delsman, 1930a; Uchida etal.. 1958 X Delsman, 1930a X Dessier, 1969 Peterson. 1956 Peterson, 1956 X Bensam, 1967 Peterson, 1956 Peterson, 1956 Peterson, 1956 Peterson, 1956 Peterson, 1956 Peterson, 1956 Peterson, 1956 Peterson, 1956 Peterson, 1956 Peterson, 1956 de Ciechomski, 1968 Jones etal., 1978 Jones etal.. 1978 Peterson, 1956 Mito, 1961a; Brownell, 1979; Russell, 1976 Jones etal., 1978 de Ciechomski, 1965 Bolin. 1936a; Ahlstrom, 1965; Ahlstrom, unpublished X D'Ancona, 1931a; Saville, 1964; Marchal, 1966 X X Orellana and Balbontin, 1983; Fischer, 1958b; Einarsson and Rojas de Mendiola, 1963 X Miller etal., 1979 X Delsman, 1931; John, 1951a X Delsman, 1931 X Delsman, 1931; John, 1951a X Delsman, 1929a X John, 1951a X Peterson, 1956 X Phonlor, 1978 X Simpson, 1959 X Delsman, 1932a X Delsman, 1932 X Jones and Menon, 1950 X Delsman, 1932a X Takita, 1967 X Delsman, 1932b X X X X X X X X X X X X X X 126 ONTOGENY AND SYSTEMATICS OF FISHES -AHLSTROM SYMPOSIUM place the clupeomorpha nearer to the next most advanced clade, the Euteleostei, than to the next least advanced clade, the Elo- pomorpha. Evidence from eggs and larvae Relevant ontogenetic evidence concerning the relationships of the Clupeiformes is meager. Elopiform eggs are unknown. Anguilliform eggs resemble clupeid eggs in having perivitelline spaces, segmented yolks, and may have oil droplets. Eel eggs can be much larger than herring eggs: 5.5 mm diameter in A/m- raena. 2.43 mm in an anguillid (Ahlstrom and Moser, 1980). Osteoglossomorphs have pelagic or demersal eggs which may be 0.5-4.0 mm in diameter, may be dark blue, and may have a very wide perivitelline space as in Hiodon (Breder and Rosen, 1966). The coincidence of demersal adhesive eggs in both the osteoglossomorphs and the Dorosomatinae is extremely un- likely to be a shared derived character from a common ancestor. Clupeid and anguillid eggs are considered unspecialized relative to eggs of the higher teleosts (Ahlstrom and Moser, 1980). Very little else may be said. Perhaps electron microscopy will reveal patterns of chorion sculpturing which will be informative. The larvae of Clupeiformes are unspecialized and undergo a fairly uneventful metamorphosis. The migration of the dorsal fin during transformation also occurs in the elopiforms. The larva of Chanos. a primitive gonorynchiform (Fink and Fink, 1981), superficially resembles clupeids or engraulids but appar- ently does not have the same migration of the dorsal fin (Rich- ards, this volume). If the Elopomorpha and the Clupeomorpha share a common ancestor it is possible that the Clupeomorpha retained the un- specialized, rapidly developing larvae while the adults evolved towards a specialized schooling planktivore body plan. The lep- tocephalus found in the elopiforms, albuliforms, and anguilli- forms could have evolved for dispersal or to reduce predation or to take advantage of larval drift the way Angntlla does in the North Atlantic and the way herring do in the North Atlantic with their circuit of migration (Cushing, 1977). The leptoceph- alus could have arisen in the common ancestor of anguilliforms and elopiforms or in parallel, in response to the same selective influence, after the adult eels had begun their divergence from the still unspecialized elopiform fishes. The leptocephalus is considered a specialized character by Forey (1973a), who sug- gested that it arose before the elopid-albulid dichotomy. Trans- forming elopoid leptocephali resemble transforming clupeiform larvae (A/e^a/ops— Harrington, 1958: Plate 1; f/ops— Sato and Yasuda, 1980: Fig. 1; ,4//)j//a-Hildebrand, 1963b: Fig. 23). The egg and larval evidence thus is consistent with a rela- tionship between the Elopomorpha and the Clupeomorpha based on primitive characters but is not helpful in aligning this Di- vision (J. S. Nelson's usage, 1976) closer to any other. Summary and recommendations The eggs and early larval stages of the Clupeiformes provide many taxonomic characters with potential value for testing phy- logenetic hypotheses. Most of the discrete characters, such as number of oil globules, have more than two states and the continuous characters, such as degree of egg eccentricity, have at least a moderate range of values. Although the fraction of species whose eggs and larvae have been described is low and the descriptions are uneven in quality and not distributed uni- formly among taxa, egg and larval characters appear consistent within genera. Within nominal subfamilies they are not consis- tent, but the subfamilies show parallel trends in adult characters and, in addition, the distribution of genera in higher taxa is not yet agreed upon by all workers. Most descriptions of clupeiform larvae have been to enable identification of regional species. Differences between larvae usually involve subtle features of pigmentation or morphome- try, or counts of meristic characters which converge with the meristics of the adult. Phylogenetically significant characters such as ephemeral dentition, osteological development, and the comparative ontogeny of characters used in the taxonomy of the adults are rarely mentioned. Future descriptions of eggs and larvae should address system- atic characters as well as those needed for identification. Eggs and larvae of many species should be redescribed to give com- plete series through metamorphosis. Ontogenetic characters should be used in revisions of the group. Classifications of the Clupeiformes which are based on just a few characters should be tested by comparing the ontogeny of those characters because there are many apparently parallel trends in the group. Addi- tional studies of the physiology and ecology of the eggs and larvae should be done to determine the functional significance of observed characters. It would also be useful to perform quan- titative phenetic and cladistic analyses now of the Clupeiformes for those regions or taxa for which information is already fairly complete. National Marine Fisheries Service, Southeast Fisheries Center, 75 Virginia Beach Drive, Miami, Florida 33149. Ostariophysi: Development and Relationships L. A. FUIMAN OSTARIOPHYSI, as regarded here, include all fishes whose 3 orders, about 55 families, and more than 5,000 species, there- four or five anteriormost vertebrae are modified to form by accounting for over 70% of the world's freshwater fish species, an otophysic connection, the Weberian apparatus (Rosen and Oslariophysans occupy most freshwater habitats worldwide, from Greenwood, 1970). These primarily freshwater fishes comprise torrential Himalayan streams to still tropical lakes, as well as FUIMAN: OSTARIOPHYSI 127 Fig. 6 1 . Egg of Clenolucius hujela ( 1 8 hours poslfertilization) show- ing the membranous pedestal by which the egg attaches to plants. Pho- tograph by H.-J. Franke. coastal marine waters (the latter by a few characids, cyprinids, and aspredinids, as well as all ariid and plotosid catfishes). The presence of a Webenan apparatus has overshadowed the suite of remaining diagnostic characters for the group which includes an axe-shaped endochondral portion of the metapterygoid, an- teriorly bifurcate pelvic girdle, second hypural fused to the com- pound terminal centrum, and elongate olfactory tracts (all de- tailed by Fink and Fink, 1981). Additional characters include a pheromone-mediated alarm reaction and homy dermal pro- jections called unculi (Roberts, 1982b). According to the classification of Fink and Fink (1981), the orders of Ostariophysi (their Otophysi) are: Cypriniformes, Characi formes, and Siluriformes (the latter including Siluroidei and Gymnotoidei). Cypriniforms (with over 1,800 species in 5 families) uniquely share peculiarities of the following: kineth- moid bone, palatine-mesopterygoid articulation, fifth cerato- branchial, and lateral process of the second vertebral centrum. They lack jaw teeth and an adipose fin. They are found in North America, Eurasia, and Africa. Characiforms (comprising at least 1,000 species in 14 families) are characterized by multicuspid teeth, a prootic foramen, dorsomedial opening in the posttem- poral fossa, enlarged lagenar capsule, and a gap between the compound terminal centrum and hypural 1. They occur in Af- rica, South America, and southernmost North America. Silu- roids (with about 2,000 species in 3 1 families) are distributed nearly worldwide. Although quite diverse morphologically, they commonly lack scales and several bones (including the sym- plectic, subopercle, and separate parietals). They show consid- erable fusion of portions of the first five vertebrae and pectoral and dorsal fin rays. The electrogenic gymnotoids are character- ized by an extremely long anal fin and substantial reductions or losses, such as the loss of dorsal and pelvic fins, and palatine and ectopterygoid bones. They are confined to South America and southernmost North America. Development Knowledge of the early life history stages of ostariophysans is rather spotty and concentrated on fishes from a few geographic regions. Major descriptive works cover portions of the Soviet Union (Kryzhanovskii, 1949; Kryzhanovskii et al., 1951; Kob- litskaia, 1981), Japan (Okada, 1960; Nakamura, 1969), and the United States (Jones et al., 1978; Snyder, 1981; Auer, 1982; Fuiman et al., 1983). Most of these works concentrate on cy- priniforms. Additional descriptive data are available as indi- vidual papers on Indian major carps (Cyprinidae) and Indian siluroids (reviewed by Jhingran, 1975). African and South American ostariophysan eggs and larvae remain little known. Of the six families of cypriniforms, nothing is known of the eggs and larvae of the families with fewest species, Gyrinocheili- dae and Psilorhynchidae. Catostomids are known well. Cypri- nids, cobitids, and homalopterids are known to a lesser degree. Scattered notes are available for nine characiform families but only a few descriptions of ontogeny exist. Brief descriptions of larvae of representatives from seven families of siluroids are available, and notes on eight additional families exist. Photo- graphs of larvae of two gymnotoids. Eigenmannia virescens anA Aptewnotus leptorhynchus are published (Kirschbaum and Westby, 1975; Kirschbaum and Denizot, 1975; Kirschbaum, 1984) but without morphological descriptions. Most informa- tion on ostariophysan larvae deals with external morphology. Osteological studies are few (Bertmar, 1959; Hoedeman, 1960a- d). Eggs Ostariophysan eggs vary considerably in their morphology and the habitat they occupy. Most are spherical, demersal, 1 to 5 mm in diameter, with pale yellow, somewhat granular yolk Table 26. Larval Characters of Major Groups of Ostariophysans. Cypnniformes Characiformes Siluroidei Gymnotoidei Size at hatching (mm XL) Yolk-sac shape Gap between yolk sac and anus Barbels: Presence Timing of development Size at finfold absorption (mm TL) 2-10 pyriform or tubular absent present or absent late or early 15-25 2-5 elliptical present absent 10-20 3-8 elliptical present present early 11-23 elliptical absent absent 15 128 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM A FUIMAN: OSTARIOPHYSI 129 Ssur Fig. 62. Representative cypriniform larvae. (A-C) Cyprinidae: (A) Tribolodon hakonensis (UMMZ 212151) 9.2 mm TL; (B) Semotilus alromaculatus 8.6 mm TL; (C) Barbiis { = Capoela) tilteya (UMMZ 212148) 6.0 mm TL; (D, E) Cobitidae; (D) Misgurnus fossiHs 6.9 mm TL, after Kryzhanovskii (1949); (E) Acanthophthalmus cf kuhni 4.0 mm TL (specimen from S. S. Boggs). lacking oil globules. Eggs may be strongly adhesive (e.g., Cy- priniformes: Nemacheilus [=Barbatula] torn [Kobayasi and Moriyana, 1957]; Characiformes: Gymnocon'mhus tenictzi [pers. obs.]; Siluriformes: Loricana calaphracta [pers. obs.]), nonad- hesive (e.g., Cypriniformes: Clenopharyngodon idclla [Inaba et al., 1957]; Siluriformes: Tandanm landanus [Lake. 1967]), or weakly adhesive (e.g., Cypriniformes: Catoslomus commersoni [pers. obs.]; Characiformes: Scrrasalmm nattercn [pers. obs.]; Siluriformes: Baganus hagarius [David, 1961]). Adhesive fila- ments or other apparent modifications of the egg surface are almost entirely unknown. Representatives of outgroups (Gonoi^nchiformes, Clupeo- morpha, "Salmoniformes," and Osteoglossomorpha) share the spherical egg with yellow, granular or segmented yolk. Their eggs are pelagic or demersal, usually 1 .0 to 1.3 mm in diameter. adhesive (in Osmerus) or nonadhesive (in Chanos. Alosa. and Hiodon). without oil globules (Chanos) or with one to several (in Alosa and Osmerus). Exceptions to this characterization of ostariophysan eggs exist. Among cypriniforms, the cyprinid subfamily Acheilognathinae (Gosline, 1978) exhibits elliptical to pyriform eggs which are deposited in the mantle cavity of a bivalve mollusc (Kryzhan- ovskii et al., 1951; Nakamura, 1969; Makeeva, 1976). Their irregular shape may be the important mechanism preventing the eggs from being expelled. Some cyprinid eggs are pelagic (e.g., Hypophthalmichthys molitrix [Nakamura, 1969; Koblit- skaia, 1981]) and have a larger diameter (ca. 5 to 6 mm) due to the considerable perivitelline space. Only one ostariophysan, the cypriniform Cobitis biwae, was reported to have 12 to 13 small oil globules in the yolk (Okada and Seiishi, 1938; Okada, Fig. 63. Representative cypriniform larvae (continued). Catostomidae: Hypentetium etowanum (upper) 13.1 mm and (lower) 15.0 mm TL. 130 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM /JS-^y Fig. 64. Representative characiform larvae. Serrasalmidae: Serrasalmus nattereri (UMMZ 211677) 8.2 mm TL (upper). Characidae: Hy- phessobrycon cf. callistus (UMMZ 21 1676) 6.6 mm TL (lower). 1960), but this is in doubt (N. Komada, pers. comm.) and has not since been confirmed. Characiform eggs are poorly known; most information is from the aquarium hobby literature. Known characid (sensu Gery, 1977) eggs are small (0.8 to 1.2 mm). However other families have eggs between 2 and 4 mm (e.g., Alestidae, Anostomidae, Curimatidae, Hepsetidae, Serrasalmidae). Apparently most species have eggs that adhere to plants. Franke (1981) described adhesive threads (gallertigen Klebfdden) on the surface of the egg of Ctenolucius hujeta (Ctenoluciidae) and noted that this was the mechanism by which they attached to plants. My ex- amination of eggs supplied by Dr. Franke found the adhesive structure to be a membranous pedestal rather than adhesive threads (Fig. 6 1 ). This is the only known chorionic modification of ostariophysans. Most siluroids have demersal, medium sized eggs ( 1 to 4 mm). Some are tended by one or both parents [e.g., Clarias batrachus (Mookerjee, 1946; Mookerjee and Mazumdar, 1950), Ictalurus punctatus (Tin, 1982c)]; others are not given parental care [e.g., Clarias gariepinus (HoW, 1968; Bruton, 1979), Pangasius sutchi (Varikul and Boonsom, 1969)]. The eggs are typically spherical; however, Clarias eggs are often slightly elliptical (Mookerjee, 1946; Greenwood, 1955; Bruton, 1979). Some callichthyids de- posit small eggs (ca. 1.0 mm) in a foam nest on the surface of still waters (Kryzhanovskii, 1949). Parents in several families carry their eggs. Some loricariids (e.g., Loricaria spp.) carry a mass of eggs by means of fleshy appendages of the lower lip. Aspredo laevis eggs apparently are attached by vascularized stalks to the venter of the female (Wyman, 1859). Finally, ariids are oral incubators with perhaps the largest eggs of all oviparous teleosts (10 to 25 mm) (Chidambaram, 1942; Gudger, 1912, 1916, 1918; and other authors). Although yolk is usually yellow to slightly orange or brown, several species have unmistakably green yolk [e.g., Bagarius bagarius (David, 1961), Clarias ba- trachus (Mookerjee, 1946; Mookerjee and Mazumdar, 1950), Heteropneustes fossilis (Pal and Khan, 1969), Loricariichthys sp. (Taylor, 1983), Phractura ansorgei (Foersch, 1966)]. At least one siluroid, the silurid Ompok bimaculatus, has reddish brown yolk (Chaudhuri, 1962). A few species have a jelly-like coat surrounding the chorion [e.g., Bagarius bagarius (David, 1961), Parasilurus asotus (Kryzhanovskii et al., 1951), Phractura an- sorgei (Foersch, 1966), Trachycorystes insignis (Burgess, 1982)]. Larvae Most ostariophysans hatch in an altricial state at about the time when pectoral buds form, but before the head becomes free from the yolk sac and retinal pigment develops, although there is variability in the exact stage. The yolk sac is usually large and cumbersome, enforcing a stationary existence during the first days, either on the substrate (most commonly) or at- tached to plants by means of a cephalic adhesive mechanism (found in most characiforms and a few cyprinids, but structur- ally diflTerent in these groups). Caudal fin rays diflierentiate first, followed by nearly simultaneous formation of dorsal and anal fin rays. Pectoral and pelvic fin rays develop near the end of the larval period. The gonorynchiform Chanos hatches at about the same stage of development as ostariophysans, but Atosa and Osmerus hatch somewhat later (i.e., pectoral buds and retinal pigment are clearly developed). These outgroups generally have pelagic larvae at hatching. Fin rays in Chanos develop in the FUIMAN: OSTARIOPHYSI 131 same order as described above, but the sequence differs for Alosa and again for Osmerus. Within Ostariophysi, cypriniform larvae (Figs. 62, 63) are largest at hatching (Table 26), the largest sizes represented most- ly by catostomids. The pyriform yolk sac extends from below the head posteriorly to the anus (Fig. 62a). Barbels, when pres- ent, develop very late in Cyprinidae but early in Cobitoidea (sensit Sawada, 1982). Cyprinids display considerable variation in the elaboration of the larval circulatory system. Temporary networks of blood vessels invade portions of the finfolds and the surface of the yolk sac in a variety of patterns to form the larval respiratory system (Kryzhanovskii, 1947). Cobitoideans usually have greatly expanded finfolds, especially those of the pectoral buds. Pronounced external gill filaments are known in the cobitine genera Coto/5 (Kryzhanovskii, 1949;Okada, 1960; Sterba, 1962), Lepidocephaliis (Bhimachar and David. 1945), and A/;5^r«wi (Kryzhanovskii, 1949; Okada, 1960), but not in the non-cobitine cobitoidean genera Botta. Lefua, or Nemach- eilus, nor in other ostariophysans. Cyprinids with cephalic ad- hesive glands include: Ahramis brama (Penaz and Gajdusek, 1979); Brachydanio rerio (Frank. 1978); Cypri niis carpio (Hoda and Tsukahara, 1971); Danio malabancus (Jones, 1938); and Notemigonuscrysoleucas (Snyder tXa\., \911\ Loosetal., 1979). In characiforms, the yolk sac is short and rounded, not ex- tending to the anus posteriorly (Fig. 64). Most known characids (sensii stricto) and a hepsetid (Bertmar, 1959; Budgett, 1902. 1 903), erythrinid (de Azevedo and Gomes, 1 942), and curimatid (de Azevedo et al., 1938) have a temporary larval cephalic ad- hesive organ (more distinct than the apparent glandular mech- anism in cyprinids). Those without such an organ mclude: Ser- rasalmus nattereri (pers. obs.), Metynnis maciilatiis (Azuma, 1982), and Brycinus longipinnis (Frank, 1972). The adipose fin appears to develop de novo toward the end of the larval period, not as a remnant of the median finfold. However, the small size of the adipose fin and lack of specimens, photographs, illustra- tions, and descnptions of late larval characiforms prevents ver- ification of this inference. Although few species are known as larvae, Siluroidei may contain the greatest diversity of larval characters among Ostar- iophysi (Fig. 65). Most siluroids hatch as altricial larvae with a physiognomy similar to that of characiforms. Ictalurids are more precocial and lack a postlarval (sensu Hubbs, 1 943) phase. Ariids (Gudger, 1918; Ward, 1957) and some loricariids (Lopez and Machado, 1975; Machado and Lopez, 1975) hatch in a highly precocial state, resembling the adult in many aspects of external morphology but retaining a large yolk sac (Fig. 65C). In most families, barbels are usually present at hatching or soon there- after (Fig. 65a). Cephalic adhesive organs are usually absent, but at least one loricariid (Ancistrus sp.) possesses these (Franke. 1979). Clarias gariepinus (=C. mossambicus) and Ompok bi- maculatus have an adhesive organ on the venter of the yolk sac (Greenwood, 1955, 1956; Chaudhuri, 1962; Holl, 1968;Bruton, 1979). The adipose fin is clearly a remnant of the median fin- fold, as in "'salmoniforms." Larvae of a single gymnotoid, Ei- genmannia virescens. are known (Fig. 65D, E; Table 26; Kirsch- baum and Balon, in prep.). Relationships The Ostariophysi are thought to be the sister group of the Gonorynchiformes (Greenwood et al., 1966; Rosen and Green- wood, 1970; Gosline, 1971; Fink and Fink, 1981). The next closest relatives are Clupeiformes (Gosline, 1971) or "Salmon- iformes" (Greenwood et al., 1966; Fink and Weitzman, 1982). All concepts of Ostariophysi (those with a Weberian appa- ratus) recognize four major groupings, "cyprinoids," "chara- coids," "gymnotoids," and "siluroids." The traditional view of relationships holds that "characoids" are the ancestral stock, giving rise to the remaining lineages, with "gymnotoids" being modified "characoids," and "cyprinoids" being the closest rel- atives of the "characoids" plus "gymnotoids." Fink and Fink (1981) gave a detailed history of the classification schemes for the Ostariophysi and their relatives as an introduction to their work on the subject, which is the only attempt to reconstruct the phylogeny on the basis of a large set of data ( 1 27 characters). Their proposed cladistic phylogeny differs significantly from the traditional one by aligning "gymnotoids" with "siluroids" as the Siluriformes (Fig. 66). Developmental characters in systematics Few attempts have been made to apply developmental char- acters to the systematics of ostariophysans. Kryzhanovskii (1947) grouped cyprinids into four subfamilies according to details of the larval respiratory system. He also included characters re- lating to reproductive guild (later elaborated in Kryzhanovskii, 1948), original (ontogenetically) position of the mouth, and rel- ative size of the pectoral buds. He supported these subfamilial designations with experimental results on the morphology and viability of larvae produced by artificial hybridizations within and among the proposed subfamilies. Nakamura (1969) dealt with the cyprinids of Japan. In his English summary, he stated that currently proposed closely re- lated forms (meaning genera, species, and subspecies) have sim- ilar life history characteristics. He noted a few exceptions, such as similar (as adults) species oi Moroco whose early larvae differ morphologically and ecologically. In contrast, he noted that the eggs and early larvae of Ctenopharyngodon idella and Hypoph- thalmichthys molitri.x were very similar although the species were placed in different subfamilies. He used differences in egg and larval morphology to support the previously uncertain sep- aration of the genera Squalidus and Gnathopogon. In a similar survey. Loos and Fuiman (1978) attempted to characterize the subgenera of the New World cyprinid genus Notropis in terms of their egg and larval morphology. However, they found substantial variability within the established sub- genera and were unable to characterize them precisely. Each of these attempts to apply developmental characters to systematics was concerned only with establishing group mem- bership and not with determining relationships among the groups. Further, none of the work was based on a large data set nor was it approached in a rigorous manner. The difficulties encountered by Nakamura ( 1 969), and especially by Loos and Fuiman (1978), probably were due to the apparently convergent ecomorpho- types expressed by unrelated taxa. The low taxonomic level investigated, combined with the morphological similarity im- plied by von Baer's law, probably accounted for much of the remaining difficulty in detecting consistent differences among taxa. Fink and Fink's (1981) classification is based largely on os- teological characters. The great size and diversity of Ostario- physi make a detailed study of developmental osteology and concomitant investigations of bone homologies impractical at this time. Yet, available information permits a preliminary eval- 132 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM D FUIMAN: OSTARIOPHYSI 133 Fig. 65. Representative siluriform larvae. (A-B) Clariidae: Clanas gariepinus (British Museum of Natural History, uncataloged) (A) 6.6 mm and (B) 8.4 mm TL; (C) Loricariidae: Ancistrus spinosus (UMMZ 212152) 8.3 mm TL; (D-E) Rhamphichthyidae; Eigenmannia virescens (D) 5.0 and (E) 8.1 mm TL. uation of relationships based on developmental characters. The following analysis attempts to evaluate the contribution of se- lected developmental characters to ostariophysan systematics by constructing an independent assessment of phylogeny based on developmental characters. That the assessment should be independent was attested by Moser and Ahlstrom (1974): "we are increasingly impressed with the functional independence of larval and adult characters. It is apparent that the world of the larvae and the world of the adults are two quite separate evo- lutionary theaters." Representative ontogenetic series of all families of ostario- physans are nearly impossible to obtain because of the large size and wide geographic distribution of the group and the dearth of ichthyologists studying larvae. Consequently, the analyses employed here were based on specimens generated from labo- ratory breeding experiments, wild-caught material, and data published in apparently accurate accounts of ontogeny. Species used in the analyses included four outgroups to the Ostariophysi (Gonorynchiformes, Clupeomorpha, "Salmoniformes," and Osteoglossomorpha), all characiforms and siluriforms with suf- ficient morphometric and developmental data for analysis, and a sample of five species from the most primitive cypriniform family, Cyprinidae. These cyprinid species possess different combinations of larval characters (determined by their location on a Wagner tree generated for 33 larval cyprinids [Fuiman, 1983a]). Although not used directly, incomplete data on ap- proximately 85 additional non-cyprinid ostariophysans provid- ed corroborative information. Species included in the analysis of relationships and their sources are listed below. Initials denote specimens borrowed from, or information provided by: Florida State Board of Con- servation (FSBC), University of Michigan Museum of Zoology (UMMZ), or Frank Kirschbaum (FK). OsTEOGLOSSiFORMEs: Hiodofi tergisus [Snyder and Douglas (1978); Wallus (1981, pers. comm.)]. Salmoniformes: Osmerus mordax [Cooper ( 1 978); Tin ( 1 982b)]. Clupeiformes: Alosa pseudoharengus [Jones et al. (1978); Tin (1982a)]. Gonorynchiformes: Chanos chanos [Chaudhuri et al. (1978); Liaoet al. (1979); Miller et al. (1979)]. Cypriniformes: Cyprinidae— Cvpn>!Wicarp/o [UMMZ 21 1678; Hoda and Tsukahara (1971); Nakamura (1969); Okada (I960)]; Leiiciscus cephaliis [Cemy (1977); Kryzhanovskii (1949); Penaz (1968); Prokes and Penaz (1980)]; Opsan- ichthys unciroslris [Kryzhanovskii et al. (1951); Makeeva and Ryabov (1973); Nakamura (1951, 1969)]; Parabramts pekmensis [Institute of Hydrobiology (1976); Kryzhanov- skii et al. (1951)]; Squalidus gracilis [Nakamura (1969)]. Characiformes: Alestidae— .-l/eirw haremose [Durand and Loubens (1971 )]. Erythrinidae— //op/Zaj^ malabaricus [FSBC 8962, 8963, 9593; de Azevedo and Gomes (1942); Hensley (1976); Moreira ( 1 920); von Ihering et al. ( 1 928)]. Charac- idae— Hyphessobrycon cf. callistiis [UMMZ 21 1676], Ser- r&sdAmiddie— Serrasalmus nattereri [UMMZ 21 1677; Azu- ma(1975)]. Siluriformes: Siluroidei: Ba.gn6.aQ — Mystus seenghala [Saigal and Motwani (1962)]; Rita rila [Karamchandani and Mot- Cypriniformes Characiformes Siluroidei Gymnotoidei Fig. 66. Cladogram of ostariophysan relationships derived from adult characters by Fink and Fink (1981). Stem lengths imply no special significance. 134 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Squalidus CYPRINIFORMES Ictalurus 21 Eigenmannia *28 SILURIFORMES CHARACIFORMES Fig. 67. Wagner tree of ostariophysan phylogeny based on larval characters. Stem lengths are proportional to the number of character-state changes on a given stem. wani (1955)]. Clariidae— Ctor/a^ batrachus [UMMZ 1 86690. 209039; Devaraj et al. (1972); Mookerjee (1946); Mook- erjee and Mazumdar (1950)]. Ictaluridae— /rta/Mr«5 neb- ulosus [Armstrong ( 1 962); Tin ( 1 982c)]. Pangasiidae-Paw- gasius sutchi [Varikul and Boonsom (1969)]. Sisoiidae— Bagarius bagarius [David (1961)]. Gymnotoidei; Rham- ph'ichVnyiddie— Eigenmannia virescens [FK, Kirschbaum and Westby (1975)]. Phylogenetic methods The phylogenetic reconstruction based on developmental characters was generated by the cladistic Wagner tree method (described by Kluge and Farris. 1969; Farris. 1970; Lundberg, 1972; and Jensen, 1981). Characters were chosen by virtue of their availability in published accounts. Nearly all were recorded as continuous measures, but individual modes with their neigh- boring values and disjunct portions of distributions separated Table 27. Ranges of Values for Coded Character States of 16 Ostariophysans. Character numbers correspond to those given in the text. Primitive states are given in boldface type. \*V\z\rz\(^\PT Character slate number a b c d e f 1 2.27-3.42 3.74-3.74 4.74-4.90 2 3.93-5.06 5.65-6.28 7.06-7.40 8.86-8.86 3 1.35-1.53 1.92-2.58 2.98-4.03 4 1.19-1.19 2.09-2.09 2.46-3.39 3.78-5.14 5 0.97-1.61 2.18-2.18 6 2.06-2.46 2.62-2.86 3.25-3.31 7 0.13-0.28 0.36-0.52 8 0.14-0.28 0.42-0.58 0.71-0.82 9 0.72-1.00 1.29-1.29 10 1.26-1.68 1.97-2.11 11 -1.22—0.94 -0.67—0.23 0.06-0.06 12 0.70-1.04 1.15-1.40 13 0.22-0.39 0.54-0.84 14 15.3-19.0 21.6-22.5 25.0-26.3 28.5-30.3 32.7-32.7 15 8.0-8.0 12.0-20.0 25.5-25.6 29.0-29.0 38.7-38.7 45.9-45.9 16 0.28-0.29 0.39-0.44 0.52-0.70 0.81-0.81 17 0.35-0.35 0.52-0.55 0.70-O.96 1.06-1.10 18 0.22-0.38 0.42-0.56 19 0.0-0.05 0.12-0.28 0.44-0.44 20 0.0-0.03 0.07-0.15 0.22-0.33 21 0-0 1-1 22 0-0 1-1 23 0-0 1-1 24 0-0 1-1 FUIMAN: OSTARIOPHYSI 135 Table 28. Character-State Changes on Stems Leading to Hypothetical Ancestors (Nodes) and Terminal Taxa on the Wagner Tree of Ostariophysi. Numbered character states correspond to those given in Table 27 , Uniquely derived, unreversed character states are given in boldface type. Reversed characters are noted by (r). Node numbers correspond to those given in Fig. 67. Node Characler state 1 8c, lib, 12b, 14b, 20b, 20c 24b 2 6b 3 14d, 15b, 18b 4 13a, 21a 5 18a(r), 20b(r) 6 16b 7 17b 8 14c, 20a(r), 23b 9 6b, 19a. 24a(r) 10 3b, lla(r), I2a(r). 14b 11 lla(r), 22b 12 6b, 6c 13 3b, 4c, 12b(r), 14a, 14b, 14c 14 3a, 4b, 18a(r), 20b(r) 15 10b, lib, 24a(r) 16 12b. 17b. 14b(r) 17 2b, 6b, 7a. 8b(r) 18 8a, 16b. 20b{r) 19 12b, 16a, 18a(r), 20a(r) 20 2a(r), 6a(r). lOa(r) 2 1 5b, 6c, 7b(r), 8b(r), 9b, 1 1 c, 1 5c, 17a, 1 7b. 1 9a 22 lOa(r), 15c, 15d 23 14b(r) 24 lb, 2c, I7d 25 6a(r), 1 la(r), 15c(r), 16d, 19c 26 Ic, 15e, 19a, 20b(r). 22a 27 2d, 8a, 14b(r), 14c(r), 17b. 17c(r). 18a(r) 28 4a, 4b, 6c, 10b, 15f by measurable gaps were coded individually. Characters were polarized by outgroup comparison (Table 27). The evolutional^ transformation series for each continuous, multiple state char- acter was assumed to be linear (i.e.. with one or two adjacent states for a given state). Consequently, a character coded with n states had n - 1 different changes from one state to another, disregarding the direction of change. These transitions were termed "two-state factors." All two-state factors and their states for each species were generated by the FACTOR computing program (Estabrook et al., 1976). The output from this program included an input file for the WAGNER 78 computing program which was used to construct Wagner trees. The data deck was resequenced and a new Wagner tree generated several times in order to identify the shortest (most parsimonious) tree (Jensen, 1981). Characters Morphomelhc characters.— To develop morphometric charac- ters for phylogenetic analysis, the following lengths were mea- sured along the longitudinal axis of the fish; total length, preanal length, head length, and eye diameter. Two vertical measure- ments, head depth and body depth at anus, were meant to rep- resent size and shape in the dorso-ventral direction. All mea- surements were defined by Fuiman (1979). They were made reasonably independent of one another by subtracting preanal length from total length to yield peduncle length, and head length from preanal length to yield tnank length. Peduncle length, trunk length, head length, eye diameter, body depth, and head depth comprised the basic morphometric characters. (0 o a CO 0) E 3 3- 2- 1- \ZZ} Cypriniformes I I Characiformes ^H Siluroidei rXI Gymnotoidei 0.25 0.45 0.65 0.85 Yolk-Sac Shape (depth/length) Fig. 68. Frequency distribution of yolk-sac shape for recently hatched ostariophysan species. Body dimensions of larvae are strongly influenced by allom- etry (Fuiman. 1983b). Such measures cannot be expressed as simple proportions, because the proportions are not constant within a species throughout the larval period. The effect of size on shape must be eliminated in comparisons of shape. Further, any single measure which accounts for size in one taxon may be an inappropriate measure of size in a distantly related taxon. Within-group principal component analysis can be used to ex- tract a size component, PCI (Humphries et al., 1981), that is a linear combination of several variables, each containing infor- mation on size and shape. Thus, PC 1 includes more information on size than any single measure and is a better comparison across taxa. Univariate and multivariate methods of allometry relate dis- tance measures log-linearly (Huxley, 1932; Jolicoer, 1963). Thus, a within-species principal component analysis of the logarithms of the six basic morphometric characters, based on the covari- ance matrix, was performed to extract the size component (PCI ). The extreme PCI scores for all taxa were compared and two values (0.00 and 0.60), one near each end of the larval period, were chosen as standard sizes for comparing morphometry. The six morphometric measures were reconstructed for each of these sizes by means of the regressions of the logarithm of the char- acter on PC 1 . By selecting two sizes to compare, the phylogenetic analysis included information on changing shape (allometry) as well as static shape. The final 1 2 character values were recorded as predicted lengths (in mm) for each morphometric measure at each of 2 standard sizes. However, body depth at the anus contained no discontinuous, phyletic variability. The final mor- phometric characters were; (Characters 1 and 2) Peduncle length (smaller and larger standard size, respectively), (3 and 4) Trunk length, (5 and 6) Head length, (7 and 8) Eye length, (9 and 10) Head depth. Three additional morphometric characters were 136 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM 0.9 E E c O) c (V -J 0) o c 3 T3 (U Q. CI3 Cypriniformes I 1 Characiformes fTTTTI Siluroldel — - Gymnotoidel — I 1 1 1 1 1 1 — 0.0 0.2 0.4 0.6 Size (PCI score) Fig. 69. Morphometric characters important in defining major groups of ostariophysan larvae. Shaded areas and individual Imes enclose all regression-predicted values at two standard sizes (0.0 and 0.6) of a given taxon. included: (11) Size at hatching (PCI score at total length for hatching, based on the regression of PCI on the logarithm of total length), (12) Size at complete finfold absorption (PCI score at total length for complete finfold absorption, based on the regression of PCI on the logarithm of total length), (13) Yolk- sac shape (ratio of the greatest vertical length [depth] of the yolk sac to its greatest horizontal length in recently hatched individ- uals). Meristic characters.— These include: ( 1 4) Preanal myomeres (all myomeres at least partly anterior to a vertical line projected from the anus, including an occipital segment) and (15) Postanal myomeres (all myomeres entirely posterior to a vertical line projected from the anus, including a urostylar segment). Missing myomere data for Hoplias malabancus were taken from vertebral counts made from radiographs of adults (UMMZ 66435). The one-to-one ontogenetic relationship of myosepta to neural spines in monospondylous fishes (Lauder, 1980) per- mitted estimation of myomere number from vertebral number only by inclusion of myomeres for an occipital segment, a uro- stylar segment, and the four (five in siluroids) obscured We- berian vertebrae (Fuiman, 1982a). Ontogenetic characters. —Size, rather than chronological age, is most closely related to development (Gerking and Rausch, 1 979). Thus, total length at the onset of selected developmental events was recorded. To compare these sizes among species with dif- fering initial lengths and ranges of lengths for the larval period. the logarithm of the hatching length was subtracted from the logarithm of the length at a given event. This difference was divided by the difference of the logarithms of length at complete finfold absorption and at hatching (the criteria used here to delimit the larval period). The resultant character was the per- centage of the larval period that occurred prior to the event, an estimate of relative timing of the event. When characters were present at hatching or did not develop until after complete fin- fold absorption they were coded as 0.00 or 1.10, respectively. The following events were recorded: (16) Anal fin rays (first distinct ray), ( 1 7) All median fin rays (all median fin rays present, finfolds may persist, fin margins may be incomplete), (18) Yolk absorption (complete absorption of yolk), (19) Head straight (head free from yolk sac and not deflected downward), (20) Eye pigment (first uniform pigmentation of retina). Presence/absence characters— Presence (coded as I ) or absence (0) of the following structures at any time during the larval period was recorded: (21) Jaw teeth (teeth on the premaxilla, maxilla, or dentary), (22) Adipose fin, (23) Caudal spot (con- gregation of melanophores at the base of the caudal fin forming a distinct spot), (24) Lateral stripe (melanophores on the mid- lateral myoseptum forming a continuous, longitudinal stripe). Phylogenetic results The Wagner tree (Fig. 67, Table 28) contains 101 steps for the 46 two-state factors ("characters"). Members of each major Table 29. Distribution Statistics of Preanal. Postanal, and Total Myomere Number for Ostariophysan Larvae. Values are based on means for each species. Preanal myomeres Postanal myomeres ' Including Cyprinidae, Calostomidae, and Cobiloidea. Total myomeres Taxon species Mean Extremes Mean Extremes Mean Extremes Cypriniformes' 52 29.4 18.5-38.8 12.4 7.0-18.8 41.7 32.8-50.7 Characiformes 4 25.5 17.8-32.7 15.2 8.0-20.0 42.0 36.1-50.0 Siluroidei 6 19.7 15.3-26.3 25.9 16.5-38.7 45.4 33.0-65.0 Gymnotoidei 1 17.3 45.9 63.2 FUIMAN: OSTARIOPHYSI 137 4-1 3- co .2 '5 a CO "?; 2-\ E 3 r//J Cyprinidae I I Cobitoidea ^B Siluroidei ^ ^. i 1 r 0.0 0.4 0.8 1.2 Barbel Formation (onset as percentage of larval period) Fig. 70. Frequency dislnbution of relative timing of" barbel forma- tion in ostanophysan species. Cyprinids are represented by 1 barbelled species, not all of which are discussed in the text. taxon (Cypriniformes. Characiformes, Siluroidei) are placed near one another, but larval characters are insufficient to demonstrate the monophyly of characiforms or siluroids. The largest number of primitive characters is found in Hoplias (Characiformes), but the cypriniform lineage differs from Hoplias by only three char- acter state changes (node 3). As suggested by Fink and Fink (1981), the gymnotoids are most closely related to siluroids (node 26). The cyprinifoim lineage (node 4) is united by two unreversed synapomorphies: an elongate yolk sac (Figs. 62A and 68) and the absence of jaw teeth. Cypriniforms and characiforms uniquely share large eyes at the larger standard size (PCI = 0.6; Fig. 69). This character reverses to a plesiomorphous condition for the siluriform lineage. Synapomorphies of siluriforms include a long peduncle at the larger size (Fig. 69) (a unique state for the group, except for a single reversal in Bagarius), short head at the larger size (highly homoplasious), and small eyes at the smaller size (PCI = 0.0; Fig. 69) (unique except for a reversal in Ictalurus). The gymnotoid, Eigenmannia (node 28), expresses six auta- pomorphies, two unique and two occurring in only one other place on the tree. The uniquely derived conditions are a short trunk at the larger size (Fig. 69) and numerous postanal myo- meres (Table 29). Several morphometric characters make valuable contribu- tions to the phylogenetic reconstruction. The axial measure- ments (head, trunk, and peduncle lengths) exhibit a clear trend for increasing head and peduncle lengths at the expense of trunk length through the cypriniform - characiform ^ siluroid - gymnotoid phyletic sequence. A portion of the variation in pe- duncle size is attributable to migration of the anus anteriad in this phyletic sequence, as evidenced by decreasing preanal and increasing postanal myomere counts (Table 29). However, the remaining peduncle variation and that of the head length are the result of allometry. In Fink and Fink's (1981) study, a single character involving the evolution of a new structure, a pair of barbels, conflicted with their adult-based cladogram. Ontogenetic evidence sup- ports their contention that the presence or absence of barbels is a poor indicator of relationship in ostariophysans. An ontoge- netic character for timing of barbel development (constructed in the same manner as described earlier for other ontogenetic characters) displays two distinct modes (Fig. 70). Cyprinids de- velop barbels during the latter third of the larval period, often after finfold absorption (i.e., as juveniles). Siluroids and co- bitoideans' do so during the first third of the larval period, sometimes prior to hatching. Although the sample size of cob- itoideans is small, it appears that they develop barbels somewhat later than the siluroids. Thus, although barbels are present in adults of all three groups, there is an important difference in these structures between the groups: heterochrony. That het- erochrony is a major cause of evolutionary change was amply attested by Gould (1977). Heterochrony in barbels may be an important consideration for classification within siluroids. The number of pairs of barbels (usually counted in the adult stage) is an important character for recognizing siluroid families. At least one pangasiid, Silonia silondia. has been described in which the larvae have three pairs of barbels (nasal, maxillary, and mandibular) that gradually be- come smaller until only one pair of minute maxillary barbels are present on the surface of adults (Karamchandani and Mot- wani, 1956). The phylogenetic analysis presented here is based on devel- opmental characters. It shows general congruence with the most thoroughly researched adult-based cladogram (Fink and Fink, 1981); however, larval characters alone are not as informative as adult characters. Larval characters support the new idea that gymnotoids are more closely related to siluroids than to char- aciforms. Characiforms appear to be primitive ostariophysans by virtue of the basal location of the relatively primitive char- aciform Hoplias. The apparent paraphyly of characiforms and siluroids is due to the lack of shared characters for each of these groups and would be altered by the reasonable addition of the numerous adult autapomorphies discussed by Fink and Fink (1981). Once monophyly is demonstrated by adding adult char- acters, Hoplias would probably occupy a basal position (with respect to the other three characiforms examined here) on a characiform lineage. However, the position of this lineage with respect to that of the cypriniforms may or may not agree with Fink and Fink's (1981) adult-based cladogram. School of Natural Resources, S. T. Dana Building, University of Michigan, Ann Arbor, Michigan 48109. ' Cobitoideans included here and in Fig. 70 were: Cobitidae— Bo/;a .vafir/i; (Changjiang, 1976); Cobilis taenia (Chyung. 1961; Koblitskaia, 1981; Kokhanova, 1957; Kryzhanovskii, 1949; Kryzhanovskii et al., 1951; Menasse, 1970); Mtsgurnus anguillicaudalus {Chyung,. 1961; Ko- bayasi and Moriyana, 1957; Okada, 1960; Okada and Seiishi, 1938; Suzuki, 1955, 1968); Homalopteridae— A'emac/jei/jis dorsalis (Kry- zhanovskii, 1949). Gonorynchiformes: Development and Relationships W. J. Richards THE Gonorynchiformes is a small group of fishes which have been allied with the clupeiforms or salmoniforms and most recently have been placed as a lineage, within the ostariophysan group, which includes also the Cypriniformes, Characiformes, and Siluriformes (Fink and Fink, 1981). The group is comprised of seven genera classified in about four or five families. The most widely known species is Chanos chanos Forsskil placed in the monotypic family Chanidae. The Gonorynchidae is a marine family of one genus Gonorymchus and several species found in tropical waters of all but the western Atlantic and eastern Pacific. The remaining twelve or so species are African freshwater forms in the genera Kneria. Parakneria. Grassei- chthys and Phractolaemus, which may represent two or three families. The eariy life history of Chanos is very well known because of the extensive culturing; Gonorymchus is poorly known. The early life histories of the freshwater species are unknown. Pellegrin (1935) notes that young specimens of Cromeria nt- lotica have a superficial resemblance to young Albula. It is ap- parent that this resemblance is to the shape of juveniles and not to a leptocephalus stage. Several subsequent papers have erro- neously reported that Pellegrin said that Cromeria resembled larval Albula. Development The early life history of Chanos chanos, the milkfish, has been described by Delsman (1926d, 1929b). Since Chanos is an im- portant aquaculture organism, several recent papers have de- scribed various aspects of development, among them the de- scription by Liao et al. (1979) is the most complete. Miller et al. (1979) provides a good account for separating them from common marine larvae. To summarize, the eggs and larvae superficially resemble clupeids and engraulids but differ in sev- eral trenchant characters. The eggs as described by Delsman ( 1 929b) are spherical, 1 .2 mm in diameter, lack oil droplets and have a weakly segmented yolk which may be similar to the granular yolks seen in ostariophysans. Yolk-sac larvae have me- lanophores scattered over the body and fin folds and a myomere formula of 34 -I- 10 (preanal and postanal). As development Fig. 71. Lateral and ventral views from top to bottom: Chanos chanos. 1 1.7 mm SL from Kumano, Tanegashima collected August 19, 1978, drawn by J. C. Javech; and Gonorymchus abrevialus. 12.8 mm SL from R/V Shoyo Maru station 25, 35°05'N, 144°24.3'E. collected on November 10, 1963; drawn by J. C. Javech. 138 RICHARDS: GONORYNCHIFORMES 139 progresses, the melanophores collect along the dorsal and ven- tral midlines of the trunk. In larvae 10-15 mm SL (Fig. 71) pigmentation is variable with melanophores on the dorsal mid- line varying from one to many and melanophores on the lateral line varying from none to many. The ventral midline has a continuous streak of melanophores in sharp contrast to clupeids and engraulids which have melanophores laterally on each side of the gut thus presenting two parallel streaks in ventral view. The anal fin of Chanos originates beneath the dorsal fin as in engraulids. In Hawaiian waters meristics separate Chanos from Gonorynchiis and other clupeids and engraulids. Chanos has 40-46 vertebrae [44-46 according to Miller et al. (1979) and 40-45 according to Senta and Kumagai (1977)]. Dorsal rays are 14-16, anal rays 8-11, pectoral rays 17 and pelvic rays 10-12 (Miller et al., 1979). Much less is known about the early life history stages of Gon- orynchus. Furukawa (1951) described the larvae of G. ahbrev- latus and illustrated 18 and 23 mm specimens. He based his identification on dorsal (1 1-1 2) and anal (7-8) fin rays, vertebral counts (55) and the posterior position of the dorsal and anal fins. Hattori (1964) illustrated and briefly described a series of G. ahbreviatus from 8.6 to 90.5 mm. He noted that the positions of the dorsal and anal fins do not shift during development. Mito (1966) illustrates two larval G. ahbreviatus. I examined a series of G. abbreviatus specimens and one is illustrated here (Fig. 7 1 ). The larvae resemble clupeids with the wide separation of the dorsal and anal fin. Pigment occurs dorsally and ventrally on the caudal peduncle and extends posteriad into the bases of the procurrenl caudal rays. Internal pigment occurs above the hindgut and behind the brain. A few external melanophores are present on the top of the head. Additional external mela- nophores appear with growth. These include a series which de- velops as lateral spots increasing in number with growth. In a few specimens examined a 15.9 mm larva had one spot and these increased in number to 18. At 23 mm SL pigment also appeared on the opercle and ventral rim of the orbit. The pelvic fin is discernible as a bud in small larvae but fin rays are not defined until 18 mm SL. A swimbladder is not discernible on any of the specimens as it is in clupeids and Chanos. Relationships The relationships of the Gonorynchiformes have been dis- cussed most recently by Fink and Fink (1981). They conclude that this order is the sister group of the Otophysi (the taxon which includes fishes with the Weberian apparatus). Chanos and Gonorynchiis larvae more closely and superficially resemble clupeoid larvae than any other group. This matter should be thoroughly investigated when early life history aspects of the freshwater species become better known. It will be interesting to see if those larvae resemble the marine species or freshwater Otophysi. National Marine Fisheries Service, Southeast Fisheries Center, 75 Virginia Beach Drive, Miami, Florida 33149. Salmoniforms: Introduction W. L. Fink ORIGINALLY a major portion of the Protacanthopterygii of Greenwood, et al. (1966), the order Salmoniformes is now the only portion left in that group, and the former term has ceased to have a useful function. This erosion of the Pro- tacanthoptergyii has resulted from the search for and taxonomic recognition of natural groups of primitive euteleosts, a practice that has and is continuing to have profound effects on fish clas- sification at all levels. This part of the symposium, concentrating on the "salmoniforms," places its participants in the middle of a continually changing set of problems, some of which have been longstanding. One of the questions we address here is whether the Salmoniformes as conceived by Greenwood et al. is itself useful any more, and if not, what are the relationships of the formerly included groups. In the years since it was delin- eated, the Salmoniformes has undergone attrition, most notably at the hands of Rosen (1973). Of particular concern to us is whether there is one large monophyletic unit which can be called Salmoniformes, as maintained by Rosen (1974), or whether there are several units, as suggested by Fink and Weitzman (1982), thus reciuiring us to modify our conclusions and clas- sifications. The basic questions are these; (1) What are the re- lationships of the Esocoidei (sensu Rosen, 1974), both to one another and to other primitive euteleosts? (2) What are the relationships of the Ostariophysi, (sensu Rosen and Greenwood, 1970)? Do these fishes lie above or below the Esocoidei in the phylogeny? (3) What is the pattern of relationships among the traditionally recognized "salmoniform" taxa, exclusive of the Esocoidei and Ostariophysi? Is this a natural division? (4) What are the phylogenetic relationships of and within the Argenti- noidei (sensu Greenwood and Rosen, 1971)? (5) What are the phylogenetic relationships of and within the Osmeroidei? (6) What are the phylogenetic relationships of and within the Sal- monidae? (7) Where does Lepidogalaxias belong? (8) What are the interrelationships within the stomiiform fishes? (9) What of the Myclophoidei, as recognized by Greenwood, et al. (1966)? This "group" has been most recently addressed by Rosen (1973) in his discussion on the Eurypterygii and Neoteleostei. Parts of these groups overlap into areas covered by this particular part of the symposium, such as placement of giganturids, and other parts into non-"salmoniform" portions such as that on myctophi forms. In many ways this symposium is a report on the state of the science of fish classification, will summarize current ideas of relationships and, especially, will point to where the greatest need for further research lies. Museum of Zoology, University of Michigan, Ann Arbor, Michigan 48109. Esocoidei: Development and Relationships F. D. Martin THE Esocoidei consist of two families, Esocidae and Um- bridae, with one and three genera respectively (Nelson, 1976). Table 30 lists all currently accepted species and gives their geographic ranges. All recent classifications consider the esocoids as members of the Salmoniformes (Greenwood et al., 1966; Gosline, 1971; Rosen, 1974; Nelson, 1976; and others). All esocoid fishes live in freshwater and occur in temperate and arctic waters of the Northern Hemisphere. All species are pred- atory with Esox being primarily piscivorous. They are distin- guished from other salmoniform fishes by the lack of the meso- coracoid, lack of pyloric caeca, a single rudimentary arch over PUl, and a single uroneural (Rosen, 1974). Table 31 gives de- velopmental features that characterize esocoid fishes and con- trasts them with Salmonidae and Osmeridae. Development Eggs are demersal and adhesive in most species (Breder and Rosen, 1966) but Esox niger eggs become buoyant at later stages of development and are not adhesive after water hardening (Jones et al., 1978). Eggs are of moderate size (1.0 to 2.2 mm usually) (Jones et al., 1978) and are either scattered as by Esox or are in nests as with Umbra and Novumhra (Breder and Rosen, 1 966). Table 30. Genera and Species of Esocoid Fishes and Geograph- ical Ranges. Esocidae Esox E. lucius E. reicherli E. masquinongy E. niger E. americanus Umbridae Novumhra N. hiibbsi Umbra v. krameri V. linu U. pygmaea Datlia D. pectoralis D. asmirabilis Holarctic (Grossman in Lee et al.. 1980). Amur River region of Siberia (Berg, 1948). Eastern North America, primarily Great Lakes and Upper Mississippi drainage (Grossman m Lee et al., 1980). East Goast drainage of North America, also lower Mississippi drainage (Grossman in Leeet al., 1980). Eastern half of North America (Grossman m Leeetal., 1980). Olympic Peninsula of Washington State (Meldnm m Lee el al., 1980). Middle and lower Danube System and lower Dniester River (Berg, 1948). Southern Ganada and Gentral United States (Gilbert m Leeetal., 1980). Southeastern New York to Northern Florida. mostly on Goastal Plain (Gilbert in Lee et al., 1980). Arctic and sub-Arctic Alaska and eastern tip of Sibena (Rohde in Lee et al., 1980). Amguema River basin of Siberia (Gheresh- nev and Balushkin, 1980). Multiple oil droplets occur with a unique set of movements producing alternating clustering and dispersion as ontogeny pro- ceeds (Malloy and Martin, 1982). Larvae of nearly all species are known, and developmental series have been described and illustrated. Figs. 72 and 73 show representative larvae oi Esox and Umbra. Those described hatch relatively undeveloped, with head flexed over and attached to the large yolk sac; the eyes are unpigmented. In all species the notochord is stout and reaches nearly to the margin of the caudal finfold. During flexion the notochord extends well beyond the developing hypurals and may form a separate lobe to the de- veloping caudal fin until the hypurals are complete. In Umbra and Esox the pectoral fin is the first to begin differentiation (but not form rays) with the pelvic fin the last to develop fin rays. All median fins differentiate more or less simultaneously with caudal starting ditTerentiation slightly ahead of the others. Changes in body form are gradual with no noticable point of metamorphosis. Before fin differentiation is complete the body Fig. 72. Development of Esox niger from hatching to juvenile. Lengths arc total lengths. (From Mansueti and Hardy, 1967.) 140 MARTIN: ESOCOIDEI 141 Common Cardinal Hepatic Vitelline Vein -'Sublntestinal Vitelline Vein Common Cardinal, /Hepatic Vitelline Vein Fig. 73. Early yolk-sac and late yolk-sac larvae of Umbra pygmaea. (From Wang and Kemehan, 1979.) Heart 5.4 mm TL -'Sublntestinal Vitelline Vein Fig. 74. Schematic representations of the vitelline venous systems of Esox americanus (upper) and Umbra pygmaea (lower)— based in part on figure from Wang and Kemehan, 1979. Table 31. A Comparison of Egos and Larvae of Esocoid, Salmonoid and Osmeroid Fishes. Unless otherwise noted information on Umbridae and Esocidae taken from Malloy and Martin (1982). Egg Demersal Adhesive Oil droplets Size In nests Embryo and yolk-sac larva Head deflexed, adherent to yolk-sac Eye pigmented at hatching Vitelline circulation Common cardinals Hepatic vitelline vein Sublntestinal vitelline vein Sublntestinal v. v. forming rete Hepatic v. v. forming rete Larva Vertebrae (myomeres) Adipose fin Dorsal origin over or behind anus Notochord forming a urostyle extending length of hypural complex past hypurals Juvenile and adult Pyloric caeca Anterior constriction of vertebra Pharyngobranchial 1 Epurals Hypurals Neural spine on preural 1 Neural spine on preural 2 -1- -1- '-I- ' + + multiple M-2.2mm -1- or '-± multiple M.9-'3.4mm '± (mostly -) '•■ '-multiple or * '-single 'M. 5-7.0 mm -single or " '^multiple -1 mm '-I- 1 _ ' + 1 + -I- + -t- -I- '32-""'42 -I- + + + + + '43-67 -I- -I- 3.4 + 9. I0*_ 9.10 + 9+ 10_ 9.10_ 9.10+ _ "•46-75 3.18 + 4.18_ 2_ 2.15 + ? ? ? 9 9 '55-70 "13-222 "0-11 20*« + 20*** _ + 20 _ 20_ - - + -1- 0-2 2 "2 or 3 "2 or 3 5 or 6 6 "7 "6 -1- + "-or reduced "-or reduced fully fully reduced or "fully developed developed not developed • Present bul does not run on surface of yolk sac. •" In Novumbra and Daltia only present in midabdominal region of juv ••• When present ther« is also posterior constriction. ' Breder and Rosen, 1966. 'Cooper. 1978 ' Rajagopal. 1979. ^ Watling and Brown, 1955. • Baugh. 1980 • Jones ctal.. 1978. 'Carbine. 1944. " Uach. 1923. 'Soin. 1966. '"Kunz. 1966. ' Bigelow and Schroeder, 1963. ' Fuiman. 1982b. 'Nelson. 1972. ' Auer, 1981- ^ Yanagawa. 1 978. " Scott and Crossman. 1973. ' Hart. 1973. ■Nagiec, 1979. ' Greenwood and Rosen. 1 97 1 . •Cavender, 1969. 142 ONTOGENY AND SYSTEMATICS OF FISHES -AHLSTROM SYMPOSIUM form is basically that of the adult. Guts are simple with no elaborations in all species. At hatching Umbra has a shorter gut and fewer myomeres than Esox and this is reflected in there being 5 myomeres between the yolk sac and the anus in newly hatching U. pygmaea and 12 in E. americanus (Malloy and Martin, 1982). Relationships Malloy and Martin (1982) point out three ontogenetic char- acteristics shared by Esox and Umbra, which indicate close relationship. The position of the heart at the time of formation is on the yolk sac anterior to and left of the head. All other fish for which position of the forming heart is noted have it forming under the head in the pericardial cavity or, as in the Atherini- formes, near the midline and anterior to the head. The yolk-sac circulatory pattern consists of paired simple common cardinals, a posterior rete formed by the subintestinal vitelline vein and paired or single hepatic vitelline veins which enter the rete before the subintestinal vitelline vein joins the common cardinals at the heart (see Fig. 74). This differs from all other salmoniform fish for which the pattern is described (Kunz, 1 964; Soin, 1 966). The oil droplets go through a predictable series of clustering and dispersion. Oil droplet movement of this sort has only been documented previously by Ahlslrom ( 1 968) for bathylagid smelts of the genera Bathylagus and Leuroglossus. McDowall (1969) recognized a salmonoid-osmeroid-esocoid lineage but states "Where esocoids fit into this series of sub- orders and families is not clear to me." Rosen (1973) likewise considers the esocoids and salmonoids to probably be closely related but considers this alignment to be provisional. Fink and Weitzman (1982), in contrast, state that they find no evidence to consider the esocoids closely related to the other Protacan- thopterygii (sensii Rosen, 1 974), which are the Agentinoidei and Salmonoidei (including the Salmonoidea plus Osmeroidea). Fink and Weitzman list the esocoids as sedis mutahilis at the euteleostean level or as the sister group to all other euteleosts. Soin (1980), on the basis of egg development patterns, feels that the esocoid fish are incorrectly placed as a suborder of the Sal- moniformes, however he gives no guidance as to correct place- ment. While the ontogenetic evidence presented in Table 30 is not conclusive it suggests that there is a large difference between the esocoids and the Salmonoidei and this is consistent with the opinions of Fink and Weitzman. The vertebrae of Umbrids have a pronounced anterior con- striction, giving them an asymmetrical appearance, however Novumbra and Dallia show this characteristic only while young and most noticeably in the mid-abdominal region. In Esox the vertebrae are either unconstricted or are constricted both an- teriorly and posteriorly so that they appear symmetrical (Cav- ender, 1969). Other differences between the Esocidae and the Umbridae are seen in the Umbridae having nine or fewer bran- chiostegals, fewer infraorbitals, no supratemporals or intercalars and usually fewer than 41 vertebrae (Wilson and Veilleux, 1982). Chesapeake Biological Laboratory, University of Maryland, Box 38, Solomons, Maryland 20688. Salmonidae: Development and Relationships A. W. Kendall, Jr. and R. J. Behnke SALMONIDS (whitefishes, ciscoes, grayling, trout, and salm- on) are highly important in terms of aesthetic appreciation, commercial and recreational value, and scientific study. Studies of the development of salmonids from hatching until the time of yolk depletion, and of the relationships among subfamilies and genera have been largely neglected [see review of systematics by Dorofeyeva et al. (1980)] despite the large body of literature on early embryological development and relationships among species and populations. Salmonids all spawn in fresh or brack- ish water, some are anadromous while others are strictly fresh- water. The family is composed of about 10 genera in three subfamilies: Coregoninae, Thymallinae, and Salmoninae (Table 32) (Nelson, 1976). Along with a precise homing ability, salmonids tend to form genetically isolated populations. They seem to be able to occupy new niches and habitats as these become available in the cold temperate parts of the Northern Hemisphere. One result of this adaptability is the existence of taxonomic problems mainly at the species-population levels (Utter, 1981). Development Post-hatching development of salmonids has been little stud- ied (Table 33), and only a superficial analysis of comparative developmental stages has been attempted (Soin, 1980). Thy- mallus and the salmonines share apparently advanced features of development such as large yolk sac with an extensive vitelline circulatory system and development of rather uniform intense pigment, while coregonines develop larvae that are more typical of other freshwater fishes (Faber, 1970). Thymallus seems inter- mediate between the coregonines with a "normal" larval stage and the salmonines in which the larval stage is largely bypassed (the young have fully formed fins by the time the yolk is ab- sorbed). Parr marks (vertical blotches or bars of pigment over the trunk of juveniles) are present in all salmonids except Cor- egomts a.nd StenodushuX are not seen injuveniles of other fishes. Norden (1961) incorrectly considered the early stages of Core- goniis artedii as figured by Fish (1932) to be similar to those of Thymallus arclicus. He also stated that "the development of KENDALL AND BEHNKE: SALMONIDAE 143 Table 32. Characters that vary among the Salmonid Subfamilies. Subfamily Character Coregoninae Thymallinae Satmoninae General Genera Coregonus, Prosopium. Steno- Thymallus Brachymystax. Hucho, Salvelinus. dus Salmo, Parasatmo, Oncorhvnchi Species 30 4 32 Habitat freshwater, few anadromous freshwater freshwater and anadromous Egg size 1.8-3.7 mm 2.5 mm 3.7-6.8 mm Diploid chromosome num- 64-82 102 52-92 bers Dorsal fin rays 10-15 17-25 8-12 Dentition' Tooth character narrow, sharp, 2-3 sections uniform in size vary in size Maxillary toothless toothed toothed Dentary minute teeth restricted to ante- narrow, teeth of uniform size numerous teeth of varying size all rior end all along bone along bone Vomer small and toothless (except in Stenodus and some Corego- nus) small, with teeth long, with teeth Premaxillary small large large Caudal skeleton- Epurals 3 3 2-3^ Stegural little developed little developed well developed Neural and hemal spine ex- little moderate large pansion Urodermal present absent absent Neural spine on PU, absent absent present Neural spine on PU, not fully developed not fully developed fully developed'' Cranial osteology' Orbitosphenoid present absent present Suprapreopercular absent absent present Panetals meet at midline yes yes no Hypethmoid present absent usually absent Basisphenoid usually absent present present Uppermost orbital' present present absent ' Vladykov (li)70). ■'Cavender (1970)- ' Nordcn (1961 1, * Some vanation withm Salmonmae in these Iwo characters. Those with 2 cpurais usually have most extensive neural spine development. ^ Sometimes erroneously termed dcrmosphenotic; sometimes present in Salmoninae; see Behnke (1968, p 9-10). the young grayling has much in common with that of both the coregonines and salmonines" (Norden, 1961:743). Among the coregonines, larvae of Prosopium (Faber, 1970; Auer, 1982), L«/c;c7!r/!V5 (Fish, 1932; Faber, 1970; Auer, 1982), and Coregonus (¥'\%\\. 1932; Faber, 1970; Auer, 1982) have been illustrated and briefly described. All show similar larval mor- phology (Fig. 75). They are rather slender with a long preanal finfold— the yolk being confined to the anterior trunk region. The yolk-sac length is <35% total length (TL), eye diameter is <7% TL, and body depth at anus is usually < 10% TL (Auer, 1982). The yolk is exhausted before any of the fins, except the caudal, possess full complements of rays. Prosopium eggs have multiple oil globules, while Leucichthys and Coregonus eggs have a single oil globule (Auer, 1982). Pigment in preflexion and flexion larvae is mainly associated with the dorsal and ven- tral midlines. Later, the body becomes more uniformly pig- mented. Prosopium develops parr marks during the juvenile period. Larvae oi Stenodus are undescribed and they may differ from those described above, since adults of this genus appear quite divergent from the others in this subfamily. Early development of Thymallus thymallus has been fully described (Penaz, 1975). They hatch with a large, anteriorly placed yolk sac that is covered by a rather extensive vitelline circulatory system, and the preanal and postanal finfolds are about equal in length (Fig. 75). The yolk sac is exhausted during notochord flexion and by that time some fin rays have developed in all of the fins. The larvae are rather heavily pigmented during this period. When the fins have developed their adult comple- ment of rays, the fish appear like juveniles and parr marks begin to form. Early development of all the salmonine genera and most sub- genera is known, although several are inadequately described (Table 33). Described development of all salmonines is quite similar (Figs. 76, 77). Their eggs are among the largest of all teleosts. They all hatch with large yolk sacs and well developed vitelline circulatory systems. The preanal finfold is shorter than the postanal finfold (except in Hucho where they are about equal). The preanal finfold extends somewhat down the poste- rior of the yolk sac in Oncorhynchus. The notochord is slightly flexed and some caudal rays are present. Yolk-sac length is 144 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Table 33. Meristic Values and References to Descriptions of Larvae of Salmonids. Total reported ranges of meristic values are given, although the extremes of the ranges may be rarely observed. Subfamily Genus Subgenus Ranges of meristic values References with illustrations of flexion stage larvae Verte- Dorsal brae' fin- Pec- Tolal Lateral Branch i toral Pelvic gill line ostcgal fin fin rakers scales rays Primar\' source Coregoninae Stenodus Prosopium Coregonus Leucichlhys Coregonus Thymallinae Thymallus Salmoninae Brachymystax Hucho Hucho Parahucho Salvelinus Sahelinus Baione Crislivomer Salmo Salino Salmothymus Acantholingua Plalysalmo Parasalmo Oncorhvnchus 64-69 12-19 15-18 16-17 11 19-24 90-110 Faber (1970), Auer( 1982) 50-65 10-15 10-14 13-18 9-12 11-44 9-12 Scott and Cross- man (1973) 50-108 6-10 Scott and Cross- man (1973) Fish (1932); Faber (1970). 50-67 8-15 9-16 13-18 8-13 21-64 58-110 7-10 Scott and Cross- Auer(1982) man (1973) Fish (1932), Faber (1970), 55-64 10-13 9-14 14-17 11-12 15-78 70-102 6-10 Scott and Cross- Auer(1982) Penaz(1975) Smol'yanov (1961) Balon(1956) Balon (1980) 58-62 17-25 11-15 14-16 10-11 16-33 81-103 7-9 man (1973) Scott and Cross- man (1973) 58-62 12-15 11-14 15-18 64-71 12-14 11-13 15-18 57-62 12-14 12-14 14-17 9-10 20-30 120-150 10-13 Behnke(1968) and original 10 10-17 120-150 9-12 Behnke(1968) and original 9 14-20 110-120 9-12 Behnke(1968) and original 57-71 10-12 8-10 14-16 9-11 11-51 105-152 10-15 Scott and Cross- man (1973) Balon (1980), Auer( 1982). 57-62 10-14 9-13 11-14 8-9 14-22 110-130 9-13 Scott and Cross- Martinez (1983) man (1973) Fish (1932), Balon (1980). 61-69 8-10 8-10 12-17 9-10 16-26 116-138 10-14 Scott and Cross- Auer(1982) Auer(1982). Martinez (1983) Auer(1982), Martinez (1983) Auer(1982) man (1973) 54-62 10-15 8-13 12-16 9-10 14-25 100-130 10-12 Behnke(1968) and original 56-60 13-15 11-13 12-14 9-10 25-32 100-115 10-12 Behnke(1968) and original 52-59 11-13 10-12 11-13 9-10 18-22 95-110 9-11 Behnke(1968) and original 57-59 13 11 14 9 23-24 109-110 10-11 Behnke(1968) and original 55-67 8-12 8-12 11-17 9-10 14-28 100-150 9-13 Scott and Cross- man (1973) 61-75 9-16 12-19 11-21 9-11 18-43 120-160 11-19 Scott and Cross- man (1973) Overall ranges 50-75 8-25 -19 11-21 8-13 10-78 50-160 6-19 ' Vanalions exist in the literature in how many of last 3 upturned vertebrae are counted; some authors omit the last 3 upturned vertebrae. ' Includes rudiments where specified. A variation of 2-3 rays may result from different methods of counting (whether unbranched or rudimentary rays are included). >35% TL, eye diameter >7% TL, and body depth at anus usually > 10% TL (Auer, 1982). Pigmentation is unifoimly heavy at hatching or later in the yolk-sac stage. The median fins de- velop rays before the paired fins. By the time the yolk is absorbed the finrays have completed foimation and the fish takes on a juvenile appearance. Thus, the yolk remains a source of nutri- tion throughout the larval stage. Relationships Although salmonids are considered to be living representa- tives of the basal stock from which euteleostean evolution pro- ceeded, there is no clear consensus on their relationships to other fishes. Since there are differing opinions on the relationships between the major teleostean lineages (i.e., the divisions of Greenwood et al.; 1966), it is difficult to select representatives of outgroups to compare with the salmonids. Recent studies (Rosen, 1974; Fink and Weitzman, 1982; Fink, this volume) have pointed out that the Protacanthopterygii and even the Salmoniformes are probably not monophyletic taxa. The sal- monids along with the galaxioids, osmeroids, and argentinoids, may form a group (Salmonae) that is the primitive sister group of the neoteleostei. However, the relationships among these groups is not clear, and the salmonids may be closer to the neoteleostei than to these other groups with which they have frequently been aligned (Fink and Weitzman, 1982; Lauder and Liem, 1983; Fink, this volume). Some primitive teleost traits KENDALL AND BEHNKE: SALMONIDAE 145 Fig. 75. Flexion stage larvae of: (A) Coregonus (Leucichthys) artedii (17.5 mm); (B) Coregonus (Coregonus) clupeaformis (18.5 mm); (C) Thymatlus ihymallus {\6.0 mm). A and B from Fish (1932), C from Penaz (1975). Table 34. Characters that vary among the Coregonine Genera and Subgenera (sg) mainly from Norden (1961) and Cavender (1970). Coregonu. Prosopium Character Coregonus (sg) Leucichthys (sg) Stenodus Species 8 17 1 1 Habitat Some occasionally anad- Several anadromous Freshwater Anadromous romous Basibranchial plate Absent Absent Present .Absent Parietal bones meet along Yes Yes Yes No: narrowly separated midline Postorbitals in contact with Yes Yes Yes No preopercle Parr marks Absent Absent Present in some Absent Flaps between nostnls 2 2 1 2 Mouth size Small Moderately large Small Large Teeth Weak or none Weak or none Weak or none Many, small Mouth position Subterminal Supenor or terminal Subterminal Terminal Vomer Small, toothed in some Small, toothed in some Small, toothless Large, toothed First supraorbital Moderate Moderate Short Long Supraethmoid Short Short Long Short 146 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Fig. 76. Flexion stage larvae of: (A) Brachymystax lenox(\1.2 mm); (B) Hucho (Hucho) hucho (20.8 mm); (C) Salvelinus (Sahelinus) alpinus (19.8 mm); (D) Sahelinus (Cnslivomer) namaycush (approx. 20.4 mm). A from Smoryanov (1961), B from Balon (1956), C and D from Balon (1980). Fig. 77. Flexion stage larvae of; (A) Sahelinus (Batone) fontmalis ( 1 4.0 mm); (B) Parasahno gairdnert ( 1 4.0 mm); (C) Parasalmo darki (14.2 mm); (D) Salmo trutta (14.0 mm); (E) Oncorhynchus tshawytscha (25.0 mm). A-D from Martinez (1983), E original. KENDALL AND BEHNKE: SALMONIDAE 147 148 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM large mouth w 7i3 many small teeth n 753 ^large toothed vomer ^ N 720 no teeth on vomer N 720 / mostly: small mouth n 7S3 small teeth on vomer of young only n=720 parr marks absent n 743 two flaps between nostril n 71 enlarged first supraorbital n 71 Joss of basibranchial plate n 71 / \\ N 727 slightly notched ethmoid cartilage light spots 1 Palatine vomer strong ascending pre- teeth form z^ maxillary process J n j" shaped "C gap between palatine I band n 753^v vomer teeth 1 Thymallinae N 72S notched ethmoid \ cartilage N 732 palatine vomer teeth form a "T" / Coregoninae N 679 no teeth on maxillary c 9 one urodermal c 11 small neural spine on PU^ N-753 < 16 dorsal rays N-750 general loss of teeth c 11 small neural spine on PU-, N-679 > 16 dorsal rays N-e79 no orbitosphenoid increase in size and amount of yolk in egg bypass larval stage Salmonidae 100) • N 739 • •N 752 ♦IM 743 *N 752 • N 752 • *C 27 • • N 752 • •rj 752 • *N 752 • "N 752 • •C27 «»C 11 • *c n *?C9 Tetraploid karyotype ( 2n chromosomes - ' Axillary pelvic process Three upturned caudal vertebrae ( two ural centra) Parr marks in juveniles Three post cleithra Mesopterygoid toothless Last four hemal spines and parhypural fit together ( peg and socket) Adipose fin present Oviducts incomplete or absent Mesocoracoid present Opisthotic present Principal caudal rays = 19 Three epurals Full neural spine on PU, Two hypurais (ventral) on U^, 4 hypurals (dorsal) on U, . 1 long, 2 short uroneurals N 726 blunt pointed ethmoid cartilage N 732 gap between palatine vomer teeth M '53 no ascending premaxillary process N 753 postorbitals contact preopercular N 753 opisthotic touches prootic N 728 reduced dorsal fontanelles in adult well developed stegural expanded caudal neural and hemal spines neural spine on PU. large neural spine on PU, N-679 parietals separated by supraoccipital N 739 small scales ("> 100 in lateral Ime) N-679 suprapreopercular present N 731 curved preopercular N 736 dorsal rays < 16 e-9 reduction or loss of hypethmoid B - Behnke 1968 N - Norden 1961 C - Cavender. 1970 H = Hol£ik,1982 (number refers to page in above references) * Salmonidae (synapomorph for family) "" Salmonoidei (synapomorph for suborder) ■•• Shared primitive (plesiomorph) character with other "primitive" teleosts Fig. 78. Hypothesis of relationships among extant saimonid genera. Groupings and branching points are based largely on a consensus of recent literature and are not the result of a strict cladistic analysis. possessed by salmonids include lack of oviducts, presence of abdominal pores, and three upturned caudal vertebrae sup- porting the hypurals. Salmonids are autapomorphic with about twice the DNA content of other '^salmoniform" families, ap- parently the result of having a common tetraploid ancestor. The salmonids possess an adipose fin, a mesocoracoid, pyloric caeca, and the vestige of a spiral valve intestine. The gill membranes extend far forward free from the isthmus and there is a pelvic axillary process. Two shared derived features of the salmonids and neoteleostei are: 1) the articulation of both the basioccipital and exoccipital with the first vertebra, and 2) the presence of a medial cartilage between the ethmoid and premaxilla (Fink and Weilzman, 1982). Although it is not possible at present to perform a meaningful cladistic analysis of the salmonids, some evidence is available in the literature which can contribute to such an analysis (Fig. 78). Cavender (1970) compared the osteology of leptolepids. extinct fish thought to represent the basal teleost condition, with that of the salmonids. He found several characters that indicated 1) that the salmonids are monophyletic, and 2) how the three subfamilies of salmonids are interrelated. The coregonines ap- peared to be most similar to the leptolepids, the thymallines more derived than the coregonines, and the salmonines more derived than the thymallines. Reshelnikov (1975). on the basis of several types of characters, suggested elevating the subfamilies to familial status. Coregoninae contains about 30 species in three genera. They are mainly freshwater, and produce rather small eggs, compared to those of the other two subfamilies. They share several ad- vanced characters with the other subfamilies, indicating that salmonids are monophyletic, but lack a number of advanced character states possessed by the other two subfamilies, as these branched oflTafter the coregonines. Within the coregonines, Pro- sopium seems least diverged (Table 34). Sienodus shows several, possibly secondarily derived character states concordant with feeding on large active prey (expanded dentition, large mouth). Coregoniis, which seems to be a sister group to Stenodus. is separated into two subgenera: Leucichthys with adaptations for plankton feeding, and C orego nus y^h\ch are mainly benthic feed- ers. Thymallinae contains one genus, Thymallus, with about four species in freshwater of the colder parts of the Northern Hemi- KENDALL AND BEHNKE: SALMONIDAE 149 Table 35. Characters that vary among the Salmonine Genera. Characters Brachvmyslax Hucho Salvetinus Salmo' Parasalmo^ Oncorhynchus^ Subgenera Hucho. Para- hucho Sahelinus, Baione. Crisli- vomer Salmo, Salmo- ihy/mts. Acan- iholingua. Ptatvsahno Species 2 3-5 8 8 5 6 Habitat freshwater freshwater and freshwater and freshwater and freshwater and usually anadro- anadromous anadromous anadromous anadromous mous Mouth size small large large large large large Teeth on shaft of no no no yes yes yes vomer Palatine-vomer- U-shaped band U-shaped band teeth narrowly teeth narrowly teeth narrowly teeth widely sepa- ine teeth separated separated separated rated Postorbitals con- no no no no no yes tact preopercle Supraethmoid long, with nu- broad, with nu- long, with nu- notched poste- notched poste- deeply notched shape merous poste- rior projec- tions merous short posterior pro- jections merous poste- rior projec- tions riorly riorly posteriorly Ascending pre- intermediate intermediate extended and intermediate intermediate none maxillary pro- sized sized well developed sized sized cess Opisthotic touch- no no no no no yes es prootic Dorsal fonta- persistent persistent persistent persistent persistent reduced in adult- nelles Egg size 4-5 mm large 4-5 mm 5-7 mm large large Diploid chromo- 92 84 78-84 56, 80-82' 56-70 52-74 somes Dark spots-light yes yes no yes" yes yes background ' There is lack of agreement on the relationships between these laxa; e.g., some consider Parasalmo a subgenus in Salmo. while others would also consider Oncorhynchus a subgenus of Salmo. * Retained in O. ma.sou * Salmo salar has 56-60 diploid chromosomes- * Salmo marmoratus and S platycephalus have no dark spots. sphere. They have several character states that seem advanced over those seen in coregonines. They are moderate-sized, gen- eralized insectivores (Table 32). Salmoninae contains four to six genera, depending on opin- ions over the relationships among the species in Salmo, Par- asalmo, and Oncorhynchus (Table 35). These seem to be the most advanced of the salmonids, and share several character states that are derived compared to the other two subfamilies (Table 35). Holcik (1982) presented evidence which suggests that the genera Hucho, Brachymysla.x, and Salvelinus form one lineage; Parasalmo and Salmo another; and Oncorhynchus a third. Salmonines are mainly active predators and most tend toward an anadromous life histoi7. Early life history and developmental information should con- tribute to the rigorous analysis of characters that will be required to validate the foregoing hypotheses about relationships. Such information is not presently available in the literature, but should be readily obtainable, since so many of these fishes are routinely reared in laboratories and hatcheines. Developmental infor- mation seems particularly promising in this family, since a wide range of the life history patterns are present and larvae can be superficially grouped according to their representative subfam- ilies. (A.W.K.) Northwest and Alaska Fisheries Center, 2725 MoNTLAKE Blvd. E., Seattle, Washington 98112 and (R.J.B.) Department of Fishery and Wildlife Biology, Colorado State University, Fort Collins, Colorado 80523. Southern Hemisphere Freshwater Salmoniforms: Development and Relationships R. M. McDowALL SEVERAL family-level groups of diadromous salmoniform fishes are found in cool-temperate southern hemisphere fresh waters, forming an obvious ecological counterpart to the northern cool-temperate Salmonidae, Osmeridae, Plecoglossi- dae, Salangidae, etc. With the exception of a single species, in a high elevation lake in New Caledonia, they are all south of about latitude 28°S. They occupy all of the main land masses (Australia, New Zealand, South America, South Africa) and some of the more distant southern islands (Lord Howe. Chat- hams, Aucklands, Campbell, Falklands). Diagnostic familial and generic characters are listed in Table 36. Familial arrangement of these fish varies from including all in a single purportedly monophyletic family Galaxiidae (Nelson, 1972), through two families in separate sub-orders (Rosen, 1974) to four families in one or two suborders. There are two obvious and widely accepted familial groupings: Galaxiidae— Aplochi- tonidae and Retropinnidae — Prototroctidae (McDowall, 1969). The most recent view (Fink and Weitzman, 1982) suggests that these four family level taxa are possibly all of osmeroid deri- vation agreeing with my own evaluation (McDowall, 1 969), and in contrast with Rosen (1974 — he links galaxiids and aplochi- tonids with salmonoids; retropinnids and prototroctids with osmeroids). The southern taxa are all clearly primitive prota- canthopterygians of salmoniform type. Beyond that little can be said other than that a further search of additional character complexes is needed to clarify relationships. Within-family relationships are little studied. Three of the southern families (Retropinnidae, Prototroctidae, Aplochiton- idae) can be dealt with more simply than the fourth (Galaxiidae). Retropinnidae (Australia and New Zealand— see McDowall, 1979). — four species in two genera: Present state of knowledge does not permit explicit recognition of affinities. Elongation of the alveolar process in the premaxilla of Stokellia anisodon is an advanced character which leaves three species of Retropinna with the primitive condition (alveolar process short, maxilla sometimes toothed). Stokellia also has unossified gill rakers (an "advanced" but "loss" condition) and high scale count (100 compared with 70 or less in Retropinna— which is the derived condition?) Prototroctidae (Australia and New Zealand— see McDowall. 1976).— Two species in one genus. Two congeneric species pose no phylogenetic problems. The only significant question to ask is "How do these species relate to the Retropinnidae?" Answers to this question have not yet been sought. Aplochitonidae (Tasmania and South America— see McDowall. 197 la).— Three (perhaps four) species in two genera. Mono- phyly of the Aplochitonidae (Aplochiton and Lovettia) should not be assumed. Inclusion of Lovettia in the Gala.xias-Aplo- chiton assemblage is supported by characters in Table 36 but Lovettia has such reduced osteology that a search for characters in other structural systems is needed before its relationships can be clarified. Inclusion oi Lovettia in the Aplochitonidae is based, in part, on history (it has always been there!) and in part, on the fact that it is a "galaxioid" with the dorsal fin over the pelvics and an adipose fin present (like Aplochiton and unlike Gala.xias). Galaxiidae (.Australia. New Caledonia, Lord Howe. New Zea- land. South America. South Africa). S'w genera with 37 species distributed as follows: Gala.xias, 24— all areas but New Cale- donia; Paragala.xias, 4— Tasmania; Neochanna, 3 — New Zea- land; Gala.xiella, 3 — Australia; Brachygala.xias, 2 — South America; NesogalcLxias. 1 —New Caledonia. This larger and more complex family offers scope for phylogenetic analysis that has had little attention. Relationships Previous studies of within-family relationships have been based on morphological similarity (McDowall, 1 970), phenetics based on muscle myogens (Mitchell and Scott, 1979), or den- drograms derived from cluster analysis of morphometric or me- ristic data (Campos, 1979). Johnson et al. (1981, 1983) have sought to establish relationship on the basis of karyotypes and multivariate analysis of morphometric and meristic characters in the diverse Tasmanian fauna. The only attempt at a "strictly phylogenetic" interpretation of within-family relationships, by Rosen (1978), was based on misinterpretation of character states and a limited perception of variation in the family, and achieved nothing (McDowall, 1980). A broad and strictly phylogenetic analysis of galaxiid inter-relationships is not yet available and probably depends on examination of additional character complexes. On the basis of out-group comparisons (all salmonoid— os- meroid— galaxioid families have members that are diadromous) it is my view that diadromy in the Galaxiidae is a primitive character. It is represented in at least six species. Diadromous species tend to be large and generalised in char- acter, but with specific adaptations to habitats occupied during freshwater life. Vertebral numbers are high (> 60) and ray counts in pelvic (7) and caudal (16) fins very stable. There are indications of close relationships with diadromous stocks, e.g., Gala.xias maculatus seems likely to be a neotenous derivative of some other diadromous galaxiid; distinctive ju- venile colour patterns may relate G. argenteus to G. fasciatus and perhaps G. truttaceus. There are numerous landlocked populations of diadromous species, and present interpretations are that several species are derived by isolation following landlocking, e.g., G. auratus (landlocked) derived from G. truttaceus (diadromous) in Tas- mania; G. gracilis from G. maculatus in New Zealand. Wholly freshwater species tend to be the more specialised members, in which there is often dwarfing, reduced vertebral counts, greater meristic instability, as well as the loss of the distinctive marine juvenile stage. Some freshwater groups have 150 McDOWALL: SALMONIFORMS 151 not yet recognised origins within the diadromous stocks and there is identifiable speciation related to known geo-tectonic events. The relationships of some of the more distinctive species groups— Neochanna (New Zealand), Galaxiella (Australia), and including geographical outliers like Gala.xias zebratus (South Africa) and Nesogalaxias neocaledonicus (New Caledonia)— re- main obscure. Previous inclusion of Australian and South American species in Brachygalaxias is ill-founded, on present data, and confuses the understanding of relationships. An interesting phylogenetic problem in the Galaxiidae in- volves the diminutive Tasmanian Paragalaxias. with four species in high elevation lakes that probably pre-date Pleistocene gla- ciations. Paragalaxias is distinctive in having the dorsal fin origin only a little behind the pelvic bases. In this regard it resembles aplochitonids differing from all other galaxiids in which the dorsal origin is close to the level of the vent/anal origin. Thus is Paragalaxias a galaxiid in which the dorsal fin has migrated forwards, the resemblance to Aplochiton being con- vergent or is it an aplochitonid in which the anterior dorsal fin position is primitive but in which the adipose fin has been lost? Examination of additional character complexes in which gal- axiids and aplochitonids differ is needed to clarify this question. The preceding discussion makes it evident that relationships between and within the southern diadromous salmoniforms re- main in need of clarification. Only the Galaxiidae is large and diverse enough to provide fertile ground for a study of within- family phylogeny. In all the families, species and characters are conservative in nature and lack distinctive or extreme speci- alisation. Inter-specific differences tend to be expressed as changes in meristic characters (like vertebral and fin ray counts), often to presence/absence character states (pyloric caeca, canine teeth) and sometimes to distinctive and stable differences in colour patterning. There are few readily evident characters that are indicative of major phyletic lineages. Possibly investigation of laterosensory papillary rows will be informative. At present, establishment of phylogenies appears difficult. A study of re- lationships using DNA hybridisation techniques (Sibley and Ahlquist, 1981) is at present in early planning stages. Life History Patterns and Reproduction In general life history patterns are understood although details are sparse. There are broad similarities in patterns. Retropinnidae.— Aspects of early life history have been de- scribed by Milward {1966 — Retropinna sewon/— Australia), Jolly (1967-«. retropinna— N.Z.) and McMillan (1961— Sto- kellia anisodon—N.Z.). The eggs are tiny— 0.5 to 0.6 mm in lacustrine R. retropinna, 0.95 mm in R. semom. They are de- mersal and adhesive, spherical, without distinctive features. They are a pale straw colour. They are deposited on sandy bottoms in lower river reaches or estuaries (around lake shores in land- locked populations), where development occurs; development is relatively slow ( 10-20 days) and description of development shows nothing distinctive (Fig. 79). Newly hatched larvae in some species go to sea. In others they are lacustrine or riverine. Larvae at hatching are small (2-5 mm), very slender and elon- gated, the yolk sac with a single oil globule, and situated ante- riorly beneath the opercular openings/pectoral fins. The gut is long, the vent at about 70% of length. A continuous finfold encompasses the trunk. Pectoral fin buds are present. Newly hatched larvae are positively phototropic. Pigmentation and later development are undescribed. Juveniles from a summer- Table 36. Character States in Principal Genera of Southern Freshwater Salmoniforms. (* except Paragalaxias; + present, - ab- sent; u uniserial; m muUiserial; 1 parhypural + hypurals; 2 tubercles in Lmetlia may not be comparable with others). Figures are "usual" al- though variants are known. The divergent galaxiid genera are excluded (Paragalaxias, Galaxiella, Neochanna, etc.). Proto- Retro- Galaxi- Irocli- pinni- idae Aplochilonidae dae dac Ga- Lovel- Prolo- Retro- Characters laxtas Aplochiton tia Iroctes pinna Dorsal fin Over pelvics x X X X Over anal X* X Adipose - + + + + Scales - - - + + Homy keel - - - + + Cucumber odour - - - + + Pyloric caeca 4- -1- + - - Vomerine shaft long - long short short Vomerine teeth - - - + + Basi branchial - - - -1- + teeth Palatine teeth - - - + + Mesopterygoidal u u u m m teeth Extrascapular - - - + -1- Ectopterygoid ~ slender splint ~ + + Coracoid-cleithrum - - - + + process Posterior pubic - - - + -1- symphysis Pubic foramen - - - + + Caudal skeleton 1 +5 1 + 5 1 +5 1 +6 1 +6 Branched caudal 14 14 14 16 16 rays Nuptial tubercles - - + + + Ovaries both both both left left autumn spawning may return to fresh water the following spring and are transparent and elongate; mostly mature adults return one year later (age about 2 years) to spawn and die (see Jolly, 1967; McMillan, 1961; Milward, 1966). Protolroclidae.-lAXXXe is known of this family, with one species extinct the other rare. McDowall (1976) and Berra (1982) have described what is known of life histories. The eggs are small (~ 1 mm) round and demersal, and are probably deposited in upstream fresh waters. The larvae are not known but believed to be carried to estuaries or the sea to develop, probably for about six months, and return to freshwater in spring as slender transparent juveniles (Fig. 80). Aplochilonidae. — \n Lovettia, mature adults migrate from the sea in spring to spawn in fresh water, and are strongly dimorphic. The male's reproductive opening migrates forward to the isth- mus and the opercular flaps become elongated and papillated. Fecundity is very low (=1 50-200). The tiny eggs (= I mm) are demersal and spherical, and are attached in clusters to hard surfaces (logs, stones, etc.) taking up to 23 days to hatch, and the larvae drift downstream to sea. The post spawning adults die. The life cycle is essentially annual. Larvae at hatching are 152 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Fig. 79. Young of Relropinna retropmna. 35 mm (above); and Aplochilon sp., 24 mm (below). 5-7 mm long with little yolk anteriorly below pectoral fins. They are very elongate, the vent posterior at more than 75% body length, the trunk encompassed by a low finfold from head around tail to yolk sac. Small pedunculate pectoral fins occur. Pigmen- tation is confined to the eyes and a narrow band in mid-ventral between head and vent. Newly hatched larvae disperse to sea and are not fiirther studied (see Blackburn, 1950). Aplochilon taeniatus is recorded spawning in ft-esh water dur- ing winter, the small (1.5 mm), spherical eggs being demersal and attached to firm benthic objects, fecundity 2,500-3,000 and development about 20 days. The larvae are very elongate and slender with a yolk sac beneath the pectoral fins. The vent is at about 75% of body length. A finfold encompasses the trunk and tail. Campos (1969) shows a single large melanophore just in front of the vent. His figure of a larva presumably 8 mm long (he states 80 mm) shows a series of melanophores along the abdomen and a few on the lower caudal peduncle. Recent col- lections of larval Aplochiton from Fiordo Aisen in southern Chile show that some movement to sea occurs. At a length of 24 mm the late larva has well differentiated rays in the dorsal, caudal, anal and pectoral fins and distinct pelvic fin buds are evident (Fig. 79). An adipose fin is also differentiated. Pigmen- Fig. 80. Young of Galaxias maculatus. 14.5 mm (above); and Prolotrocles maraena 35 mm (below). McDOWALL: SALMONIFORMS 153 tation is sparse, limited to spaced melanophores along the ab- domen. The larva remams very elongate, the vent at about 85% of total length. Eigenmann ( 1 928) reported that A. manmis (=A. taenialus) spawns in the sea but this has never been corroborated (see Campos. 1969). Galaxiidae. — Diadromous species: Spawning is usually in fresh- water. Eggs of Galaxiasfasciatus are deposited in autumn-win- ter on stream-side forest debris during floods and develop out of water, hatching when re-immersed during a subsequent flood. The larvae go to sea on hatching, returning in spring as elongate, transparent juveniles about 45 mm long. A minor metamor- phosis involves shrinkage at freshwater entry. The eggs are of moderate size (~2 mm) and number many thousands; devel- opment takes about 30 days. Most other diadromous species have unobserved habits. G. maciilatus spawns in tidal estuaries where streamside vegetation is inundated at high spring tides and development takes place between successive series of spring tides. Most adults die after spawning and larval life is marine. The eggs are simple, spherical, demersal and adhesive, varying from 1-2 mm diameter and more or less colourless. Benzie (1968a) described eggs of G. maculatus as "finely etched." Lar- vae at hatching have a well developed yolk sac. with a single oil globule, the sac below and behind the pectoral fins. The larvae are slender and elongate at hatching, 7-8 mm long, and have the finfold continuous from about mid dorsal around tail to yolk sac. The vent is posterior, at about 75% of total length. Non-diadromous species: Most species in the family are non- diadromous (31 of 37 species). Those known spawn on sub- strates near adult habitats and the pelagic "whitebait" juvenile stage is omitted. Eggs are laid in aggregations (G. vulgaris). Lar- vae on hatching, where described, resemble those of G. mac- ulatus. Galaxiella pusilla is distinctive in being sexually dimorphic, spawning in pairs, the females laying eggs individually on stream vegetation. Individual placement of eggs is also reported for Brachygalaxias bullocki. The ability to aestivate is recorded for some species (Neochanna. New Zealand) and spawning follows restoration of water. It is suspected in others (Galaxiella. Aus- tralia; Brachygalaxias. Chile) and may involve drought survival of eggs (see Benzie, 1968a, b; Backhouse and Vanner, 1978; Cadwallader, 1976; Campos, 1972; McDowall 1968b, 1978; McDowall et al., 1975; Mitchell and Penlington, 1982). Little IS known about the marine larval/juvenile life of any of these southern salmoniforms. Small numbers of Galaxias larvae (Fig. 80) have been collected at sea (McDowall et al., 1975), as have a few, usually pre-migratory Retropinna. The presence of a pelagic-living, transparent, elongate, migratory juvenile seems to be common to most species that are marine or lacustrine at some slage— Galaxias. Retropinna. Prototroctes. Aplochiton. This is likely to have little phylogenetic significance but to relate more to their pelagic, oceanic habits. These small fish resemble many other unrelated fish with pelagic juveniles. The marine, pelagic phase is followed in all instances by a minor metamorphosis on entry to fresh water. Principally this involves rapid assumption of pigmentation and in some species a distinct change in body form. Shrinkage is recorded in a few species. Identification of oceanic larvae and juveniles to family is assisted by dorsal fin position and the early development of an adipose fin in all but galaxiids. The elongate form with the vent at about 75% of total length is helpful. Differences have been recorded in pigment patterns between some of the diadromous galaxiid juveniles although insufficiently to use as diagnostic differences (McDowall and Eldon. 1980). Meristic differences between species are of little value for specific identification ow- ing to their wide ranges and latitudinal variability. Identification remains a difficulty and improvement will depend on the capture and examination of additional material. Fisheries Research Division, Ministry of Agriculture and Fisheries, Christchurch, New Zealand. Osmeridae: Development and Relationships M. E. Hearne OSMERIDAE, the true smelts, are a small family of northern hemisphere salmoniform fishes. The family includes 2 subfamilies, 6 genera, 10 species, and 13 forms (monotypic and subspecies). They have marine, anadromous or landlocked and freshwater life histories in the Pacific, Arctic and Atlantic oceans and their drainages (McAllister, 1963). These silvery tasty little fishes are captured by both recreational and commercial pur- suits along the open coast beaches and rivers during their spawn- ing runs. Development The smelts are highly selective spawners, choosing to spawn on very specific sub-tidal areas, beaches and rivers. Some species spawn in the daytime, and some spawn at night. The eggs of osmerids possess an adhesive membrane that attaches to sand grains and plant material. This anchor membrane results from the ruptunng of an outer "chorion" during spawning, which turns out and onto the substrate. This adaptation for demersal spawning is observed in all 10 species of osmerids (Hamanda, 1961; Thompson et al., 1936; Morris, 1951; McAllister, 1963; Simonsen, 1978; DeLacy and Batts, 1963; Hearne, 1983). The first description of smelt development was made by Eh- renbaum (1894) for the Elbe River smelt, Osmerus eperlans illustrating embryological stages, yolk-sac larva, transforming larva, and the juvenile. Up to now, the yolk-sac stage of many of these species has been at least illustrated or photographed. 154 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Fig. 81. (A) Yolk-sac larvae of Spirinchus starksi. Osmeridae, 7.4 mm, from Morris (1951); (B) Yolk-sac larvae of Plecoglossus altivelis. Plecoglossidae, ca. 6.0 mm, from Okada (1960); (C) Post yolk-sac larva of Salangichlhys inicrodon (Salangidae), ca. 7.0 mm, drawn from two specimens in CAS 504 1 5. Ahlstrom (pers. comm.) deteimined that, in general, osmerid larvae were unique from other elongate larvae in the California Current system by having a single mid-ventral row of mela- nophores below the gut. Based on all the available larval de- scriptions for osmerids, including the Atlantic forms, this single row of melanophores appears to be a hallmark of the family. Listed in Table 37 are sources of larval and juvenile descriptions for the ten species of smelts. These descriptions use various characteristics for each species and are not comparative in design. Melanophore counts are referred to by Yapchionges (1949), Follett (1952), Simonsen (1978), Morris (1951), Dryfoos (1965) and Moulton (1970). Myomere counts were used by Delacy and Batts ( 1 963). Cooper (1978) and Morris (1951) used both myomere and melanophore counts. Larval osmerids have the following external features in com- mon: elongate body shape; gut about 75% body length; mouth sub-terminal; head dorso-ventrally flattened; lower jaw not well- developed in early larvae; conspicuous choroid fissure in ventral third of eye with ventral rim of clear choroid tissue; stalked pectorals, stalk becoming more pronounced in late larvae; yolk sac positioned 6-12 myomeres posterior to the pectoral base; finfold extending from midbrain area to tail, from mid-yolk sac to anus, and from anus to tail; no dorsal melanophores; scattered melanophores (20-50) on ventral half of yolk sac; 0-2 mela- nophores on posterior ventral half of yolk sac; single row of melanophores along ventral midline of gut, sometimes extend- ing into finfold; 1-3 melanophores on dorsal surface of gut at the anal bend; single row of melanophores on ventral midline of tail; conspicuous opaque liver ventral to foregut (Ehrenbaum, 1894; Yapchionges, 1949; Morris, 1 95 l;DeLacy and Batts 1963; Dryfoos, 1965; Eldridge, 1970; Blackburn, 1973; Cooper, 1978; Heame, 1983). A comparative study of four of the species oflTOregon (Heame, 1983) used ventral melanophore counts and myomere counts in an attempt to characterize the larvae of these species. Ten- dencies in these counts showed Hypomesus pretiosus and Spi- rinchus starksi to have high ventral melanophore counts while Spirinchus thaleichthys and Thaleichthys pacificus have lower melanophore counts. Myomere counts showed tendencies that further separated each similarly pigmented pair. Table 37, Sources of Larval and Juvenile Descriptions of Smelts, (x no description found.) Taxon Larvae Juveniles Hypomesus pretiosus Hypomesus transpacificus Spirmchus lanceolatus Spirinchus starksi Spirinchus thaleichthys Thaleichthys pacificus Alios merus elongatus Mallotus villosus Osmerus mordax Osmerus eperlanus Yapchionges, 1949 X Hikita, 1958 Morris, 1951 Dryfoos, 1965; Moulton, 1970 DeLacy and Batts, 1963 X Schmidt, 1906c Cooper, 1978 Ehrenbaum, 1894 Follett, 1952 Simonsen, 1978 Hikita, 1958 Heame, 1983 Simonsen, 1978 Baraclough, 1964 Heame, 1983 Templeman, 1948 Cooper, 1978 Ehrenbaum, 1894 HEARNE: OSMERIDAE 155 The transformational stages of osmerids are not fully known, since complete developmental series have not been reported for all of the species. However, it is apparent from rearing studies (Morris, 1951; Cooper, 1978) that caudal flexion occurs after yolk absorption and along with median fin formation. The pelvic fins arise from the ventral body musculature as prominent buds after the median fin rays have formed, and appear stalked, be- coming inserted as the ventral musculature joins ventrally. The pectoral fins are present at hatching and remain pedunculate until postflexion stages acquire adult-like pigmentation. During flexion an additional series of melanophores forms along the ventro-lateral edge of the body musculature and ap- pears as a double row of spots from ventral view. There are also count differences between the species in these secondary me- lanophores (Heame, 1983), and they may aid in identification of flexion and postflexion stages. The postflexion stages of two species of osmerids have been erroneously described as new species belonging to other families by Chapman ( 1 939). Hubbs (1951) has shown that one of these smelts, placed in the family Paralepididae as Lestidium parn. is actually a late postflexion stage of Thaleichthys pacificus. and the other one, placed in the family Sudidae as Sudis squamosa, is a postflexion Mallotiis villosus. The blackened gut cavities of the postflexion stages of these two species, lend a distinct re- semblance to the midwater-inhabiting sudids and paralepidids, and also suggest a unique departure from the developmental trend of the other species that may warrant the use of the term "pre-juvenile" as defined by Hubbs (1943). Relationships In a recent statement on classification, Rosen ( 1974) proposed an infraorder Salmonae to include two suborders, the Argen- tinoidei and Salmonoidei, the Osmeridae being placed in the latter under the superfamily Osmeroidea (with the Plecoglos- sidae, Retropinnidae, and Salangidae). On the basis of embry- ological and larval features, Soin (1980) characterized different types of salmoniform fishes. He placed the Piecoglossidae and Osmeridae in the same category based on similar egg mor- phology (presence of an anchor membrane), degree of devel- opment at time of hatching and at time of yolk absorption. In a study of stomiiform fishes using adult characters. Fink and Weitzman (1982) placed the families Osmeridae, Salangidae, Piecoglossidae, Retropinnidae, and Galaxiidae all together as "unresolved sister taxa." The larvae of osmehds (Spin nchus slarksi. Fig. 8 1 A) are strik- ingly similar to larval plecoglossids (Plecoglossus alttvelis. Fig. 8 1 B). The yolk sac of these two families is positioned such that its posterior edge is near myomere 11-12. The plecoglossids also have a single median ventral row of melanophores and, as development proceeds, another latero-ventral row of spots ap- pears along the ventral edge of the body musculature, just as in osmerid development. Photographs of the yolk-sac stage of Salangichthys microdon, Salangidae, (Okada, 1960: pi. 17) show that the yolk-sac mor- phology is different than in the Osmeridae and Piecoglossidae. The yolk sac of Salangtchthys microdon is co-extensive with the undersurface of the gut and is more oblong shaped (pyriform) than the more rounded, anteriorly placed yolk sac of the os- merids and plecoglossids. The post yolk-sac larvae of salangids (Fig. 81C) are nearly identical to those of osmerids and pleco- glossids, exhibiting the single median ventral row of melano- phores. Also, the eggs of salangids are different than the osmerid- plecoglossid type by having, instead of an anchor membrane, an anchoring structure that is composed of various kinds of filaments that turn out and onto the substrate (Wakiya and Takahashi, 1913). Larval development is not yet documented for the Sundasalangidae, however adults of this minute family of salangoid fishes have ventral pigment patterns (Roberts 1981: fig. 1) that are strikingly similar to the postflexion pigment pat- terns of osmerids. The same ventral pigment patterns (single ventral midline, paired latero-ventral melanophores) can also be seen in adults of Salangidae (Okada, 1960). One interpretation may be that the similarities in ventral pigment patterns and egg morphology may be the retention of a trait of an ancestor common to the Osmeridae, Piecoglossidae, and Salangidae, and give support to theories arising from sys- tematic observations of adult salmonoids that these families are closely related to each other and not to the other salmoniform families. 184 Day Street, San Francisco, California 94131. Argentinoidei: Development and Relationships E. H. Ahlstrom, H. G. Moser and D. M. Cohen THE argentinoid fishes as here discussed have been consid- ered a suborder by Cohen ( 1 964b) and many other authors and a super-family of an expanded suborder that also includes the alepocephaloids by Greenwood and Rosen (1971). The latter group is not treated at length in this book, because little infor- mation on alepocephaloid ELH stages has appeared since Beebe's (1933a) survey which showed they hatch from large eggs and have direct development. The argentinoids sensu strictu appear to be monophyletic on the basis of four derived characters. One character concerns the development of rays in the finfold of the larva and is described later in this paper. A second character is the development of pustules on the inner surface of the chorion (not known for opisthoproctids). A third character relates to the swimbladder, which, when present, is served by a unique kind of rete mirabile, first described by Fange (1958) and further investigated by Marshall (1960) who named these structures micro-retia mirabilia. A fourth unique character, and one which never has been adequately studied and documented, is the ten- dency in the group for the vomer and palatines to assume the functions of the premaxillary and maxillary. 156 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Table 38. Literature References for Ontogenetic Stages of Argentinoids. Species Egg Transformation stage Argentinidae Argentina elongata A. silus A. sphyraena Glossanodon leioglossus G. polli G. semifascialus Microstoma microstoma Nansenia groenlandica N. oblita Xenopthahnichthys danae Bathylagidae Balhylagus antarcticus B. euryops B. longirostris B. nigrigenys B. ochotensis B. schmidti B. stilbius B. wesethi Opisthoproctidae Balhylychnops exitis Dolichopteryx spp. Dolichopteryx longipes Macropinna microstoma Opisthoproctus grimaldii Rhynchohyalus natalensis H'interia telescopa Robertson, 1975a Schmidt, 1906c Sanzo, 193 Id Sanzo, 193 Id Nishimura, 1966 Sanzo, 193 Id Sanzo, 193 Id Yefremenko, 1982 Pertseva-Ostroumova and Rass, 1973 Ahlstrom, 1969 Ahlstrom, 1969 Ahlstrom, 1969 Holt, 1898; Schmidt, 1906c Schmidt, 1906c, Sanzo, 193 Id Schmidt, 1918, Sanzo. 193 Id Nishimura, 1966 Lo Bianco, 1903; Schmidt, 19 If Sanzo, 193 Id Schmidt, 1918 Schmidt, 1918; Sanzo, 193 Id Yefremenko, 1979a, 1983 Brauer, 1906; TSmng, 1931 Ehrenbaum, 1905-09; Murray and Hjort. 1912; Roule and Angel, 1930; Beebe, 1933b Pertseva-Ostroumova and Rass, 1973 Ahlstrom, 1972b Dunn, 1983a Ahlstrom, 1965, 1972b Ahlstrom, 1965, 1972b Roule and Angel, 1930 Beebe, 1933a Chapman, 1939 Schmidt, 1918 Bertelsen et al., 1965 Belyanina, 1982b Schmidt. 1906 Sanzo, 1931d Poll, 1953 Nishimura, 1966 Schmidt, 1918 Schmidt, 1918 Schmidt, 1918 Bertelsen, 1958 TSning, 1931 Beebe, 1933b Ahlstrom, 1972b Dunn, 1983a Cohen, 1960 Although now there seems to be general agreement as to the genera to be included in the group, their internal arrangement is an unsettled matter. Opinions range from those of C. L. Hubbs (1953), who relegated all to a single family, to those of Chapman (1948 and papers cited therein), who advocated eight different families. Subsequently Cohen (1964b) classified the group in three families using inadequately evaluated characters. Family Argentinidae (most genera are probably worldwide): Subfamily Argentininae (benthopelagic. outer shelf to slope): Argentina (12 species) and Glossanodon (seven or more species). Subfamily Microstomatinae (mesopelagic)': Microstoma (one or two species), Nansenia ( 1 3 species) and Xenophthalmichthys (one or two species). Family Bathylagidae (meso-to bathypelagic): Bathylagns (including Leiiroglossiis and Therobromus; about a dozen to 1 5 species; several species in the Arctic and Antarctic). Family Opisthoproctidae (mesopelagic): Group 1: Macropinna (one species; restricted to N. Pacific and east- em S. Pacific), ' Herein considered a distinct family. Opisthoproctus (two species), Rhynchohyalus (one species; Atlantic and Indian Oceans) and H'interia (one species). Group II: Balhylychnops (one or more species), and Dolichopteryx (perhaps half a dozen species). An alternate arrangement presented by Greenwood and Ro- sen (1971) and essentially based on inadequately evaluated char- acters in the branchial arches and caudal fin skeleton proposed two families within a superfamily Argentinoidea: Family Ar- gentinidae and Family Bathylagidae with Subfamily Bathyla- ginae (including Microstomatidae) and Subfamily Opistho- proctinae. Unanswered questions concerning the systematics of the group are numerous and exist at all levels. Following is a summary. ( 1 ) What are the external relationships of the argentinoids? (2) How many distinct lineages exist within the group, how should they best be arranged with respect to each other, and how many families should be recognized? (3) Do Argentina and Glossan- odon constitute a monophyletic group? If not, where does each belong? (4) How many genera should be recognized among the bathylagids? (5) Within the opisthoproct group do the elongate species in the Bathylychnops-Dolichopieryx group and the short- bodied species in the Opisthoproctus group constitute mono- phyletic lineages and if so should they be named? (6) Since species complements of genera are inadequately known, espe- AHLSTROM ET AL.: ARGENTINOIDEI 157 Table 39. Characters of the Eggs of Argentinoidei. Number of Distribution of Diameter of Species Diameter oil globules oil globules oil globules Source Argentina stalls 1.31-1.66 vegetal pole 0.27-0.46 Original Argentina siliis 3.0-3.5 vegetal pole 0.95-1.16 Schmidt, 1906c Argentina sphyraena (Mediterranean) 1.60-1.68 vegetal pole 0.44 Sanzo, 193 Id (North Sea) 1.70-1.85 vegetal pole 0.37-0.47 Schmidt, 1906c Argentina elongata 1.67-1.80 vegetal pole 0.35-0.45 Robertson, 1975a Glossanodon leioglossus 1.44-1.52 vegetal pole 0.36 Sanzo, 193 Id Glossanodon semifasciatus 1.5-1.6 vegetal pole 0.36 Nishimura, 1966 Microstoma microstoma (Atlantic) 1.60-1.72 vegetal pole 0.48-0.52 Sanzo, 193 Id (Pacific) 2.05-2.38 vegetal pole 0.49-0.82 Original Nansenia Candida 1.39-1.56 vegetal pole 0.41-0.49 Original Nansenia crassa 1.05-1.30 vegetal pole 0.30-0.35 Original Nansenia ohlita 1.39-1.56 vegetal pole 0.40-0.53 Sanzo, 193 Id Bathylagiis antarclicus 1.8-2.2 3-8 * 0.2-0.3 Yefremenko, 1982 Bathylagus schmidti 1.65-1.90 up to 9 * Ahlstrom, 1969 Bathylagiis slilhiiis 1.01-1.21 15-25 *■ Ahlstrom, 1969 Bathylagus urotranus 1.03-1.21 15-25 * Pertseva-Ostroumova and Rass, 1973, and original Bathylagus ochotensis 0.92-1.1 many to two clumps ** Original Balhylagits wesclhi 0.90-1.10 12-20 ** Ahlstrom, 1969 Bathylagus nigrigenys 0.83-1.09 12-20 ** Pertseva-Ostroumova and Rass, 1973, and original ' First grouped at vegetal pole, then move to beneath embryo, then coalesce to one at each equatonal pole. • Numerous globules at vegetal pole then coalesce to one clump at each equatonal pole. cially the mesopelagic ones, do presently available early life history specimens help define the species composition of argen- tinoid genera? Development Eggs are known for 1 3 species of argentinoids and larvae for 22 species (Table 38). We present in this paper eggs of 5 ad- ditional argentinoid species and larvae of 8 additional species. These are: eggs and larvae oi Argentina sialis. Microstoma sp., Nansenia Candida and N. crassa; larvae only for Bathylagus argyrogaster. B. bencoides. B. pacificus and Balhylychnops ex- ilis: eggs only for Bathylagus ochotensis. Eggs The eggs of argentinoids are pelagic, round, have a moderate to narrow perivitelline space, segmented yolk and a chorion with distinctive pustules on the inner surface (Table 39, Fig. 82). Egg diameters and oil globule characters are given in Table 39. Argentinoid larvae hatch as relatively undifferentiated yolk- sac larvae, regardless of egg size. That is, yolk-sac larvae of A. silus at 7.5 mm, newly hatched from eggs 3.0-3.5 mm diameter, are at about the same stage of development as 3 mm bathylagid yolk-sac larvae which hatch from 1 mm eggs. In most marine fishes larger eggs produce more highly differentiated hatchlings. Larvae Body form. — Argentinid and bathylagid larvae are slender, those of microstomatids are deeper-bodied, and opisthoproctids have a wide variety of body shapes ranging from the slender larvae of Balhylychnops to the deep-bodied Opisthoproctus (Table 40, Figs. 83-87). The gut is elongate and straight in argentinids and bathylagids, with the exception of B. milleri where the gut is straight but only about half the body length. In argentinids the gut is lined with transverse rugae for almost the entire length. In most bath- ylagids the gut has two distinct sections: an anterior section with longitudinal internal ridges, separated by a valve from a shorter posterior section with transverse rugae. The anterior section in B. hericoides and B. longirosths is markedly smaller in diameter compared with other species. Larvae of j5. wesethi. B. nigrigenys and B. argyrogasterhave transverse rugae along the entire length of the gut and the anterior section is relatively larger in diameter and thin-walled. Also the posterior section is subdivided by a second valve. B. ochotensis larvae develop a similar structure. The gut in microstomatid larvae is long, but anteriorly has an elongate S-shaped fold that lies flat on the left side (Fig. 84). The lumen of the anterior folded section is characterized by longitudinal ridges whereas the posterior straight section has transverse rugae. The short pyloric section has longitudinal ridges. Schmidt (1918) shows the gut extended beyond the finfold mar- gin in Nansenia ohlita and trailing in early stage Microstoma microstoma larvae but we have not seen this in any specimens of these genera. In opisthoproctids the gut is elongate in Balhylychnops and Dolichoptery.x and relatively shorter in the deeper-bodied gen- era, Macropinna. Rhyncholyalus and Opisthoproctus. In all gen- era there is a sac-like stomach, which exits through a constricted pyloric section to the intestine. In Balhylychnops and Doll- chopteryx the sac is elongate and pointed at its tip whereas in the other genera it is more rounded in form. The sac lies on the left side, except in Balhylychnops where it lies on the right. In the latter genus the pyloric constriction leads into a short but prominent bulbous section. DoHchopteryx is similar but lacks 158 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Fig, 82. Eggs of argentinoids. (A) Argentina sialus. 1.5 mm, CalCOFI 5103, Sta. 1 17.35; (B) Microstoma sp., 2.2 mm, CalCOFI 751 1, Sta. 87.90; (C) Nansenia Candida. 1.4 mm, CalCOFI Sta. 60.90; (D) N. crassa. 1.5 mm, CalCOFI; (E) Bathylagus stilhius. 1.1 mm, from Ahlstrom (1969); (F) B. schmidli. 1.8 mm. from Ahlstrom (1969); (G) B. ochotensis. 1.1 mm, CalCOFI 5002 Sta. 60.90; (H) B. weselhi. 1.0 mm, Ahlstrom (1969); (I) B. nigrigenys. 0.96 mm. CalCOFI 5106 Sta. 157.20. the post-pyloric bulb. In Macropinna and Opisthoproctus there is a straight section leading posteriorly from the pylorus, which ends in an S-shaped fold and an enlarged rectal bulb, the latter described by Bertelsen and Munk (1964). The anterior section including the sac and pyloiojs have longitudinal internal ridges while sections posterior to this have transverse rugae. In late larval stages the entire section posterior to the pylorus becomes part of the S-shaped coil. The head is relatively small in argentinids and has a rounded blunted anterior profile (Fig. 83, Table 40). It is slightly larger in most microstomatids, with the exception oi Microstoma sp. (Pacific form) which has a small head. In most microstomatids the head has a rounded, blunted anterior profile and is bent slightly downward from the longitudinal axis. In both families the eye is either round or slightly ellipsoidal. In bathylagids the head is moderate in size but highly various in shape (Figs. 85, 86; Table 40). The snout is generally longer than in Argentinidae and Microstomatidae. Eye shape and structure vary greatly within the bathylagids. Bathylagus milleri has a large, nearly round eye in contrast to AHLSTROM ET AL.: ARGENTINOIDEI 159 v^^&^ t&. **J isf *tii "jv tn^ 'zr • •^^^I''' ."»• *. *S ^^ S -S^* » ^i^^-^j*** « #»♦ "^ »? PIt^i Fig. 83. Larvae of Argentinidae. (A) Argentina stalls. 7.0 mm. CalCOFI 5103 Sta. 1 17.35; (B) A. stalls. 9.0 mm, CalCOH 5104 Sta. 97.40; (C) .-1. stalls. 17.5 mm, CalCOFI 5103 Sta. 120.35; (D) A. stalls. 21.0 mm, CalCOFI 5105 Sta. 123.40; (E) A. silus. 32.5 mm. redrawn from Schmidt (1906c); (F) A. sphyraena. 19.2 mm, ibid; (G) Glossanodon semifasciatus, 12.5 mm, from Nishimura (1966). AHLSTROM ET AL.: ARGENTINOIDEI 161 Table 40. Comparative Morphometry of Aroentinoid Larvae. Mean values (%) of body proportions for three ontogenetic stages (preflexion- flexion-postflexion) are listed. Snoul-anus Eye stalk Snout-anal Snout-dorsal Snout-pelvic distance Head length Head width Eye length length Body depth fin distance fin distance fin distance Species Body length Body length Head length Head length Head length Body length Body length Body length Body length A rgentma sialis 76-78-84 17-21-22 54-44-41 28-24-24 — 9-10-10 0-78-81 0-46-47 0-0-49 Microstoma microstoma ?-?-80 7-7-23 7-7-45 7-7-27 — 7-7-13 7-7-80 7-7-68 0-7-64 Microstoma sp. (Pacific) 76-79-80 17-19-19 53-49-44 31-29-27 _ 8-10-10 0-78-81 0-70-72 0-64-67 Nansenia Candida 74-77-82 21-25-26 60-50-44 36-28-28 — 12-14-16 0-75-82 0-54-58 0-56-61 Nansenia crassa 74-78-80 22-25-28 58-50-44 36-29-24 _ 10-12-15 0-76-80 0-52-57 0-56-60 Nansenia groenlandica 7-78-80 7-27-25 7-50-42 7-21-23 — 7-15-15 7-77-80 7-52-52 7-54-57 Xenophthatmichthys danae ?-?-82 7-7-24 7-7-48 7-7-21 — 7-7-12 7-7-86 7-7-74 ''-7-52 Bathylagus milleri 59-57-61 20-19-26 56-54-52 31-27-26 _ 9-9-15 0-0-71 0-0-50 0-0-45 Bathylagus schmidli 72-76-78 16-19-22 50-52-46 39-26-25 .04-0-0 7-8-10 0-0-79 0-0-57 0-0-55 Bathylagus slilbius 74-77-80 20-22-24 54-53-47 32-25-20 .03-0-0 8-10-13 0-0-79 0-0-57 0-0-55 Bathylagus urotranus 78-82-81 20-24-28 56-53-46 27-18-21 .03-0-0 10-10-12 0-0-81 0-0-61 0-0-59 Bathylagtis pacificus 76-85-81 22-24-25 39-42-44 29-22-18 28-29-20 8-10-13 0-81-80 0-49-48 0-51-51 Bathylagus curyops 78-80-82 18-20-20 46-50-50 31-26-25 10-7-3 10-11-12 0-78-80 0-45-48 0-0-47 Bathylagus bericoides 84-85-89 25-26-26 34-38-36 27-25-22 60-64-36 8-8-9 0-83-88 0-0-52 0-0-53 Bathylagus longiroslris 85-88-92 26-27-25 34-34-34 24-20-19 54-48-27 8-10-10 0-88-90 0-0-53 0-0-57 Bathylagus ochotensis 81-85-90 20-23-23 44-44-44 32-21-21 17-15-15 8-10-11 0-83-87 0-53-54 0-56-56 Bathylagus wesethi 79-89-94 13-26-27 59-53-50 27-16-13 — 9-14-16 0-85-90 0-58-60 0-57-59 Bathylagus nigrigenys 80-86-93 20-29-28 78-60-53 30-18-14 — 12-16-18 0-86-90 0-57-60 0-0-60 Bathylychnops exilis 7-80-82 7-21-22 7-42-38 7-22-18 — 7-8-7 7-82-84 7-71-73 7-66-67 Dolichopleryx longipes 7-74-75 7-24-26 7-44-34 7-22-16 — 7-8-10 7-0-77 7-0-71 7-62-62 Macropinna microstoma 7-64-59 7-26-35 7-52-47 7-22-21 — 7-15-21 7-0-70 7-0-66 7-43-48 Opisthoproctus soleatus 7-7-80 7-7-37 7-7-46 7-7-18 - 7-7-18 7-7-83 7-7-63 7-7-40 other species which have relatively smaller, more elliptical eyes. Eyes are sessile in B. milleri and in the B. wesetht group but are stalked to some degree in all other species known. In B. slilbius and relatives (B. urotranus, and B. schmidli') the stalks are short and found only in early larvae. Stalks are longer and persist into later larval stages in other species, reaching a ma.ximum of 65% of the head length in B. bericoides. In opisthoproctids the head is moderate in size in the slender forms, Bathylychnops and Dolichopleryx, and longer and more massive, with a pronounced hump or bend at the nape, in the deep-bodied genera. All genera have an elongate snout and Bath- ylychnops has a unique triangular flap at its tip. Bathylychnops has round eyes that are rotated slightly dorsoanteriad. In the other genera, the eyes are tubular and directed dorsally, even in the smallest larvae available. Eye diverticulae with associated accessory retinae, characteristic of opisthoproctid adults, begin to form at the end of the larval period. Fins —A major feature of all argentinoid larvae is the devel- opment of a prominent median finfold in which the dorsal and anal fins develop, connected to the trunk by a series of hyaline strands (Figs. 83-87). The first fins to form are the pectorals. In argentinids and bathylagids they are relatively small and de- velop rays late in the larval period. Microstomatid and opis- thoproctid pectoral fins are generally larger; however, there is a wide size range, from relatively small fins in Microstoma to large, fan-like fins in some species of Nansenia (e.g., N. groen- landica) to very elongate pectorals in Dolichopleryx binocularis. Ossification of rays begins earlier in these groups, usually before notochord flexion. After the pectorals, the caudal fin is usually the next to form. In argentinids notochord flexion and development of principal caudal rays occurs at a size about midway in larval growth whereas in opisthoproctids this occurs earlier in the larval pe- riod. In bathylagids the process is somewhat delayed and in some species (e.g., B. euryops. B. milleri) notochord flexion may not be completed until near the end of the larval period. The dorsal and anal fins begin to form at about the stage of notochord flexion in all argentinoids except opisthoproctids, where notochord flexion slightly precedes the appearance of dorsal and anal fins. The anal fin begins forming far posteriad in argentinoids, just posterior to the anus or the point of de- flection of the free terminal gut section. In B. milleri and in the deep-bodied opisthoproctids with coiled guts there is a space between the anus and the anal fin origin. The position of the dorsal fin is varied among argentinoids and forms in the larvae in approximately the same position that it will occupy in the adult. The fin has its most anteriad location in Argentina where its origin is well forward of the midpoint of the body (Fig. 83). The extreme case is found in A. silus where snout to dorsal origin is about 38% of the body length in larvae and about 43% in adults. In most bathylagids the dorsal origin is slightly anterior to mid-body. The exceptions are B. slilbius and relatives, where the dorsal origin is slightly posterior to mid-body, and B. wesethi and relatives where it is located still further posteriad. Fig. 84. Larvae of Microstomatidae. (A) Microstoma microstoma. 1 1.0 mm, from Schmidt (1918); (B) Microstoma sp., 12.0 mm, CalCOFl 5 104 Sta. 90.52; (C) Nansenia Candida. 8.4 mm, CalCOFl 5007 Sta. 1 00.70; (D) N. crassa. 8.5 mm, CalCGR 5 103 Sta. 1 37.50; (E) N. groenlandica, 10.0 mm, from Schmidt (1918); (F) N. oblita, 9.0 mm, ibid; (G) Xenopthalmichthys danae. 16.5 mm, from Bertelsen (1958). 162 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Fig. 85. Larvae of Balhylagus. (A) B. milleri. 27.5 mm, CalCOFl 5106 Sta. 70.60, dorsal view of 9.5 mm specimen at left; (B) B. schmidli. 31.5 mm, CalCOFI Northern Holiday Exped. Sta. 31; (C) B. snlbiits, 23.2 mm, CalCOH 4905 Sta. 1 1 1.38, dorsal view of 8.5 mm specimen at left;(D) B. pacificus. 21.4 mm, CalCOFI 7905 Sta. 63.60; (E) B. euryops 24.0 mm, dorsal view of 14.0 mm specimen at left, from Tuning (1931); (F) B. antarcticus. 26.5 mm, from Yefremenko (1983). AHLSTROM ET AL.: ARGENTINOIDEI Table 41. Meristics of Argentinoid Fishes. 163 Branchiostegal Dorsal Anal Pectoral Pelvic Procurrenl Species Venebrae rays fin rays fin rays lin rays fin rays caudal fin rays Argentina altceae 43-46 5 11-13 13-15 16-18 10-12 australiae 50-53 5 10-12 12-13 13-14 11-13 hrucei 45-47 5 10-12 11-13 18-20 13-14 elongata 52-55 5 10-12 11-14 13-16 11-12 euchus 48-49 5 12 13-15 16-18 10-11 georgei 48-51 5 10-12 10-13 16-19 12-14 kagoshimae 51-52 5 10-12 11-13 15-17 11-12 sialis 47-51 5 10-13 12-15 11-18 10-12 12-Hl silus 65-70 6 11-13 11-17 15-18 12-13 sphyraena 46-55 6 10-12 11-15 12-15 10-12 stewarti 53-54 5 10-12 12-13 18-21 13-15 striata 48-52 5 10-12 11-14 18-21 11-15 10-1-9 Glossanodon leiglossus 49-51 5 12-14 10-13 19-22 11-12 tmeatus 4 11-13 15 18-21 11-13 mildredae 50-52 5 13 13 23 12-13 polli 5 12-14 11-14 19-22 12-13 pygmaeus 43-44 5 10-12 11-13 12-14 10-12 semifasciatus 49 5 11-13 11-13 18-21 10-12 struhsakeri 51-53 12-14 12-13 23-25 13-15 Microstoma microstoma 45-47 3-4 11-12 8-9 8 9-11 11 + 11 sp. (Pacific) 49-50 4 9-11 7-8 11 9 10- -lH-10 Xenophthalmichthys danae 3 10-12 9-10 7 8-9 10-1-9 Nansenia atlanlica 41-42 4 9-10 8-9 12-13 10-11 ardesiaca 46-48 4 9-10 9-10 11-14 10-12 Candida 44-47 3 9-10 8-9 9-11 9-11 11-1-14 crassa 43-46 4 9-10 8-9 11-13 10-11 groenlandica 42-45 3 9-10 8-10 11-13 10-12 ohiita 42-45 4 10-11 9-10 10-11 10-11 Bathylagus amarcticus 2 9-11 21-25 9-10 argyrogasler 2 12 14-15 8 bericoides 48-53 -) 10-11 18-22 10-12 9-10 euryops 44-46 2 9-11 16-19 7-12 7-9 greyae 2 11-13 13 12-13 10-11 longirostris 48-51 2 10-12 19-21 9-12 9-10 mtlleri 51-55 2 6-9 20-28 11-16 6-8 16- -18-1-15-17 nigrigenys 41 2 11-12 14-17 10 8-10 ochotensis 47-49 2 9-12 12-15 9-11 9-10 13- -14+15-16 pacificus 45-49 2 8-9 15-22 7-11 7-10 13+13-14 schniidti 47-52 2 10-11 11-14 8-9 8-9 16- -17+16 stilhius 38-42 2 9-11 11-14 8-11 8-10 12- -16+13-15 urolranus 39-42 2 9-10 10-11 9-11 7-8 12- -14+12-13 weselhi 43-46 2 12-13 14-16 10-11 9-11 14-15-1-14-15 Dolichopteryx anascopa 2 10 12 14 12 bmocularis 2 15 11 14 9 hrachyrhynchus 2 13 12 13 8 longipes 41-44 2 10-11 8-9 13 8-9 Bathylychnops exilis 81-84 2 14-16 13-14 12-13 7 RhynJichyalus natalensis 4 10-12 10 19-20 11-12 Macropinna microstoma 36 3 11-12 14 17-19 10 Winteria telescopa 8 8 12-14 9 164 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Table 41. Continued. Species Vertebrae Branchios rays legal Dorsal fin rays Anal tin rays Pectoral fin rays Pelvic fin rays Procurrent caudal fin rays Opisthoproctus grimaldii soleatus 31 2 2 12-14 10-12 8 13-14 11 9-11 10 The dorsal fin forms in a variety of positions among micro- stomatids. In most species of Nansenia. the dorsal fin originates slightly posterior to mid-body, although in some species (e.g., A', groenlandica), its origin is slightly anterior to mid-body. The dorsal origin is further posteriad in Microstoma. In M. microsto- ma predorsal length is about 67-68% of the body length and assumes a more anterior position in adults (ca. 63%). In larvae of the Pacific species predorsal length is about 75% of the body length, and is slightly more posteriad in adults. In adult Xen- ophthalijuchthys the dorsal origin is at mid-body; however, in the 16.5 mm specimen from the Atlantic (Bertelsen, 1958) pre- dorsal length is 62% of the body length. In our single larva (12.2 mm) from the Pacific predorsal length is 75% of body length, indicating a marked anteriad migration during ontogeny or strong allometric growth posterior to the dorsal fin. Alternatively, the Pacific form may prove to be distinct when adult specimens are captured. The dorsal fin in opisthoproctids is located posteriad on the body. This is most marked in the slender forms, Bathylychnops and Dolichopteryx. and reaches an extreme in D. hinocularis where predorsal length is greater than Vj of the body length. In the deep-bodied genera the dorsal origin is posterior to mid- body, but less so than in the slender-bodied forms. The pelvic fins are the last fins to form in most argentinoids, usually late in the larval period. The exception is opisthoproctids where the pelvic fins form early in the larval period. In argen- tinids, bathylagids and microstomatids the pelvic fins form at about mid-body, below the dorsal fin. In the slender opistho- proctid genera the pelvics form well back on the body, but anterior to the dorsal fin. Among the deep-bodied genera, Op- isthoproctus forms the pelvics far back on the body, beneath the dorsal fin. In Rhynchohyalus and Macropinna the pelvics de- velop just posterior to mid-body and anterior to the dorsal fin. In the larvae the fins are elevated to the sides of the body. This position persists in juvenile and adult Macropinna where the fins are located just behind and below the pectoral fin bases. The pelvic fins become elongate in Dolichopteryx and the deep- bodied genera. The pelvic fin base is pedunculate in opistho- proctid larvae, a condition that persists into the adults of some genera, notably Dolichopteryx. Argentinoids, except Microsto- ma. Xenophthalmichthys and some species of Dolichopteryx. develop adipose fins late in the larval period. A summary of meristics of argentinoids is given in Table 41. The sequence of ossification of fins and other skeletal elements o( Bathylagus schmidti is described by Dunn (1983a). Pigmentation. — \n argentinids, pigmentation consists of a series of 6-8 ventral trunk blotches that extend from the pectoral fin base to the end of the gut (Fig. 83). The series is continued posteriorly as I or 2 median ventral blotches and ends as a large blotch at the caudal region. The number of blotches is constant for each species, as is the sequence of formation. In Argentina sialis and Glossanodon the ventral blotches expand dorsally as lateral bars, but this does not occur in A. silus and A. sphyraena. These latter species differ additionally in lacking the internal head pigment which develops in A. sialis and Glossanodon lar- vae. A feature common to most microstomatid larvae is a heavy line of embedded pigment above the gut (Fig. 84). In Micro- stoma this pigment continues forward to the gill arches and within the head anteriorly to the snout. In Nansenia, head pig- mentation is superficial, or concentrated ventrally on the head. In Microstoma, an embedded dorsal line of pigment is located posterior to the dorsal fin. Dorsal pigmentation in Nansenia may take the form of a series of embedded blotches (e.g., N. crassa) or an embedded line of melanophores running the length of the body (e.g., N. ohlita). Most microstomatids have con- spicuous melanistic pigment associated with the caudal fin re- gion. A notable feature oi Microstoma and some Nansenia (e.g., N. crassa) is the presence of heavy melanistic pigment at the curve of the gut loop. Our single damaged specimen of Xenoph- thalmichthys (12.2 mm) has pigmentation similar to Micro- stoma but lacks the posterior dorsal body pigment and has a series of slanted melanophores along the hypaxial myosepta. Pigment patterns in bathylagids may be grouped into two categories— those species with large isolated melanophores (Fig. 85) and those with linear series of smaller melanophores (Fig. 86). Bathylagus milleri has a unique pattern of opposing dorsal and ventral midline melanophores, large melanophores on the head and pectoral fin base and a large lateral blotch on the notochord tip. Bathylagus stilbius and B. urotranus develop a series of 5-6 melanophores on each side of the posterior section of the gut. A single large melanophore, is found on the lower trunk midway between the pectoral fin and the anus and the head has mela- nophores, chiefly on the upper and lower jaws and opercle (Fig. 85). B. schmidti differs in having a series of lower trunk blotches and 1 or 2 postanal lateral blotches. Bathylagus euryops has a series of 3-6 melanophores on the lateral surface of the gut and 3-5 large melanophores on the lateral surface of the trunk (Fig. 85). Other pigmentation consists Fig. 86. Larvae of Bathylagus. (A) B. hericoides. \1 .1 mm, Dana Sta. 4007, dorsal view ofl 1.8 mm specimen at left; (B) B. longirostris. 20.1 mm, SIO/STOW XIII Exped., dorsal view of 12.4 mm specimen at left; (C) B. ocholensis. 21.5 mm, CalCOFl 5106 Sta. 77.65, dorsal view of 8.5 mm specimen at left; (D) B. wesetlu. 1 1.3 mm. from Ahlstrom (1972b), dorsal view of 8.5 mm specimen at left; (E) B. mgrigenys. 21.8 mm, SIO Shellback Exped. Sta. 92, dorsal view of 8.7 mm specimen at left; (F) B. argyrogaster. 17.1 mm, Dana Sta. 4003. AHLSTROM ET AL.: ARGENTINOIDEI 165 166 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM AHLSTROM ET AL.: ARGENTINOIDEI Table 42. Characters used in Analysis of Four Argentinoid Groups. 167 Character number Dervied character state Outgroup 1 Accessory cartilage at posterior tip ceratobr. 5 2 PU, + U, fused 3 Light organs present 4 Frontals fused 5 Epibr. 4 with one post. art. surface 6 Larval gut with stomach 7 Pelvic fins form early and large 8 Swimbladder absent 9 Urodermal absent 10 LL scales extend onto caudal fin 1 1 Larval gut folded 12 Extrascapular attached to pterotic 13 Uncinate process lacking on epibr. 4 14 Pectoral fin forms early and large Osmerids Greenwood and Rosen, 1971 Teleosts in general Goody, 1969 Teleosts in general Bertelsen and Munk, 1964 Teleosts in general Cohen, 1964b Osmerids Greenwood and Rosen, 1971 Osmerids This paper, Heame (this volume) Osmerids This paper, Heame (this volume) Teleosts in general Cohen, 1964b Teleosts in general Greenwood and Rosen, 1971 Teleosts in general Osmerids Heame (this volume) Teleosts in general Chapman, 1942 Osmerids Greenwood and Rosen, 1971 Osmerids This paper, Heame (this volume) of a line of small melanophores above and below the notochord tip, a patch of melanophores on the opercle and groups of small melanophores on the upper and lower jaws. Balhylagns anlarcti- cus has 3 lateral gut spots, a large lateral trunk melanophore at the I0lh-I2th myomere, and head and notochord pigment sim- ilar to that of B. euryops. Early larvae of B. pacificus have a large lateral blotch at mid-body and another one posteriad on the body. Initially these melanophores are located at the junction of the gut and body but in later larvae are located on the trunk. Later a 3rd blotch forms midway between these two. A 4th lateral trunk blotch forms in some late larval specimens between the pectoral fin and the large mid-body blotch and melanophores form lateral to the liver and at the free terminal section of the gut. Head and notochord pigment is similar to B. euryops and B. antarcticus. Bathylagiis hericoides is unusual in having only a series of as many as 18 lateral gut melanophores (Fig. 86). Late postflexion larvae develop pigment on the lower jaw, isthmus, opercle, pec- toral fin base and lateral caudal peduncle. Bathylagus longiros- tris develops a heavier pattern of pigmentation, beginning with a series of small melanophores on the posterior section of the gut in early larvae. Also in preflexion larvae a series of rect- angular-shaped melanophores develops on the hypaxial myo- meres. Later in the larval period the lateral gut series is extended forward along the entire gut, although with wider spacing than on the posterior gut section. Also, the epaxial myomeres develop rectangular-shaped melanophores, beginning posteriorly and ac- cruing anteriorly. The head develops pigmentation from the opercle to the jaws (Fig. 86). Bathylagus ochotensis develops a similar pigment pattern except that the melanophores on the posterior gut section are comparatively larger and fewer, the anterior region of the gut lacks melanophores and the epaxial myomere series is limited to the posterior region. Larvae of B. wesethi. B. nigrigenys and B. argyrogaster have a similar pigment pattern that differs markedly from that of other Bathylagus (Fig. 86). Initially there is a series of paired melanophores dorsolateral to the gut, extending from the pec- toral fin base to the terminal section. These remain throughout the larval period but become embedded and obscured in late larvae. Bathylagus nigrigenys begins with about 8 pairs, which increase to 10, whereas B. wesethi begins with 6 pairs and has 7-8 during most of the larval period. Both species develop pig- ment at the notochord tip; B. wesethi has a dorsal and ventral spot, while B. nigrigenys has only a ventral spot. At notochord flexion a series of melanophores appears along the hypaxial region of the body and, soon after, a series develops along the epaxial myomeres. More lateral series are added and in late larvae the entire body is covered. Melanophores also form in the median finfold of advanced larvae. Initially head pigmen- tation consists of melanophores on the opercle and jaws but in later larvae the entire head is covered. Opisthoproctid larvae have distinctive and, in some genera, heavy pigment patterns (Fig. 87). Bathylychnops has a dorsal series of 6 large paired blotches that permeate the musculature, bridge across the longitudinal septum and expand onto the fin- fold. A series of 8 large ventrolateral blotches alternate with those of the dorsal series, with the exception that the postanal blotch lies opposite the dorsal blotch and expands to form a band. A large blotch covers the base of the caudal fin. The head IS heavily pigmented with superficial melanophores on the bran- chiostegals, urohyal and lateral brain and deeply embedded me- lanophores in the snout, jaws, cheek and ventral brain region. The lower limbs of the gill arches and their filaments are heavily pigmented as are both the pectoral and pelvic fin bases. The species of Dolichopteryx have lateral series of melano- phores above the gut and some species develop serial melano- phores on the hypaxial myomeres (Fig. 87). Head pigment con- sists of melanophores on the jaws, gill arches and, in most species, the internal snout region. Macropmna develops a series of slant- ed melanophores, one on each hypaxial myomere, and a heavy embedded blotch at the pelvic fin base, that expands both dorsad and ventrad as a band. The caudal fin base has a large blotch Fig. 87. Larvae of Opisthoproctidae. (A) Bathylychnops exilis. 1 5.6 mm. CalCOFI 7203 Sta. 67.80; (B) Ventral view of above; (C) Dolichopteryx hinoculans. 58.0 mm, redrawn from Roule and Angel (1930); (D) Afacropinna microstoma. 1 1.7 mm, CalCOFI 7412 Sta. 120.50; (E) Ventral view of above; (F) Rhynchohyalus natalensis. 23.0 mm, from Bertelsen et al. (1965); (G) Opisthoproctus grimaldii. 14.0 mm from Schmidt (1918). 168 ARGENTINIDAE ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM MICROSTOMATIDAE BATHYLACIDAE OPISTHOPROCTIDAE 3,4,5,6,7, (13), (14) Fig. 88. Cladogram showing the distribution of character states in four nominal families of argentinoid fishes. Numbers refer to characters in Tables 42 and 43. Parentheses mdicate character reversals. and in the gut region there is pigment above the terminal section and ventral to the liver. Head pigment is confined to the lower jaw. The pigment pattern of Rhynchohyalus as described by Bertelsen et al. (1965) consists of a series of four dusky bars beginning at the pelvic fin and ending at the caudal fin base. Embedded beneath these is a layer of diffuse melanophores which becomes denser toward the caudal. The pectoral fin bases are pigmented and in the ventral region there are melanophores on the isthmus and gut. The anal light organ is covered with a melanistic sheath. The late larval specimen of Opisthoproctus grimaldii illustrated by Schmidt (1918) shows a diffuse covering of melanophores over the body and a dusky bar extending down from the dorsal fin. A 10 mm larva of O. so/eat us in our col- lection has a pigment pattern similar to Macropuma, with a series of slanted melanophores on the hypaxial myomeres, embedded blotches at the pelvic and caudal fin bases, pigment at the liver and ventrally at the angles of the lower jaw. Transformation stage In argentinids transformation from larva to demersal juvenile is a prolonged process and pelagic juveniles with the retained larval pigment blotches or bars have been reported many times (see Cohen, 1958; Nishimura, 1966). Morphological changes (e.g., deepening of the body, prolongation of the snout, eye enlargement) and the masking of the larval pigment occur grad- ually. The beginning of this stage may be defined by the folding of the anterior gut region to form a stomach. This occurs at 25- 30 mm in Argentina sialis, but has not been documented for other species. Pelagic juveniles of Glossanodon and A. sialis develop a silvery stripe at the lateral line region. This has not been reported for pelagic juveniles of A. silus and A. sphyraena and may afford an additional character for separating Argentina into two groups. The end of the pelagic juvenile stage, marked by the development of scales and silvery integument, is attained Table 43. Distribution of Char.acter States in Folir Nominal F.MHILIES OF Argentinoid Fishes. Direction of transformation A * B. Character Opistho- number Argenlinidae Micrstomatidae Bathy[ag]dae proctidae 1 B B B A 2 B B B A 3 A A A B 4 A A A B 5 A A A B 6 A A A B 7 A A A B 8 A A B A 9 A A B A 10 A B A A II A B A A 12 B A A A 13 A B B B 14 A B A B AHLSTROM ET AL.: ARGENTINOIDEI 169 at various lengths by different species. Schmidt (1906c) reports complete transformation at about 50 mm in A. sphyraena and at a much larger size in A. silus. Size at completion of trans- formation in Glossanodon species is also in the 50-100 mm size range (Nishimura, 1966). Microstomatids develop a lustrous guanine layer on the in- tegument in late larvae and some species develop distinct ju- venile pigmentation. In Mil rosloma juvenWcs the region of the body from the dorsal fin origin posteriad is more darkly pig- mented than the rest of the body, and grades to a solid black pigment at the caudal fin base. Juveniles of some Nansenia species develop heavy melanistic pigment at the base of the caudal fin and often at the base of the adipose fin (Schmidt, 1918; Kawaguchi and Butler, in press). Bathylagids have a direct transformation and undergo a marked morphological change from the slender larval form to the ju- venile form, characterized by a large head and eyes and deeper body. The gut becomes coiled and covered by a black peritoneal sheath. The head becomes heavily pigmented but the body is slower to develop the black pigment characteristic of all Bath- ylagiis species (other than the B. stilbnis group) and, in species such as B. euryops and B. nulleri. the large larval melanophores are visible in specimens up to 30 mm and 50 mm respectively. In the deep-bodied opisthoproctid genera transformation to the juvenile stage is marked by deepening of the body and at- tainment of melanistic integument and large scales. Cohen ( 1 960) described the large (up to 124 mm) transitional specimens of Bathylychnops which are semi-transparent and retain the large larval pigment blotches. Sexually mature specimens of Doli- chopteryx are semi-transparent, have a membranous body en- velope, poorly developed musculature, an exposed gut covered only by peritoneum, weakly attached fins and melanistic pig- ment of the type usually associated with larvae (Cohen, 1960). Relationships Our survey of argentinoid ontogenetic characters provides insight into some of the systematic questions posed at the be- ginning of the paper. A close relationship between argentinoids and alepocephaloids is not supported since the latter hatch from large eggs (estimated at 3-4 mm based on size of yolk-sac lar- vae), have direct development, and share no specialized onto- genetic characters with argentinoids. Four major argentinoid lineages can be defined by specializations of the eggs and larvae and thus four families recognized: Argentinidae, Microstoma- tidae, Bathylagidae, and Opisthoproctidae. Argentina and Glos- sanodon have generalized larvae except that all known species have distinct lateral series of melanistic blotches or bands, not found elsewhere among argentinoids. The pattern of banding does not separate the two genera. All known bathylagid eggs have multiple oil globules. A num- ber of bathylagid groups are apparent from larval characters: 1) niillcri, 2) slilhms-schmidti-iirotranus, 3) euryops- pad ficus-ant- arcticus, 4) hericoides-longirostris, 5) wesethi-argyrogaster-ni- grigenys. Of these groups, stilbius-schmidti-urotranus has the most generalized morphology and pigmentation, lending no support for its separation as a distinct genus. Opisthoproctid larvae share a number of neotenic features, including a saccular stomach. Except for body shape, Dolichop- teryx shares more derived larval characters with the deep-bodied genera than with Bathylychnops. and the latter has a number of characters unique to opisthoproctids. Division of the family based on body shape is not supported by ontogenetic evidence. Ontogeny offers little information on species composition of genera, because only a fraction of argentinoid eggs and larvae are known. However, egg and larval characters clearly separate Atlantic and Pacific Microstoma as distinct species. Bathylagits hericoides larvae from the Atlantic and Pacific are indistinguish- able. The same is true for B. longirostris from all oceans. Bath- ylagus nigrigenys and B. argyrogaster larvae are indistinguish- able, lending support for a single circumtropical species. Bathylagus stilbiiis eggs and larvae are indistinguishable from those of B. urotranus. We have attempted to analyze the distribution among four nominal groups of argentinoids, of 14 characters, four of which are taken from developmental stages and 10 from the adult (Table 42). We have used teleosts in general and osmerids as our outgroup following Fink and Weitzman ( 1 982). Distribution of character states are presented in Table 43. A possible arrangement of groups based on the fewest number of character reversals is presented in Figure 88. Opisthoproc- tidae appears to be a well-founded family. More precise inter- pretation of the inter-relationships and nomenclatural ranking for argentinids, microstomatids, and bathylagids requires ad- ditional data. (H.G.M.) Southwest Fisheries Center, P.O. Box 271, La JoLLA, California 92038; (D.M.C.) Natural History Museum Los Angeles County 900 E.xposition Boule- vard, Los Angeles, California 90036. Stomiatoidea: Development K. Kawaguchi and H. G. Moser FISHES of this group of midwater predators are characterized by their dark coloration, serial photophores, large jaws, fang-like teeth, and chin barbels. Traditionally they have been grouped in six families allied to the lightfishes and hatchetfishes (Weitzman, 1974), and together are now considered monophy- Ictic and given ordinal status (Rosen, 1 973; Fink and Weitzman, 1982). Fink (this volume) gives evidence for reducing the six stomiatoid families to one. Because knowledge of stomiatoid ontogeny lags far behind that of the adults, for convenience of discussion we use Weitzman's (1974) grouping of the families 170 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Fig. 89. Larvae of Slomias and Chauliodus. (A) 5. alnventer. 4.6 mm, CalCOFI 7501 Sta. 97.60; (B) 5. atriventer, 10.0 mm, CalCOH 6604 Sta. 107.65; (C) 5. atmenler. 22.2 mm; CalCOH 6604 Sta. 107.65; (D) S. ferox. 30 mm, from Ege, (1918); (E) C. sham. 6.0 mm; from Mito (1961a); (F) C macouni. 15.0 mm, CalCOH 6204 Sta. 60.60; (G) C. macouni. 45.2 mm, CalCOH 5707 Sta. 67.60. Astronesthidae, Stomiatidae, Chauliodontidae. Melanostomia- tidae, Maiacosteidae, and Idiacanthidae, in the Superfamily Sto- miatoidea. Eggs Eggs are known for Chauliodus, Stomias, and Tactostoma and have in common a round shape, smooth chorion, and seg- mented yolk, Chauliodus eggs have a wide perivitelline space and lack an oil globule. Egg diameters are: C sloani. 2.2-2.5 mm (Sanzo, 1931d); C. barbatus, 3.1-3.6 mm (Pertseva-Os- troumova and Rass, 1973); C macouni. 2.7-3.1 mm, with an initial yolk diameter of 1.3-1.5 mm (original data). Mito (196 la) described an egg, referred to C. sloani. l.\l mm in diameter with no oil globule but with a second membrane. Stomias eggs have a second membrane, a single oil globule and the following diameters: S. colubrinus. 1.3-1.5 mm, with inner membrane 1.05-1.1 mm (Pertseva-Ostroumova and Rass, 1973); S. atri- venter. 0.88-0.92 mm, inner membrane diameter is 0.82-0.84 KAWAGUCHI AND MOSER: STOMIATOIDEA 171 Table 44. Meristic Counts of Stomiatoid Genera. Most frequent count or range is followed by overall range or infrequent count in parentheses. Data from Gibbs (1964a,b), Gibbs et al. (1983), Morrow (1964a, b, c). Morrow and Gibbs (1964), Bolin (1939a), Imai (1941). onginal counts. Vertebrae Fin rays Family and genus Dorsal Anal Pectoral Pelvic Stomiatidae Macrostomias 164 13,14 16(15-18) 7(6) 4 Stomias 64-83 17-20(16-22) 19-21 (18-25) 6-7 (6-9) 5(4) Chauliodontidae Chautiodus 51-62 6,7 (5-7) 10-12(10-13) 12,13(11-14) 7 (6-8) Astronesthidae Astronesthes 46-58 15(10-21) 12-22 8 (5-9) 7 (6-8) Borostomias 53-55 13(10-14) 13-16(10-19) 7 (6-9) 7 Hetempholus 66 11 (13) 12-15(17) 7 7 Neoneslhes 53 9-11 (12) 25-27 (22-28) 8(7) 7 (6-8) Rhadineslhes 67 11 (12.13) 18(19-21) 7 (6-8) 7 Melanostomiatidae Bathophilus 38-45 (33-50) 13-16(9-18) 15-16(9-18) 1-37 11-16(4-26) Chiroslomias 54-55 18-20 22-26 6 7 Echwstoma 57-59 11-14(11-16) 13-18(13-19) 1 + 3 8 Eiistomias 56-69 21-25(20-30) 32-46 0-13 7 (6-8) Ftagellostomias 65 16(14-17) 23-25(21-26) I + 8-9 + I1 7 Grainmalostomias 50-56 18-21 21-23(20-24) 4-11 7-8 Leptostomias 77-80 (75-83) 16-22 20-29 10(9-11) 7(8) Melanostomias 50-57 12-17 16-20 5(4-6) 7(8) Opostoimas 60 21 24 1+4 8 Pachystomias 48 22(21-24) 27 (25-29) 5-6 8-9 (7) Photonectes 49-64 15-24 17-24 0-3 7(6) Tactosloma 80-82 14-16 19-22 8-10 Thysanaclis 61 17-18 21-25 1+10,11 7 Trigonolampa 61-62 19-20(18) 18(19) 5 7 Malacosteidae Anslostomias 44-56 18-26 24-32 6-10(3-17) 6 Malacosteus 49 14-19(20) 17-21 (23) 3-4 (5) 6 Photostomias 52-58 22-28 25-32 6 Idiacanthidae Idiacamhus 79-85 54-74 34 (33-39) 6 mm, oil globule diameter of 0.20-0.25 mm, initial yolk diameter of 0.70 mm (original data). Tactosloma macropus eggs have a single membrane, 1.44-1.54 mm in diameter, an oil globule 0.30-0.40 mm in diameter and an initial yolk diameter of 0.78- 0.80 mm (original data). Eggs of C. macouni and 5. athventer are illustrated in Matarese and Sandknop (this volume). Larvae Larvae of Stomiatoidea occur in the upper water column, some at the surface. In most groups the larvae are elongate, have a large head, elliptical eyes that protrude slightly from the dorsal head profile, an elongate, straight gut (trailing from the body in some species), a well developed finfold, large paddle-shaped pectoral fins that lack rays until transformation, and late-form- ing pelvic fins. Melanophore patterns provide a useful set of characters and genera usually have a distinct pattern. The larval melanophores are retained in a subcutaneous position in trans- forming specimens and provide a means for identifying larvae. During transformation, photophores form simultaneously and initially are unpigmented. Counts of fin rays, vertebrae, and photophores are summarized in Tables 44 and 45. Stomiatidae (Fig. 89). — Larvae of five species are known (Table 46). Larvae are 3-4 mm at hatching and have an elongate yolk sac. The slender body is round in cross-section, but becomes slightly deeper by late postflexion. The head is relatively small with a slightly flattened snout. The eyes are elliptical. The elon- gate gut extends almost the entire length of the body and has a slightly enlarged terminal section that reaches the anal fin origin. The median finfold is small and best developed posteriorly. The opposing dorsal and anal fins develop far posteriad on the body in early postflexion larvae, but the pelvic fins do not appear until just before transformation. Late-stage embryos oi Stomias have melanophores along the dorsum, which migrate ventrad and form a distinct series be- tween the body and gut. This series extends to the tip of the notochord. The series is lost before notochord flexion but, in most species, another sparser series develops along the ventral midline of the gut, from the isthmus to the anus. 5". boa and S. fero.x develop a mid-lateral series of melanophores along the body and S. colubrimis has scattered melanophores along the entire hypaxial body region. These species also develop exten- sive dorsal and lateral head pigment. All species form scattered pigment on the dorsal, anal, and caudal fins. A 75-mm specimen (MCZ Cat. No. 59858) with an extremely slender body form (body depth 1.3% of body length) has fin and 172 ONTOGEIVY AND SYSTEMATICS OF FISHES -AHLSTROM SYMPOSIUM Table 45. Photophore Counts of Stomiatoid Genera. Most frequent count or range is followed by overall range or infrequent count in parentheses. Data sources as in Table 1. Photophore groups as defined by Morrow (1964a). Photophore groups Family and genus IP PV vav AC ov VAL Stomiatidae Macrostomias 11(12) 80-86 58-67 19-22 79-85 58-68 Stomias 9-13 32-51 5-16 14-20 32-50 4-17 Chauliodontidae Chauliodus 8-11 17-23 22-30 8-13 17-21 22-29 Astronesthidae Astronesthes 5-12 6-20 7-27 7-13 5-19 7-26 Borostomias 10-13 20-31 15-25 9-15 21-29 16-25 Heterophotus 10-11 32-35 13-14 12-15 33-36 16-20 Neonesthes 9-12 14-17 16-21 13-18 13-15 13-21 Rhadinesthes 10(6) 25(26) 20-23 16 22-24 27 Melanostomiatidae Bathophiius 5 (4-6) 12-18 11-13(11- 17) 5-7 (5-9) 13-14(10-16) 9-11 (8-17) Chirostomias 9(8) 25-27 (28) 19-20(16) 9(10) 23 (24-25) 19-20(16) Echiostoma 8 + 2 25-28 14-18 12-13(11) 24-31 13-17(18) Eustomias 7-8 (9) 27-33 (24-36) 13-17(11- 21) 17-23(15-25) 26-33 (24-37) 13-18(12-22) Flagellostomias 9-10(8) 31-34 14-16 16-18(15) 31-32(30) 14-15(12-17) Grammatostomias 7(6) 15-18 19-22 10-13 15-18 19-22 Leptostomias 10(11) 42-45 (39-48) 20-23 (24) 11-13(14) 40-43 (39-48) 20-22 (23-24) Metanostomias 8 + 2 or 3 23-30 12-15 9-11 22-28 11-15 Opostomias 4 + 4 27 17 16 27 17 Pachysiomias 8-9 14-16(17) 13-14 8-9 17-18 14-15 Photonecies 8-11 19-24, 34-38 11-15(16- 18) 10-13(9) 19-24(17), 30-36 11-14 (15-17) Tactostoma 8 46 19 12 43 18 Thysanactis 20 31-32 14-16 11-12 30-32 14-16 Trigonolampa 11 23-24 (22) 22 (24) 10-11 22-24 23-24 (26) Malacosteidae Aristoslomias 5 + 3 15-17(14-19) 15-18 9-11 (12) 16-19(14-20) 15-17(14-18) Malacosieus (Serial photophores absent or uncountable) Photostomias 5 + 2 13-16 21-25 12-15 12-17 20-23 Idiacanthidae Idiacanlhus 1P + PV = 31-36 16-18(15) 13-18 22-25 31-35(30-36) vertebral counts that match Macrostomias longibarbatus. Its morphology is that of a highly attenuate Stomias larva. Pig- mentation is restiicted to a series of small melanophores along the ventral midline of the gut. The ventral photophore rows are beginning to form. Chauliodontidae (Fig. 59). — Larvae of five species are known (Table 46). Larvae are 6-7 mm long at hatching, with an elongate yolk sac. The body is slender with a circular cross-section, and remains so throughout development. The head is relatively small, with elliptical eyes and a short, acute snout. The gut has a smaller diameter than in Stomias but is relatively longer. The short terminal section extends beyond the anal fin origin. The median finfold is small and best developed rearward on the body. The dorsal, anal, and pelvic fins form in late postflexion larvae in the adult position. A fan-shaped array of melanophores occurs in the caudal region of yolk-sac larvae but is soon lost. No other pigment develops. Larvae of some species reach 46 mm SL and there appears to be marked shrinkage at transformation. Astronesthidae (Fig. 90). — Astronesthid larvae have been illus- trated and described briefly by Roule and Angel ( 1930), Whitley (1941), Pertseva-Ostroumova and Rass (1973), and Belyanina (1982b); however only two of these were identified to genus (Table 46). We have examined more than 10 types of astro- nesthid larvae, 7 of which are listed in Table 46. Astronesthid larvae display a great variety of structure and pigmentation, but hold in common the advanced position of the dorsal fin, in contrast to other Stomiatoidea, except Chauliodus. The types differ fundamentally in gut shape and body form: Types I and II are laterally compressed, relatively deep-bodied, and have a non-trailing or slightly trailing gut with terminal section as in melanostomialids; Types III-VIl have a slender body and a trailing gut; in Types III-V the gut is deflected ventrad from the body just anterior to the anal fin base and in Type VI and VII at midbody, anterior to the dorsal fin (Figure 90). Type I (Fig. 90A). — larvae up to 26.5 mm; laterally compressed; head shallow with acute snout; eyes relatively large, slightly Fig. 90. Larvae of Astronesthidae. (A) Type I, 23.7 mm, ORl A105; (B) Type II. SIO Tasaday I A3; (C) Type IV. 33.0 mm. MCZ Cat. No. 59855; (D) Type V, 22.0 mm, Dana Sta. 3931; (E) Type VII, 28 mm, MCZ Cat. No. 59856. KAWAGUCHI AND MOSER: STOMIATOIDEA 173 174 ONTOGE^Pr' AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Table 46. Pigment Characters and Gut Structure in Larvae and Transforming Specimens of Stomiatoidea. (NT = not trailing, ST = slightly trailing, T = trailing freely). Hypaxial myoseptum Length Length of Dorsal myomere Epaxial myoseptum melanophores of larvae ti^nsfonning melanophores melanophores (no./myo- Gut Species (mm) specimens (mm) (no./myomere) (no. /myoseptum) seplum) stnicture Source Stomiatidae Stomias boa _ 38 NT Sanzo, 1912a Stomias boa 10.4-30.4 41.5 NT Sanzo, 193 Id Stomias boa 9.0-32 — NT Ege, 1918 Stomias ferox 9.0-44 — NT Ege, 1918 Stomias colubrinus 3.3-16 — NT Pertseva-Ostroumova and Rass, 1973 Stomias alriventer 4.6-32 — NT original Macrostomias longibarbatus - 75 NT original Chauliodontidae Chauliodus sloani 33.6 41.6 NT Sanzo, 1915a Chauliodus sloani 5.7-41.6 27.1 NT Sanzo, 193 Id Chauliodus sloani 2.1 — NT Mito, 1961a Chauliodus danae 22.5-25 — NT Belyanina. 1977 Chauliodus macouni 38.0-49 35-44 NT Belyanina, 1977 Chauliodus mmimus 23.5-35 — NT Belyanina, 1977 Chauliodus pammelas 10.6-40 _ NT Belyanina, 1977 Chauliodus sloani 7.4-35 27-34.2 NT Belyanina, 1977 Chauliodus macouni 5.6-46 - NT original Astronesthidae Unidentified 14.0-23 — T, NT Roule and Angel, 1930 Astronesthes lupina 20 — T Whitley, 1941 Boroslomias panamense 5.0-17 — T Pertseva-Ostroumova and Rass, 1973 Unidentified 16 — 7 + + NT Belyanina, 1982b Unidentified 17.7 — 2 total T Belyanina, 1982b Type I 12.3-26.5 — several several to many several to many NT original Type II 14.9-26 29,40 ST original Type III 16.2 20.5,22.5 T original Type IV 14.4-34.5 40.5 T original TypeV 17.4-19.4 20,22 T original Type VI — 28 T original Type VII — 28 T original Melanostomiatidae Tactostoma macropus 5.0-44 49 0-1 1-3 NT original Melanostomias spilorhynchus 17 21-32 ca. 3 NT Beebeand Crane, 1939 Melanostomias biseriatus — 21-25 ca. 3 NT Beebe and Crane, 1939 Melanostomias valdiviae — 25 2-3 NT original Melanostomias sp. 13.4-17.2 16.4-22 2-4 NT original Echiosloma tanneri 20,25 — 2-5 NT Beebe and Crane, 1939 Echiostoma sp.? 13.8 _ 2-4 NT Belyanina, 1982b Echiostoma barbatum — 34 1-2 NT original Photonectes dinema — 24 and > 1 (?) 3-4 NT Beebe and Crane, 1939 Photonectes leucospilus — 25 and > 1(?) 3-4 NT Beebe and Crane, 1939 Photonectes albipinnis — 16-22 2-3 NT original Photonectes sp. 11.0-12.5 — 4-5 NT original Photonectes parvimanus 12.0-26 25 3-6 3-4 NT Beebe and Crane, 1 939 Photonectes parvimanus 27 — 3-4 2-4 NT original Photonectes parvimanus — 28 1-2 2-4 NT original Photonectes sp. 5.4-22.2 — ca. 7 5-7 NT original Opostomias mitsuii 15.0-21 1 0-1 (2-3 posteri- orly) 1-2 (3-5 post.) NT original Flagellostomias boureei 20.0-21 34,39 1 1 NT Beebeand Crane, 1939 Flagellostomias boureei 10.8-36.4 — 1 1-2 NT original Odontostomias micropogon — 42 1 1-? 2-4 NT Beebe and Crane, 1939 Leptostomias gladiator 12.0-30 38-45 1 + several 1-5 2-4 NT Beebe and Crane, 1939 Lepiostomias gracilis — 37.8 1 + 1-5 5-7 6-9 NT original Leptostomias sp. 25 — 1 + 1-3 4-5 4-6 NT original Bathophilus nigerrimus 11.6 21.7 1 or > NT Sanzo, 1915a Bathophilus nigerrimus 5.9, 14.0 19.2-21.7 1 or > NT Sanzo, 193 Id KAWAGUCHI AND MOSER: STOMIATOIDEA 175 Table 46. Continued. Hypaxial myoseptum Length Length of DoPial myomere Epaxial myoseptum melanophores of larvae transforming melanophores melanophores (no./myo- Gut Species (mm) specimens (mm) (no, myomere) (no./myoseptum) septum) structure Source Bathophilus metallicus 25 3 or > NT Beebe and Crane, 1939 Bathophilus sp. 11, 12 1 or > NT Beebe and Crane, 1939 Bathophilus sp. 7 1 or > NT Beebe and Crane, 1939 Bathophilus sp. 15 _ (?) NT Rouleand Angel. 1930 Bathophilus sp. 18.2 — 1 or > NT de Sylva and Scotten, 1972 Bathophilus filifer 4-10 - 1 or > NT Pertseva-Ostroumova and Rass 1973 Bathophilus brevis 15.7 — 1 or > NT original Bathophilus Jlemingi 2.9-23.8 — 1 to several NT original Euslomias sp. 33 _ 7 total T Regan, 1916 Eustomias sp. 13 — 7 total T Beebe and Crane, 1939 Eustomias spp. (4 types) 6.0-45 - 5-1 1 total T original Malacosteidae Aristostomias scintillans 4.3-47 45 14 total to many T original Photostomias guernei 20.0-27.5 30,31 8 pairs total T original Unidentified 12 _ 1 2 total T Beebe and Crane, 1939 Unidentified 34.5 - T original Idiacanthidae Idiacanthus fasciola 16.0-28 35-48 1 T Beebe, 1934 Idiacanthus sp. 7.0-39 — 1 T Pertseva-Ostroumova and Rass, 1973 Idiacanthus antrostomus 4.5-71 67-> 1 T original elliptical; gut moderately slender, thin-walled; finfold moderate; pigment pattern consists entirely of minute melanophores, in- creasing in number with development, principally in the ex- paxial and hypaxial myosepta; other pigment above brain, paired internal streaks in snout, melanophores in dorsal and ventral finfold, dorsal fin base, and on posterior half of gut. Type II (Fig. 90B). — larvae reach at least 26 mm; deep-bodied and laterally compressed in late-stage larvae; head deep; eyes small, slightly elliptical; gut slightly trailing and with larger di- ameter than in Type I; dorsal finfold relatively deep; pigment above brain, along lower jaw and at angular and gular region; blotch at posterior margin of superior hypural complex and one midway out on inferior group of caudal rays; fin ray and ver- tebral counts and photophore counts match Astronesthes gem- mifer. Type III. — larvae reach at least 16.2 mm; body slender; head and eyes moderate in size; eyes elliptical; slender gut trails free from body at anal fin origin; finfold moderately developed, ex- cept posterior to dorsal fin the finfold appears as an enlarged adipose fin; pigment restricted to a series of melanophores along lower jaw and between upper and lower hypural complexes; counts match Astronesthes richardsoni. Type IV (Fig. 90C).— lai^ae reach 40 mm; morphology similar to Type III, except head relatively longer and eyes almost round; gut with leaf-like appendages on trailing section; pigment re- stricted to postorbital blotch and interorbital band; fin and ver- tebral counts and photophore arrangement match Heterophotus. Type V (Fig. 90D). — larvae reach about 20 mm; morphology as in Types III and IV; eyes slightly elliptical; pigment heavy; melanophores on head, lateral to posterior brain region, on snout and lower jaw symphysis; lateral surface of body covered with an irregular pattern of large melanophores; melanophores on trailing gut. Pertseva-Ostroumova and Rass (1973) identified larvae of this type as Borostomias panamense. Type VI. — specimen transfoiming at 28 mm; morphology sim- ilar to Types II-V, except trailing gut deflected from body far in advance of anal fin origin; eyes elliptical; dorsal finfold highly developed and ventral finfold anterior to anal fin is rudder-like; pigment lacking; meristics indicate it is in the genus Astro- nesthes. Type VII (Fig. 90E). — specimen transforming at 28 mm; mor- phology similar to Type VI; dorsal and anal fins supported on cartilaginous pedestals; a series of 4 melanophores along hori- zontal septum; some melanophores on anterior region of dorsal and anal fin bases and on preanal finfold. Whitley (1941) de- scribed a larva similar to this as Astronesthes lupina. Melanostomiatidae (Figs. 91-92). — Larvae have been identified for 10 of the 15 genera (Table 46). Bathophilus was the first to be identified (Sanzo, 1915a). The only comprehensive work on melanostomiatid ontogeny is that of Beebe and Crane (1939) who identified larvae of 8 genera and 5 species by the use of transforming series. Since then, the only other melanostomiatid larvae that have been described are Bathophilus filifer {Pertseva- Ostroumova and Rass, 1973), Bathophilus sp. (de Sylva and Scotten, 1972), and Echiosloma (?) sp. (Belyanina, 1982b). De- 176 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM scriptions of Opostomias and Tactostoma are included in this paper. Larvae of Tactostoma were initially identified by E. H. Ahlstrom. Larval representatives of the 10 genera are highly various in form and pigmentation, however, with the exception of Euslo- mias, they share the following structural features: body elliptical in cross-section; head laterally compressed; eyes small and el- liptical; gut terminated in an elongate muscular bulb that may extend beyond the anal fin origin but not beyond the margin of the finfold; dorsal and anal fins form in adult position posteriorly on the body; body pigment consists of one or more melano- phores dorsal to each myomere, one or more melanophores on the hypaxial myosepta and, in some genera, on the epaxial my- osepta. Dorsal and lateral pigmentation tends to be heavier in forms with higher meristic counts. The genera differ principally in body size, relative body depth, relative head size, jaw size, gut diameter, size and shape of the terminal gut section, finfold height, and pigment pattern. Present knowledge indicates that genera apparently have dis- tinct facies, tentative descriptions of which are presented below. Confirmation awaits identification of additional species. Tactostoma (Fig. 91 A). — larvae reach 44 mm in length; body extremely slender; head flat and elongate initially, becoming less flat and relatively smaller with development; eye size moderate; gut slender; finfold moderate; pectoral fin lost at transformation; early larvae develop one melanophore per myomere along dor- sum and 1-3 melanophores on the hypaxial myosepta; post- flexion larvae gradually lose the dorsal melanophores and then the hypaxial myosepta pigment, in contrast with other genera in which body pigment increases with development; pigment on lower jaw symphysis, isthmus, pectoral fin base, cleithrum, and above gut terminus; dorsal and ventral pigment accentuated at caudal peduncle. Melanostomias (Fig. 91B).— transforming specimens as small as 16.4 mm; body slender; head small; snout short; eye size moderate; gut slender; finfold relatively small; one melanophore per myomere along dorsum in one form and in another form the zone between the 7th- 10th myomere and the dorsal fin lacks dorsal pigment; 2-3 melanophores in hypaxial myosepta; pig- ment above and below head, below liver, on terminal gut sec- tion, and along finfold margins. Larvae tentatively identified as Echiostoma have similar characters (Table 46). Photonectes (Fig. 9 IC). — larvae of different forms transform at sizes between 16 and 28 mm; body somewhat deep; head size and snout length moderate; eyes small, highly elliptical; several forms of dorsal myomere pigment ( 1 melanophore per myomere in Subgenus Photonectes and 3-7 per myomere in Subgenus Trachinostomias); hypaxial myosepta with 2-7 melanophores depending on form (Table 46); extensive pattern of minute me- lanophores on head, finfold, and median fins. Flagellostomias (¥\g. 9 ID). — larvae may reach 30-40 mm; body somewhat deep; head large, deep, with steeply sloping snout and large jaws; eyes small; gut diameter relatively large; finfolds large, accentuating body depth; one large melanophore per myo- mere along dorsum; 1-3 melanophores in hypaxial myosepta; some scattered lateral melanophores in median fin region; other pigment scant; a few melanophores in head region, some on finfold in posterior gut region, and on dorsal and anal fins. Opostomias (Fig. 9 IE). — body moderately deep; head large, deep posteriorly with elongate sloping snout; eyes small; gut slender; finfold large; one melanophore per myomere along dorsum; 1- 2 melanophores in hypaxial myosepta; epaxial and hypaxial myosepta below dorsal fin base have several melanophores, giving this region a banded appearance; melanophores on dorsal head region, gill arch and gut terminus. Leplostomias (Fig. 91F). — larvae may reach about 40 mm; body somewhat deep; head moderately large, deep; eyes small; gut slender; finfold moderate; pigmentation heavy; one large me- lanophore and 1-5 smaller ones per myomere along dorsum; numerous melanophores on epaxial and hypaxial myosepta, increasing with development to completely outline myosepta; pigment extensive on dorsal and ventral head regions, on gill arches; pigment below liver, on finfold margins, above gut ter- minus and on dorsal and anal fins. Bathophilus (Figs. 92A-C). — larvae transform at 25 mm or less; deep-bodied compared with other genera; head and jaws large; barbel forms in late postflexion larvae, particularly in B. hrevis; eye size moderate; gut large to voluminous, with highly devel- oped s-shaped terminal section; finfolds, particularly dorsal, large; one or several melanophores per myomere along dorsum and an opposing series of melanophores along ventral surface of myomeres; no lateral pigment; head, finfolds and median fins pigmented. Eustomias (Fig. 92D). — larvae of some species reach 45 mm; body slender, and round in cross-section; head elongate and flat with large spatulate snout; large jaws; eyes moderate in size, slightly elliptical to round; gut slender, deflected ventrad at anal fin origin and trailing from body; body pigment consists of 5- 1 1 large melanophores along the dorsal midline; usually pigment at lower jaw symphysis. Malacosteidae (Fig. 9iA — Larvae of this group have not been described, although the 12-mm larva illustrated by Beebe and Crane (1939) and referred to "lEustomias" is apparently Ar- istostomias. We have examined larval series and transforming specimens of A. scintillans and Photostomias guernei (Table 46). Aristostomias scintillans (Fig. 93A). —larvae reach 47 mm length; body slender; head large, flat; snout elongate; jaws large; eyes slightly elliptical; opercle markedly reduced; gut slender, de- flected ventrad at anal fin origin and trailing from body; finfold moderate; dorsal and anal fins form in adult position at about flexion stage; pelvics form late; initial pigment pattern is a series of paired melanophores along the dorsum, beginning with 14 Fig. 91. Larvae of Melanostomiatidae. (A) Tactostoma macropiis, CalCOFI Norpac Sta. 14; (B) Melanostomias sp., 16.0 mm, ORI KH73- 2, Sta. 49-7; (C) Photonectes sp., 22.2 mm, SWFC, Albacore Oceanography Cruise 71, Sta. 99; (D) Ftagetloslomias boureii. 36.4 mm. SIO Cat. No. 73-329, Tasaday I, Tow 42; (E) Opostomias mitsiiii. 1 5.0 mm, ORI KH 73-2 Sta. 2-3; (F) Leptoslomias sp., 24.5 mm, MCZ Cat No. 59857. KAWAGUCHI AND MOSER: STOMIATOIDEA 177 178 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM KAWAGUCHI AND MOSER: STOMIATOIDEA 179 Fig. 93. Larvae of Malacosteidae. (A) Aristostomias scinlillans. 34.7 mm. CalCOFI 5008 Sta. 70.30; (B) Photostomias sp., 26.7 mm. ORI KH 73-5 Sla. 55-13. Bn 24-12; (C) Malacosteidae, 34.5 mm. from Moser (1981). Fig. 92. Larvae of Melanostomiatidae. (A) Bathophilus flemingi. 25.5 mm. CalCOFI 4910. Sta. 80.137; (B) B hrevis. 15.7 mm. ORI KH 81-1, Sta. 17; (C) B. nigernmus, 21.7 mm, redrawn from Sanzo (1931d); (D) Eustomias sp. 33 mm, redrawn from Regan (1916). 180 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Fig. 94. Larva of Idiacanlhus anirostoinus. 55 mm. CalCOFI 6207 Sta. 90.120. KAWAGUCHI AND MOSER: STOMIATOIDEA 181 pairs and increasing in numbers with development to cover the entire dorsum; paired ventral series develop, initially poste- riorly, and increase in numbers so that all myomeres have me- lanophores on the ventral surface; pigment on brain, snout, lower jaw, gular-isthmus region, otic region, caudal fin, and in vague rings along trailing gut. Ahstostomias larvae were iden- tified initially by E. H. Ahlstrom. Photostomias giternei (Fig. 93B). — larvae reach about 30 mm; morphology similar to A. scintillans except eyes smaller and narrower and pelvic fins somewhat elongate; body pigment con- sists of a series of 8 minute dorsal melanophore pairs and 8 slightly larger opposing pairs along the ventral surfaces of the myomeres; melanophores at lower jaw symphysis, large mela- nophore on each pectoral fin base, sparse melanistic rings along trailing gut. Malacosteid C (Fig. 93C). — intact specimen (captured by Dr. Richard Harbison, WHOl) has morphological and meristic characters of malacosteid larvae but lacks pigment except on the extensive gut. Shallow capture locality of this specimen and our capture of large A. scintillans larvae in MANTA nets in- dicates late-stage malacosteid larvae have a shallow distribution in the water column. Idiacanihidae (Fig. 94j. — Brauer ( 1 906, 1 908) described the re- markable larvae of Idiacanthus and named them Slylophthal- mus paradoxus. Beebe ( 1 934) correctly identified the larvae and described them in detail. Idiacanthus larvae are extremely slen- der, reaching a length of 35-70 mm depending on the species. Other characteristics are: elongate and extremely flat head; el- liptical eyes on long stalks with cartilaginous supporting rods; stalk length up to 27% of body length in /. antrostonms (Weihs and Moser, 1981); gut slender, deflected at anal fin origin and trailing; finfold small; dorsal fin begins forming in preflexion larvae; dorsal fin larger than anal fin and slightly in advance of it in postfiexion larvae; during transformation, rays added se- quentially anteriad so that in adults the dorsal extends about -A of the body length and the anal about 'A; pectoral fins well developed but lost at transformation and pelvic fins develop in transforming females, but not at all in males; pigment pattern consists of a melanophore on the posterior margin of each hy- paxial myomere, spreading into the myosepta when expanded, several elongate internal blotches in the isthmus region, and a series of melanophores along the trailing gut; adult males of /. fasciola reach 32-42 mm SL, lack teeth and paired fins and have relatively larger eyes and an enormous luminous gland. Relationships Information on larval characters of 18 of the 26 stomiatoid genera recognized by Fink (this volume), representing all 6 of the families recognized by Weitzman (1974), permits some pre- liminary generalizations and conclusions: (1) Larvae of Sto- miatidae and Chauliodontidae are similar in morphology and are distinct from other stomiatoids. Pigmentation provides fur- ther evidence of this; Chaidiodus larvae are unique among known stomiatoids in lacking pigment after the yolk-sac stage and the median series of gut melanophores of Stomias also appear to be unique. (2) Larvae of Astronesthidae are diverse in mor- phology and pigmentation and most of the larval specializations that appear in other stomiatoid families are found among as- tronesthid genera. Larval specializations of some genera (e.g., ornamented trailing gut, trailing gut deflected at mid-body, rud- der-like finfolds) are not found elsewhere in Stomiatoidea. Het- erogeneity of larval characters in Astronesthidae supports Fink's view that the group is paraphyletic. (3) In the Melanostomia- tidae, larvae of Melanostomias. Photonectes. Echiostoma. Oposlomias. Flagellostomias. Odontostomias and Leptostomias are similar in morphology, have paired melanophore series on the dorsum, and differ chiefly in head size, body depth, and in the extent of myosepta pigment. Tactostoma larvae have the characters of this group of genera except that the body is ex- tremely slender and the pigmentation is lost in the postfiexion stage. Larvae of Battiophilus difler from those of the above group in a number of characters (voluminous gut with specialized terminal section, melanophore series on the ventral surface of the myomeres, lack of myosepta pigment). Larvae of Eustomias are different from all known larvae of Melanostomiatidae in having a trailing gut, flat head and snout, and a pigment pattern consisting of a median series of up to 11 large melanophores on the dorsum. Except for this latter feature, Eustomias larvae are similar to those of Malacosteidae. (4) Idiacanthus larvae have a combination of characters unique among stomiatoids. The stalked eyes are autapomorphic. Larval characters provide no support for Fink's hypothesis that this genus is closely related to Tactostoma. Ocean Research Institute, University of Tokyo, 1-15-1, MiNAMiDAi, Nakano-ku, Tokyo 164, Japan, and Na- tional Marine Fisheries Service, Southwest Fisheries Center, 8604 La Jolla Shores Drive, La Jolla, Calif- ornia 92038. Stomii forms: Relationships W. L. Fink STOMIIFORMS are well known as a major component of the midwater oceanic fauna. Past concepts of their rela- tionships to other primitive euleleosts were reviewed by Fink and Weitzman (1982), but in brief in this century, they have been considered isospondyls (Parr, 1927; Regan, 1923; Morrow, 1964) or, more recently, salmoniform protacanthopterygians (Greenwood et al., 1966). In 1973, Rosen placed these fishes as a separate order (Stomiatiformes) within the Neoteleostei, as 182 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM STERNOPTYCHIDAE GONOSTOMATIDAE PHOTICHTHYIDAE ASTRONESTHIDAE IDIACANTHIDAE MALACOSTEIDAE MELANOSTOMIIDAE STOMIIDAE CHAULIODONTIDAE Fig. 95. Weitzman's (1974) hypothesis of relationships of the sto- miiform fishes. The Gonostomatidae and Stemoptychidae comprise the Gonostomata and the remaining families comprise the Photichthya. sister group to the Eurypterygii. Fink and Weitzman (1982) agreed with this placement, provided more characters to sub- stantiate it, and demonstrated monophyly of the stomiiforms. Steyskal (1980) has presented arguments that the root of the family-group names demands that these be altered from Sto- miatidae and Stomiatiformes to Stomiidae and Stomiiformes, respectively, and I use these forms throughout this paper. As recognized by Weitzman (1974), there are two major sto- miiform lineages, Gonostomata and Photichthya, both classified at infraordinal rank, with families Gonostomatidae and Ster- noptychidae in the former and families Photichthyidae, Sto- miidae, Chauliodontidae, Astronesthidae, Melanostomiidae, Malacosteidae, and Idiacanthidae in the latter (Fig. 95). I have no disagreement with Weitzman's hypotheses of monophyly of the Stemoptychidae, but our recent work on Diplophos (Fink and Weitzman, 1982) caused us to question the monophyly of the Gonostomatidae and Photichthyidae, and my work on the barbelled stomiiforms, comprising the remaining families, has cast doubt on the entire traditional arrangement of the included 26 genera as well as on the monophyly of the Photichthya. I have found features which support new hypotheses of relation- ship within the stomiiforms and will present some of these ideas below. Some are more tentative than others. Weitzman is cur- rently working on the genera he placed in the Gonostomatidae and Photichthyidae. First, I have found no evidence that Diplophos is the sister group of any other genus of stomiiform and it may be, as Fink and Weitzman (1982) suggested, the sister group of the rest of the order. Specializations in the adductor muscles indicate that Diplophos GONOSTOMA Cyclothone Margrethia BONAPARTIA Triplophos STERNOPTYCHIDAE PHOTICHTHYA Fig. 96. Hypotheses of stomiiforms as discussed herein. See text for explanation. Gonostoma. Cyclothone, Margrethia. and Bonapartia form a monophyletic group, but what relationships within that group are I cannot say, and presumably this will be treated by Weitz- man. These hypotheses would cause a redefinition of the Gon- ostomatidae, restricting it to the four genera mentioned just above. Relationships of Triplophos are also unclear, and there is evidence in the hyoid apparatus that it may be related to some of the "photichthyans," rather than the gonostomatids, as Weitzman ( 1 974) supposed. Weitzman ( 1974) established mon- ophyly of the Stemoptychidae, and 1 have nothing to add to his conclusions. Nevertheless, since he did not deal with monophyly of the Gonostomatidae or with the sister group relationship of the Stemoptychidae, there is no current evidence that the latter is more closely related to some subset of the former, and I leave that part of the phylogeny unresolved. These hypotheses are summarized in Fig. 96. See also the paper by Ahlstrom, Rich- ards, and Weitzman (this volume) on the Gonostomatidae, Ster- noptychidae and other stomiiforms. Within the "Photichthya," we have the same problem as with the Gonostomatidae; that is, there is a diagnosable monophy- letic unit (the barbelled forms) and an undiagnosed grade group, the Photichthyidae. My own efforts have been on the barbelled forms, currently distributed in six families, as listed above. There have been no strictly phylogenetic studies of relationships within the group, but they were examined in a traditional sense by Parr (1927), Regan and Trewavas ( 1 929, 1 930), and Beebe and Crane (1939). FINK: STOMIIFORMS 183 My hypotheses are based on a study of 330 characters, mostly taken from the skeleton, but with some from the head muscles, photophores, and other parts of the soft anatomy. The conclu- sions are presented in Fig. 97. Traditional families are not rec- ognizable in this scheme of relationships. Evidence for the arrangement of the genera is presented else- where (Fink, in prep.), but some characters will be discussed below, particularly those relevant to some of the larger portions of the tree or in areas that might seem controversial to some readers. For ease of communication, I will state here that my choice of classification for this group is an expansion of the traditional Stomiidae of Regan and Trewavas (see Fig. 97). Monophyly of the Stomiidae is established on the basis of up to 1 7 characters, including 1 ) presence of a mental barbel, 2) 5 hypurals in the caudal skeleton rather than 6, 3) lack of gill rakers in adults, 4) a divided geniohyoideus muscle, and 5) a portion of the adductor mandibulae inserting on the postorbital photophore. The Astronesthidae, as most recently discussed by Weitzman (1967), consisted oi Astronesthes. Boroslomias. Heterophotus. Neoncsthcs, and Rhadinesthes. As can be seen in Fig. 97, the group is clearly not monophyletic. Neonesthes is the sister group of all other stomiids, a hypothesis borne out by many characters shared by the remaining stomiid genera, including lack of tooth- plates on basibranchial 1, epibranchial 4, and on the posterior edges of gill arches 1-4, and presence of rector muscles attaching to the fifth ceratobranchial. The several equally parsimonious constructions of stomiid relationships leave an unresolved tri- chotomy at the next level, there being insufficient evidence re- garding the positions of Aslronesthes. Boroslomias. and the re- maining stomiids. This problem will be further discussed by Fink (in prep.). The remaining stomiids are united by such traits as lack of toothplates on basibranchial 3 and position of the basihyal- hypohyal ligament, as well as specializations of the dorsal and anal fin skeletons. At this point there lies another unresolved trichotomy, involving the groups Heterophotus plus Rhadi- nesthes, Slomias plus Chauliodus, and the remaining stomiids. Heterophotus and Rhadinesthes are documented as sister taxa by several characters, including an elongate dorsal spine on the cleithrum and a preopercle that is narrow at the area of the symplectic-hyomandibular joint. That Chauliodus and Stomias are sister taxa is supported by numerous characters, including a nasal bone which forms a cup-like wall to the nasal capsule; distribution of the palatine teeth into two areas, one anterior and one well posterior; branchiostegals deeply bifurcated dor- sally; and a distinct hexagonal pigment pattern in the skin. I do not recognize the genus Macrostomias since work in progress shows that those species are the sister group to a derived group within Stomias. The remaining genera, comprising the traditional families Melanostomiidae. Malacosteidae, and Idiacanthidae, are united by presence of many features, including no more than one pair of toothplates associated with any basibranchial ossification, and reduction of the distal radials of the pectoral fins. As postulated by Regan and Trewavas (1930), I have also found that Chirostomias and Tngonolampa are sister taxa based on features such as fusion of the bilateral toothplates of basi- branchials 2 and 3 and reduction of the supramaxiUa to a sliver of bone. These genera are the sister group to the remaining genera, a hypothesis supported by several characters, including fewer than 6 branchiostegals articulating with the posterior cer- Neonesthes astronesthes borostomias Heterophotus Rhadinesthes Chauliodus Stomias Chirostomias Trigonolampa Thysanactis leptostomias Opostomias Odontostomias Flagellostomias Photonectes Echiostoma Melanostomias Idiacanthus Tactostoma Grammatostomias Bathophilus eustomias Aristostomias Malacosteus Pachystomias Photostohias Fig. 97. Hypothesis of relationships within the Stomiidae, as dis- cussed herein. atohyal ossification, 3 or fewer distal pectoral fin radials, and presence of a modification of the anterior pectoral fin rays into a structure I call the "rod-ray complex." For the remaining genera, I will concentrate on establishing the major lineages as monophyletic and on areas that affect traditional familial classifications of the group, particularly the relationships of the "malacosteids" and Idiacanthus. One monophyletic group is comprised of Flagellostomias. Leptostomias. Odontostomias. Opostomias. and Thysanactis. Among the diagnostic features are fusion of the distal cartila- ginous tips of the lateral ethmoid and supraethmoid, and an elongate opercular process of the hyomandibula. The remaining genera are supported as monophyletic by nu- merous characters, among them being lack of a retroarticular (also lacking in Trigonolampa), and the form of the articulation of the interhyal. The latter element articulates anterior to the front margin of the cartilage between the hyomandibula and symplectic and is bound to the metapterygoid by a ligament from the anterior margin of the interhyal. The Malacosteidae has traditionally been comprised of three genera, Aristostomias. Malacosteus. and Photostomias, all of which lack a floor to the mouth. The evidence shows that Pachy- 184 ONTOGENY AND SYSTEMATICS OF FISHES -AHLSTROM SYMPOSIUM stomias also belongs to this group, and not with the other "me- lanostomiids." This finding is not particularly radical, since other authors have noted the close morphological resemblance of that genus to the other three and indeed, it has been kept out of the Malacosteidae mostly because the mouth floor is still present, though thin, in members of the genus. The data are insufficient to allow an unambiguous resolution of the interre- lationships of these genera, but numerous characters support the monophyly of the assemblage, including the suborbital pho- tophore being ventral or posteroventral to the eye and the car- tilage of the palatine arch being interrupted between the pos- terior margin of the palatine and the rest of the arch. Idiacanthus has usually been placed in a family by itself as was done, for example, by Beebe (1934), primarily on the basis of the specialized stalked-eyed larval stages and the degree of sexual dimorphism. Beebe recognized that the genus was "closely related to the Melanostomiatidae," as did Gibbs (1964b). Nei- ther author suggested more precise relationships, and Beebe and Crane (1939) showed Idiacanthus in a large multichotomy in their figure of "relationships." Regan and Trewavas ( 1 930) con- sidered Idiacanthus to belong with Melanostomias, Echiosto- ma. and Photonectes. but did not say precisely where. My data support placement of the genus as sister group to Tactostoma, a genus described in 1 939. These two are then related to a group of genera as shown in Fig. 97. Note that Melanostomias and Echiostoma are excluded, being the sister group of the entire assemblage. I am confident of the placement of Idiacanthus and Tactostoma together, based on an array of characters, including reduction of the basihyal to a thin, cylindrical element, origin of the dorsal section of the medial division of the adductor mandibulae muscle anterior to the insertion of the levator arcus palatini muscle, and an extremely elongate body. But I am not particularly confident in the placement of these two genera with the others, even though the data appear impressive at first glance. This lack of confidence is attributable to the fact that most of those characters change at least three times in the entire tree, leaving but one, lack of a posttemporal bone, as the only un- reversed character supporting the hypothesis. Another possibility is that Idiacanthus and Tactostoma are the sister group of Melanostomias and Echiostoma. as suggested in part by Regan and Trewavas (1930), apparently based on the close morphological resemblance of Idiacanthus with the latter two genera. Such a hypothesis would require some additional reversals or independent losses, but as just noted, most of these characters change several times even in the most parsimonious tree. This part of the total phylogeny deserves more critical examination, and it is hoped that larval specializations will be found which will be found which will cause one hypothesis to be clearly preferred over the other. Regarding classification of the stomiiform fishes, it appears that most of the traditional groups will cease to be recognized, a move that was initiated by Weitzman (1974). A period of flux should be expected until his curtent work is completed, but such temporary instability is the current state of teleostean classifi- cation at all levels, as phylogenetic methodology is applied with increasing frequency. One might expect, however, that classi- fication within the Stomiiformes will be stable sooner than that in many other groups, because phylogenetic methods already have been applied to it for several years. I will not present a classification here, but I do provide such for the Stomiidae in my revision of the group (Fink, in prep). In summary, there is still much to be done in unravelling the phylogenetic history of the main lineages of stomiiform fishes. 1 have outlined above areas where our knowledge is either in- complete or poorly developed, and these should be the areas where workers now concentrate their attention— to establish monophyletic groups among the "primitive" stomiiforms and to critically reexamine some of the hypotheses I have produced within the barbelled stomiiforms. Some of this work is under- way, using adult and sub-adult specimens, but the usefulness of larvae is as yet unknown. The data presented in Ahlstrom's (1974) work on patterns of metamorphosis in "gonostomatid" fishes corroborate, when analyzed by phylogenetic methods, the placement by Weitzman (1974) of many of those genera in an expanded Stemoptychidae. An example of this is the presence of photophores in clusters with common bases in those fishes recognized by Weitzman as stemoptychids. Kawaguchi and Moser (this volume) present the most comprehensive infor- mation to date of stomiid larvae. Their data indicate that there should be a plethora of characters for phylogenetic analysis and that study of larvae should indeed prove useful in testing hy- potheses of stomiid relationships. However, even a cursory ex- amination of their data indicates that, as with characters in adults, there appears to be a high degree of homoplasy. This is an interesting phenomenon deserving further study. Museum of Zoology, University of Michigan, Ann Arbor, Michigan 48109. Families Gonostomatidae, Stemoptychidae, and Associated Stomiiform Groups: Development and Relationships E. H. Ahlstrom, W. J. Richards and S. H. Weitzman A summary of known information about the larvae and re- formation, both published and unpublished, gleaned from early lationships of the stomiiforms with elongate gill rakers in life history stages and from adults. We also append some ten- adults was published by Ahlstrom (1974). The present paper is tative new hypotheses of relationships within this "group" of an addendum to that contribution and includes additional in- stomiiforms. AHLSTROM ET AL.: GONOSTOMATIDAE, STERNOPTYCHIDAE 185 Table 47. Summary of Diagnostic Characters for Eggs of Certain Stomiiform Fishes. Illus- Species Egg diameter Oil globule Diameter Yolk Special features trated Source Argyropelecus 0.92-1.04 1 0.26-0.28 segmented large oil globule Yes Sanzo, 1928 hemigymniis Ichthyococcus 0.80 1 0.24 segmented large oil globule Yes Sanzo, 1930b ovalus Maurolicus 1.63 1 0.25 segmented hexagonal pattern Yes Mito, 1961a muellen 1.32-1.58 1 0.26-0.28 segmented on shell Yes Sanzo, 193 Id 1 inciguerna 0.58-0.74 none irregularly thin inner shell Yes Ahlstrom and lucc'lia segmented membrane Counts, 1958 powenae 0.75-0.85 1 0.17-0.19 segmented no thin inner shell membrane Yes Ahlstrom and Counts, 1958 nimhana 0.64-0.72 none irregularly segmented thin inner shell membrane No Ahlstrom and Counts, 1958 atlenuata 0.84-0.92 1 0.18-0.195 segmented no thin inner shell membrane No Sanzo, 193 Id Gonosloma 0.80-0.81 1 0.20-0.21 — — No Sanzo, 193 Id denudatum Ahlstrom (1974:672) favored recognition of one family for those stomiiforms with elongate gill rakers in adults. According to the rules of priority this would be the Stemoptychidae. Weitz- man (1974:338) recognized three families, Gonostomalidae, Photichthyidae, and Stemoptychidae, for the same stomiiforms, the last family including the "maurolicin" genera formerly as- signed to the Gonostomalidae and the deep-bodied stemop- tychids traditionally assigned to the family. In a phylogenetic or cladislic analysis this elongate gill raker bearing "group," if recognized as a single family, is paraphyletic if one considers certain of its subgroups as equivalent or higher taxonomic cat- egories. For example, recognition of Ahlstrom's Stemoptychi- dae, which would include the Stomiidae, a monophyletic group with its members having a median barbel attached to the ventral surface of the head in association with the hyoid bone and lacking elongate gill rakers in adults, is incompatible with a phylogenetic classification based on nested monophyletic groups, since the Stomiidae is the sister group of another group within Ahlstrom's Stemoptychidae. Furthermore, the character used here to "define" the paraphyletic Stemoptychidae, the presence of elongate gill rakers in adults, is excellent for use in a key for identification purposes, but cannot be used as a synapomorphy relating these fishes because it is primitive for stomiiforms. Ahlstrom's Stemoptychidae is undefinable in a phylogenetic analysis based on the information at hand. A resolution of the use of familial and subordinal names in stomiiform fishes must await completion of ongoing phylogenetic studies of these fishes. Because these studies are incomplete, it is difficult to make recommendations for names of certain stomiiform subgroups. Among the stomiiforms with elongate gill rakers in adults, the "family" problem is more complex than that recognized by Ahlstrom (1974) or Weitzman (1974). We here recognize two family names but these apply to only some of the 24 genera listed below. We recognize the Stemoptychidae of Weitzman (1974) and the Gonostomatidae in a new and restricted sense. See discussion below. The stomiiforms discussed here include the following 24 gen- era, listed alphabetically, which have been variously recognized as belonging to the families Gonostomatidae, Stemoptychidae, Maurolicidae, and Photichthyidae: Araiophos Grey (two species), Argyripnus Gilbert and Cramer (four, possibly a few more), Argyropelecus Cocco (about sev- en), Bonapartia Goode and Bean (one). Cyclolhone Goode and Bean (twelve). Danaphos Bruun (one, possibly two), Dip- lophos GUnther (two), Gonostoma Rafinesque (six), Ichthyo- coccus Bonaparte (three), Manducus Goode and Bean (two),' Margrethia Jespersen and Tuning (one, possibly two), Mau- rolicus Cocco (one, possibly two). Photichthys Hutton (one), Pollichthys Grey (one), Polyipnus Giinther (about sixteen), Polymetme McCulloch (one, possibly four), Sonoda Grey (two), Sternoptyx Hermann (two or three), Thorophos Bruun (two, including Neophos Myers), Triplophos Brauer (one), Valen- ciennellus Jordan and Evermann (one, possibly two), Vinci- guerria Jordan and Evermann (five), Woodsia Grey (one), and Yarella Goode and Bean (one). ' Grey (1964:88) recognized Manducus Goode and Bean, 1896 as a junior synonym of Diptophos GxmXheT, 1873 because, as she stated ". . . the differences appear to be of a specific rather than a generic nature . . ." This was in the context of the kinds of differences Grey noted separating other species of "gonostomatids." She did recognize both as subgenera ot Diphphos. We recognize both as genera. The species were most recently reviewed by Mukhacheva (1978) who recognized four species, D. maderensis (Johnson), D. rebamsi Krefft and Farm, D. greyae R. K. Johnson, and D. taenia Giinther. We have examined all four species and find that D. taenia and D. rebamsi have the cartilages of the two medial proximal pectoral radials, radials III and IV in the terminology of Fink and Weitzman (1982:66), fused while retaining two bony elements separate as reported for D. taenia by Fink and Weitzman (1982:65-67). Furthermore, one of the distal radials is out of line, not in a single series in these two species. These characters are specialized for these species. In Manducus maderensis and A/, greyae there are four completely distinct proximal radials and the distal radials are all in a simple straight series. Because the pectoral radial morphology in Diplo- phos taenia and D. rebamsi may be an intermediate stage of a transition series between radials such as are found in Manducus maderensis and M. greyae and those in the "photichthyid" genera, we recognize Man- ducus &% a genus and apparent sister group of the "photichthyid" genera as well as the Stomiidae, nearly all of which have the radials 111 and IV completely fused to one bone. A few stomiids have an apparent neo- morph condition in which the third proximal radial is divided into two radials, giving a total of four proximal radials. See also text discussion. 186 ONTOGENY AND SYSTEMATICS OF FISHES- AHLSTROM SYMPOSIUM Table 48. Summary of Meristic Characters for Adults of Certain Stomiiform Fishes. No. species Fin rays Branchi- oslegal rays No. of vertebrae Genera Dorsal Anal Pectoral Pelvic No. of gill rakers Araiophos 2 13-20 20-29 16-18 5 9-11 43-45 2-3 + 12-19 = 14-22 Argyripnus 4+ 11-12 11-15 + 8-12 = 22-29 15-19 6-7 8-10 41-46 4-7 + 12-19= 16-26 Argyropelecus 7 (8)9(10) 6-8 + 5-6=11 -13 10-11 7 7 34-40 15-24 Bonapartia 1 17-20 29-31 14-16 7-8 13-16 37 5-6 + 11-12= 16-18 Cydothone 12 + 12-15 16-21 9-13 6-7 10-14 29-33 4-10 + 9-18= 14-27 Danaphos 1 6 24-25 13-14 6 9-10 38 2 + 11-13 = 13-15 Diplophos 2 10-13 47-69 8-9 7 10-14 44-94 3 + 7-9= 10-12 Gonostoina 6 10-18 21-31 9-13 6-8 10-13 37-40 5-11 + 10-17= 15-27 Ichlhyococcus 3 10-15 13-17 7-8 6-7 11-12 38^7 7-11 + 15-26 = 22-37 Manducus 2 11-13 36-59 9-11 8 11-14 63-76 3-5 + 8-10 = 12-14 Margrethia 1 15-16 21-26 13-15 8 13 34 5 + 10-11 = 15-16 Mauro/icus 1 9-12 8-10 + 11-15 = 19-27 17-20 6-7 9-10 33-35 4-8 + 17-22 = 22-30 Photkhthys 1 12-13 23-26 9 6-7 20-21 51 4-5 + 11 = 15-16 Pollichlhys 1 10-12 22-30 8 6-7 11-12 40 4-5 + 11-12= 15-17 Polyipnus 17 10-17 13-19 12-16 7 9 31-36 10-28 Polymetme 3 11-13 24-33 9-11 7 (8?) 12-14 44-45 5-8 + 9-12= 15-19 Sonoda 2 8-9 8-10 + 14-16 = 22-25 13-15 6 8-10 40? 3-5 + 15-18= 18-21 Sternoptyx 3 8-11 14-16 10-11 7 7 28-31 7-9 Triplophos 1 10-12 53-63 9-11 6-7 11-14 ca 60 9 + 14-16 = 23-25 Thorophos 2 8 38 13 7 7-8 40-45 5 + 13-14= 18-19 Vatenciennellus 2 or 3 7-12 22-25 12-13 6-9 9-10 32-33? 2-3 +12= 14-15 Vinciguerha 4 13-16 12-17 9-10 7 10-12 38-42 3 + 11-23-11 = 15-33 Woodsia 1 11-12 14 9-10 7-8 17 42-45 3-5 + 13 = 16-18 Yarella ■> 14-16(17) (28)29- U 8-10 6-7 13-16 45-54 6-7 + 12-16= 18-22 Table 49. Position of the Dorsal and Anal Fin and Condition of the Adipose Fin in Certain Stomiiform Fishes. Dorsal &n position Genus Adipose fin Anal origin in advance of dorsal fin. Dorsal origin opposite 5th or 6th anal ray Anal origin opposite dorsal origin Anal origin opposite last dorsal fin ray Anal origin well in advance of dorsal by 9 rays Anal origin opposite dorsal fin or slightly behind Anal origin behind dorsal fin Anal origin beneath 5th ray or behind dorsal fin Anal origin opposite or 3-4 rays in advance of dorsal origin Anal origin behind dorsal fin by a space = '/2 dor- sal base Anal origin beneath 3rd from last or last dorsal fin ray Anal origin beneath 5th dorsal fin ray Anal origin beneath last dorsal fin ray Anal origin behind dorsal fin Anal origin beneath 3rd dorsal fin ray Anal origin usually beneath middle of dorsal fin Anal origin beneath end of dorsal fin Anal origin in advance of dorsal. Dorsal origin above 5th anal ray Anal ongin opposite dorsal origin Anal origin beneath end of dorsal fin Anal origin in advance of dorsal origin by 3 or 4 rays Anal origin 1 or 2 rays in advance of dorsal origin Anal ongin beneath middle of dorsal fin Anal origin behind middle of dorsal fin by dis- tance about = dorsal base Anal origin beneath middle of dorsal fin Anal opposite dorsal at 8 mm, adult position at 1 1 m Anal origin opposite dorsal origin Anal origin behind dorsal fin Same as adult Same as adult Same as adult Anal origin beneath end of or behind dorsal fin Same as adult Anal origin behind dorsal fin Unknown Araiophos Argyripnus Argyropelecus Bonapartia Cydothone Danaphos Diplophos Gonostoma Ichthyococcus Manducus Margrethia Maurolicus Photichthys Pollichlhys Polyipnus Polymetme Sonoda Sternoptyx Triplophos Thorophos Valenciennellm Vinciguerria Woodsia Yarella Present or ab- sent Present Present or ab- sent Absent Absent Absent Absent Present or ab- sent Present Absent Same as adult Present Anal origin beneath middle of dorsal Present fin, advances to adult condition as juveniles Unknown Present Anal origin advances forward beneath Present dorsal fin Same as adult Present or ab sent Unknown Present Unknown Absent Anal origin behind dorsal fin Present Unknown Absent Unknown Present or ab sent Same as adult Present Same as adult Present Same as adult Present Same as adult Absent AHLSTROM ET AL.: GONOSTOMATIDAE, STERNOPTYCHIDAE 187 Table 50. Dernition of Alphabetical Symbols used for Designating Photophores in Deep Bodied Sternoptychids and Other Stomiiform Fishes. Other slomiiforms Deep bodied slemoplychids SO Symphyseal photophores (organs) located at tip of lower jaw. Orb Photophores associated with the eye located ante- rior and posterior of orb it. Op Photophores on opercle series generally three, cod- ed as follows 1/(1 -I- 1). Br(BRP) Photophores located on the branchiostegal mem- branes. Is(I) Photophores located on the isthmus. IP Photophores of the ventral series found from the isthmus to the base of the pectoral fin. PV Photophores of the ventral series found from the pectoral fin base to the pelvic (ventral) fin base. VAV Photophores of the ventral series found from the pelvic (ventral) fin base to the anal fin base. AC Photophores of the ventral senes found from the anal fin base to caudal fin base of the ventral se- ries. IC Summary of photophores of the ventral series from the isthmus to caudal fin base (IP + PV + VAV + AC). IV Summary of photophores of the ventral series from isthmus to pelvic (ventral) fin base (IP + PV). OV Photophores of the lateral series from the opercle to pelvic (ventral) fin base. VA(VALA) Photophores of the lateral series from the pelvic (ventral) fin base to the anal fin base. OAA Summary of photophores of OV plus VA series. OA(OAB) Summary of lateral photophores from the opercle to anal fin base (OV + VA). OAC(OC) Entire lateral series on body sides just dorsal to ventral series and extending from opercular border, or just medial to it, over anal fin to cau- dal fin base. ODM Photophores (organs) found dorsal to the lateral midline (found only in Gonosloma gracile). SO PO PTO PRO Br Is AB PAN AN SC SAB SP L SAN Subopercle photophore which is equivalent to pos- teriomost photophore in opercular series of gon- ostomatids. Photophore located anterior to orbit. Photophore located posterior to orbit and may be equivalent to upper photophore of opercular se- ries of gonostomatids. Preopercular photophore, used for an PO photo- phore dorsal to ventral limb or preopercle. Same as gonostomatid definition. Same as gonostomatid definition. Photophores of ventral series located abdominally between pectoral fin base and pelvic fin base and equivalent to PV in gonostomatids, plus a few posterior photophores of the IP series. Photophores found anterior to anal fin and may be equivalent to VAV or VA in gonostomatids. Photophores found above anal fin. Photophores found on lower (sub) caudal peduncle. Together with AN group may be equivalent to AC in gonostomatids. Photophores located above (supra) to the abdomi- nal series and may be equivalent to VA in gon- ostomatids. Photophores located above (supra) the pectoral fin and may be equivalent to OV in gonostomatids. Photophore located laterally above PAN (found only in Polyipnus). Photophores located above (supra) to anal photo- phores and equivalent to part of AC series. Some genera are extremely rare (i.e., Thorophos and Sonoda) while Others represent the most abundant vertebrate animals on earth (Cyclothone and I'incigiierria). Developmental information has been published for 16 of these genera (12 prior to Ahlstrom, 1974; 3 by Ahlstrom, 1974; and one by Ozawa, 1976). Development Eggs.— Eggs were desciibed for Argyropelecus hemigymnus by Sanzo (1928); for Ichthyococcus ovatus by Sanzo (1930b); for Maurolicus muelleri by Sanzo (193 Id), Mito (1961a). and Oki- yama (1971); for Vinciguerna lucetia. V. poweriae. and I', nim- baria by Ahlstrom and Counts ( 1 958); for V. attenuata by Sanzo (193 Id); and for Gonostomadenudatumby Sanzo (\9'i\d). Oth- er accounts provide minimal details of ovarian eggs of other species. The details of egg characters are summarized in Table 47. Larvae. — Much has been accomplished for the identification of the larvae of these stomiiform genera and now descriptions are available for all except Manducus. Triplophos, Polymetme, Pho- tichthys, Thorophos, and Sonoda. The larvae tentatively iden- tified as Polymetme by Ahlstrom ( 1 974), on further examination by one of us (Richards), were determined to be Pollichthys. One stomiiform larval form has been described but not assigned to a genus [designated "Maurolicine Alpha" by Ahlstrom (1974: 670)]. It presumably is the larva of some stemoptychid (as de- fined by Weitzman, 1974). Descriptive details and illustrations of several species were given by Ahlstrom (1974). Here we pro- vide new or additional data including characters useful in iden- tifying these larvae and illustrations of all the species described to date, including some illustrated for the first time. The identification of stomiiform larvae with elongate gill rak- ers as adults requires a knowledge of developmental data from larvae, juveniles, and data from adults of the following char- acters: counts of fin rays, teeth, and other meristic characters as photophores; patterns of photophore development; and distri- butions (patterns) of dark chromatophores (dark pigment cells). With those sets of data, nearly all species should be identifiable at least to genus, and in cases of complete data, to species. A 188 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Table 51. Photophore DrsTRiBUTiON in Certain Stomtiform Genera. Refer to text and Table 50 for definition of codes. Photo- phores in No. of group of rows so orb OP BR IS IV VAV AC glands Araiophos 1 No 1 1 5-7 Yes (2) + (3) + 3-4 + (2) = 10-11 3-5 6-8 Yes Argyripnus 2 No 1 3 6 Yes (6) + (10) (18-28) (4-5) + (12-18) = 35-51 Yes Argyropelecus 2 No 2 2 6 Yes 18 4 10 Yes Bonapartia 1 Yes 1 3 11-13 No 14-16 5-6 18-20 No Cydothone 2 No 1 2 8-11 No 12-14 4-5 12-16 No Danaphos 2 No 1 2-3 6 Yes 18 5 22-26 Yes Diplophus 3 + Yes 1 3 7-12 + 0-3 9 Yes 33-*9 13-17 33^9 No Gonostoma 2 Yes 1 2-3 No 11-16 3-10 15-23 No Ichthyococcus 2 No 2 3 11-12 Yes 25-28 9-14 12-14 No Manducus 2 + Yes 1 3 8-13 Yes 30-33 12-14 28-39 No Margrethia 1 No 1 3 9-12 No 13-15 4 17 No Maurolicus 2 Yes 1 3 (6) Yes (6) + (12-13) = 18-19 (6) 1 + (14-18) + (7- = 22-27 9) Yes Pholichlhys 2 Yes 2 3 17-18 Yes 10 + 14-15 = 24-25 15-17 16-18 Yes Poltichthys 2 Yes 2 3 8 Yes 21-23 7-9 18-21 No Polyipnus 2 No 2 2 6 Yes 16 5 10-18 Yes Polymetme 2 Yes 1 3 9-10 Yes 19-21 7-8 21-25 No Sonoda 2 No 1 3 6-7 Yes 6 + 10= 16 7-8 (16-21) + (19-24) (5-6) + (5-6) + (5- = 36-43 or -6) Yes Siernoptyx 2 No 2 2 3 Yes 15 3 7 Yes Tnplophos 2 + 3 or 4 Yes 1 3 8-13 Yes 24-30 5-7 35-41 No Thorophos 2 Yes & no 1 3 6 Yes 17 5 13-15 Yes Valenciennellus 2 No 1 3 6 Yes (3 + (4) + (16-17) = 23-24 (4H5) 3-6 or 9-17 Yes Vincignerna 2 Yes or no 2 3 7-9 Yes 21-24 7-11 12-15 No Woodsia 2 Yes 2 3 14 Yes 25 11-12 12 No Yarella 2+ sev Yes 1 3 11-13 Yes 23-25 9-12 20-28 No summary of several meristic characters for genera is given in Table 48. The position of the dorsal and anal fins is also a helpful aid, but caution must be used since their positions relative to other structures may change with growth. Also, the presence or absence of the adipose fin is helpful, but again, caution is in order because this fin is fragile and often damaged or lost due to contact with a net. These fin features are given in Table 49. Of special importance in identifying lai^ae and adults is the distribution and patterns of the photophores. This includes the number in each series, the patterns of their distribution in re- lation to each other, and especially the sequence of development which Ahlstrom (1974) stressed. Some confusion appears in the literature because more than one alphanumeric code has been developed to indicate, in some cases, the same sets of photo- phores in different stomiiform groups. A further complication is that the deep-bodied stemoptychids have a different code because of their altered body shape as adults and homologies were considered uncertain. Weitzman (1974:461), because he united the "maurolicin" and deep-bodied stomiiforms as one family considered the different termmologies "artificial" and as obscuring homologies. He therefore discussed and presented a synonymy of stomiiform photophores. We have defined the alphabetical codes in Table 50 and included what we believe are equivalent photophores in stomiiforms. In this code, par- enthetical numbers indicate photophores found in common glands whereas non-parenthetical numbers indicate that the photophores are single. The distribution of photophores for each genus is given in Table 51. Table 52 provides sequences of photophore formation for Bonapartia, Margrethia, and Gon- ostoma. Table 53 provides similar information for Araiophos. Maurolicus, Danaphos, Valenciennellus. and Argyripnus; while Table 54 provides similar data for Polyipnus. Argyropelecus, and Siernoptyx. Diagnostic pigmentation and morphometric characters are summarized in Table 55. Illustrations (Figs. 98 to 104) are provided for the genera for which larvae are known and for many of the known species. In addition, the following authors provide specific information which will aid in larval identifications: Jespersen and TSnmg (1919, 1926), Sanzo (193 Id), Ahlstrom and Counts (1958), Ahlstrom and Moser (1969), Ozawa (1976), Grey ( 1 964), Badcock and Merrett (1972), Kawaguchi and Marumo (1967), Okiyama (1971), Badcock ( 1 982), Rudometkina (1981), Gorbunova (1981), Mukhacheva (1964), and Ahlstrom (1974). Relationships There has been a dichotomy of opinions about the interre- lationships of the genera and the use of family names based on the use of larval versus adult morphological characters. Ahl- strom (1974:670-672) presented his views on this group based on larval characters, principally the mode of photophore for- mation. The suggested relationships resulting from his analysis contrasted in part with those of Weitzman (1974:472), whose views were based on study of adult osteology and soft anatomy. Both Ahlstrom and Weitzman in addition to their own data, AHLSTROM ET AL.: GONOSTOMATIDAE, STERNOPTYCHIDAE 189 Table 52. Sequence of Photophore Formation in Bonapania. Margrethia. and Gonostoma. BR pv VAV AC Bonapartia pedaliota Margrclhia obtusirostra Gonostoma etongatum Gonostoma demidalum Gonostoma gracile Gonostoma ehelingi Gonostoma bathyphilum Gonostoma allanlicum adult 9.5 11.5 12.0 14.0 15.0 16.0 23.0 adult 5.8 6.4 8.0 11.3 15.0 adult 6.0 7.5 7.9 10.2 13.0 14.0 16.7 22.5 adult 18.25 19.0 20.75 24.75 29.65 34.0 39.0 adult 15.5-5-17.0 20.0 22.0 adult 13.8 15.0 adult 11.0 14.8 adult 12.0 13.0 14.5 17.8 18.8 23.8 I 1 1 1 1 I 1 I 1 1 1 1 1 1 1 1 1 I I n I 1 I 1 1 1 1 1 11-13 2 3 4 5 5 6 II 9-12 2 6 9 9 2/1 2 2 3 9 9 I 3 5 9 9 9 2 9 9 9 4 9 4 9 9 14-15 3 5 5 10 9 II 14 13-15 2 6 10 14 14 15 5 4 10 11 II II 15 15-16 1 2 3 6 14 16 16 13-15 13 14 15 7 9 11-12 5 10 15-16 1 2 13 16 16 5-<6) 2 2 4 3 5 5 4 2 4 4 4 (4)-5 2 3 2/3 4 5 5 1 3 5 5 5 4-5 5 4 10 4-5 2 5 3 5 5 16-18 + 2-3 3 + 1 I + 1 5 + 2 14 + 2 13-14 + 3-4 1 + 2 1 + 2 5 + 3 11+4 21-23 1 + 1 + 22 17-20 + 2 + 3 3 + 3 11+3 15 + 5 15 + 5 17-19 17 18 19 20-21 19 1 19 19 II- Grey, 1964 Grey, 1964 Original Grey, 1964 Grey, 1964 Original Jespersen and TSning, 1919 Grey, 1964 Grey, 1964 Ahlstrom, 1974 Ahlstrom, 1974 Ahlstrom, 1974 Ahlstrom, 1974 Ahlstrom, 1974 13-15 Grey, 1964 Ahlstrom, 1974 Ahlstrom, 1974 Original Ahlstrom, 1974 Ahlstrom, 1974 Grey, 1964 Jespersen and Tuning, 1919 13 Grey, 1964 13-15 Grey, 1964 Sanzo, 1912b Sanzo, 1912b Sanzo, 1912b Sanzo, 1912b Sanzo, 1912b 13 Sanzo, 1912b 13 Sanzo, 1912b 1 2 + 6-7 Kawaguchi and Marumo, 1 967 Kawaguchi and Marumo, 1967 Kawaguchi and Marumo, 1967 12 + 4 Kawaguchi and Marumo, 1967 21 Grey, 1964 Ahlstrom, 1974 Ahlstrom, 1974 14 Grey, 1964 Ahlstrom, 1974 Ahlstrom, 1974 13 Grey, 1964 Ahlstrom, 1974 Ahlstrom, 1974 Ahlstrom, 1974 Original Ahlstrom, 1974 13 Original used the results of photophore anatomy research by Bassot ( 1 966, 1971) to support their conclusions. These results seemingly completely supported Weitzman's referral of genera to family groups and agreed with Ahlstrom except for placement of three genera — C'lr/or/iowc, Diplophos (including Mandncus), and Tnplophos. One of us (Weitzman), continues to study relationships of the stomiiforms with elongate gill rakers in adults and we offer the following analysis as a current comment on the status of our knowledge of these fishes. The two concepts of relationships by Ahlstrom and Weitzman may be compared as follows: Ahlstrom (1974:670-672) stressed relationships of taxa based on photo- phore patterns and development. Ahlstrom (1974:672) consid- ered the stomiiforms with elongate gill rakers in adults as a group comprised of three groups of genera, with any subdivision being into two subfamilies based on photophores occurring in- dividually or in clustered groups. These groups of genera in- clude: (I) Those with individual separate photophores, most of the photophores developing simultaneously and initiated as a "white" photophore stage. This group includes Manducus, Dip- lophos, Cyclothone. Yarella, Pollichthys, V'inciguerha. Wood- sia, Ichlhyococcus. and presumably Triplophos and Polymetme, 190 ONTOGENY AND SYSTEMATICS OF FISHES -AHLSTROM SYMPOSIUM Table 53. Sequence of Photophore Formation in Araiophos, Maurolicus, Danaphos, Valenciennellus, and Argyripnus. ORB OP so BR IP PV VAV AC OA Source Araiophos adult 1 1 (6) (2) (3) + 3- (3) (2) + 2 + (2) No Ahlstrotn and Moser, 1969 eastropas 4 + (2) 11.2 (3) (2) - Ahlstrom and Moser, 1969 Maurolicus adult 1 3 1 (6) (6) (12) (6) 3/(4) + (8) (2) + 7 Ahlstrom, 1974 muelleri 5.5 (1/2) Ahlstrom, 1974 6.2 (2) (2) Ahlstrom, 1974 6.5 (2) (4) Ahlstrom, 1974 6.7 1 (3) (5) Ahlstrom, 1974 6.9 1 (4) (8) Ahlstrom, 1974 7.5 1 (4) I (9) + (2) + Ahlstrom, 1974 8.6 2 (5) (3) (12) (2) + (3) + (3) Ahlstrom, 1974 9.0 2 (5) (3) (11) (2) + (3) + (3) 1 Ahlstrom, 1974 9.7 3 (5) (5) (11) (3) + (4) + (6) (2)+ 1 Ahlstrom, 1974 10.8 3 (6) (5) (12) (4) + (5) + (6) (2) + 2 Ahlstrom, 1974 13.5 3 (6) (6) (12) (6) + (9) + (7) (2) + 6 Ahlstrom, 1974 Danaphos adult 1 3 (6) (3) + (4) (11) (5) (3) + 16 + 6 Ahlstrom, 1974 oculatus (4)+ 1 Ahlstrom, 1974 16.5 (2) Ahlstrom, 1974 16.5 (3) (3) Ahlstrom, 1974 19.2 (4) (10) Ahlstrom, 1974 21.0 1 1 (5) (2) + (4) (10/11) (2) + + + Ahlstrom, 1974 21.3 1 1 (4/5) (3) + (4) (10) (3) + + (2) + Ahlstrom, 1974 21.8 I 2 (5) (3) + (4) (11) (2) (3) + 8 + (4) + 2 Ahlstrom, 1974 24.2 I 2 (6) (3) + (4) (11) (2) (3) + 9 + (4) + 2 Ahlstrom, 1974 Valenannellus adult 1 3 (6) (3) + (4) (16-17) (4-5) (3) + (3) + (3) + (2) + 3 Ahlstrom, 1974 thpunculatus (2) + (4) 7.8 Original 8.6 (3) (3) Ahlstrom, 1974 9.5 (4) (6) Ahlstrom, 1974 11.0 (4) (10) Original 12.0 (4) (13) (2) Ahlstrom, 1974 12.4 1 (5) (15) (2) Original 13.0 1 (5) (2) (15) (2) Original 13.2 (4) (14) (3) Ahlstrom, 1974 14.0 1 (5) (4) (15) (5) Original 17.0 1 2 (4-5) (3) + (4) (15) (5) (3) + (3) + + (3) + (4) (2) Grey, ige-) Argyripmis adult 1 3 (6) (6) (10) (26) (5) + (17) (3) + 4 Badcock and Merrett, 1972 atlamicus 18.7 1 2 (6) (3) (10) (3) (4) + (4) Badcock and Merrett, 1972 16.8 1 2 (6) (3) (10) (2) (4) + (3) Badcock and Merrett, 1972 although their development is not known. (2) Those with in- dividual, separate photophores that have a gradual, protracted metamorphosis. This group includes Bonapartia, Margrelhia. and Gonostoma. (3) Those with some individual photophores but some or most of the photophores with common bases [ac- tually a common lumen, during development at least] and hav- ing a gradual, protracted metamorphosis. This group includes Araiophos, Maurolicus, Danaphos, Valenciennellus, .Argyrip- nus, Polyipnus, Argyropelecus, Sternoptyx, and presumably Thorophos and Sonoda ahhough their development is unknown. Groups (1) and (2) comprised the subfamily Gonostomatinae and Group (3) comprised the Stemoptychinae in Ahlstrom's concept. Group (3) is equivalent to Weitzman's Stemoptychi- dae. The genus Gonostoma was considered "pivotal" by Ahl- strom; that is, its relationships could be with either the gonos- tomatines or the stemoptychines of his concept. In Ahlstrom's conclusions, the photophore pattern of Group (1) is most like that of the stomiid groups discussed by Fink in this volume. Weitzman's classification (1974) concentrated in most detail on a hypothesis of phylogenetic relationships within the family Stemoptychidae as he defined it. Weitzman (1974) pointed out that more detailed studies should be conducted on other sto- miiform genera in the future, but he did discuss their possible relationships. Based on the number of proximal pectoral-fin radials, he established two infraorders for stomiiform fishes. Members of the Infraorder Gonostomata were considered to have four proximal pectoral-fin radials (except Cyclothone with one). This infraorder was divided into two families based prin- cipally on Bassot's photophore findings; Family Gonostomati- dae with Beta type photophores comprised of Diplophos in- cluding Manducus), Triplophos, Bonapartia, Margrethia. Gonostoma, and Cyclothone and the Family Stemoptychidae with Alpha type photophores comprised of Thorophos, Araio- phos, Maurolicus, Danaphos, Valenciennellus, Argyripnus, Son- oda, Polyipnus, Argyropelecus, and Sternoptyx. The problem with Weitzman's Gonostomata is that it was based on a prim- itive character for the stomiiforms, four pectoral-fin radials, and this character cannot be used as a synapomorphy to define a subgroup of stomiiforms. The non-stemoptychid and non-gon- ostomatid genera, along with the stomiiform families possessing barbels originating from the hyoid bone and lacking elongate gill rakers in the adults (the Stomiidae of Fink, this volume). AHLSTROM ET AL.: GONOSTOMATIDAE, STERNOPTYCHIDAE 191 Table 54. Sequence of Photophore Formation in Polyipnvs. Arcyropelecvs and Sternoptyx. OP PRO + Size PC PTO BR IS SO SP AB SAB PAN AN SAN LSC Source Polyipnus polli adult 1 1 6 6 1 + 1 3 10 3 5 8 3 14 Baird, 1971 4.3 2 0+ 1 1 Original 4.8 1 4 2 1 + 1 3 Original 5.5 1 1 6 4 1 + 1 2 8 Original 6.0 1 1 6 6 1 + 1 2 10 1 Original 7.5 1 1 6 6 1 + 1 2 10 3 Original 9.0 1 1 6 6 1 + 1 2 10 3 2 2 Original 9.6 1 1 6 6 1 + 1 2 10 1 3 2 2 Original 13.5 1 1 6 6 1 + 1 3 10 3 5 4 14 Original 15.3 1 1 6 6 1 + 1 3 10 3 5 4 1 14 Original 17.0 1 1 6 6 1 + 1 3 10 3 5 4 2 14 Original 18.4 1 1 6 6 1 + 1 3 10 3 5 6 3 14 Onginal 23.5 1 1 6 6 1 + 1 3 10 3 5 7 3 14 Original Argyropelecus adult 1 1 6 6 1 + 1 2 12 6 4 6 4 Baird, 1971 hemigymnus 10.92 4 6 + 1 7 1 2 Sanzo, 193 Id 9.92 6 6 0+ 1 9 2 3 Sanzo. 193 Id 7.84 1 6 6 1 + 1 2 12 3 4 Sanzo. 193 Id 11.20 1 1 6 6 1 + 1 2 12 2 3 4 4 Sanzo, 193 Id Arygropelecus sp. adult 1 1 6 6 1 + 1 2 12 6 4 6 4 Baird, 1971 4.5 + Original 9.5 6 6 + 1 6 1 Original 9.5 6 6 + 1 8 3 Original 7.0 1 6 6 1 + 1 2 12 3 3 Original 7.0 1 6 6 1 + 1 2 10 3 4 Original 7.4 1 1 6 6 1 + 1 2 12 4 4 4 3 Original 10.0 1 1 6 6 1 + 1 2 12 5 4 5 4 Original Sternoplyx sp. adult 1 1 3 5 1 + 1 3 10 3 3 I 4 Baird, 1971 4.8 + 1 Original 7.5 + 1 Original 7.8 1 2 3 + 1 4 Onginal 8.1 1 2 4 + 1 2 7 Original 7.6 1 2 5 0+ 1 3 10 1 3 1 Original Table 55. Diagnostic Pigment Characters and Unusual Morphometric Features of Some Stomiiform Larvae. Genus/species Diagnostic character Diplophos taenia Bonapartia pedaliota Margrelhia obtusirostre Gonostoma Cyclolhone Yarella blackfordi Pollichthys mauli Vincigucrria Woodsia nonsuchae Ichlhvococcus ovatus Pigment spots on dorsal and ventral midline. Extremely elongated larvae. Similar to Gonostoma but lacks deep pigment spot behind eyes and has pigment on medial portion of caudal Ijeduncle. A distinct vertical streak of pigment on caudal peduncle in most specimens. All species usually have deep pigment spot behind eyes. Specific differences among the species are as follows: G. elongatum. G. gracile and G. ehelingi lack pigment on caudal peduncle; G. bathyphilum has pigment spots on dorsal edge of caudal peduncle; G. atlanticum has pigment over medial portion of caudal peduncle (closely resembles Cyclothone in ventral pigmentation and swimbladder position); G. denudatum has broad streak of pigment diagonally over caudal fin base from dorsal caudal peduncle to base of lower caudal fin rays. A distinct, dark streak or intense melanophore over and parallel to the parhypural on the caudal fin base, pigmentation over gut and along ventral margin of tail and a conspicuous swimbladder. Myosepta pigmented over caudal peduncle giving chevron appearance. No pigment except for the eyes. Very similar to Vinciguerria in other aspects. All species have medial or ventral margin caudal pigment spot. I '. nimbaria and V. lucetia have the caudal pigment spot restncted to the ventral margin of the caudal fin base and pigment above the anal fin. V. attenuata and I '. poweriae has the caudal pigment spot in a medial position and no pigment above the anal fin. r. attentuata has pigment over the airbladder which is lacking in C. poweriae. V. poweriae has a struc- ture above the anal papilla which may appear as pigment. V. mabahiss is similar to V. nimbaria and is restricted to the Red Sea (Johnson and Feltes, 1984). Melanophores profusely distributed on all myomeres below the lateral midline. Broad pigment band along roof of mouth continuous with trunk pigment. Also has a trailing gut and elongated rays on pectoral fin, both of which may be missing. Pigment profusely distributed on all myomeres below the lateral midline. Elongate pectoral fin rays and a trailing gut, both of which may be missing. 192 ONTOGENY AND SYSTEM ATICS OF FISHES -AHLSTROM SYMPOSIUM Fig. 98. Lateral views from top to bottom: Diplophos taenia 22.0 mm SL, R/V OREGON II Cr. 126, Sta. 36754, 27''30'N, 092°30'W, May 10, 1982, drawn by J. C. Javech; I'lncigiierria lucelia 9.0 mm SL modified after Ahlstrom and Counts (1958); I'lnciguerna powcnae 1 1.5 mm SL, R/V OREGON II Cr. 126, Sta. 36746, 27°59.9'N, 088°00'W, May 8, 1982, drawn by J. C. Javech; and linciguerna atlenuata 9.7 mm SL modified after Jespersen and Tuning (1926). Fig. 99. Lateral views from top to bottom: Pollichlhvs niauli 14.5 mm SL, R/V OREGON II Cr. 126, Sta. 36688. 26°00.5'N. 0.88°00.4W, April 20, 1982, drawn by J. C. Javech; Yarella blackfordi 23.5 mm SL, R/V OREGON II Cr. 126, Sta 36752, 27°30'N, 094°30.3'W, May 9, 1982, drawn by J. C. Javech; Woodsia nonsuchae 1 1.5 mm SL, Eastropac, Sta. 75.225, drawn by J. C. Javech; and Ichthyococcus ovalus 18.1 mm SL, R/V OREGON II Cr. 126, Sta. 36746, 27°59.9'N, 0.88°00'W, May 8, 1982, drawn by J. C. Javech. AHLSTROM ET AL.: GONOSTOMATIDAE, STERNOPTYCHIDAE 193 Fig. 100. Lateral views from top to bottom: Bonapar/ia pedaliota 1 1 .5 mm SL. R/V OREGON II Cr. 1 26, Sta. 36688. 26°00.5'N, 088''00.4'W, April 20, 1982, drawn by J. C. Javech; Margrethia obtusirostra 6.7 mm SL, R/V OREGON II Cr. 126, Sla. 36773. 26°00.rN, 094°00.2'W, May 23, 1982, drawn by J. C. Javech. were placed in the Infraorder Photichthya. Nearly all have three, or rarely fewer, proximal pectoral-fin radials, a specialized char- acter which can be used as a synapomorphy uniting this group. As noted above, there are a few exceptions which bear four proximal radials but these appear to be either reversals or are neomoiphic. Within the Infraorder Photichthya the stomiiform genera with elongate gill rakers in adults were placed in the Family Photichthyidae comprised of the genera Polymetme. Yarella. Pollichthys. Pholichlhys. Vinciguerria. Woodsia, and Ichthyococcm. This placement was done on the basis of the presence of Type Gamma photophores in at least most of the genera, a specialization for the group (as well as for at least some of the stomiid genera) and therefore a synapomoiTahy. The pres- ence of elongate gill rakers in this group is not a synapomorphy because it is primitive for the group. Essentially, Ahlstrom and Weitzman disagreed on the rela- tionships of three genera. Alhstrom's Group ( 1 ) was mostly equivalent to Weitzman's Photichthyidae but included three genera, Cyclothone, Dtplophos (including Manducus), and Trip- lophos. placed in the Gonostomatidae by Weitzman. Otherwise, Weitzman's Gonostomatidae was equivalent to Ahlstrom's Group (2). Based on evidence available to Ahlstrom and Weitz- man, on some supplementary evidence provided by Fink and Weitzman (1982), and on some of our own data, we here present a somewhat different arrangement based on a more rigorous phylogenetic analysis than done by Weitzman (1974). It turns out that Weitzman's analysis of the Stemoptychidae and its genera is consistently phylogenetic but that of outgroup sto- miiforms is not. Ahlstrom (1974) did not attempt to analyze his groups phylogenetically. The evidence available now seems to resolve the conflict between Ahlstrom (1974) and Weitzman (1974). However, we would note that the analysis below is to be regarded as a guide to future studies rather than any sort of well-corroborated phylogeny. Parts, at least, of the arrangement need much additional study. Furthermore, the relationships of the genera in contention by Ahlstrom and Weitzman are still not fully clear. Some of these genera, Manducus, Diplophos. and perhaps Triplophos, are relatively primitive within stomiiforms with few characters specialized beyond the stomiiform level. This makes placing them in stomiiform subgroups difficult. Cy- clothone is more derived but retains certain primitive stomi- iform features and its relationship, although in our view is un- doubtedly with the gonostomatids, at this time is somewhat uncertain because our data are not fully analyzed. The conflict between Ahlstrom (1974) and Weitzman (1974) arose in part because they both utilized one or the other of certain characters. Type Beta photophores and "white" pho- tophore development, as though they were shared specialized characters, synapomorphies indicating relationships. Instead, we believe these features are plesiomorphous for stomiiform subgroups and cannot be used to support a hypothesis of rela- tionships among stomiiform genera. Our current analysis is as follows. Fink and Weitzman (1982:69-75) list and discuss eight syn- apomorphies for stomiiform fishes. One of these, stomiiform- type photophores, was described in some detail based in part on Bassott (1966, 1971). Bassot (1966:574-576), Weitzman ( 1974:338), and Fink and Weitzman ( 1 982:70) recognized Type Beta photophores as primitive for stomiiforms. Bassot (1966, 1971) recognized two other types of photophores. Type Alpha 194 ONTOGENY AND SYSTEMATICS OF FISHES- AHLSTROM SYMPOSIUM Fig. 101. Lateral view from top to bottom: Gonosloma bathyphilum 1 1 .0 mm SL modified after Ahlstrom ( 1 974); Gonostoma elongatum 9.8 mm SL modified after Ahlstrom (1974); Gonostoma ebeUngi 15.0 mm SL modified after Ahlstrom (1974); Gonostoma atlanticum 12.0 mm SL modified after Ahlstrom (1974); Gonostoma denudalum 20.7 mm SL modified after Sanzo (193 Id). AHLSTROM ET AL.: GONOSTOMATIDAE, STERNOPTYCHIDAE 195 Fig. 102. Cydothone signata 9.0 mm SL, drawn by H. Orr. and Type Gamma, as being more specialized. This recognition, although not stated by these authors, is based on a concept that Types Alpha and Gamma photophores of some stomiiformes appear to be elaborations of Type Beta photophores. In other words, their particular features appear to be developmental ter- minal additions to Type Beta photophores and are therefore available for use as synapomorphies for stomiiform subgroups. Although more detailed analyses of these features are needed, for the sake of discussion we here accept that Type Beta pho- tophores are primitive for stomiiforms. Weitzman (1974:338), on the basis of outgroup comparison (not described or discussed in his text), considered four proximal pectoral-fin radials to be primitive for stomiiforms, their re- duction to three or fewer to be specialized. We see no reason to change that analysis. Thus three or fewer proximal pectoral- fin radials are available as synapomorphous characters for sto- miiform subgroups. Ahlstrom ( 1 974:660) described what can be labeled as "white" photophore development in which most, or at least the ventral series of photophores, are "laid down initially during a white photophore stage [before black pigment develops] and only a few photophores are late forming." One form or another of "white" photophore development is common to all stomiiforms except those including the gonostomatid genera Bonapariia. Margrethia. and Gonostoma, and the stemoptychids of Weitz- man ( 1 974). Members of these gonostomatid and stemoptychid genera have a protracted metamorphosis from the larval stage as well as a gradual, more extended photophore formation. This latter type of photophore development appears to be an elab- oration of "white" photophore development and thus we con- sider white photophore development primitive with respect to the more complicated forms having prolonged photophore de- velopment. Again, much information of an anatomical and de- velopmental nature remains to be gathered from the process of photophore development. If "white" photophore development and Type Beta photo- phores are primitive in regard to stomiiform subgroups and therefore unavailable as synapomorphies for stomiiform subgroups, then the conflict regarding the distribution of char- acters among taxa between Ahlstrom (1974) and Weitzman (1974) disappears in a phylogenetic analysis by somewhat al- tering certain of the groups of both authors as follows. In our tentative scheme of relationships, Weitzman's Ster- noptychidae and Ahlstrom's Group (2) genera (Ahlstrom, 1 974: 671), Bonapariia, Margrethia, and Gonostoma, the Gonosto- matidae in the strictest sense, are united by a synapomorphy consisting of a specialized form of prolonged metamorphosis and photophore development described by Ahlstrom (1974: 660-661). See also Tables 52-54 herein. These three gonosto- matid genera and Cydothone apparently share derived char- acters of the jaws and associated head parts which will be ex- plained in a later contribution. These four genera retain the primitive Type Beta photophores, a character relating stomi- iforms only at the ordinal level. In our opinion these four genera constitute the Gonostomatidae and Cydothone may have lost prolonged photophore development through paedomorphic re- versal associated with the small size of most of its members, a situation needing further study. The Stemoptychidae have specialized Type Alpha photo- phores and the several other synapomorphies listed by Weitz- man ( 1 974:446-448). In addition they apparently share a unique photophore growth pattern previously unrecorded. One of us (Weitzman) has been studying photophore development in re- lation to phylogenetic studies in stomiiforms and has found that each cluster or group of photophores of the stemoptychids ap- pears to develop by budding from one single photophore rather than by fusion at a later growth stage of separately developed photophores. This is a terminal developmental addition in pho- tophore ontogeny and both outgroup comparison and devel- opmental information indicate that this pattern of photophore formation is a specialization in comparison to the simpler ap- pearance of single, separate body photophores (usually one per scale in any given series found in other stomiiforms). This growth character appears to be present in all stemoptychid genera for which we have developmental information. It is therefore a likely synapomorphy for the group. Manducus (based on the type species, Gonostoma maderense Johnson) is a primitive stomiiform, having ordinal-level char- acters with no known specialized characters except the absence of an adipose fin and a short neural spine on the preural centmm. The latter may be a primitive rather than a specialized stomi- iform feature. Diplophos (based on the type species Diplophos taenia Gunther) appears to have a transitional stage pectoral radial morphology between Manducus on the one hand and the Photichthyidae of Weitzman (1974) (an ill-defined group) and the Stomiidae on the other. In Manducus the cartilages and bones of proximal pectoral-fin radials III and IV remain separate whereas Diplophos has the cartilages, but not the bones, of the two elements fused. Fink and Weitzman (1982:65-67). In the "photichthyids" and stomiids the cartilages and bones of the two medial pectoral-fin radials are fused. This represents the terminal condition in the transition series except that in some genera there is a reversal of radial numbers and in Eustomias there occurs a further specialized, reduced pectoral-fin radial 196 ONTOGENY AND SYSTEM ATICS OF FISHES -AHLSTROM SYMPOSIUM AHLSTROM ET AL.: GONOSTOMATIDAE, STERNOPTYCHIDAE 197 Fig. 104. Lateral views from top to bottom: Polyipnus polli 5.2 mm SL R/V GERONIMO Cr. 2, Sta. 155, 05°28S, 01°120'E, August 21, 1963, drawn by J. C. Javech; Argyropelecus hemigy'mnus 7.8 mm SL modified after Sanzo (193 Id); and Sternoptyx sp. 8.8 mm SL. drawn by H. C. Orr. Fig. 103. Lateral views from top to bottom: Araiophos eastropas 8.8 mm SL modified after Ahlstrom and Moser (1969); Maurolicus muelleri 10.8 mm SL modified after Ahlstrom (1974); Danaphos oculatus middle metamorphosis modified after Ahlstrom (1974); Valenaennettus tri- putulutatus middle metamorphosis modified after Ahlstrom (1974); Argyripnus atlanticus 18.7 mm SL modified after Badcock and Merrett (1972); and maurolicine Alpha 7.5 mm SL modified after Ahlstrom (1974). 198 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM condition. The "photichthyids" and stomiids have specialized Type Gamma photophores, although it is not known that all genera in these groups have Type Gamma photophores; this is a problem for further investigation. Manducus and Diplophos retain Type Beta photophores and all of these fishes apparently retain "white" photophore development of one kind or another. These two characters are only useful at the ordinal level as synapomorphies. Again, further research on "white" photo- phore formation is needed since there appears to be more than one form of this development. The monotypic Triplophos may or may not be related to Manducus and/or Diplophos. Triplophos has a variety of derived features not shared by Manducus or Diplophos. However, this tells us nothing about its possible relationships with these gen- era. Triplophos has four proximal pectoral-fin radials but with some reduction in radial IV, Type Beta photophores, and prob- ably "white" photophore development, the last two characters synapomorphous only at the ordinal level. Four pectoral-fin radials are not a synapomorphy for stomiiforms at any level since the feature is found in most teleost outgroups. Triplophos appears to be a primitive stomiiform with certain autapomorph- ic features associated with an elongate body. Its relationships are uncertain and there may be indications in the head and pectoral girdle anatomy of a relationship with certain photich- thyid genera. The problem needs much study. Cyclothone retains Type Beta photophores and "white" pho- tophore development but has its own specialized features such as only one pectoral-fin radial. It has a modified head and jaws, which resemble and are, in our opinion, synapomorphous with those of Gonostoma. The single pectoral-fin radial might be thought of as a terminal stage in a transition series from Man- ducus (with four pectoral-fin radials) to Diplophos to some mem- bers of the "Photichthyidae" and then to Cyclothone. However, Cyclothone does not have specialized Type Gamma photo- phores of the "photichthyid" genera. The phylogenetic rela- tionships of Cyclothone may not be certain as yet, but in many respects it bears a resemblance to the three gonostomatid genera and we favor its placement with these genera. See also discussion above. Although we have perhaps resolved the differences between Ahlstrom (1974) and Weitzman (1974), we have not achieved a useful phylogeny of most stomiiform groups. Rather, we have attempted to outline certain suggested hypotheses of relation- ships to be investigated in the future with additional data. Adult morphological data of the kind used by Weitzman to define and relate the stemoptychid genera are available in abundance and may be useful for other stomiiform groups. A closer look at growth stages with the specific purpose of looking for possible developmental specializations and terminal additions to char- acters found in outgroups should greatly aid in delineating re- lationships among the stomiiform genera. However, problems associated with a high percentage of homoplasy can be expected for some groups. The answers to problems of stomiiform in- terrelationships will not come easily. Consideration of certain features is in order. For example, larvae of Diplophos superficially resemble those of Chauliodus with their prolonged development to a large larval size and great elongation with bodies that are circular in cross section. Are these convergent larval specializations or primitive stomiiform features found only in certain stomiiform genera? The ventral pigmentation on the body of developing Diplophos resembles that of developing paralepidids and myctophoids. Is this a prim- itive stomiiform feature of Diplophos shared with certain sto- miiform outgroups or a gross convergence of pigment patterns? Woodsia and Ichthyococcus share with certain stomiid genera (for example, Eustomias) such developmental features as elon- gate pectoral-fin rays, trailing guts, pigmentation patterns, and bodies with a circular cross section. Some, if not all, of these may be shared larval specializations. But again, independent appearance of these characters indicated by a high degree of homoplasy may be a vexing problem. Larvae of other genera such as Vinciguerria. Pollichthys, and Cyclothone have body shapes and certain other features that closely, but presumably superficially, resemble those of clupeoid larvae. Detailed com- parisons of these similarities may possibly distinguish between homology and convergence among these taxa. In summary, a future phylogenetic analysis based on much additional data may clear up many of the problems of stomi- iform generic relationships. However, at present we are left with numerous phylogenetic problems and assignment of certain gen- era to family-level groups at this time would be misleading. The above analysis retains Weitzman's Stemoptychidae. It restricts the Gonostomatidae to the genera Bonapartia. Margrethia, and Gonostoma. and we recommend the inclusion of Cyclothone. The other groups of non-stomiid stomiiforms remain unclear as to family relationships. We agree with Fink and Weitzman (1982) that Manducus and Diplophos are primitive stomiiforms, but we cannot provide a stable classification for Manducus. Diplophos. and Triplophos. Manducus and Diplophos might seem to be sister taxa because of their similarity of appearance. How- ever, they share no known specialized character or characters that would unite them as a stomiiform subgroup except the absence of an adipose fin and possibly a short neural spine on the preural centrum. Currently all their other shared characters seem primitive for stomiiforms. Further analysis of this situa- tion is needed. Triplophos is again very much like a primitive stomiiform in its head especially, but it has a number of specialized stomiiform features as listed by Grey (1964:106) and may show some re- lationship to some of the "photichthyid" genera. That the genera classified in the "Photichthyidae" by Weitz- man (1974) form some kind of related group seems reasonable. However, relationships among these genera are not known. That these "photichthyid" genera are related to Diplophos is possible, and that the stomiids are related to the "photichthyids" is, in our view, very probable. The larval specializations of Woodsia and Ichthyococcus noted above, may be important here because they may be synapomorphies relating these genera to the sto- miids. Until the developmental and adult morphological features of many stomiiform genera are analyzed in detail, certain aspects of their developmental stages outlined, and detailed outgroup analysis performed on all putatively useful characters, we can make no certain predictions about relationships and classifi- cation. (W.J.R.) National Marine Fisheries Service, Southeast Fisheries Center, 75 Virginia Beach Drive, Miami, Florida 33149; (S.H.W.) Division of Fishes, National Museum of Natural History, Smithsonian Institution, Washington, D.C. 20560. Giganturidae: Development and Relationships R. K. Johnson THE Giganturidae contains two highly-specialized bathy- pelagic species placed in two monotypic genera; Gigantura chum Brauer, 1901 and Rosaura mdica (Brauer, 1901). Adults now placed in Rosaura were formerly recognized as Bathyleptus Walters, 1961. Morphological specializations of giganturids are sufficiently divergent and numerous that the group has usually been accorded subordinal or ordinal status somewhere within the group now recognized as basal neoteleosts (Stomiiformes + "Aulopiformes" + "Myctophiformes," see Rosen, 1973; John- son, 1982; Fink and Weitzman, 1982). Giganturids are oceanic and deep mesopelagic or bathypelagic as juveniles and adults. Most hauls successful for juveniles and adults have been at depths in excess of 500 m (with closing net captures as deep as 2,000-2,500 m). There is no evidence for diel vertical migration. G. chum is tropical, R. indica tropical- subtropical (sensu Johnson, 1982; 185). Giganturids are un- known from the Southern Ocean, Pacific Subarctic, temperate North Atlantic (including Mediterranean), and only a single specimen (G. chum') is known from the eastern tropical Pacific. Giganturids are relatively large-bodied with adults of Rosaura achieving more than 220 mm SL, adults of Gigantura more than 1 70 mm SL. Giganturids are well-known swallowers with greatly expandable pouchlike stomachs. Most identifiable gut contents have been fishes, often single large fish ingested whole (e.g., Regan, 1925). Transformed giganturids are distinguished from most or all other teleosts by the following combination of characters; (A) eyes tubular, directed straight forward, in parallel with main axis of body; (B) gape of mouth extends far behind eye; teeth fang-like, unbarbed, recurved, depressible; teeth bi- serial on each jaw, a medial row of enlarged canines and a lateral, more irregular row of smaller canines; anteriormost canine in each jaw recurving anteriad; (C) bases of pectoral fins nearly horizontal, above the gill openings; pectoral fins with a very high fin-ray count, 37 to 43 in Rosaura. 30-33 in Gigantura; (D) caudal forked, middle rays of lower lobe lengthened enor- mously; in one 120.3 mm SL specimen of G. chuni the fila- mentous extension of the lower caudal lobe adds 243 mm to the length of the fish; (E) skin loose, scaleless, with a thick layer of mesenchymal jelly adding substantially to an overall char- acteristic flabbiness; (F) stomach a thickwalled blind pouch, giving rise to the intestine ventrally, near midline; intestine passing laterad and dorsad, to right, continuing along dorsal contour of stomach until finally turning ventrad behind poste- rior terminus of stomach and ending at anal papilla; (G) lack of pelvic fins, dorsal adipose fin, branchiostegal rays, gill rakers; loss of most of gill arch elements on arches I-III, but with strong, recurved teeth on 3rd pharyngobranchial (pb) and 4th pb tooth- plate; loss of numerous other skeletal elements (cf Regan, 1 925; Walters, 1961, 1964; Rosen, 1973); and (H) considerable con- solidation of caudal fin skeleton with two presumably com- pound hypurals (Rosen, 1973). Development Eggs of giganturids are unknown. Larvae are known for both species but only the larva of Rosaura (a single 8.4 mm specimen. Fig. 105) has been illustrated (Tucker, 1954). For both species larvae have commonly been taken in the upper 100 m. The distributional ranges of larvae and adults are coextensive and there is no evidence for seasonality in reproductive effort (with only ca 400 known larval specimens, the data are far from complete). The sexes are separate and according to Clarke and Wagner (1976) the females may reach twice the size of males, although available data are sparse. Osteological examination has been confined to adults except for those elements visible and described in Tucker's (1954) astonishingly detailed decrip- tion of the holotype of Rosaura rotunda. Development is direct but transformation is abrupt with the change from larval to adult morphology occurring over the approximate size range of 30-40 mm SL in Gigantura and 40-60 mm SL in Rosaura. Transformation series are now known for both species (only 8 transforming specimens of Gigantura are known, for Rosaura the count stands at 34) but these results remain unpublished. The interim account below is thus based on work in progress. Gross aspect (Fig. 105). — "Rosaura" larvae are short, deep, glo- bose, translucent and virtually colorless. The forehead is steep, the eyes small, round and directed laterad. The snout is pointed. The body is deepest at a vertical through the center of the opercle. The pectoral insertion is nearly vertical. A dorsal adi- pose and distinct partly-stalked 5-rayed pelvic fins are present. Large, readily visible, rather platelike branchiostegal rays are present. Raptorial jaw teeth are present in the smallest known larvae (4 mm SL). Teeth on the jaws are biserial with an inner series of prominent canines and an outer series of shorter more broadbased teeth on the premaxillaries and dentaries. There are 2-4 recurved smaller fangs on the basihyal. The maxillary is included in the gape but is edentulous. The abdominal body wall is nearly transparent and balloonlike, enclosing an expan- sive gut cavity. The body form remains essentially unchanged over a period of larval growth extending to ca 30 mm SL (Gi- gantura) and to ca 35 mm SL {Rosaura). when transformation begins. Changes during transformation are striking, as described below. At all stages— larvae, transforming specimens, and ju- veniles and adults— the species can be distinguished on the basis of relative depth of the caudal peduncle. The value of this char- acter varies ontogenetically but the relative peduncle depth is always greater in Gigantura. Meristic characters.— Courtis of fin rays do not differ between larvae and adults except that semi-stalked pelvic fins (5 rayed) are universally present in larvae and early transforming speci- mens but are completely lost during transformation. Values for anal-fin ray counts (8 to 10 in G. chuni. II to 14 in i?. indica) and pectoral-fin ray counts (30 to 33 in G. chuni. 36 to 42 in R. indica) separate the two species without overlap. Dorsal-fin ray counts ( 1 6 to 19) have the same range in both species. The caudal is the first fin to form; it is asymmetric with 10 -t- 6(7) principle caudal rays and (3)4(5) procurrent caudal rays above and below. Next to form, in order, are the dorsal + anal fins, pelvic fins, and pectoral fins (the dorsalmost pectoral rays begin 199 200 ONTOGENY AND SYSTEMATICS OF RSHES-AHLSTROM SYMPOSIUM ."//,/., Fig. 105. Giganturidae. (Upper) Larva oi Rosaura indica. 8.4 mm SL (=holotype oi Rosaura rotunda from Tucker, 1954). (Lower) Adult Rosaura indica. 182 mm SL (from Berry and Perkms, 1966). to differentiate in larvae as small as 5.5 mm SL. but the ventral- most pectoral rays are the last fin rays to be formed). The pelvic fins appear just below the dorsal-fin origin and do not greatly shift in relative position until transformation. A dorsal finfold connects the incipient dorsal fin with the caudal fin in small larvae, but loses this connection in larvae larger than 6 mm SL. and shrinks in extent but remains as a highly visible adipose fin until transformation, when it is resorbed. Peritoneal pigment sections. — A single peritoneal pigment sec- tion characterizes the larvae of both species. This section lies just above and posterior to the dorsal transverse limb of the intestine. The section is never paired as in synodontoids and remains proportionately constant in size throughout larval life and is represented in adults as a small, intensely-black oval pigment patch above the stomach (growth of the section ap- parently ceases at about the onset of transformation, but the section apparently remains in both juveniles and adults of both species). The dense brown or black pigment enclosing the gut is not derived from this peritoneal pigment section, as is true for many "inioms" (see Johnson, 1 982) but develops separately dunng transformation (as in Aleptsaurus and Omosudis, Was- sersug and Johnson. 1976). Other pigmentation. — \n both species pigmentation in larvae occurs in three areas (other than the peritoneal section), the eyes, over the optic lobes, and on the sides of the body posterior to the dorsal-fin base. In some but not all pre-transformation specimens of Gigantura, very small punctate melanophores ap- pear over the still otherwise essentially transparent lateral ab- dominal body wall. Gut morphology.— The stomach is enlarged and sac-like. The mtestine leaves the pyloric region of the stomach, descends round the left margin of the abdominal cavity, crosses trans- versely upon the ventral body wall, reascends the right side and then turns again, descending abruptly and obliquely down and posteriad to the vent. Transformation —Changes during transformation are numer- ous and striking: (A) Body shape. The body changes in shape from short, rotund and deep, rather as in some ceratioid larvae JOHNSON: GIGANTURIDAE 201 (Bertelsen, 1 95 1 ) or the larvae of certain scopelarchids (Johnson, 1974b, 1982) to the elongate, shallow, slender shape of the gi- ganturids. The head while still massive is proportionately much less so ('/« vs 'A SL in Rosarua) and the dorsal head profile is essentially horizontal rather than steeply oblique (Fig. 105). (B) Eyes. Eyes in larvae are round, small and directed laterad; eyes in adults are fully tubular and directed rostrad. (C) Fins. Dis- tinct, partly-stalked, 5-rayed pelvic fins are present in larvae, resorbed or shed during transformation, and lacking in adults. The line of insertion of the pectoral-fin rays is obliquely vertical in larvae, essentially horizontal in adults. In larvae the pectoral insertion is behind the gill slit, in adults (especially prominent in Gigantura) the pectoral insertion is substantially above the gill slit. A distinct dorsal adipose fin is present in larvae, absent in adults. Procurrent caudal fin rays number (3)4(5) in larvae and are prominent, in adults procurrent caudal rays are fre- quently embedded in the skin, difficult to see, and number (0)1(2,3). (D) Teeth. Among the most striking changes occurring dunng transformation is the total loss of all larval teeth (in- cluding basihyal teeth). Transforming specimens are character- ized by a scalloped, irregularly-emarginate jaw edge (upper and lower) which is edentulous. None of the 40 known transforming specimens shows development of adult teeth and the smallest known post-transformation specimen (36.4 mm SL, G. chuni; 47.9 mm SL, Rosaura indica) possess a full complement of adult teeth. (E) Color. Larvae are essentially translucent with very little development of pigment, adults are entirely blackish/ brown (often with the development of an iridescent finish in Gigantura). Onset of transformation is indicated by the "sud- den" widespread development of pigmentation. (F) Loss of skel- etal elements. Larvae possess at least the following skeletal ele- ments not seen in adults: symplectic, coracoid, cleithrum, posttemporal, supracleithrum, branchiostegal rays. Relationships The first association of "Rosaura" with the giganturids was by Ahlstrom and Berry about 1960 (letters and mss material made available by H. G. Moser) with the first published sug- gestion made in Berry and Perkins (1966). Key characters sug- gesting relationship included the very high pectoral-fin ray count and the highly unusual 10-1-6(7) distribution of principle caudal rays, apparently unique to "Rosaura" and the giganturids. The disparities between "Rosaura" larvae and adult giganturids— briefly outlined above— left doubt in many minds, but the cap- ture of essentially complete transformation series (to be de- scribed and illustrated in detail elsewhere) make it unquestion- able that "Rosaura" is the larval form of the giganturids. With a caudal peduncle depth of ca 9.9% of SL (Tucker, 1954:168) there is likewise no doubt that the type of Rosaura rotunda represents a larva of "Balhyleptus," requiring recognition of the more elongate, shallow-bodied species as Rosaura indica (Brauer, 1901). The deeper-bodied species is Gigantura chuni Brauer, 1901 (other species have been described but the characters used to distinguish them do not work, nor has other evidence been found to support the hypothesis of more than two species). Of the two, Walters (1961, 1 964) argued for the more apomorphous condition of Gigantura but his characters need to be re-exam- ined in light of outgroup comparisons and in conjunction with other characters. Vanous authors have allied giganturids with such disparate groups as Stylephoridae, Saccopharyngiformes and "... a line [leading] from a subiniomous group such as the esocoids toward the synodontoid inioms, and this line later may have given rise to the Cetunculi . . ." (Walters, 1961). Rosen (1973:438-441) has offered evidence that the original placement by Regan (1925: 57) of giganturids with synodontoids was correct. Rosen calls particular attention to similarities in upper jaw and infraorbital configuration with synodontoids and the presence of a retractor dorsalis (=RAB in Rosen, 1973; see Winterbottom, 1 974b) mus- cle configration state characteristic of the synodontoid/alepi- sauroid line (Johnson, 1982:85, 95). An important character (Johnson, 1982:71; Okiyama, this volume) uniting synodon- toids with alepisauroids is the presence in larvae of multiple (3 or more) peritoneal pigment sections. Uniting synodontids and harpadontids (sensu Sulak, 1977) is the fact that in larvae of these fishes the sections are paired . . . and not connected over the gut. The condition in "Rosaura" is that seen in aulopids, chlorophthalmids, primitive scopelarchids, and ipnopids, viz. a single section situated over the gut. This is the state thought primitive for inioms. Also distinguishing the giganturids is a unique conformation of the gut. In larvae the gut arises from the pylorus, descends round the left margin of the abdominal cavity, crosses transversely midventrally, reascends the right side, turns abruptly mediad, then turns again, descending abruptly and obliquely to the vent. In adults the intestine arises mid- ventrally, makes a few small twists, ascends the right side, and passes posteriad above the dorsal contour of the expanded stom- ach, only descending to the vent posterior to the terminus of the stomach. In all the inioms I have examined the intestine arises midventrally and passes essentially straight back to the vent along the midventral wall of the abdominal cavity. For the time being, the available evidence suggests that the giganturids are neoteleosts (retractor dorsalis muscle), allied with the inioms (discrete peritoneal pigment section), diverging early from the rest and acquiring characters making them among the most specialized and distinctive of teleosts. Field Museum of Natural History, Roosevelt Road at Lake Shore Drive, Chicago, Illinois 60605. Basal Euteleosts: Relationships W. L. Fink AS mentioned in the introduction to this section of the sym- posium, the order Salmoniformes has had a history of attrition, such that today I would recognize it as coextensive with the Salmonidae. Previously included taxa are now scat- tered, primarily as unresolved lineages at or near the base of the Euteleostei. What follows is a preliminary analysis, a sketch of alternative hypotheses of interrelationships of the basal eu- teleosts. Fully resolving these problems will take more time and more material than I have had available to me, and I hope that work stimulated by this symposium will provide insights which have not been forthcoming using traditional material and char- acters. Unfortunately, very little information of a comparative nature is available on the larvae of basal euteleosts, and when these larvae have been discussed, only rarely have characters or char- acter transformations useful at large clade levels been men- tioned. Since adult specimens are more easily available in most collections, that is what I have relied on, with examination of larvae when possible. Results The Euteleostei is a large group of modem teleosts which is poorly diagnosed in terms of unique traits, and most more phy- logenetically advanced members lack some of the diagnostic characters. Patterson and Rosen (1977) considered the following as euteleostean traits: 1) an adipose fin, 2) nuptial tubercles, and 3) an anterior membraneous component to the first uroneural. Near the "base" of the Euteleostei, Fink and Weitzman(1982) recognized several lineages, including the Esocoidei, Ostario- physi, Argentinoidei, Osmeroidei, Salmonidae, and Neoteleos- tei. All were considered monophyletic, but the interrelations of these large clades were left unresolved (Fig. 106). Below is a review of each of the groups, with new information included when possible. Esocoidei or Esocae.—The^ fishes have been a continuing problem for ichthyologists. They are considered as euteleosts on the basis of an anterior membraneous component to the first uroneural, although it is not extensive. No esocoids can have an adipose fin as the dorsal fin is posteriorly situated. Neither do they have breeding tubercles. Rosen (1974) provided diag- nostic characters documenting monophyly of the group. Fink and Weitzman (1982) suggested that esocoids could be the sister group of all other euteleosts based on the lack in the latter of a toothplate on the 4th basibranchial, a bone which is present in esocoids and other primitive teleosts (see those authors for a discussion of the distribution of this character). Wilson and Veilleux ( 1 982) have recently reviewed interrelationships in the Umbridae, and they place Umbra and Dallia as sister taxa, with Novumbra as their sister group; all these together are placed as the sister group of Esox. This corroborates the hypothesis of Nelson (1972). Rosen (1974) considered Lepidogalaxias to be a member of this assemblage, which he termed the Esocae. Fink and Weitz- man (1982) questioned that hypothesis, leaving the genus un- placed. I have further comments and a new hypothesis of its relationships below. I have nothing to add to what Fink and Weitzman (1982) did with esocoids sensu stricto. and until more is forthcoming, consider them the likely sister group to other euteleosts. Ostariophysi— In terms of numbers of species and morpholog- ical diversity, this is the dominant basal euteleostean group. Fink and Weitzman (1982) did not consider the relations of these fishes to other euteleosts, primarily because their survey was intended to establish the placement of stomiiforms, and there was no evidence suggesting relationship between the two groups. No phylogenetic examination of ostariophysan rela- tionships to other teleosts has been done since Rosen and Green- wood (1970) expanded traditional concepts of the group by adding the previously protacanthopterygian gonorynchifonns. Fink and Fink ( 1 98 1 ) examined relationships within the group, placing siluroids and gymnotoids as sister taxa (order Siluri- formes), these the sister taxon of characiforms, and these to- gether the sister group of cypriniforms (the Otophysi, inclusive); sister group relationship of the gonorynchiforms to the Otophysi was corroborated. This entire assemblage was considered mono- phyletic on the basis of numerous characters, including lack of a dermopalatine, unique gasbladder morphology, specializa- tions of the vertebrae, and adductor mandibulae anatomy. Argentinoidei.— Cvetn'wood and Rosen (1971) combined the alepocephaloid and argentinoid fishes into an expanded Argen- tinoidei, in the Salmoniformes. Fink and Weitzman (1982) agreed with the combination of the two groups and used the formal subordinal name to include both subgroups. However, Fink and Weitzman (1982) were unable to provide evidence bearing on relationships of these fishes, even though their cladogram (Fig. 23, Fig. 106 herein) showed them as the sister group of the osmeroids. I have similarly been unable to place them, in part because of lack of adequate material. Osmeroidei— Thii group, which includes the northern and southern smelts, galaxiids (here including Lovettia and Aplo- chiton), Plecoglossus, and salangids, can be diagnosed as mono- phyletic based on several characters, including presence of one or more rows of teeth near the medial border of the mesopter- ygoid, loss or appearance late in ontogeny of the articular bone, and presence of a foramen in the posterior plate of the pelvic bone. Some subgroups of osmeroids have lost various of these diagnostic characters, but the patterns of loss allow other fea- tures to provide evidence of relationship in the group. Nevertheless, relationships within the suborder remain prob- lematical. The following review is based upon examination of specimens, the literature, and the contributions to this sym- posium. Incidentally. I have not attempted to diagnose the var- ious genera, but McDowall's comments (this volume and 1 969) indicate that such needs to be done. The phylogenetic hypoth- 202 FINK: BASAL EUTELEOSTS 203 ESOCAE OSTARIOPHYSI ARGENTINOIDEI OSMEROIDEI NEOTELEOSTEI SALMONIDAE STOMIIFORMES AULOPIFORMES MYCTOPHIFORMES ACANTHOMORPHA Fig. 106. The hypothesis of relationships suggested by Fink and Weitzman ( 1 982) for the basal euteleosts. eses and data are included in Fig. 107 and its caption. The data used in this analysis were chosen partly because they have been used traditionally in osmeroid systematics but I have little con- fidence in some of them; as a result this analysis represents a preliminary sketch of a more detailed study. The most striking thing about osmeroid systematics is that we still have questions about some very basic things, such as the status of the Osmeridae. As noted by Nelson (1970) and Rosen (1974), no evidence has ever been presented that the family is a monophyletic group. Indeed, it seems quite possible that Plecoglossus could be more closely related to some "os- merids" than to others, and this would render the family para- phyletic. A minimal requirement of any future work on system- atics of the group should be documentation of whether it is natural. Fig. 107. Alternate cladograms of relationships within the Osme- roidei. The bottom figure represents the hypothesis supported when all characters are given equal weight and paedomorphic traits are consid- ered homologous. The top figure represents the hypothesis which con- siders the paedomorphic reductive traits of salangids and galaxiids as non-homologous. For discussion, see text. The supporting characters are listed below, with the derived condition indicated by a 1 , the primitive by a 0. Each character number is indicated on the cladogram where it is in the derived state. Dark squares indicate unique appearance of a trait; empty squares indicate multiple appear- 18-20 6ALAXIIDAE lovettia Aplochiton Retropinna Stokellia Prototroctes SALANGIDAE Plecoglossus "OSMERIDAE" 2>29 -J-.GALAXIIDAE lovettia 'Aplochiton ■SALANGIDAE -Retropinna •Stokellia Prototroctes Plecoglossus 'OSMERIDAE" ance of a trait; triangles indicate a trait that is reversed at a lower level of generality; and circles indicate those characters in the reversed state. 1 . Posterior shaft of vomer (0) long ( 1 ) shori. 2. Articular bone (0) present and fused with angular (1) absent or greatly reduced. 3. Meso- pterygoid teeth (0) over much of bone ventral surface (1) restricted to medial border of ventral surface or lacking. 4. Pelvic foramen (0) absent ( 1 ) present. 5. Anchor membrane of egg (0) absent ( 1 ) present. 6. Caudal skeleton fusion patterns (0) none or rudimentary neural arches fusing with centrum and then, if at all, to the uroneural ( 1 ) rudimentary neural arch fusing with uroneural first, then these to the centrum. 7. Infraorbital sensory canals (0) curved posterodorsally ( 1 ) curved posteroventrally. 8. Mesocoracoid (0) present ( 1 ) absent. 9. Dorsal fin position (0) forward (1) posterior. 10. Principal caudal fin rays (0) 10/9(1) 9/9 or fewer. 1 1. Palatine teeth (0) present (1) absent. 12. Ectopterygoid bone (0) present ( 1 ) absent. 1 3. Extrascapular (0) present ( I ) absent. 14. Coracoid-cleith- rum process (0) present (1) absent. 15. Posterior pubic symphysis (0) present ( 1 ) absent. 1 6. Scales (0) present ( 1 ) absent. 1 7. Vomerine teeth (0) present (1) absent. 18. Posterior border of bones of suspensorium (0) smooth (1) deeply incised or emarginate. 19. Principal caudal fin rays (0) 9/9(1) 8/8. 20. Hypural number (0) 6 (1) 5. 21. Infraorbital sensory canals (0) not extending to preopercle (1) extending to pre- opercle. 22. Ceratohyal ventral border (0) more or less straight, bran- chiostegals along most of its length ( 1 ) deeply concave anterioriy, bran- chiostegals restricted to area posterior to concavity. 23. Homy abdominal keel (0) not present ( 1 ) present. 24. Ovaries (0) both present ( 1 ) left only. 25. Ectopterygoid bone (0) posterior to autopalatine (1) ventral to au- topalatine (coded as present in Stokellia based on McDowall, 1969). 26. Cucumber odor (0) absent (1) present. 27. Basioccipital lateral pegs (0) none (1) present. 28. Lateral hyomandibular spur (0) not present (1) present. 29. Caudal fin posterior border (0) deeply forked (1) rounded or emarginate. 30. Adipose fin (0) present ( 1 ) absent. 3 1 . Mesopterygoid teeth (see also Character 3) (0) restricted to ventromedial area of bone ( 1 ) absent. 204 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Salangids have been associated in the past with various mem- bers of the osmeroid assemblage, but even this was questioned by Nelson (1970). Rosen (1974) presented evidence from the caudal skeleton which shows that salangids are osmeroids, but no evidence about their placement within the group has been presented to date. Fink and Weitzman (1982) agreed with Rosen and placed the Salangidae as incertae sedis in the Osmeroidei. What little evidence I have been able to find about the rela- tionships of salangids is equivocal. If examined by a standard parsimony procedure, as represented by the Wagner analysis shown in Fig. 107 (bottom), the numerous reductive traits of salangids place them within the "southern smelt" plus galaxiid assemblage. On the other hand, salangids share with Plecoglos- siis and the "osmerids" a complex caudal skeleton character involving fusion of uroneural 1 to a compound centrum made upofPUl, Ul,and U2, followed ontogenetically in some forms by fusion of rudimentary neural arches with the uroneural por- tion of the complex. This latter character is in contrast to the autogenous uroneurals of most galaxiids, the "southern smelts," and other primitive teleosts. Further, when uroneurals and ru- dimentary neural arches are fused in galaxiids, the fusion se- quence is rudimentary neural arch to the compound centrum, followed by fusion with the uroneural, rather than the reverse. The hypothesis that emerges from these observations is illus- trated in Fig. 107 (top), showing salangids, Plecoglossus. and "osmerids" in an unresolved trichotomy. For further discussion of caudal fin morphology, see Greenwood and Rosen (1971), Rosen (1974), and Fink and Weitzman (1982). Any choice of these alternate hypotheses of salangid relation- ships would rest on whether or not one wished to accept the numerous reductive traits that unite the salangids with the "southern smelts" and galaxiids as homologues. Such choice is based on criteria which cannot be discussed in detail at this point due to space restrictions, but I have commented elsewhere (Fink, 1982) on hypothesis choice forced by confrontation with apparent paedomorphosis. In this case, for example, some of the general morphological attributes that salangids share with the members of those groups differ when examined in detail. Although this lack of close correspondence in similarity is cer- tainly no guarantee that the reductions are not homologous, it does raise the issue. Further, the highly developed caudal skel- eton of salangids is identical to that of "osmerids," and thus more differentiated than that of either the southern smelts or galaxiids. This incongruity in degree of morphological differ- entiation suggests that in this case, one should be cautious in assuming homology in the reductive process and search for other, non-reductive characters to resolve possible misplace- ments. The family Sundasalangidae is not accepted herein because in every case in which Roberts (1981) contrasted sundasalangids and salangids, the character for salangids was primitive. I suggest that recognition of family rank for Sundasalanx^^oviXd probably render the Salangidae paraphyletic and thus defined only by the absence of characters present in Sundasalanx. This is unac- ceptable both because it forces recognition of a group based on characters its members lack and because it artificially breaks up a group all of whose members share a unique evolutionary his- tory. Regarding the "southern smelt assemblage" (including gal- axiids, but excluding salangids), I am less pessimistic than McDowall (this volume). I have taken the liberty of using the data he has presented and combined them with my own limited survey of specimens and the literature to produce the hypotheses shown in Fig. 107. The group can be diagnosed by presence of a posteroventral deflection of the infraorbital sensory canal (Nel- son, 1972) and 9/9 or fewer principal caudal-fin rays (vs a pos- terodorsal curvature of the canal and 10/9 rays in outgroups). Several characters support the placement of Retropinna and Prototrocles as sister taxa including presence of an abdominal homy keel, loss of the right ovary, and ceratohyal morphology. I have no specimens o( Stokellia on hand, but McDowall's work ( 1979) clearly shows that the genus is diagnosable and that it is related to Retropinna and Prototrocles. Unfortunately, when contrasted with Stokellia, it is not clear that Retropinna is di- agnosable, since the latter is then differentiated by primitive characters present in other taxa. Relationship among Aplochiton, Lovettia and the galaxiids is supported by numerous characters, as shown in Fig. 107. I have been unable to find any features that link the former two genera together, however, and more work needs to be done with them. Galaxiids themselves can be shown to be monophyletic based on such characters as basioccipital "pegs" extending lateral to the anterior centrum (McDowall, 1969, Figs. 2B, lOA, but note lack of "pegs" in G. paucispondylus. Fig. 1 OB). In summary, it is suggested that the broad outlines of rela- tionships among the osmeroids are beginning to emerge, much as suggested by Gosline (1960a), with a "southern smelt" as- semblage and an "osmerid" assemblage. Interrelationships within these groups remain problematical, the most obvious problems being establishment of the natural groups within the "osmerids" and placement of the salangids. Salmontds. — M.onox>\\y\y of this group is based primarily on a single character, apparent polyploidy of the karyotype (Gold, 1979). Several investigators have studied interrelationships of salmonids, most notably Behnke (1968) and Norden (1961), but these works were not phylogenetic and changes can be expected. I have examined phylogeny within the group only to establish polarities for characters relevant to relationships with other te- leosts. Regarding the latter relationships, there have been several opinions, with most workers approaching salmonids with an eye to finding ancestors ofother groups (see, e.g., Gosline, 1960, Diagram 2). The only phylogenetic analysis to date is that of Rosen (1974), which was discussed by Fink and Weitzman ( 1 982). The latter authors presented data which they considered suggestive of neoteleostean relationship for salmonids: presence in some members of paired cartilages anterior to the ethmoid region (resembling the median rostral cartilage of neoteleosts) and the exoccipital forming part of the occipital condyle. The anterior cartilages were reported by Fink and Weitzman (1982) to be prominent in Prosoplum, an observation which I can confirm from additional specimens. In addition, examination ofsmall juvenile cichlids shows that the rostral cartilage appears to develop ontogenetically from bilateral cartilage bodies which fuse at the midline; this is suggestive of corroboration of Fink and Weitzman's (1982) hypothesis that the rostral cartilage evolved from paired cartilages anterior to the ethmoid region like those in Prosoplum. More work needs to be done on the homology of "accessory" ethmoid cartilages, using double stain- ing techniques and histology on a wide variety of teleosts. 1 can also add to what Fink and Weitzman ( 1 982) noted about the occipital condyle. 1 have confirmed that the exoccipital forms part of the condyle in Thymallus and "salmonins." This mor- phology is also present in Prosoplum. but is lacking in other FINK: BASAL EUTELEOSTS 205 coregonins. In a number of features, including the morphology of the nares, Prosopium stands as the sister group of other cor- egonins, and this, plus the presence in the outgroup Salmoninae and Thymallus of exoccipital participation in the condyle, im- plies that phylogenetically derived coregonins have secondarily lost that morphology. As noted by Fink and Weitzman (1982), the condyle structure as found in salmonids is found also in neoteleosts. It is also present in Lepidogalaxias (see below) and in some osteoglossomorphs. I do not wish to belabor the possible importance of this character, especially since more careful on- togenetic and morphological studies need to be done and other characteristics evaluated. A few observations from my survey of salmonids may be added here. I have found but two characters in the literature which diagnose the coregonins; one of these needs modification and the other needs to be more concisely put. Lack of maxillary teeth has been used to diagnose the group, relative to other salmonids (Norden, 1 96 1 ), but this needs to be emended to lack of the teeth in adults, since I have found maxillary teeth in Prosopium of around 19 mm SL. I have not yet examined spec- imens this small of other coregonins so do not know the gen- erality of this primitive state. The other character is reduction in the teeth in general; this needs to be quantified relative to the outgroups. The salmonins and Thymallus can be placed together based on lack of ossification of the supraethmoid (hypethmoid of Nor- den, 1961; Behnke, 1968), and apparently on yolk character- istics, and larval size (Kendall and Behnke, this volume). Re- garding other relationships within salmonids, I have nothing to add. Lepidogalaxias.— The position of Lepidogalaxias is controver- sial. I remain unconvinced by Rosen's (1974) hypothesis that the genus belongs with the esocoids. When I previously dis- cussed this genus (Fink and Weitzman, 1982), I had not seen any specimens, but R. M. McDowall has generously made sev- eral available for dissection and clearing and staining. There is no question that this little fish is a potpourii of contradictory and reductive characters and it is no wonder that it has been so difficult to place. Pursuing the potential of relationship of this species to galaxiids, extensive comparisons with members of that group have been made. Lepidogalaxias shares a host of reductive characters with galaxiids. NVhile these may indeed be synapomorphous traits, in cases where extensive paedomor- phosis is suspected, and this appears to be so in the morpho- logical similarities involved, one hopes to find some innovative, non-reductive characters which supply evidence for grouping. I have found two such characters which suggest that Lepido- galaxias is related to neither esocoids nor osmeroids, but rather may be the sister group of the Neoteleostei, as diagnosed by Rosen (1973) and Fink and Weitzman ( 1 982). This is supported by the presence in Lepidogalaxias of two non-reductive traits, a retractor dorsalis muscle and occipital condyle composed of both the basioccipital and exoccipital bones. As discussed just above and by Fink and Weitzman ( 1 982), the latter trait is also shared with salmonids. Lepidogalaxias lacks a rostral cartilage or its homologue and type 4 teeth (hinged teeth with a posterior axis of rotation. Fink, 1981) and this would prevent its place- ment within the neoteleostean assemblage. Placing Lepidoga- laxias as the neoteleostean sister group and leaving salmonids as their sister taxon presumes either that rostral cartilage homo- logues in the salmonids have been lost in Lepidogalaxias or are EUTELEOSTEI NEOTELEOSTEI EURYPTERYGII OSTEOGLOSSOMORPHA ELOPOMORPHA CLUPEOMORPHA ESOCOIDEI OSTARIOPHYSI ARGENTINOIDEI OSMEROIDEI SALMONIDAE LEPIDOGALAXIAS STOMIIFORMES AULOPIFORMES MYCTOPHIFORMES ACANTHOMORPHA Fig. 108. Summary cladogram of relationships and characters dis- cussed in the text. not homologues after all. This ambiguity is reflected in Fig. 108 by a trichotomy. Clearly, more work remains to be done before we can be really confident in the phylogenetic placement of this intriguing fish. Lepidogalaxias can be diagnosed by a number of characters, the most striking of which is fusion of the frontal bones into a single ossification (Rosen, 1974, Fig. 40B). In their comments on this species. Fink and Weitzman ( 1 982) noted that there was a disagreement about whether there are mesopterygoid teeth present; Rosen's statement that teeth are lacking is cortect. Stomiiformes. — Vmk and Weitzman (1982) recently examined the monophyly and relationships of stomiiforms to the other basal euteleosts and corroborated Rosen's (1973) hypothesis that they are the sister group to the rest of the Neoteleostei, removing them from the "salmoniforms." This placement is supported by several apomorphic traits, including presence of retractor dorsalis muscles and type 4 tooth attachment, as well as exoccipital participation in the cranial condyle and a rostral cartilage. Weitzman (1974) presented a hypothesis of relation- ships at the "family" level within the stomiiforms, as well as a detailed phylogeny of the Stemoptychidae. In this volume, I present a generic-level phylogeny for the barbeled stomiiforms (Family Stomiidae) and some brief comments on the "gonosto- matid-photichthyid" genera. Weitzman is currently working on relationships of the latter fishes and has made considerable com- ments in this volume (see Ahlstrom, Richards and Weitzman, this volume). 206 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Eurypterygii. — FinaWy, a few comments are due on the Myc- tophoidei of Greenwood et al. (1966). This group was disman- tled by Rosen (1973), and divided into two large groups, Au- lopiformes and Myctophiformes. These two groups, together with the Paracanlhopterygii and Acanthopterygii, were classified into a new group, Eurypterygii. Aulopiformes was placed as the sister group of all other eurypterygians, and myctophiforms as the sister group to paracanthoptergyians and acanthopterygians. All of these, together with stomiiforms, form the Neoteleostei. Fink and Weitzman (1982) tentatively accepted monophyly of the Eurypterygii based on the presence in its members of a toothplate fused with the third epibranchial. Aulopiformes con- tains a large number of families, including the Giganturidae, covered in this portion of the symposium. About the latter family I have little to say except that my own dissections cor- roborate Rosen's placement of it. Summary A summary of the hypotheses I have discussed above is given in Fig. 108. The most striking aspect of it is the degree of un- certainty about relationships among the clades. This may be in part due to the limitations of my study, but it does seem to me to be a fair summary of the status of well corroborated hypoth- eses we now have about this level of teleostean phylogeny. There are certainly other arrangements that can be made, depending on which characters one wishes to stress, and none of these should be discarded out of hand. As examples, I will cite two characters and their implications. First, lack of the posterior shaft of the vomer suggests that salmonids and osmeroids are sister taxa. Appropriate outgroups have the shaft ranging from "moderate" (e.g., Chanos) to "elon- gate" (argentinoids). My own opinion, based on occipital con- dyle structure of salmonids, is that the reduction in vomer length has occurred independently in the two lineages (it has also been reversed within both); the ultimate value of the occipital char- acter remains to be seen. The second character, presence of breeding tubercles, is now considered a euteleostean trait. Note, however, that tubercles are lacking in esocoids and argentinoids but are present in os- tariophysans, osmeroids, and salmonids, indicating that these three clades form a monophyletic group. Again, there are char- acters that contradict this grouping, but it nevertheless is worthy of consideration. It is always frustrating when one sets out to solve a particular problem and then comes to the end of the allotted time without a resolution. Although I have been able to shed some light on several problems relevant to the goals of this part of the sym- posium, I have not been able to unravel the interrelationships among the major basal euteleostean clades. Clearly more work is needed, especially with character suites which have been tra- ditionally neglected. Almost all of our concepts of relationships at this level are based on features of the adult caudal skeleton and branchial basket. Some work on soft anatomy, particularly the muscles of the head, has been informative at these levels and one hopes that other parts of the soft anatomy will be equally profitable. One area virtually untouched is larval anatomy. It might be expected that not many important features will be found because of the preponderance of primitive characters in larvae. But larval characters have proven useful, as is shown by the ontogenetic transformation in tooth types in stomiiforms (from type 4 to type 3; see Fink, 1981) as well as the specialized fin traits discussed by Ahlstrom et al. (this volume) for argen- tinoids. It is in both these areas, ontogenetic character trans- formations and presence of specializations for larval life, that study of larval fishes promises rewards. The inclusion of larval morphology in studies of higher level relationships should pro- vide a richer data base than we currently have and perhaps will reveal some crucial characters for resolving the basic questions I have addressed above. This symposium has already stimulated in a major way the examination of larvae for phylogenetic anal- yses, and I predict that it, combined with the new ways now emerging of analyzing ontogenetic information, will mark a new phase in the modem study of fish classification. Museum of Zoology, University of Michigan, Ann Arbor, Michigan 48109. Myctophiformes: Development M. Okiyama MYCTOPHIFORMES is currently adopted as a distinct order with intermediate affinity between the lower and higher teleost groups, whereas no one feature would satisfac- torily separate all of them from all Salmoniformes (Gosline et al., 1966). Except Rosen (1973), recent workers agree well with the familial composition of this order despite slight differences in the familial or subordinal definition. Table 56 shows the recent classification given by Johnson (1982) based on the most comprehensive knowledge now avail- able. Important points of this scheme are the exclusion of Sco- pelarchidae from Alepisauroidei and Pseudotrichonotidae from Myctophiformes. Further details in this connection will be men- tioned again in my paper on relationships (this volume). Exploitation of the vast hydrosphere covering the pelagic as well as benthic habitat between the surface and abyssal or ul- traabyssal plain by diversified members of this group is doubt- lessly the important aspect in discussing the ontogenetic prob- lems of the myctophiform lineage. Of the five suborders, Myctophoidei and Alepisauroidei are exclusively pelagic and the remaining are demersal including secondary pelagic genera such as Parasudis and Harpadon. Synchronous hermaphrodit- ism is common to the deep-water and offshore forms belonging to Chlorophthalmoidei and Alepisauroidei with the single ex- ception of Bathysauridae in Synodontoidei (Table 56). In general, the systematics of this order are rather well under- stood except for several families or genera. As is clearly shown OKIYAMA: MYCTOPHIFORMES 207 Table 56. Systematic Status and the Current Knowledge on Early Life Stages in Myctophiformes. Suborder and family No. Reproduc- _ tion'' Information species Eggs Larvae Main sources 7 + G _(.C +_|.d Okiyama (1974b) Aulopoidei Aulopidae Myctophoidei Myctophidae" Neoscopelidae Chlorophthalmoidei Chlorophthalmidae Ipnopidae Notosudidae Scopelarchidae" Synodontoidei Balhysauridae Harpadontidae Synodontidae Alepisauroidei Alcpisauridae Anotopteridae Evermannellidae" Omosudidae Paralepididae Aulopus Diaphus. etc. Neoscopelus Scopelengys Solivomer Chlorophthalmus Parasudis Bathysauropsis Ipnops Bathytyphlops Bathymicrops Bathypterois Ahliesaurus Scopelosaurus Luciosudis Scopelarchus, etc. Bathysaurus Harpadon Saurida Synodus Trachinocephalus Alepisaurus Anotopterus Evermannella. etc. Omosudis Paralepis Notolepis Mautichthys Lestidium Lestidiops Unasudis Lestrolepis Stemonosudis Macroparalepis Dolichosudis Sudis Ca. 300 3 2 1 18 + 2 3 3 2 2 18 ■> 13 1 17 2 4 15 Ca. 30 1 2 1 7 1 5 3 1 4 20 4 3 13 7 1 2 G G 7 7 H H 7 H H H H H H H H H G G G G H H H H H H H H H H H H H H H + (+) ( + ) ( + ) ( + ) ( + ) ( + ) ( + ) + + + + + + Moser and Ahlstrom (1970, 1974) + Okiyama (1974b) + Okiyama ( 1974b), Butler and Ahlstrom ( 1 976) + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + Tuning (1918) Okiyama (1981) Okiyama (1972), Parin and Belyamna (1972) Okiyama (this study) Sanzo (1938b). Okiyama ( 1 974b) Bertelsenet al. (1976), Ozawa (1978) Bertelsen et al. (1976), Ozawa (1978) Bertelsen et al. (1976) Johnson (1974b, 1982) Marshall (1961), Rosen (1971). Johnson (1974a) Okiyama (1979b) Mito (1961a), Okiyama (1974b). Ozawa (1983) Gibbs(1959), Okiyama (1974b), Ozawa (1983) Okiyama (1974b) Rofen (1966b) Okiyama (this study) Johnson (1982) Ege (1958). Rofen (1966b), Belyanina (1981) Ege (1930, 1957), Rofen (1966a) Rofen (1966a) Rofen (1966a) Rofen (1966a) Rofen (1966a) Rofen (1966a) Rofen (1966a) Rofen (1966a) Sanzo (1917). Rofen (1966a), Shores (1969), Belyamna (1981) For the details, see relevant section. ""G: gonochonsm; H: hermaphroditism, early developmental stages is available at least for a single species. Parentheses indicate information available for transparent ovanan eggs. '^ Double crosses mean that a series of in Table 56. information on the reproduction and development is abundant even for the deep-water species contrary to the situation of about 20 years ago (Gosline et al., 1966). General larval characteristics of this order were summarized by Ahl- strom and Moser ( 1 976). Selected meristic characters including many original data are given in Table 57. Aulopidae (Fig. I09A-B).— This bottom-fish family is generally considered the most primitive representative of the order. Its systematics are inadequately known; at least seven nominal and two undescribed species (Yamakawa, pers. comm.) occur in the warm waters of the world except for the Indian Ocean. Complete early life history series including egg stages are known only for Aulopus japonicus (Okiyama. 1974b, 1980). Fragmen- tary larval accounts are also available for some unidentifiable species. Suggested dichotomy in the larval morphology in this family (Okiyama, 1974b) is apparently wrong due to the erro- neous identification of the early stages oi " Aulopus filamento- sus" in Sanzo ( 1 938b) and TSning (1918), which are now ascribed to Bathypterois of the Ipnopidae. Eggs of .4. japonicus are spherical (1.18-1.14 mm in diame- ter), pelagic, transparent, without an oil globule, and with ir- regularly raised meshes on the chorion surface. Similar features are not present in the matured ovarian eggs of A. filamentosus ( 1.36-1 .44 mm in diameter) with numerous oil globules (Sanzo, 1938b). The known larvae differ in gut structure, size of the prominent pigment section and relative width of the slightly narrow eyes. However, the followmg features are shared in com- mon: single prominent peritoneal pigment section located at the middle or slightly anterior region of the body; gently curved 208 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Table 57. Selected Meristic Characters of Myctophiform Genera. Suborder and family Genus Dorsal Anal Pectoral Pelvic Branchiostegals Vertebrae Aulopoidei Aulopidae Aulopus 14-22 8-14 11-14 9 10-17 36-53 Myctophoidei Mytophidae* Diaphus, etc. 10-26 12-27 0-22 8 6-12 28-45 Neoscopelidae Neoscopelus 11-13 10-13 15-19 8-9 8-9 30-31 Scopelengys 11-13 12-14 12-17 7-8 8 29-35 Solivomer 12-14 9-11 14-16 8 9-11 40-41 Chlorophthalmoidei Chlorophthalmidae Chlorophthatmus 9-13 7-11 15-19 8-9 8 40-50 Parasudis 10 8-9 17 9 8 38-39 Balhysauropsis 10-12 10-11 17-24 9 8-9 44-56 Ipnopidae Ipnops 8-11 11-19 12-16 8 9-12 54-61 Balhytyphtops 11-13 13-17 12-15 8 14-17 62-66 Balhymicrops 8-10 9-15 9-10 7-8 8-10 65-80 Bathyplerois 12-16 7-13 13-22 8-9 10-14 49-65 Notosudidae Ahliesurus 9-11 17-21 10-12 9 10 42-50 Scopelosaurus 9-13 15-21 10-15 9-10 10 53-67 Luciosudis 10-13 17-20 12-14 9-10 10 57-59 Scopelarchidae* Scopelarchus. etc. 5-10 17-39 18-28 9 8 40-65 Synodontoidei Bathysauridae Bathysaurus 15-18 11-14 15-17 7-8 8-12 50-63 Harpadontidae Harpadon 10-15 11-15 11-13 9 16-26 39-56 Sauhda 10-13 9-13 11-16 9 13-16 43-67 Synodontidae Synodus 10-15 8-15 10-15 8 12-18 49-65 Trachinocephalus 11-13 14-16 11-13 8 14 54-58 Alepisauroidei Alepisauridae Alepisaurus 29-49 11-19 12-16 7-10 7 47-51 Anotopteridae Anotoplerns 14-16 12-15 9-11 8 78-83 Evermannellidae* Evermannelta. etc. 10-13 26-37 11-13 9 8 45-54 Omosudidae Omosudis 9-12 14-16 11-13 8 8 39-41 Paralepididae Paralepis 9-12 20-26 14-17 8 8 60-77 Notolepis 8-11 23-34 9-13 8-9 8 74-90 Maulichlhys 10-12 22-24 15-17 9 8 64-65 Lestidium 9-11 26-33 11-13 9 8 75-91 Lestidiops 8-13 25-35 10-13 6? 8 75-100 Uncisudis 10-11 25-31 11-13 9 8 75-79 Lestrolepis 9-11 31-44 10-12 8 8 82-98 Stemonosudis 7-12 29-50 10-13 8-9 8 84-121 Macroparalepis 11-14 21-32 10-12 9 8 80-110 Dolichosudis 10 36-37 11-12 9 8 101 Sudis 12-16 21-24 13-15 9 8 52-61 • For the details, see relevant section. head profile; short fins; anus far fiDrward with wide preanai interspace; anteriorly placed dorsal and pelvic fins. A size series of A. japonicus reveals the gradual and direct development, with scant pigmentation throughout the pelagic stages; melanophores are restricted to the eyes and the caudal and postanal regions, other than the peritoneal section which increases in size in older larvae. Sequence of fin formation is C-D-A-P,-?,. Full ray com- plements are visible at about 13.3 mm, but vertebral ossification is delayed until about 20 mm, the smallest bottom specimen available in my collection. Ontogeny of the upper jaw bones is remarkable in possessing maxillary teeth (1-3) in larvae smaller than 1 1 mm. Two supramaxillaries, peculiar to this family, are ossifying in metamorphosed juveniles. Myctophidae (see Moser, Ahlstrom. Paxton, this volume). Neoscopelidae (Fig. 709C-D^. — Systematics of this deep-sea pe- lagic and benthopelagic family are well understood (Butler and Ahlstrom, 1976; Nafpaktitis, 1977), except for 5o/;vc)Wfr which is restricted to the tropical Western Pacific. The remaining two genera are known from the world oceans. Developing eggs are unknown. Mature ovarian eggs of Neoscopelus macrolepidotus (0.83-0.98 mm in diameter) contain a large single oil globule of 0.39-0.61 mm (Maruyama, 1970). Advanced larval stages have been described and illustrated for Neoscopelus sp. (Oki- yama, 1974b) and two species of Scopelengys (Butler and Ahl- strom, 1976). They are characterized by large fan-shaped pec- toral fins, large head with blunt snout tip, small round eyes, laterally compressed deep body, and an oval patch of melano- phores in the peritoneum, distinct from the solid peritoneal pigment sections of most other myctophiforms. All fins differ- entiate rapidly with the possible sequence as P,-D-A-C-P,, full counts being attained at a small size (less than 10 mm). Pig- mentation is clearly difl^erent between the two genera. Scope- lengys lacks the pigment patch lying along the dorsum of the rectum in Neoscopelus. Scopelengys uniquely develops a hori- OKIYAMA: MYCTOPHIFORMES Table 58. Comparison of the Larval Characters Among Four Genera of the Ipnopidae. 209 Characters Ipnops Bathytyphlops Bathymicrops Bathypterois Head profile slung down; flat top slightly slung down; flat top slung down; flat top slung down; flat top Pectoral fin bilobed; rays long elongated; fan-shaped elongated elongated; fan-shaped Gut size short short long long Anus position; close to pelvic fin pelvic fin pelvic fin; slightly anal fin Anus-anal fin space wide wide wide narrow Peritoneal pigment section absent single absent *numerous (12-20) or absent Body pigment (melanophores) scant scant abundant scant Possible sequence of fin P, C-A-D-Pj P, C-A-D-P, P,-C-A-D-P, P,-C-A D-P, formation Transformation complete ca. 42 mm SL 43-93 mm SL 70-90 mm SL ca. 42-43 mm SL ' Details are mentioned in the text. zontal pigment bar across the head. Small preopercular spines are known only in Neoscopelus whereas a long snout is peculiar to Scopelengys. Chlorophthalmidae (Fig. ]09E-F).— Of Ihe three genera of this benthic family, the cosmopolitan Chlorophthalmus is particu- larly diverse and abundant. Extensive revision of this genus is needed, since there are many undescribed species from the West- em Pacific and the known species can be divided into two dis- tinct groups, each warranting generic status (Doi and Okamura, 1983). Eggs are not known. Despite the abundance of adults, few larvae have been reported. Complete developmental series are available for only C. agassizi (Tamng. 1918). Known larvae of other species such as C. mento, C. prondens and Chlorophthal- mus spp. (Pertseva-Ostroumova and Rass, 1973; Miller et al., 1979; Okiyama, unpubl.) show close resemblance to C. agassizi having the extremely short gut with large preanal interspace, a similar pigment pattern composed of a single peritoneal pigment section lying at the pectoral fin base and a melanophore at the hypural complex, short fins and anteriorly placed dorsal and pelvic fins (as in Aulopidae). There are possible specific differ- ences in the size at appearance of the peritoneal pigment section (ca. 7 mm in C. prondens vs 5-6.6 mm in C. mento) and in the arrangement of the few small melanophores on the dorsal and ventral margin of the tail near the notochord tip in early larvae. Meristic characters are useful in discriminating the particular species or species groups, although early developmental stages are usually very difficult to identify to species. Larval osteology was studied in detail for C. agassizi (Rosen, 1971) but the sequence of fin formation is not clear except that the pectoral fin develops early. Principal changes during the gradual metamorphosis include the rotation of the eyes dorsally which takes place at sizes less than 40 mm (Ahlstrom, 1972a). Unusual larvae with a pigmentation pattern similar to the above described forms are found in ORl collections from the Kuroshio area (Fig. 1 09 A). These are distinct in that the head is markedly depressed, bowed with duckbilled appearance, and a single peritoneal pigment section is large enough to cover the dorsal half of the short gut. Their meristic characters (ca. 42 myomeres and ca. 1 7 pectoral rays) suggest a possible affinity with Chlorophthalmus (sensu lato). These two larval types seem to substantiate the suggested dichotomy of this genus. No in- formation is available for larvae of the other two genera (Par- asiidis and Bathysauropsis). Ipnopidae (Fig. 1 1 OA-E). — Four benthic genera compose this family which has been variously classified (e.g.. Nielsen. 1966; Sulak. 1977). Despite their deep-sea mode of existence, larval stages of all genera have been mostly obtained from the surface waters. Developing eggs are not known. Mature ovarian eggs are known for all genera with virtually identical features such as a spherical shape, diameter of about 1.0-1.2 mm, and the presence of a single large oil globule (Nielsen, 1966; Sulak, 1977 and pers. comm.; Merrett, 1980). Although intergeneric differ- ences of the early larval stages are remarkable (Table 58). they share several conspicuous characters including the more or less hung-down head profile and the elongated precocious pectoral fins. At metamorphosis these become less prominent in asso- ciation with the drastic change in the mouth size from moderate to huge and the appearance of heavy body pigmentation. Two larvae (13.9, 10.6 mm) are known for Ipnops: the larger specimen referred to /. agassizi was described in considerable detail and illustrated (Okiyama, 1981). The smaller one may be /. meadi in view of its higher anal ray count (ca. 13). A divided pectoral fin with elongated upper rays is peculiar to this genus (Table 58). Principal changes at metamorphosis include the de- velopment of the unique eye plaque, a depressed head with straight profile, and the disappearance of the peculiar feature of the pectoral fins along with the loss of several rays. Metamor- phosis may be rapid, but the smallest benthic juvenile of 40 mm still bears the immature eye plaque (Sulak, 1977). Bathytyphlops includes only two species, B. sewelli and S. marionae (Merrett. 1980). A larva of this genus was first de- scribed under the name Macristiella perlucens of uncertain af- finity (Berry and Robins, 1967). The known "Macristiella" (19 specimens, 7-43 mm) are all referable to B. marionae except for the 37 mm larva from the Indian Ocean and the smallest specimen (Parin and Belyanina, 1972). The Indian Ocean spec- imen may be identified as B. sewelli on the basis of the higher anal ray count (18), a unique character for this species. Early stage larvae have little melanistic pigmentation, but some bluish or violet coloration is present on the fins and var- ious body parts in living specimens (Berry and Robins, 1967). Preserved individuals sometimes retain this feature, usually on the large pectoral or pelvic fins. Reduction of the relative size of eyes, and the loss or replacement of the teeth as well as gill rakers are among the major changes at metamorphosis, in ad- dition to those common to the family. Otherwise, larval de- velopment is rather direct and the relative position of the fins and the anus changes little throughout ontogeny. The osteology 210 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM OKIYAMA: MYCTOPH I FORMES 211 Fig. 109. (A) Aulopus japonicus. 1 1.5 mm SL, from Okiyama (1974b); (B) Aulopus sp., 12.3 mm, from Okiyama (1974b); (C) Neoscopelus sp., 7.9 mm, from southwestern Japan, Ocean Research Institute (ORI) collection; (D) Scopelengys dispar, 6.3 mm, from Okiyama (1974b); (E) Chlorophlhalmus sp., 17.1 mm, from Indian Ocean, ORI collection; (F) Chlorophthalnms (?) sp., 7.5 mm, from Kuroshio waters off Japan, ORI collection. of both larvae and adults is well known (Okiyama. 1972; Parin and Belyanina, 1972; Sulak, 1977). Bathymicrops represents the deepest living myctophiform. Two species. B. regis and B. brevianalis, are known from ex- tremely limited material from 4225-5900 m (Nielsen, 1966; Merrett and Marshall, 1981). Pelagic eggs are unknown. A total of five larvae and juveniles (13.0-70.0 mm) are available; the smallest two larvae (13.0, 14.7 mm) from Hawaiian waters are unidentifiable; a 20 mm larva from the North Atlantic (=Sto- miatella B in Roule and Angel, 1930: PI. 1, Fig. 7) is ascribed to B. regis; the largest two juveniles (62.5, 70.0 mm) from the tropical Pacific are tentatively identified as B. brevianalis. Despite conspicuous variation among specimens, scattered melanophore patches and an extremely slender body are diag- nostic for this genus. The precocious pectoral fins are greatly elongated even in the smallest larva, but the raised bases of the dorsal and anal fins and the prominent finfolds are peculiar to the advanced stages, which also have reduced eye size and a slightly shorter gut. Size at metamorphosis is unusually large, attaining 70-90 mm. Bathypterois is the most speciose genus in this family. Three subgenera (Benthosaurus, Bathypterois and Bathycygnus) and 18 species are currently included (Sulak, 1977). Known bathy- metric ranges are 250-5,990 m. Published information of the developmental stages is scant. Pelagic eggs are not known. A single larva of 14. 1 mm (Okiyama, 1974b) was identified as B. {Bathycygnus) longipes by Sulak (1977). As stated before, the known early stages of "Aulopus filamentosus" are all referable to those of Bathypterois. probably B. (Bathypterois) mediter- raneus in view of their localities. Complete series of early stages are confined to this species, but at least three additional larval forms are now available. These known larvae share the distinct forward shift of the ventral hypural elements in addition to the features given in Table 58. Known larvae are provisionally divided into two groups on the basis of the peritoneal pigment sections, those with many sections and those which lack peritoneal pigment. Except for two larvae, B. (B.) longipes and B. (Benthosaurus) viridensis (33.1 mm) from the Atlantic (Fahay, 1983), all specimens have the former character state. The number of peritoneal pigment sections can be a useful tool in discriminating the lai^ae, but ranges of variation often overlap among species. A western Pa- cific form with 12-18 pigment sections bears close resemblance to B. (B.) mediterraneus larvae whereas decidedly lower myo- mere counts of the former (45-48) readily separate these two. B. viridensis larvae have, in addition to the complete absence of the peritoneal pigment sections, several peculiar features such as a slightly telescopic eye, a protruding gut, and a long anal fin and short tail. Comparison with the smallest demersal specimen (43 mm) of the same species (Sulak, 1977) indicates that prin- cipal metamorphic changes include the absorption of the pro- duced gut, lengthening of the posterior body and fin shrinkage. This may represent the most pronounced metamorphosis in this genus, since less remarkable transformation predominated in the other species. Identification of the other larval types remains to be determined. Notosudidae (Fig. 1 J lA-B). — Bertelsen et al. ( 1 976) extensively revised this oceanic midwater family, including information on early developmental stages of all species (except Scopelosaurus cradockei'). Supplemental information on the early stages is available in Ozawa (1978). Pelagic eggs are unknown. Maturing ovarian eggs of Ahliesaurus (ca. 0.3 mm in diameter) and Lu- ciosudis (0.4-0.5 mm) suggest that pelagic eggs are uncommonly small for this order. General characteristics of these larvae are extremely similar throughout the family: long, slender subcylindrical body, be- coming increasingly compressed toward the tail; markedly de- pressed head with wedge-like snout; posteriorly protruding lobes in corpus cerebelli; narrow eye with longer horizontal axis; a more or less distinct conical mass of choroid tissue on the pos- terior part of slightly stalked eye; anus at about midbody (except .4hliesaurus) widely separated from anal fin origin; slight in- crease of gut length with growth during the early larval stages; absence of the peritoneal pigment. Maxillary teeth peculiar to larvae help diagnose this family but are not unique (see, Au- lopidae). Possible sequence of fin formation is CA-D-P.-Pj, 212 ONTOGENY AND SYSTEMATICS OF FISHES -AHLSTROM SYMPOSIUM last elements being rarely visible in larvae less than 20 mm. Apart from the length at metamorphosis varying between 25 and 45 mm among species, pigmentation pattern is usually the only useful character for specific identification. Once established these pigment patterns, mostly restricted to the tail, are retained throughout the larval stages, although a few species are known to be unpigmented throughout all or part of the larval period. Scopelarchidae (see R. K. Johnson, this volume). Bathysauhdae (Fig. 1 1 IC).— This deep-water benthic family consists of two species of synchronous hermaphrodites, Bathy- saurus mollis and B. ferox (=B. agassizi) (Sulak, pers. comm.; Wenner, 1978). Pelagic eggs are unknown. Maximum size of mature ovarian eggs in B. ferox is 1.2 mm in diameter (Wenner, 1978). So- called "Macristium" forms are now proved to be larval Bath- ysaunis (Rosen, 1971; Johnson, 1974a); at least several of the five known "Macristium" larvae (20-83 mm) are positively identified with B. mollis. Morphology and osteology of these specimens have been closely studied, revealing many charac- teristic features such as unusually elongated fins, anterior place- ment of dorsal and pelvic fins, raised bases of dorsal and anal fins, long gut (coiled or uncoiled) terminating just in front of anal fin origin, six peritoneal saddle-shaped pigment sections all evenly spaced, and development of a pattern of lateral bars in some specimens. Besides this last feature, meristic differences serve to distinguish two species despite considerable variation. Metamorphosis may take place gradually at exceptionally large sizes (more than 83 mm). Accompanying changes include short- ening of fins, expansion of the gape with necessary associated changes in head bones and associated anatomy, backward shift of the dorsal fin origin, and darkening of the body surface, oral cavity and peritoneum. Harpadontidae (Fig. 1 1 ID-E). — Two genera are recently in- cluded here (Sulak, 1977; Johnson, 1982). Harpadon comprises at least four species living in nearshore waters, estuarine and relatively deep continental shelf waters of the Indo-Pacific. Crit- ical systematic revision of this genus is now in progress (Schmitz, pers. comm.). A pelagic egg referred to H. nehereus in Delsman ( 1929c) appears invalid (Delsman and Hardenberg, 1934). Early developmental stages are poorly studied; only two specimens of//, nehereus (25.2, and ca. 40 mm) have been illustrated and/ or briefly described (Delsman and Hardenberg, 1934; Okiyama, 1979b). A juvenile of 55 mm is the smallest specimen of the deep water congener, //. microchir, available in ORI collections. Early stages are readily discriminated from most other myc- tophiform larvae by the exceptionally high numbers of bran- chiostegal rays (16-27) and the following characters: elongate compressed body with large head and mouth, short snout (due to the forward shift of eyes), scant pigmentation except seven pairs of peritoneal pigment sections, the last two closer together than the others, and extension of the lateral line scales onto the caudal fin. Of these rather advanced developmental features, pigmentation pattern may be common to the earlier stages. Apparently, long pectoral and pelvic fins are peculiar to //. nehereus. Also, //. microchir is more lightly pigmented than //. nehereus at similar lengths. Metamorphosis seems gradual. If the occurrence of melano- phores over the stomach is of significance in defining this pro- cess, transformation is completed by 35 mm in //. nehereus. There are about 1 5 species of Saurida with highest diversity in the Western Pacific. Planktonic eggs are known for S. elon- gata, S. wanieso. and S. tumbil besides several unidentifiable species (Mito, 1961a; Zvjagina, 1965a; Venkataramanujan and Ramanoorthi, 1981). These are spherical, 1.0-1.3 mm in di- ameter, transparent, without oil globules and with a narrow per- ivitelline space. Hexagonal sculpturing on the chorion (0.03- 0.05 mm in mesh size) is either present (S. wanieso and S. tumbil) or absent (S. elongata). Early developmental stages are known for 9 species. Of these, complete developmental series are available for at least 4 Pacific species, S. tumbil, S. elongata. S. wanieso and S. gracilis (Dileep, 1977; Ozawa, 1983) and the Atlantic species, 5. brasiliensis (K\x(i.omtX]f.ma.. 1980). These lar- vae are extremely similar to those of Harpadon. except for the lower numbers of branchiostegals and invariably short fins. Complete absence of the preanal finfold in the early stages is peculiar to this genus (Ozawa, 1983). Except for S. brasiliensis. however, these are divided into two types on the basis of pig- mentation pattern. One of these consisting of S. gracilis and probably some Atlantic congeners is characterized by evenly spaced peritoneal pigment sections of similar size and simul- taneous differentiation. In addition, prominent pigment along the anal fin base and on the caudal fins may be diagnostic for this type. 5. gracilis larvae uniquely develop a small choroid mass on the ventral side of narrow eyes (Ozawa, 1983) while nothing is mentioned in this regard for Hawaiian larvae (Miller et al., 1979). Remaining larvae belong to the second type in which the terminal pigment section is smaller and later-ap- pearing than the anterior sections. Other pigment is also scarse or absent in this latter type, where specific differences are known in the size of pigment sections and vertebral numbers. Meta- morphosis occurs fairly gradually with considerable variation in size among species, but is usually complete before 40 mm (Gibbs, 1959). Fig. 110. (A) Ipnops agassizi. 13.9 mm SL, from Okiyama (1981); (B) Balhytyphlops manonae. 13.1 mm, from Okiyama (1972); (C) Bathymicrops brevianalis. 70.0 mm, from tropical central Pacific, ORI collection; (D) Bathypterois sp. (pigmented type), from northeast of Australia, Southwest Fisheries Center (SWFC) collection; (E): Bathypterois viridensts (unpigmented type), from Fahay (1983). Fig. 111. (A) Scopelosaurus smilhii. 1 3.4 mm SL, from southwestern Pacific, ORI collection; (B) the same, dorsal view of head; (C) Bathysaurus ferox. 33.0 mm, from Marshall (1961); (D) Harpadon nehereus. 25.2 mm, from East China Sea, ORI collection; (E) Saunda undosquamis. 15.6 mm, from Okiyama (1974b); (F) Synodus lucioceps. 10.5 mm, from California current region. SWFC collection; (G) Trachinocephatus myops. 21.3 mm, from Zvjagina (1965a). Fig. 112. (A) Atepisaurus brevirostris. 12.1 mm, from Rofen(1966b);(B)/l./era>:. 10.0 mm, from central Pacific near Hawaii, SWFC collection; (C) Anotopterus pharao, 14.2 mm, from California current region, SWFC collection; (D) Omosudis lowei (central western Atlantic specimen), 11.8 mm, from Rofen (1966b); (E-F) O. lowei. 22.5 mm, from tropical western Pacific, ORI collection, showing dorsal view of head. OKIYAMA: MYCTOPHIFORMES 213 214 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM OKIYAMA: MYCTOPHIFORMES 215 216 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Synodontidae (Fig. 1 1 IF-G). — Synodus includes about 30 species and has a circumglobal distribution with distinctly high diversity in the Indo-Pacific. Another monotypic genus of this family (Trachinocephalus) shows world-wide distribution. A recent re- vision of the Indo-Pacific Synodus (Cressey, 1981), including many new species, critically changed its systematic status. Thus, most of the known eggs and larvae are subject to nomenclatural revision. Early stages of this family can be separated from those of the previous family by the presence of the preanal finfold (Ozawa, 1983). Trachinocephalus myops larvae are distinct in possessing six pairs of large peritoneal pigment sections of uniform size, a rounded head with short snout, and additional unique pigmen- tation (Rudometkina, 1980; Ozawa, 1983). This species and most species of Synodus have an extremely elongated body. An exception is the eastern Pacific species, S. lucioceps. which has a slightly deeper body. A complete developmental series is known only for this species in Synodus; eggs are spherical, 1.33-1.44 mm in diameter, without an oil globule, with moderately broad perivitelline space and hexagonally sculptured chorion: larvae are characterized by 7 evenly spaced pairs of pigment sections formed gradually, a ventral melanophore lying at the midpoint of tail, and one near the notochord tip. As in the Harpadontidae, meristic characters and pigmenta- tion patterns are of particular aid in identifying the early stages of this family. If established pigmentation patterns are retained in the metamorphosed juveniles or adults, numbers of the per- itoneal pigment sections of all Indo-Pacific species of Synodus (Cressey, 1981) vary between and 17 with a maximum range of infraspecific variation of 0-3 in 5. binotalus and 14-17 in S. usitatus; some species appear to lack this pigment (i.e., S. kaian- us and S. binotalus), however this needs to be documented by complete developmental series. Another point of interest is the asymmetry and size disparity of the pigment pairs known in "S. variegatus" of Okiyama (1974b). Size at metamorphosis and sequence of fin formation of this family appear to be identical to those in Harpadontidae. Ozawa (1983) revealed the following pattern of fin formation: C-A-D- P,P2. Alepisauridae (Fig. //2,4-BA— This widely distributed bathy- pelagic family includes only two species, .Hepisaurus fero.x and A. brevirostris. with slightly different ranges: the latter is appar- ently absent from the North Pacific (Francis, 1981). Eggs are unknown. A series of early developmental stages of Alepisaurus sp. (6.9-17.2 mm) has been described and illustrated (Rofen, 1966b). In addition, three larvae (9.6-16.5 mm) from the col- lection of the Southwest Fisheries Center, La Jolla have different features. They share with previous specimens a large head and mouth, prominent canine teeth on the dentary, small fins in- cluding pigmented pectorals of moderate size, gently curved head profile and short gut with heavy pigmentation. The peri- toneal pigment section is indistinct. This new material is unique in having 4 small preopercular spines, pigment patches at the anal fin origin, and distinct bony ridges dorsally on the head. Judging from the locality of these specimens, near Hawaii in the North Pacific, Alepisaurus sp. larvae of Rofen (1966b) can be identified with A. brevirostris. and these with A.ferox. Metamorphosis may be gradual with possible sequence of fin formation P, C-D-A-Pj. Anotopteridae (Fig. 112C).— One world-wide species, Anotop- terus pharao, constitutes this open ocean family, uniquely lack- ing the dorsal fin. Eggs are not known. A larva (ca. 1 5 mm) has been briefly described without illustration (Nybelin, 1948): this specimen is unavailable now (Thulin, pers. comm.). Another larva of similar size (14.2 mm) is available from the collection of the Southwest Fisheries Center, La Jolla. It is characterized by a slender thin body, absence of peritoneal pigment sections, large head with pointed snout, a fleshy prolongation at the tips of both jaws, two large canine teeth on each palatine, and a fairly long gut extending beyond midbody. Pigmentation is scat- tered on various parts of body including the snout, jaw tips, dorsal midline of body, near the tail tip, and peritoneum (par- ticularly along the dorsum of gut). Except for the pectoral fin, fin aniages are lacking. A juvenile of about 50 mm illustrated in Rofen ( 1 966c) is similar to the described larva, except all fins are differentiated including the adipose fin: body pigmentation is remarkable in this juvenile. Perhaps, this species has the most direct pattern of early development in this order. Evermannellidae (see R. K. Johnson, this volume). Omosudidae (Fig. ] 12D-F).—A single mesopelagic species, Omosudis lowei. constitutes this cosmopolitan family. Pelagic eggs are not known. Excellent developmental series have been described and illustrated, chiefly based on Atlantic material ranging from 5.7 to 75.2 mm (Ege, 1958: Rofen, 1966b). Re- cently, a larva (11.5 mm) with different features was briefly described and illustrated (Belyanina, 1982b). Its locality in the tropical western Pacific is peculiar and additional specimens are available in ORI collections (pers. obs.). These have in common a very large head and mouth, stubby body, long pointed snout, straight head profile, small fins, par- ticularly the pectoral, large canine teeth on denlary and palatine, and several closely spaced peritoneal pigment sections. How- ever, trenchant morphological differences between the Atlantic and Pacific specimens are known: head smooth vs armed (along edge of preopercle and dorsum of head): pigmentation light vs dense at a similar size: pigmented band above posterior part of anal fin absent vs present. For this first character, there is a possibility that the minute preopercular spines have been over- looked in the Atlantic larvae. Sequence of fin formation known in the Atlantic specimens is C-DA-Pj-P,. Metamorphosis is gradual with possible dif- ferences in the size of completion between the two types as suggested above. The presence of two larval types is in sharp contrast with the current concept of a monotypic family. In this connection, Ege's comments ( 1 958) on the significant differences in dorsal ray numbers between the populations from the South China Sea and north Atlantic are of particular interest. Paralepididae (Fig. 1 13. 4-G}.— This oceanic pelagic family in- cludes about 1 1 genera and 50 species and constitutes the second largest group in the order after Myctophidae. Some genera are still in need of critical revision, while the two established subfamilies seem valid. Paralepidiinae includes two tribes, the Paralepidiini (3 genera) and Lestidiini (7 genera), and Sudinae has I genus (Sudis). Ege (1930) and Rofen (1966a) mcluded early larval stages in their extensive studies of this family. Eggs are not known but developmental stages are known for 9 out of 1 1 genera. Larval development of Sudis has been closely studied for 5. hyalina and S. a/ro.v (Sanzo, 1917: Shores, 1969: Belyanina, 1981). These unusual larvae are readily discrimi- nated from those of the other subfamily by the relatively short body with large head, long pectoral fins, long gut and early OKIYAMA: MYCTOPHIFORMES 217 Fig. 113. (A) Paralepis elongata. 16.7 mm SL, from Rofen (1966a); (B) Notolepis coatsi. 60.5 mm, from Efremenko (1983); (C) Leslidiops ringens. 9.4 mm, from California current region. SWFC collection; (D) the same, 28.5 mm; (E) Stemonosudis macrura. 1 1.2 mm, from Ege(1957); (F) Sudis hyalina. 16.1 mm, from Shores (1969); (G) 5. alrox. 21.5 mm, from Berry and Perkins (1966). established complement of peritoneal pigment sections, spine- tipped flanges on ventral region of preoperculum. over eye, and snout. 5". alrox has a spine-tipped flange along lower jaw. and the large spine at the preopercular angle is serrated only in 5. airo.x. The precocious pectoral fin is relatively short until about 1 5 mm in S. atrox whereas it is very long even in 8.0 mm larvae of 5. hyalina. The number of peritoneal pigment sections is 6 (5 in early larvae) in S. atrox vs 7-8 in S. hyalina. Trunk pigment is evenly distributed in S. atrox \s patchy in S. hyalina. Except for this genus, the developmental features of this fam- ily are remarkably cohesive. Known lar\ae have a very long compressed body, a short trunk in early larvae, large head in advanced larvae, elongated pointed snout with straight head profile, various numbers of peritoneal pigment sections sequen- 218 ONTOGENY AND SYSTEMATICS OF HSHES-AHLSTROM SYMPOSIUM tially formed with gradual lengthening of gut, well developed preanal finfolds and apparently precocious anal fin rays. Ad- ditionally, during ontogeny eye shape changes from ovoid to round, and body pigmentation changes from light to dense. These larvae are too similar in general appearances to determine trenchant characters that define genera or tribes. Peritoneal pig- ment sections, are of prime importance in identifying early stages, but show extreme variability with respect to their number and sequential development. Of particular interest in this connection is Notolepis. N. rtssoi develops 1 2 pigment sections, the largest number in the family except Stemonosudis (3 1 ), whereas the Antarctic congener, TV. coalsi, has only a single section which increases in size with growth (Efremenko, 1978, 1983a). Among the various genera the primary section develops at 5-10 mm and full complements are formed variously by the species be- tween 15-45 mm. Usually, metamorphosis takes place around this size accompanied by the development of a black perito- neum. In addition to the exceptionally higher number of pigment sections, Stemonosudis is peculiar in having a filamentous pro- jection on the lower jaw tip (in larvae of 5. macrura and in juveniles and adults of S. intermedia and 5. elongatd). Likewise, Uncisudis (=Pontosudis) uniquely develops an elongated pelvic fin. Patterns of melanophores are extremely diverse but of use in identifying species or species groups; pigment patches on the caudal peduncle, dorsum of body, and caudal and pectoral fins are particularly important. Rofen (1966a) suggested that the single larval character discriminating the two tribes in Parale- pidiinae, i.e., Paralepidiini and Lestidiini, is whether the rear- ward shift of the anus occurs early or late in ontogeny. Incertae ce^w. — Peculiar eggs described by Delsman (1938) and Mito (1961a) are currently considered to be those of mycto- phiform fishes other than Myctophidae (Moser and Ahlstrom, 1970). These eggs are spherical, 1.12-1.37 mm in diameter, with a single oil globule and bear numerous short appendages on the chorion. Two types are known only from Asian waters. Ocean Research Institute, University of Tokyo, MiNAMiDAi, Nakano-ku, Tokyo 164, Japan. 1-15-1, Myctophidae: Development H. G. Moser, E. H. Ahlstrom and J. R. Paxton LANTERNFISHES of the family Myctophidae are found in all oceans of the world. Some 230-250 species are arranged in 36 generic/subgeneric taxa (Table 59). All nominal species are listed in Paxton (1979). Characteristic of the family is the presence of light organs or photophores on the head and body (Fig. 1 1 4). The different patterns of photophores have been used, along with meristics (Table 60), in species diagnoses and as a basis for classification within the family since the late 1800's. Most authors have placed the Myctophidae and closely related Neoscopelidae with the families Aulopidae, Chlorophthalmidae and related families in an order or suborder variously named the Iniomi, Myctophoidea or Myctophiformes (Gosline et al., 1966; Greenwood et al., 1966; Nelson, 1976; Johnson, 1982), although Rosen ( 1973) separated the Myctophidae and Neosco- pelidae as a restricted order Myctophiformes. Moser and Ahl- strom (1970, 1972, 1974), Ahlstrom et al. (1976) and Paxton (1972) are the most recent papers considering relationships with- in the family; characteristics of larvae and bones and photo- phores of adults were primarily utilized in the respective studies. Paxton's (1972) classification, including genera recognized sub- sequently, is as follows: Subfamily Myctophinae Tribe Electronini Genera: Protomyctophum. Metelectrona- Krefftichlhys', Elect rona. Tribe Myctophini Genera: Benthosema, Diogenichlhys, Hygophum, Myc- tophum. Symbolophorus Tribe Gonichthyini Genera: Loweina, Tarletonbeania, Gonichthys, Centra- branch us Subfamily Lampanyctinae Tribe Notolychnini Genus Notolychnus Tribe Lampanyctini Genera: Taaningichthys, Lampadena, Bolinichthys. Lep- idophanes, Ceratoscopelus. Stenobrachius, Lampan- yctus, Triphoturus, Parvilux^ Tribe Diaphini Genera: Lobianchia, Diaphus, Idiolychnus* Tribe Gymnoscopelini Genera: Lampanyctodes, Gymnoscopelus, Notoscopelus, Lampichthys, Scopelopsis, Hintonia There has not been a family revision at the species level since Fraser-Brunner's (1949) study. A large number of more recent generic revisions and regional studies are currently the primary sources for species identifications; most of these have been uti- lized in compiling the generic distribution limits (Table 59). The most recent zoogeographic studies are those of Backus et al. Hulley (1981). Wisner(1963). ' Hubbs and Wisner (1964). " Nafpaktitus and Paxton (1978). MOSER ET AL.: MYCTOPHIDAE 219 Table 59. Geographic Distribution of the Genera and Subgenera of Myctophidae. References marked * are useful for the identification of species. The division of the Atlantic and Indian Oceans is arbitrarily taken at 20°E, the Indian-Pacific Ocean boundary at 130°E. No, of species Lai. extremes Krefftichthys I Atlantic Indian Pacific 34°S-60°S 43°S-66°S 34°S-72°S Protomyclophum (Protomyctophum) 7 Atlantic Indian Pacific 34°S-60°S 44°S-65°S 40'^70°S Protomyctophum (Hierops) 7 Atlantic Indian Pacific 70°N-56°S 35°S-52°S 57°N-67°S Electrona 5 Atlantic Indian Pacific 55°N-70°S 2°N-68°S 42°N-70°S Metelectrona 2 Atlantic Indian Pacific 35''S-5I°S 35°S-47°S 33°S-55°S Benthosema 5 Atlantic 80°N-38°S Indian Pacific 2rN-35°S 7I''N-42°S Diogenichthys 3 Atlantic Indian Pacific 50°N-48°S 18°N-45°S 37°N-41°S Hygophum 9-1 1 Atlantic Indian Pacific 49°N-48°S 20''N-42°S 39"'N-46°S Symbolophorus 7-9 Atlantic Indian Pacific 59''N-51°S 2I°N-41°S 50°N-59°S Myctophum 13-14 Atlantic Indian 65''N-40°S 20°N-34°S Pacific 42''N-42°S Loweina 3-4 Atlantic Indian Pacific 44°N-38°S IO°S-40'^ 32''N-40°S Tarletonheania 1-2 Atlantic Indian Pacific 50°N-30°N Gonichthys 3-4 Atlantic Indian Pacific 47°N-40°S 25°S-39'>S 3I''N-42°S Cenlrobranchus 3-4 Atlantic Indian Pacific 46°N-35°S 15°N-33°S 37°N-37°S Nololychnus I Atlantic Indian Pacific 56''N-38°S 1 rN-40°S 34'>N-44°S Lobianchia 2 Atlantic Indian Pacific 6rN-5l°S 2'>N-40°S 32°N-47°S Diaphus 65-75 Atlantic Indian Pacific 62''N-52'^ 23''N-48°S 55°N-58°S Idiolychnus 1 Atlantic Indian Pacific 13°S-24°S 2rN *Hulley (1981:12) *Hulley (1972:217); Andriashev (1962:224) Andnashev (1962:225): McGinnis (1982:1 1) *Hulley (1981:29, 19) Hulley (1972:218); *McGinnis (1982:17) ♦Andriashev (1962); *McGinnis (1982:16, 17) Nafpaktitis et al. (1977:31); *Hulley (1981:36) •Nafpaklitis and Nafpaktitis ( 1 969:7); 'McGinnis (1982:18) ♦Wisner ( 1 976:20); 'McGinnis ( 1 982: 1 8) *Hulley (1981:40, 46); *McGinnis (1982:21) Nafpaktitis and Nafpaktitis (1969:10); *McGinnis (1982:21) •Andriashev ( 1 962); Ebeling ( 1 962: 1 40); *McGinnis (1982:21) •Hulley (1981:53) •McGinnis (1982:25) •Bussing (1965:200); •McGinnis (1982:25) •Nafpaktitis et al. (1977:52); Hulley (1972:220); (the specimen from 55°S is possibly mislabeled, McGinnis, (1982:26, 29)) Kotthaus (1972:18); *Nafpaktitis and Nafpaktitis ( 1969:1 1) •Wisner (1976); Nafpaktitis et al. (1977:52); Robertson et al. (1978:302) Nafpaktitis et al. (1977:58); Hulley (1981:58) •Nafpaktitis and Nafpaktitis (1969:15) •Wisner (1976:49); Rass (1960:149) •Bekker(1965); •Nafpaktitis et al. (1977:38); •Hulley (1981:61) •Bekker (1965:80); Hulley (1972:222) •Wisner (1976); •Bekker (1965:94); McGinnis (1982:30) •Hulley (1981:101) Kotthaus (1972:27); *Nafpaktitis and Nafpaktitis (1969:29) •Wisner (1976); Frost and McCrone (1979:755); •McGinnis (1982:33) •Nafpaktitis et al. (1977:62); 'Hulley (1981:87) Nafpaktitis and Nafpaktitis (1969); •Bekker and Borodulina (1978:1 20); McGinnis ( 1 982:34) •Kawaguchi and Aioi (1972); •Wisner (1976); Kawaguchi et al. (1972:27); Paxton and Nafpaktitis (ms) •Nafpaktitis et al. (1977:85) •Bekker (1964:23); •Nafpaktitis and Nafpaktitis (1969:31) •Wisner (1976); •Bekker (1964:23); McGinnis (1982:37) •Bekker (1963:160); 'Wisner (1976:82) Nafpaktitis et al. (1977:88); Hulley (1981:107) •Bekker (1964:38) •Bekker (1964); •Wisner (1976:86); McGinnis (1982:36) •Nafpaktitis et al. (1977:91) •Bekker (1 964:5 1, 58) •Bekker (1964:58) •Nafpaktitis et al. (1977:94); 'Hulley (1972:222) Kotthaus (1972:30); McGinnis (1982:37) Ebeling (1962:141); McGinnis (1982:37) •Nafpaktitis el al. (1977); Bekker (1967:98); McGinnis (1982:51) •Nafpaktitis (1978:7); McGinnis (1982:51) 'Wisner (1976:96); McGinnis (1982:51) 'Nafpaktitis et al. (1977:158); McGinnis (1982:52) 'Nafpaktitis (1978:62, 78) 'Nafpaktitis (1978:62); McGinnis (1982:52) 'Nafpaktitis and Paxton (1978:495) 'Nafpaktitis and Paxton (1978:495-496) 220 ONTOGENY AND SYSTEMATICS OF FISHES -AHLSTROM SYMPOSIUM Table 59. CoNTrNUED. No of species Ocean Lai. exlremes References Lampanyctodes 1 Atlantic Indian Pacific I9°S-34°S 35''S 34°S-5I°S Gymnoscopelus (Gymnoscopelus) 4 Atlantic Indian Pacific 34°S-66°S 60°S-65°S 40°S-72°S Gymnoscopelus (Nasolychnus) 4-5 Atlantic Indian Pacific 34°S-57°S 24°S-65°S 40°S-70°S Scopelopsis 1 Atlantic Indian Pacific 11°S-48°S 9»S-40°S 15°S-35°S Lampichthys 1 Atlantic Indian Pacific 30°S-48°S 35'^-40°S 7°S-49°S Notoscopelus (Noloscopelus) 5 Atlantic Indian 65°N-60°S 8''S-36'^ Pacific 50°N-37°S Notoscopelus (Parieophus) 1 Atlantic Indian Pacific 50°N-2rN Hintonia 1 Atlantic Indian Pacific 39°S-48''S 47<«-51°S 40°S-50°S Lampadena (Lampadena) 8-9 Atlantic Indian Pacific 65°N-48°S 6''N-49°S 4rN-49°S Lampadena (Dorsadena) I Atlantic Indian Pacific 45°N Taaningichthys 3 Atlantic Indian Pacific 43°N-44°S 8°N-30°S 4rN-68°S Ceratoscopelus 3 Atlantic Indian 52°N-45°S 20''N-43''S Pacific 43°N-42°S Lepidophanes 2 Atlantic Indian Pacific 43''N-48°S Bolinichthys 7 Atlantic Indian Pacific 53°N-41°S 21°N-44°S 31°N-43°S Triphoturus 3-4 Atlantic Indian Pacific 8°N-I4°S 38°N-35°S Stenohrachius 2 Atlantic Indian Pacific 57<'N-30°N Parvilux 2 Atlantic Indian Pacific 40''N-14°S Lampanyctus 40 Atlantic Indian 65°N-60°S 16°N-60°S Pacific 59''N-72°S ♦Ahlstrom et al. (1976:146); Grindley and Pennth (1965:283) Paxton and Nafpaktitis (in prep.) *Wisner (1976: 1 58-1 59); McGinnis (1982:55) *Hulley (1981:254); •McGinnis (1982:59) *Andriashev (1962:267); •McGinnis (1982:59) ♦McGinnis (1982:61, 58) ♦Hulley (1981:261); (03°S, Fraser-Brunner (1931:224) presumably a waiO Smith (1 933a: 1 26); *McGinnis (1982:64) ♦Andriashev (1962); McGinnis (1982:64) *Hulley (1981:241) Legand (1967:49); McGinnis (1982:57) *Wisner (1976:222); Paxton and Nafpaktitis (in prep.) Hulley (1981:242) McGinnis (1982:57) *Wisner (1976:215); McGinnis (1982:57) ♦Nafpaktitis et al. (1977:254) Andriashev (1962:278) Nafpaktitis and Nafpaktitis (1969:35); Grindley and Penrith (1965:283) *Fujkii and Uyeno (1976); Frost and McCrone (1979:755); Collins and Baron (1981:11) •Nafpaktitis et al. (1977:257) •Hulley (1981:239) McGinnis (1982:55) •Wisner (1976:220); McGinnis (1982:55) •Kreflt (1970:285); Hulley (1981:180) •Nafpaktitis and Paxton (1968:20, 21) •Nafpaktitis and Paxton (1968:20, 21) •Coleman and Nafpaktitis (1972:2) •Hulley (1981:167); 'Davy (1972) •Nafpaktitis and Nafpaktitis (1969:40) •Davy (1972:72); •Nafpaktitis et al. ( 1 977: 1 9 1 ) •Nafpaktitis et al. (1977:243); Hulley (1981:237) •Bekker and Borodulina (1968:792); •Nafpaktitis and Nafpaktitis (1969:65) •Wisner (1976:207); Robertson et al. (1978:302) •Nafpaktitis et al. (1977:225); •Hulley (1981:223) •Nafpaktitis et al. (1977:240); •Hulley (1981:229) Kotthaus (1972:18): 'Nafpaktitis and Nafpaktitis (1969:60) •Johnson (1975:58); Nafpaktitis et al. (1977:234) Hulley (1981:205) •Nafpaktitis and Nafpaktitis (1969:51) •Wisner (1976:165) •Wisner (1976:160) •Wisner (1976:163, 164) •Nafpaktitis et al. (1977:196); •Hulley (1981:183); Zahuranec (1980) •Nafpaktitis and Nafpaktitis (1969); Kotthaus (1972:35); •McGinnis (1982:42); Zahuranec (1980) •Wisner (1976:191); McGinnis (1982:42); Zahuranec (1980) MOSER ET AL.: MYCTOPHIDAE 221 Table 60. Meristics of the Genera AND Subgenera of Myctophidae. Fin rays Branchio- Dorsal Anal Pectoral Pelvic Procurrenl caudal Vertebrae stegals Gill rakers Krefflichthys 11-14 17-19 14-16 8-9 8-9 + 7-9 36-39 6-8 + 19-23 Protomyctomphum 10-14 21*-27 14-17 8-9 7-9 + 6-9 35-41 8-10 4-7 + 14-21 P. Hierops 11-13 20-27 15-18 8 7-1 1 + 6-9 36-42 9-10 3-5 + 13-18 Electrona 12-16 18-22 11-17 8 6-10 + 6-9 33-41 7-8 3-10 + 12-25 Metelectrona 13-15 19-22 14-16 8 10 + 9 35-38 8 4-7 + 16-20 Benthoscma 11-15 16-22 10-17 8-9 7-9 + 7-9 31-37 9 3-10 + 10-21 Dwgenichlhys 10-13 14-18 10-14 7-8 7-9 + 7-9 29-34 7 2-4 + 10-12 Hygophum 10-15 18-25 12-17 8-9 6-9 + 6-9 34-40 9 3-6 + 12-16 Myclophum 11-15 16-27 12-22 7-8 7-9 + 7-9 35-46 8-9 4-8 + 10-21 Symhotophorus 12-16 18-24 12-20 8 8-10 + 7-9 36-42 9 4-7 + 12-19 Loweina 10-13 13-17 9-12 7-9 6-7 + 6-7 37-39 9 2-3 + 5-10 Tarlelonheania 11-15 16-20 11-16 8 5-8 + 5-8 40-42 8 4-6 + 10-12 Gonichlhys 10-13 17-24 11-18 6-8 5-6 + 5-6 38-41 9 3-6 + 7-12 Cenlrobranchus 9-12 16-20 11-17 8 5-7 + 5-7 35-40 7-8 Nololychnus 10-12 12-15 11-15 6-7 7-9 + 7-9 27-31 9-10 2 + 8-9 Lnlnanchta 15-18 13-15 11-15 8 5-7 + 5-6 33-35 9 4-6 + 11-16 Diaphus 10-19 11-19 9-14 8 5-8 + 5-8 31-37 8-9 4-11 + 9-21 Idwlychnus 14-15 14-16 13-15 8 34 6-7 + 14-15 Lampanyctodes 13-14 14-17 12-14 8 8-10 + 9-10 36-39 9-11 10-11 + 20-23 Gymnoscopelus 14-21 16-22 12-16 8-9 10-12 + 11-15 41-45 10 6-12 + 14-26 G. Nasolychnus 16-20 16-20 12-15 8 8-13 + 10-15 41-45 10-11 7-12 + 17-25 Scopelopsis 20-24 23-27 10-12 7-8 9-11 + 11-12 38-39 9-10 7-9 + 16-18 Lainpkhlhys 16-18 21-23 11-15 8 10 + 12 40-41 9 4-6 + 13-16 Noloscopelus 21-27 18-21 11-14 8-9 10-14 + 10-15 35-40 10 4-10 + 9-22 N. Parieophus 23-26 18-20 12-14 37-38 8-10 + 18-20 Hmtonia 14-16 12-14 13-15 8 10-11 + 13 37-39 9 6-7 + 11-14 Lampadcna 13-16 12-15 13-18 8 8 + 8-9 35-40 9 3-8 + 9-18 L. Dorsadena 14-15 12-14 15-16 8-9 4-5 + 12 Taaningichthys 11-14 11-14 12-17 8 7-10 + 6-10 34-41 8-9 2-5 + 6-14 Ceraloscppelus 13-15 13-16 12-15 8 6-7 + 6-7 35-38 9 3-5 + 9-16 Lepidophanes 11-15 13-16 11-14 8-9 6-7 + 6-8 33-37 9 3-4 + 8-1 1 Bolinichthys 11-15 11-15 11-15 8 7 + 7-8 33-36 9 3-7 + 11-17 Triphoturus 12-16 13-18 8-10 8 5-7 + 6-7 30-36 10-11 2-4 + 8-11 Slenobrachius 12-15 14-16 8-10 8 6-8 + 7-9 35-38 9-10 5-6 + 12-14 Panilux 14-17 15-18 10-13 8 8 + 8-9 35-38 10-11 4-6 + 11-15 Lampanyctus 10-19 14-21 0-17 8 6-8 + 6-8 30-40 8-11 3-8 + 9-19 • Incorrectly 15-27 in Paxlon, 1972 (1977) and Hulley (1981) on Atlantic species and McGinnis (1982) on Southern Ocean species. Most lantemfishes make extensive vertical migrations from mesopelagic depths to the upper waters at night, some reaching the surface (Paxton, 1 967). The fisheries potential of myctophids and other mesopelagic fishes has recently been reviewed (Gjo- saeter and Kawaguchi, 1980). Adults range in size from 20-300 mm (Kreflt, 1974) and have a life span of from one year in some tropical species (Clarke, 1973) to more than five years in the few temperate species that have been studied (Smoker and Pearcy, 1970; Gjosaeter, 1973: Kawaguchi and Mauchhne, 1982). Eggs Myctophids are oviparous and presumably all produce plank- tonic eggs although such have been reported for only two species. Sanzo (1939a) indicated that mature ovarian eggs of E. rissoi have the following characteristics: round shape; 0.80-0.84 mm diameter; segmented yolk; single oil globule, ca. 0.28 mm di- ameter; smooth chorion. He illustrated a planktonic egg with similar characteristics and tentatively identified it as that o( E. rissoi. Robertson (1977) described the planktonic egg of Lam- panyctodes hectoris as follows: weakly oval; long axis 0.74-0.83 mm, short axis 0.65-0.72 mm; strongly segmented yolk; single oil droplet, 0.21-0.23 mm diameter: narrow perivitelline space; chorion smooth and delicate. He based his identification on the similarity of these eggs and mature ovarian eggs of running ripe L. hectoris captured at the same time by trawl. We have observed planktonic eggs similar to those described by Robertson (1977) but have not found them with advanced embryos that could be matched with co-occurring yolk-sac myc- tophid larvae. The fact that these and other types of eggs ten- tatively identified as myctophids occur in relatively low abun- dance compared with myctophid larvae led Moserand Ahlstrom (1970) to suggest that the fragile chorion breaks in contact with plankton nets and the embryo is extruded through the mesh. Larvae Moser and Ahlstrom (1970) reviewed the literature on myc- tophid larvae: however, numerous recent contributions have advanced our knowledge of the group and are listed in Table 61. Of the 32 recognized genera of myctophids, larvae have been described for all but Hintonia. The larval stages of myc- tophids provide sets of characters that are useful at levels of systematic analysis from species separation to hypotheses of 222 ONTOGENY AND SYSTEM ATICS OF HSHES-AHLSTROM SYMPOSIUM Table 61. Summary abbreviated as follows: OF Literature Containing Illustrations of Developmental Stages of Myctophids. Frequently cited authors are Ahlstrom (A), Belyanina and Kovalevskaya (B + K), Dekhnik and Sinyukova (D + S), Moser and Ahlstrom (M + A), Pertseva-Ostroumova (P-O), Shiganova (S), Tuning (T). Species Single larval stage Multiple larval stages Transforming stage Juvenile stage Benlhosema fibulatum glaciale panamense pterota suborbitale Bolinichthys dislofax pyrsobolus Centrobranchus andrae breviroslris choerocephalus nigroocellatus Ceratoscopelus maderensis townsendi warming! Diaphus agassizii holli malayanus melapoclampus mollis pacificus rafinesquei Iheta Diogenichthys atlanticus laternalus panurgiis Electrona antarctica carhbergi rissoi subaspera Gonichthys coccoi tenuiculus Gymnoscopelus bolini braueri fraseri mcholsi opislhoplerus Hygophum atraium henoiti brunni hanseni hygomi M + A, 1974 Holt, 1898; S, 1977 M + A, 1974; P-O, 1974 P-O, 1964; M + A, 1974 M + A, 1974 P-O, 1964 P-O, 1974 P-O, 1964 M + A, 1974 P-O, 1974 M + A, 1972; S, 1977 M + A, 1974 Miller etal., 1979; Belyanina, 1982b D + S, 1966 M + A, 1974 P-O, 1964; M + A, 1974 P-O, 1964 M + A, 1974 M + A, 1974 M -I- A, 1974 M -I- A, 1974 P-O, 1964 P-O, 1964 P-O. 1974 T, 1918; Sparta, 1951; M + A, 1974 M + A, 1970 Tsokur, 1981 P-O, 1974; Badcockand Merrett, 1976; S, 1977 Holt, 1898; T, 1918 Sparta, 1951 M + A, 1970 P-O, 1974; S, 1977 P-O, 1974 P-O, 1974 M + A, 1970 - T, 1918; D -1- S, 1966 T, 1918 S, 1977 S, 1977 P-O, 1975 T, 1918 Tsokur, 1975 Sparta, 1952 S, 1977 P-O, 1975 T, 1918 Tsokur, 1975 Sparta, 1952 S, 1977 T, 1918 T, 1918; A, 1965; M + A, 1970; P-O, 1974; S, 1977 A, 1965; M + A, 1970 P-O, 1974 T, 1918 T, 1918; M + A, 1970; S, 1977 M + A, 1970 P-O, 1974 P-O, 1967; B + K, 1979 - B + K, 1979 T, 1918; Sanzo, 1939a; Sanzo, 1939a D + S, 1966; M -I- A, 1970 M + A. 1974 M -I- A, 1974 T, 1918; S, 1977; D + S. 1966 M + A, 1970 S, 1977 P-O, 1977; B + K, 1979 P-O, 1977 M + A, 1972; P-O, 1977; B + K, 1979 Yefremenko, 1977 M + A, 1970 T, 1918; S, 1974 S, 1977 T, 1918; P-O, 1974; S, 1977 M + A, 1970 S, 1977 M + A, 1972 M + A, 1970 T, 1918; S. 1974 S, 1977 T, 1918; P-O, 1974; S, 1977 Holt, 1898; T, 1918; Sparta, 1951 M + A, 1970 Tsokur, 1981 S. 1977 M + A. 1970 T, 1918; S, 1977 P-O, 1975 T, 1918 Tsokur, 1975 Sparta, 1952 S, 1977 T, 1918 T. 1918; M + A, 1970; S, 1977 M + A, 1970 T, 1918; Sanzo, 1939a; M + A, 1970 T, 1918; S, 1977 S, 1977 M + A, 1970 T, 1918; S, 1974 S, 1977 T, 1918; S, 1977 MOSER ET AL.: MYCTOPHIDAE Table 6 1 . Continued. 223 Species Single larval stage Mullipte larval stages Transforming stage Juvenile stage macrochir pro.ximum rcinhardli taanmgi Idiolychnus urolampus Kretflichthys anderssoni Lampadena luminosa urophaos Lainpanyctodes hectoris Lampanyclus achirus crocodilus jordani nohilis pusillus regalis ritleri Lampichthys procerus Lepidophanes gaussi guerjtheri Lohianchia M + A, 1974 M + A, 1974; Miller et al., 1979 M + A, 1974 M + A, 1974 M + A. 1974 M + A, 1974 M + A, 1974; Miller etal., 1979 M + A. 1974 P-O, 1964 Miller etal., 1979 M + A, 1974 M + A, 1974 M + A, 1974 M + A, 1972 S. 1975 P-O, 1974 M + A, 1970; S, 1977 Yefremenko, 1976; B + K, 1979 M + A, 1972 Ahlstrom et al., 1976 T, 1918; D + S, 1966 T, 1918; D + S, 1966 A, 1965 M + A, 1972 S, 1977 S, 1975 P-O, 1974 M + A, 1970; S, 1977 Yefremenko, 1976 M + A, 1972 Ahlstrom et al., 1976 T, 1918 T, 1918 Bolin, 1939b M + A, 1972 M + A, 1972; S, 1977 S, 1975 M + A, 1970; S, 1977 Yefremenko, 1976 Ahlstrom et al., 1976 T, 1918 T, 1918 do/Ieini M + A, 1974 T, 1918; D + S, 1966; T, 1918; S, 1977 T, 1918; S, 1977 S, 1977 gemellari Sanzo, 1931c; P-O, 1964; M + A, 1974 T, 1918 T, 1918 T, 1918 Loweina rara M + A. 1974 M + A, 1970; P-O, 1974 M + A, 1970 M -1- A, 1970 lerminata Belyanina, 1982b - — — Meleleclrona ventralis M + A, 1974 - — — — Myctophum asperum P-O. 1964; M -1- A, 1974 Imai, 1958; P-O, 1974 Imai, 1958; P-O, 1974 aurolaternatum M + A, 1974 — __ brachygnathum M -1- A, 1974 _ lychnobium M + A, 1974; P-O, 1974 — P-O, 1974 nilidulum M + A, 1974 M + A. 1970; P-O, 1974 M + A, 1970 oblusirostre M + A, 1974 _ — punctalum M + A, 1974 Sanzo, 1915b; T, 1918; S. 1977 Sanzo, 1915b; S, 1977 T, 1918; T, 1918; S, 1977 selenops M + A, 1974 — _ spinosum M + A, 1974 P-O, 1974 P-O, 1974 P-O, 1974 Nololychnus valdiviae P-O, 1964; M + A, 1974 T, 1918 T. 1918 T, 1918 Notoscopelus caudispinosus Belyanina, 1982b _ — elongatus — T, 1918 T, 1918 T, 1918 resplendens M + A, 1974 M + A. 1972; Badcock and Merrett, 1976; S, 1977 M + A, 1972; s. 1977 224 ONTOGENfY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Table 61. Continued. Species Single larval stage Multiple larval stages Transforming stage Juvenile stage Parvilux ingens M + A, 1974 - - - Prolomyclophum arcticum — T, 1918 T, 1918 T, 1918 boHni _ P-O, 1967; B + K, 1979 — chilensis M + A. 1974 — _ — crockeri — M + A, 1970 — M -1- A, 1970 normani P-O. 1967; M + A, 1974 _ _ P-O, 1967 parallelum — P-O, 1967; B -1- K, 1979 — _ subparallelum M + A, 1974 — — — tenisoni M + A, 1974 — — _ ihompsom P-O, 1964 P-O, 1967; M + A, 1970 - M -1- A, 1970 Scopelopsis nndlipiinctatus — M + A, 1972; P-O, 1972 M + A, 1972; P-O, 1972; M-F A, 1974 M + A, 1972 Stenobrachius leucopsarus P-O, 1964; M -1- A, 1974 Fast, 1960; A, 1965; A, 1972b Fast, 1960 Fast, 1960 Symbolophours hoops — P-O, 1974 — — californiense P-O, 1964; M -1- A, 1974 A, 1965; M + A, 1970; P-O, 1974 M + A, 1970; P-O, 1974 — evermanni P-O, 1964 P-O, 1974 P-O, 1974 P-O, 1974 veranyi — Sanzo, 1915b; T, 1918; D-l-S, 1966 Sanzo, 1 9 1 5b; T, 1918 Sanzo, 1 9 1 5b, T, 1918 Taaningichthys minimus - M + A, 1972 - - Tarletonbeania crenularis P-O, 1964; M ^ A, P-O, 1974 1974; A, 1965; M -i- A, 1970 Bolin, 1939b; M + A, 1970 M -1- A, 1970 Tripholurus mexicanus M + A, 1974 A, 1965; A, 1972b _ _ nigrescens Moser, 1981 - - - Op po XXvo Fig. 1 14. Hypothetical myctophid showing photophore terminology, from Paxton (1972). MOSER ET AL.: MYCTOPHIDAE 225 Fig. 1 15. Larvae of Electronini. (A) Krefftichlhys anderssoni. 15.7 mm; (B) Protomyctophum normani. 15.2 mm; (C) P. Heirops ihompsom, 13.8 mm; (D) Elcclrona rissoi. 7.9 mm; (E) £. antarclica. 12.7 mm; (F) Melelectrona ventralis, 10.3 mm. A, B, E, F from Moser and Ahlstrom (1974); C and D from Moser and Ahlstrom (1970). 226 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Table 62. Sequence of Formation of Photophores which Appear in Fourteen Genera of Myctophidae. The Bfj appear first in all genera listed. Parentheses indicate photophores appear late in larval period. PO, PO, PVO, PLO VO, AOa, AOa, Benthosema suborbitale glaciate pterola fibulalum Diogenichthys lalernalus atlanticus Myclophum spinosum lychiwbium asperum brachygnalhum obtusirosire selenops Lobianchia Diaphus theta pacificus Gymnoscopelus Lampanyctodes Scopelopsis Lainpichthys Notoscopelus Lampadena Ceratoscopelus Lepidophanes Bolinichlhys 22--2 1 1 333------33 - - - - (1) (1) (1) (1) (1) (1) ------- - --1-4 6---2 3 5--5-6- --1--3 5--2---6--46 _________ 1 ________ (5) 1 2 2 2 3 1 1 (1) (4) (3) (7) (8) (5) - (4) (1) - (9) (3) (6) - - (6) - 3 3 1 2 1 (1) ordinal relationships. One set is the size at various develop- mental milestones. Myctophid larvae hatch at about 2 mm length with a yolk-sac remnant. Notochord flexion occurs in a narrow size interval (0.5-2.0 mm) and the size at mid-flexion is typically about half the maximum larval size. Size at transformation also occurs within a short length interval, usually not exceeding 2 mm. Most myctophid species transform in the length range of 12-19 mm, although some (e.g., Electrona rissoi, Notolychnus valdiviae) are as small as 9-10 mm at transformation and some species of Symbolophorus reach about 23 mm before transfor- mation. Gymnoscopelus nicholsi has the largest larvae recorded, up to 28 mm. Head, body, and gut shape are distinctive for most species and within most genera there is a similarity of shape (Figs. 1 1 5- 124). While most myctophid larvae are moderately slender, body shape can range from highly attenuate (e.g., Hygophum reinhardti) to markedly robust (e.g., some Myctophum and Lampanyctus species). Some are deep-bodied but laterally com- pressed (e.g., Gonichthyini). Robust larvae and deep-bodied, laterally compressed forms tend to have large heads and jaws, while attenuate forms have flat heads. The eye is varied in size and shape and provides numerous characters. In the Myctophinae the eyes are elliptical in outline in contrast to most Lampanyctinae which have rounded eyes. Further specializations in Myctophinae are the presence of var- iously shaped choroid tissue on the ventral surface of the eye in most genera and eye stalks in several genera. Among 1am- panyctine genera eyes are sessile and only Lobianchia doflcini and species of Triphoturus have markedly narrowed eyes with choroid tissue. The gut has distinctive transverse rugae and ranges from short, to elongate, to trailing free from the body. In most myctophids it extends to about the midpoint of the body and is slightly S- shaped. The curvature tends to be more pronounced in taxa with short guts. In two myctophine genera (Metelectrona and some Hygophum species) the anterior section of the gut is small in diameter and opens dorsally into the relatively larger pos- terior section. In most myctophids, ray formation and ossification of fins proceeds in the following sequence: caudal, pectoral, anal, dor- sal, and pelvic. However, in some Symbolophorus species the pelvic fin forms early and ossification of rays precedes that of the anal and dorsal fins. In most species the pectoral fin is relatively small, but deep-bodied and robust forms in both Fig. 1 16. Larvae of Myctophini. (A) Benthosema glaciale. 10.5 mm; (B) B. suborbitale. 9.2 mm; (C) B. pterola. 8.5 mm; (D) B. fibulatum. 8.7 mm; (E) Diogenichthys lalernalus. 1.1 mm; (F) D. atlanticus. 8.8 mm. A-D from Moser and Ahlstrom (1974); E and F from Moser and Ahlstrom(1970). ^^JmiJdJ^i^L *ss;;;^a 228 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Fig. 1 17. Larvae of Myctophini. (A) Hygophum proximum. 8.9 mm; (B) H. taaningi. 6.8 mm; (C) H. reinhardti. 12.8 mm; (D) Symbolophorus californiense. 1 1.5 mm; (E) Myctophum punclalum. 13.6 mm; (F) M. aurolalernatum. 26.0 mm. A. B. E, F from Moser and Ahlstrom (1974); C and D from Moser and Ahlstrom (1970). MOSER ET AL.: MYCTOPHIDAE 229 subfamilies have large fins and fin bases. In Symbolophorus the fin base is uniquely shaped and in Lobianchia the fin blade has a unique shape. In two genera (Loweina. Tarletonbeania) the lowermost pectoral ray is elongate and ornamented. The finfold is enlarged in many myctophine genera and greatly enlarged in one myctophine tribe, the Gonichthyini. Myctophids, with the exception of Notolychnus and Taan- ingichthys, develop the middle branch iostegal photophore (Br,) during the larval period. It is located posteroventral to the orbit but during transformation assumes a position beneath the orbit on the branchiostegal membrane. Three myctophine genera and 1 1 lampanyctine genera develop additional photophores during the larval period; however, the Br, is always the first to develop. The larval photophore complements and the sequence of ap- pearance of constituent photophores are useful characters. Myctophid species have distinct melanophore patterns, with the exception of the large genus Diaphus, for which only a few specific patterns have been identified. Most genera may be sep- arated by overall similarity of pattern among their species and some have unique melanophore loci. There are no clear patterns for tribes or subfamilies although certain pigment loci are per- sistent in some tribes (e.g., caudal fin base spots in diaphines; dorsal midline series in gymnoscopelines). In the following summary of key larval characters, the genera are listed for convenience as in Moser and Ahlstrom (1970. 1972, 1974) and the sequence does not necessarily imply rela- tionship. Likewise, the species groups serve only to identify phenotypically similar larval types. Larvae of a majority of myc- tophid genera have a moderately slender body, a head of mod- erate size, with a slightly convex dorsal profile and a pointed snout of moderate length. Body and head shape are noted only when they depart from this morph. In Myctophinae eye shape is noted when it is markedly elliptical and size is noted only when larger or smaller than typical. In Lampanyctinae eye shape is noted only when it departs from the round condition and eye size only when larger or smaller than typical. Choroid tissue is described only when it is present. Gut length and shape are described only if there is a departure from the typical morph — a slightly S-shaped gut that extends to about midbody. The most persistent pigment locus in myctophid larvae is above or to the side of the free terminal section of the gut, thus only the lack of this pigment is noted. Larval photophores, in addition to the Br,, and their sequence of appearance are shown in Table 62. Myctophinae Krefflichthys. — Fig. 1 15 A; head small with short snout; conical choroid tissue; gut straight, extending beyond midbody; dorsal fin displaced posteriad; lateral gut and postanal median ventral melanophore series; large lateral hypural pigment patch. Protomyctophum. — ¥\%. 1 15B, C; two subgenera; head small to moderate in size; gut short, wide space between anus and anal fin; head pigment lacking except in otic region of P. Heirops chilensis; some species may have melanophores on lateral gut, above gut on trunk, above gas bladder, in postanal ventral mid- line series, prominent pigment on lateral hypural region. P. Heirops: Fig. 1 1 5C; characters similar to P. Protomyctophum except eye narrower. Eleclrona— Fig. 1 15D, E; body moderately slender to moder- atey deep; head moderately large; snout blunt or pointed; gut short, somewhat saccular, strongly S-shaped; space between anus and anal fin not as large as in Protomyctophum; three morphs. E. subaspera-E. carlsbergi: eye slightly elliptical, small lunate choroid mass in E. carlsbergi; pigment above gut; E. subaspera has pigment lateral to cleithrum. E. rissoi: Fig. 1 1 5D; head large, broad; eye very narrow; pigment at lower jaw symphysis, on pectoral fin blade. E. antarctica: Fig. 1 1 5E; body and head lat- erally compressed; gut mass protrudes ventrally from body pro- file; eye small, narrow, with bicolored elongate conical choroid mass; pigment on upper jaw, pectoral fin blade, lateral gut, lateral hypural region. Metelectrona. — Fig. 1 1 5F; body and head laterally compressed; dorsal finfold enlarged with fin base initially separated from body; lunate choroid mass; anterior gut section with small di- ameter, opening dorsally into somewhat saccular posterior sec- tion; pigment below lower jaw and on isthmus. Benthosema. — Fig. 1 16A-D; two morphs; photophores (Table 62). B. glaciale-B. sitborbitale: Fig. 1 16A, B; eyes narrow, with small lunate choroid mass; gut moderately short in preflexion larvae with space between anus and anal fin; pigment on snout, lower jaw, hindbrain, lateral and ventral cleithral region; pig- ment above gut in B. glaciate. B. pterota-B. fibulatum: Fig. 1 I6C. D; eyes less narrow than in above morph, with sliver of choroid tissue or none; gut extends to about midbody with no space between anus and anal fin; preflexion larvae with melanophore series on lateral gut and on postanal ventral midline, coalescing to a single melanophore; lateral cleithral pigment; lower jaw pigment in B. pterota. Diogenichthys. — Fig. I16E, F; eyes very narrow in preflexion stage, less so in postflexion; photophores (Table 62); pigment series on lateral gut and on postanal ventral midline, increasing with development; spot at caudal fin base; pigment on tip of lower jaw in D. laternatus; D. atlanticus has spot on trunk above terminal gut flexure and pigment on symphyseal barbel. Fig. 1 18. Larvae of Myctophum. (A) M. phengodes. 9.8 mm; (B) M. asperum. 6.8 mm; (C) M. brachygnathum. 7.5 mm; (D) M. selenops. 7.8 mm; (E) A/, spinosum, 9.0 mm. From Moser and Ahlstrom (1974). Fig. 1 19. Larvae of Gonichthyini. (A) Loweina rara. 17.6 mm; (B) Tarletonbeania crenularis. 18.9 mm; (C) Gomchthys tenutculus. 1.1 mm; (D) Centrobranchus choerocephalus. 7.3 mm. From Moser and Ahlstrom (1970). Fig. 120. Larvae of Lampanyctinae. (A) Notolychnus valdiviae. 8.7 mm; (B) Lobianchia dojleini. 8.2 mm; (C) L. gemellari. 6.7 mm; (D) Diaphus theta. 6.9 mm; (E) D. pacificus. 5.2 mm; (F) Gymnoscopelus nicholsi. 23.5 mm. A-E from Moser and Ahlstrom (1974); F from Moser and Ahlstrom (1972). Fig. 121. Larvae of Lampanyctinae. (A) Lampanyctodes hectoris. 1 3.0 mm; (B) Scopelopsis muttipunctatus. 1 3.4 mm; (C) Lampichthys procerus, 1 4.5 mm; (D) Notoscopelus resplendens. 11.2 mm; (E) Lampadena lununosa. 1 2.8 mm; (F) Taanmgichthys minimus. 1 4.4 mm. A from Ahlstrom et al. (1976); B, C. F from Moser and Ahlstrom (1972); D and E from Moser and Ahlstrom (1974). 230 ONTOGENY AND SYSTEMATICS OF FISHES -AHLSTROM SYMPOSIUM ^yig^i®^' ^^^> MOSER ET AL.: MYCTOPHIDAE 231 232 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM MOSER ET AL.: MYCTOPHIDAE 233 234 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Hygophum. — Fig. 1 17A-C; diagnostic pattern of melanophores at the cleithral symphysis and isthmus region consisting of paired pigment dashes that form a median Hne as the series extends forward on the isthmus; Br, photophore forms late in larval period; three morphs. H. proximum-H. hygomi-H. benoiti-H. hanseni-H. brunni: Fig. 1 1 7A; eye moderately narrow with con- ical choroid tissue; pigment sparse in most species with some lateral gut spots in all species; some species may have pigment on hypaxial myosepta, jaws, lateral cleithral region, base of cau- dal rays. H. atratuin-H. reinhardti: Fig. 1 1 7C; body very slen- der; head flat; eyes very narrow, on short stalks; elongate conical choroid mass; gut almost straight, small diameter; pigment se- ries along lateral gut and hypaxial myosepta; pigment at caudal fin base; pigment on lower jaw symphysis in H. atratum. H. macrochir-H. taaningi: Fig. 117B; body and head deep and laterally compressed; eyes large, relatively wide; no choroid tis- sue; anterior gut section narrow in diameter, opening dorsally into somewhat saccular posterior section; H. macrochir has pig- ment on upper and lower jaw and a patch of melanophores on posterior gut section; H. taaningi has pigment on gular region and lateral surface of cleithrum. Symbolophorus. — Fig. 1 17D; head broad, somewhat flat; eyes slightly stalked, conical choroid mass; pectoral fin large with supernumerary rays, base wing-shaped, rays ossify early; pelvic fin large, early-forming in some species; dorsal finfold well de- veloped with fin base forming in it; pigment series on lateral gut and postanal ventral midline in preflexion larvae; pigment on snout, hindbrain, lateral cleithral region, isthmus, paired fins. Myctophum.— Figs. 1 17E, F and 1 18A-E; at least five distinct morphs, all but M. aurolaternatum with enlarged fan-shaped pectoral fins, some with supernumerary rays and early ossifi- cation; conical choroid mass. M. aurolaternatum: Fig. 117F; body very slender; head somewhat flat; eyes small, on elongate stalks; gut straight, at midbody becomes trailing, extending to well beyond caudal fin; dorsal finfold well developed, fin base forms at its margin; pigment series on lateral gut, evenly dis- tributed on trailing section, except heavier near terminus; pig- ment on jaws, isthmus, opercle, branchiostegal membrane, pec- toral fin, anal fin base, caudal fin. M. nitidulum-M . punctatum: Fig. 1 1 7E; body moderately slender to slightly deep; head broad, somewhat flat in preflexion stage; eyes on short stalks; numerous small melanophores on snout, jaws, brain, isthmus, branchio- stegal membrane; two rows of melanophores on ventral surface of gut; opposing melanophores on postanal dorsal and ventral midline; pigment on pectoral fin base and blade and at base of caudal rays. M. phengodes: Fig. 1 1 8 A; body and head moder- ately deep; similar to M. nitidulum, except pigment sparse and eyes not stalked; pigment at base of pectoral fin rays. M. spi- nosum-M. lychnohium: Fig. 1 18E; head with convex dorsal pro- file and long snout giving the larva a fusiform appearance; long axis of eye rotated towards horizontal; photophores (Table 62); head heavily pigmented on jaws, brain, postorbital and oper- cular regions; pigment above gut on trunk, embedded in my- osepta in M. spinosum; opposing dorsal and ventral midline blotches, larger and more deeply embedded in M. spinosum with embedded myoseptal pigment along horizontal septum; blotch at base of caudal rays. M. asperum-M . brachygnathum- M. obtusirostre-M. selenops: Fig. 118B-D; body deep, robust; head broad, deep with convex dorsal profile and large snout; eye relatively larger than in other morphs; choroid tissue broadly conical, except in M. selenops where it is elongate and pigmented at tip; photophores (Table 62); head pigment similar to M. spinosum; most species have heavy pigment lateral to cleithra and on pectoral fin bases; all species lack trunk and tail pigment, except M. asperum which has extensive embedded myoseptal and dorsal/ventral midline blotches. Loweina. — Fig. 1 1 9 A; body and head moderately deep, laterally compressed; dorsal and anal fins displaced far posteriad; dorsal and ventral finfolds greatly enlarged and conspicuously pig- mented to produce a disc-shaped profile; eyes large; gut with expanded anterior section and enlarged terminal section; pec- toral fin large with lower-most ray elongate, ornamented with pigmented spatulations; interorbital pigment band; pigment at lateral cleithral surface, dorsal fin origin, and opposing midline blotches at caudal peduncle region. Tarletonbeania. — Fig. 1 198; similar to Loweina. except median fins displaced less posteriad; eye narrower and with lunate cho- roid mass; four melanophores on periphery of brain, two me- lanophore series on ventrum of gut. Gonichthys. — Fig. 1 1 9C; body and head deep and laterally com- pressed, leaf-like; snout large, angulate in profile; eye small with elongate conical choroid mass, pigmented at tip; enlarged dorsal and ventral finfolds; pectoral fins moderately large; pigment on snout, jaws, midline of brain, postorbital and opercular regions; pigment on lateral hindgut and on trunk above gut; series of embedded blotches on dorsal midline of body, opposing blotch- es on postanal ventral midline; large pigment patch on lateral caudal peduncle region in G. tenuiculus; heavy embedded pig- ment streak along horizontal septum in G. coccoi. Centrobranchus. — Fig. 119D; morphology similar to Gonich- thys except snout markedly blunt and rounded and terminal gut flexure less acute; two morphs. C. choerocephalus-C. breviros- tris-C. nigroocellatus: Fig. 1 19D; eye very narrow with unpig- mented choroid mass that exceeds it in length; pigment sparse; some at postorbital-opercular region, branchiostegal membrane, ventral surface of liver. C. andrae. eye wider than in above morph and with short conical choroid mass; pigment extensive, on snout, upper jaw, dorsal brain, opercle, branchiostegal mem- brane, lateral hindgut, ventral surface of liver, pectoral fin base; embedded spots along dorsal midline with opposing spots along postanal ventral midline; embedded spots along horizontal sep- tum in caudal peduncle region. Lampanyctinae Notolychnus. — Fig. 1 20A; head relatively large with moderately elongate snout; eyes usually narrow, often irregular in shape; gut short, more so in preflexion stage; no photophores, even Br, lacking; pigment on lateral hindgut, gas bladder, base of caudal rays; a persistent but sparse postanal ventral midline series. Lobianchia. — Fig. 120B, C; body deep, robust; head broad with large snout; pectoral fins large; blade wing-shaped with upper rays longer than others; photophores (Table 62); head unpig- mented; pigment on trunk, on gut below pectoral fin base, on pectoral fin base and blade, embedded in gut region anterior to pectoral fin base, along anal fin base, and at base of caudal rays; embedded melanophores in myosepta above pectoral fin be- coming extensive in postflexion stage; two morphs. L. dofleini: Fig. 120B; eye small, narrow, with lunate to squarish choroid MOSER ET AL.: MYCTOPHIDAE 235 Fig. 122. Urvae of Lampanyctmae. (A) Ceraloscopehis townsendi. 16.6 mm; (B) Lepidophanes gaussi. 13.5 mm; (C) BoUmchthvs distofax. 9.4 mm; (D) Slenohrachius leucopsarus. 10.4 mm; (E) Parvilux ingens. 14.4 mm; (F) Triphoturus mexicanus. 10.5 mm. A-E from Moser and Ahlstrom (1974); F from Ahlstrom (1972b). 236 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM mass; gradual transition from lower pectoral rays to longer upper rays. L. gemellari: Fig. 120C; eye large, almost round, choroid mass a lunate sliver; abrupt transition between lower pectoral rays and long upper rays. Diaphus. — Fig. 1 20D, E; pigment lacking on head; melanophore at anteroventral surface of liver, one or more at midgut region, one or more at base of caudal rays; gas bladder pigmented; two morphs. D. theta: Fig. 120D; body moderately slender; head moderate in size; photophores (Table 62); numerous melano- phores in postanal ventral midline series, persisting into post- flexion stage. D. pacificus: Fig. 120E; body moderately deep, somewhat robust; head moderately large; photophores (Table 62); a few melanophores in postanal ventral midline series, usu- ally coalescing to one before flexion stage. Gymnoscopelus. — Fig. 120F; photophores (Table 62); pigment above brain, at lateral cleithral region, above midgut, above gas bladder; postanal ventral midline series present but, in some species, restricted to caudal peduncle region; melanophore series on each side of dorsal midline, in most species extending be- tween caudal and dorsal fins, in others extending forward to dorsal fin origin, and in others restricted to caudal penduncle region; pigment at base of caudal rays; some species have pig- ment on lateral hypural region; lateral pigment patch at caudal peduncle in G. opisthopterus, which also has embedded mela- nophores above vertebral column. Lampanyctodes. — Fig. 121 A; photophores (Table 62); pigment above brain, at anteroventral surface of liver, above gas bladder; a postanal ventral midline series and a series on each side of dorsal midlme between dorsal and caudal fins; pigment at base of caudal rays and at lateral hypural region. Scopelopsis. — Fig. 121B; photophores (Table 62); pigment sim- ilar to Lampanyctodes except additional melanophores on hind- brain, nape, lateral cleithral region; pigment rows along dorsum irregular. Lampichthys. — Fig. 121C; photophores (Table 62); pigment similar to Scopelopsis except dorsal rows consist of large closely- spaced melanophores which at maximal development extend from caudal fin to dorsal fin origin; a short melanophore series along horizontal septum on caudal peduncle in late postflexion stage. Notoscopelus. — Fig. 12 ID; photophores (Table 62); body mod- erately deep; head moderately large; eye large; snout becomes somewhat bulbous at flexion stage; gut short in early preflexion stage, elongates to about midbody by late preflexion; pigment at tips of jaws, above brain, above gas bladder and at lateral cleithral region in early postflexion larvae; additional pigment develops below lower jaw, on hindbrain and nape; series of melanophores on each side of dorsal midline, beginning at mid- body and gradually developing along entire dorsum; series along horizontal septum and along anal fin base; pigment on base of caudal rays and on pelvic and anal rays in some species at late postflexion stage; extensive embedded myoseptal pigment on trunk or tail in postflexion stages of some species. Lampadena. — Fig. 1 2 1 E; photophores (Table 62); pigment above brain, nape, gut, gas bladder; most species have large melano- phores along dorsal midline, with opposing postanal ventral midline melanophores; some species with smaller, more nu- merous melanophores in dorsal and ventral series; embedded pigment above spinal column in some species. Taaningichthys. — Fig. 121F; body slender; lower jaw projects beyond upper; no photophores, even Br, lacking; pigment above brain, in otic region, one to several opposing melanophores at postanal dorsal and ventral midline; late postflexion larvae may develop minute melanophores along each side of dorsal midline; pigment at base of caudal rays; series of embedded melano- phores above spinal column. Ceratoscopelus. — Fig. 122A; eye elliptical in early larvae; pho- tophores (Table 62); pigment above gut; postanal ventral mid- line series in early larvae, coalesces to a single spot in postflexion larvae; C. maderensis has short series at dorsal and ventral midline in caudal peduncle region; embedded pigment above posterior region of spinal column in some species. Lepidophanes. — Fig. 122B; eye small; photophores (Table 62); usually two melanophore pairs at dorsal midline in caudal pe- duncle region and one or two ventral midline melanophores; L. gaiissi has median melanophore above hindbrain and median ventral melanophore below pectoral fin base. Bolinichthys. — Fig. 1 22C; moderately deep-bodied; snout blunt; eye large; photophores (Table 62); sparse pigment; midline spot above brain, embedded otic spot, embedded pigment above gut; some species with a sparse postanal median ventral series that coalesces to a single melanophore; B. distofa.x has a short series on horizontal septum; embedded pigment above posterior re- gion of spinal column in some species. Triphoturus. — Fig. 122F; eye elliptical with choroid mass; pig- ment at tip of lower jaw, at angular region of jaw, at lateral cleithral region; early preflexion larvae have paired lateral gut spots near pectoral fin base and at midgut; anterior pair coalesces to a median position anteroventral to liver, the posterior pair becomes dorsal to gut; pigment above gas bladder; early pre- flexion larvae have postanal median ventral series that coalesces to one or two spots; pigment along margin of preanal finfolds; a single dorsal spot at adipose fin in T. mexicanus; a series of pigment dashes on horizontal septum in T. nigrescens. Stenobrachius. — Fig. 1 22D; gut melanophores and postanal me- dian ventral series similar to Triphoturus; pigment above brain and nape in postflexion stage; late postflexion larvae have embedded melanophores in trunk myosepta and melanophore series on each side of dorsal midline. Parviln.x. — Fig. 122E; head, eyes large; tapered body; gut short Fig. 123. Larvae oi Lampanyclus. (A) L. steinbecki. 6.6 mm; CalCOH Sla. 70.200; (B) L. pusiUus. 1.1 mm; redrawn from Taaning (1918); (C) L. nobilis, 9.6 mm; SEFC, OR II 7343 Sta. 98; (D) L. par\icauda. 7.5 mm. SWFC, Eastropac Op Sta. 023; (E) L. crocodilus. 11.5 mm, redrawn from Tining (1918). MOSER ET AL.: MYCTOPHIDAE 237 Fig. 124. Larvae of Lampanyclus. (A) L. rilleri. 10.1 mm; (B) L. idostigma. 7.2 mm. CalCOFI 6002 Sta. 133.45; (C) L. regalis. 13.0 mm; (D) Lampanyctus sp., 8.7 mm; (E) L. achirus. 13.4 mm; (F^ Lampanyclus sp., 9.4 mm. A, C, D, E from Moser and Ahlstrom (1974); F from Moser(1981). MOSER ET AL.: MYCTOPHIDAE 239 in early preflexion stage, elongates to midbody by flexion stage; in postflexion stage pigment above brain, embedded in otic region, lateral to cleithrum, at anteroventral region of liver; one to several dorsal median melanophores and one ventral median melanophore at caudal peduncle. Lampanyclus. — Figs. 123. 124; body slender; head deep; gut short in early preflexion stage; during preflexion stage gut length- ens to midbody. body deepens and becomes somewhat robust in most species; pigment above brain in most species; postflex- ion larvae develop trunk myoseptal pigment that increases to cover most of the anterior trunk at transformation; at least 6 morphs. L. nohilis-L. parvicaiida-L. oinostigma-L, crocodilus- L. ritteh-L. idostigma: Figs. 123C-E. 124A, B; body and head moderately deep; eyes, jaws, pectoral fins moderate in size; pig- ment may be present at snout, lower jaw. opercle, above gut, anteroventral surface of liver, at dorsal or ventral midline on tail. L. pusillus-L. steinbecki: Fig. 123 A, B; deep, broad body and head, very robust; snout blunt; eyes large; dorsal and anal fins displaced posteriad; pectoral fins moderately large; L. pus- illus heavily pigmented on head, body, pectoral fin base; series along horizontal septum; L. steinbecki with pigment below lower jaw, on opercle. pectoral fin base; series along horizontal septum and embedded pigment on tail in postflexion larvae. L. regalis- L. ater. Fig. 1 24C; deep, large head and body; snout elongate, jaws large, teeth well developed, especially at tip of upper jaw; preopercular spines in some species; dorsal and anal fins dis- placed posteriad; pectoral fins moderate to large; pigment may be present at tips of jaws, embedded in snout, at postorbital and opercular regions, pectoral and pelvic fins; spot at adipose fin in L. regalis; one or two dorsal spots in L. ater. Information on L. ater irom H. Zadoretsky (Dept. Zoology, Univ. of Rhode Island, pers. comm.). L. achirus: Fig. 1 24E; body moderately deep; head and jaws large with snout produced into toothy ros- trum; dorsal and anal fins displaced posteriad; pectoral fins mod- erately large; pigment on tips of jaws, embedded in snout, and present at postorbital and opercular regions. L. lineatus-L. cu- prarius: body moderately elongate; snout elongate, jaws large; head pigment as in L. achirus; L. lineatus pigment consists of numerous melanophores along dorsum and ventrum and at base of caudal rays; L. cuprarius has pigment above gut and an ir- regular bar below dorsal fin. Information from H. Zadoretsky (pers. comm.). (H.G.M.) National Marine Fisheries Service, Southwest Fisheries Center, P.O. Box 271, La Jolla, California 92038; (J.R.P.) The Australian Museum, 6-8 College Street, Sydney 2000, Australia. Myctophidae: Relationships J. R. Paxton, E. H. Ahlstrom and H. G. Moser THE family Myctophidae has usually been placed in the order Myctophiformes (Iniomi. Scopeliformes) since the work of Regan (191 la), who recognized two suborders, the Mycto- phoidea and Alepisauroidea (ateleopodids, given a third sub- order, are currently placed elsewhere). The families Myctophi- dae and Neoscopelidae have long been considered close relatives; they were placed in one family until 1949 (Smith). Although Greenwood et al. (1966:371) relegated the order to a subordinal level within the Salmoniformes, they pointed out that mycto- phoids. and neoscopelids in particular, possess advanced char- acters that indicate they may be ancestral to the paracanthop- terygian radiation. Paxton (1972:54-55)considered myctophids and neoscopelids most closely related to the Chlorophthalmi- dae. with that evolutionary line of the Myctophoidea arising from an aulopid-like ancestor. Moser and Ahlstrom (1970: 141- 142) described the larval similarities in the families Chloroph- thalmidae, Neoscopelidae and Myctophidae. Family Relationships Rosen (1973, 1982) split ofl" the Myctophidae and Neosco- pelidae as a restricted order Myctophiformes which he consid- ered the primitive sister group of both the Paracanthopterygii and Acanthopterygii; the remaining myctophiform families were placed in a new order Aulopiformes. Matsuoka and Iwai (1983) found cartilage in the adipose fin of only the Myctophidae and Neoscopelidae in the five 'iniomous' families they studied. Oki- yama (1974b) studied the relationships of the suborder Mycto- phoidea (sensu Gosline et al., 1966) and based on larval peri- toneal pigment spots and the relationship of abdominal to caudal vertebrae, three familial groups were recognized: Aulopidae- Synodontidae-Bathysauridae, Chlorophthalmidae-Ipnopidae and Neoscopelidae-Myctophidae. Sulak (1977) lumped the Ipnopidae and Bathypteroidae into the Chlorophthalmidae and the Harpadontidae and Bathysauridae into the Synodontidae, considering both groups arose from the Aulopidae; he did not consider the position of the Myctophidae. Schwarzhans (1978) considered myctophids and neoscopelids most closely related and distinct from Aulopiformes on the basis of otolith mor- phology. In his excellent study of the Evermannellidae. Johnson ( 1 982) presented a rigorous analysis of 5 1 characters involving mostly adult but some larval features. He concluded that neoscopelids and myctophids are most closely related to each other, sharing eight derived character slates, but that they were the sister group of four families (Notosudidae, Scopelarchidae, Chlorophthal- midae and Ipnopidae) constituting a chlorophthalmoid group within the Myctophiformes. However, he noted only a single shared derived character in those six families, and it is shared with part of another line. Johnson (1982:95) placed the Aulo- pidae in a second line and all remaining families in the third 240 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Table 63. Characters of the Myctophidae. (0) = plesiomorphic state, (1) = apomorphic state, (2) = different or advanced apomorphic state, 1 = by outgroup comparison, 2 = raised photophore, 3 = gener- alized larva, * = discussed in text. Characters 1. Jaws long (0). moderate (1), short (2)—*. 2. Extrascapulars 2 (0), 1 from fusion (1), 1 from loss (2) — *. 3. Cleithral shelf absent (0), present (1)— 1. 4. Pre 3-9 (0?), 1-2(1?)-*. 5. Larval eyes round (0), narrow ( 1 )— 1 , 3. 6. Dn present (0?), absent (1?)-*. 7. Moderately or strongly hooked teeth in posterior dentary absent (0), present (1)-1. 8. Procurrent ventral rays 5-10 (0), 9-15 (1)-1. 9. Supramaxillary present (0), absent (1)— 1, *. 10. PO4 level (0), raised (l)-2. 1 1 . Pubic plate narrow (0), wide ( 1 )— 1 . 12. PO, and PO, level (0), raised (l)-2. 13. VO, level (0), raised (l)-2. 14. PVO horizontal (0). angled (1), vertical (2)-2. 15. Caudal luminous organs present (0), absent (1)—*. 16. AOa, level (0), raised (l)-2. 17. Pol angled (0), horizontal (l)-2, *. 18. Enlarged teeth in dentary absent (0), present (I)— I. 19. Vertebrae 28-41 (0), 41-45, (1)-1, *. 20. VO, level (0), elevated (1)- 2. 2 1 . Enlarged dentigerous area on anterior premaxillary absent (0), pres- ent (1)-1. 22. Secondary photophores absent (0). present ( 1 )— 1 . 23. Larval gut moderate (0), initially short (1), long (2) — 3, *. 24. Larval trunk myoseptal pigment absent (0), present (1)— 1, 3. 25. Slightly hooked teeth in posterior dentary absent (0), present ( 1 )— I. 26. Caudal luminous organs not sexually dimorphic (0), sexually di- morphic (1)—*. 27. Larval photophores (except Br,) absent (0), present (1)— 1, 3. *. 28. Hyomandibular foramen behind anterior head (0), in anterior head (1)-1. 29. Accessory luminous tissue absent (0), present (1)— 1. 30. Caudal luminous organs any other state (0), homogeneous and translucent ( 1)—*. 3 1 . Procurrent ventral rays without hooks (0), with hooks ( 1 )— 1 . 32. Procurrent dorsal rays without hooks (0), with hooks (1)— 1. 33. Crescent of white tissue on posterior iris absent (0), present (1) — 1. 34. Pol 0(0), 1 (1), 2-3 (2)- 2, *. 35. Dorsal process of opercular head of hyomandibula absent (0), pres- ent (1)-1. 36. SAOs weakly angled (0), strongly angled (I) — 2, *. 37. Larval eyes moderate (0), very large (1)— I, 3. 38. PLO level with PVO, (0), above PVO, (l)-2. 39. SAO 2, close to VO and AO series (0), 2-3 above VO and AO series (1)— 2. 40. Larval pectoral fin moderate (0), large (1)— 3, *. 41. Mouth terminal (0), subtcrminal (1)— 1. 42. Antorbital broad (0), thin (1)— 1. 43. Larval fin fold small (0), extensive (1)— 1, 3. 44. PLO below (0) opposite or proximate to upper pectoral base (1), far above upper pectoral base (2)— 2. 45. Lower pharyngeal teeth conical (0), pegs or plates (1)— 1. 46. Nasal trough-shaped (0), convex (1)— 1. 47. Larval lower pectoral ray not elongate (0), elongate (1)— 1, 3. 48. Gill rakers lathe-like (0), as tooth plates (1)— 1. 49. Dorsal hypurals 4 (0), 3-2 (1). 1 (2)- I. 50. Coracoid fenestra present (0), absent (1)— 1. 51. Double row of isthmus pigment in larvae absent (0), present (1) — 1, 3. 52. Premaxillary teeth conical (0), flattened (1)— 1. 53. Larval pectoral base fan-shaped (0), wing shaped (1)— 1, 3. 54. Larval head pigment present (0), absent (1)— 1, 3. Table 63. Continued. 55. Larval choroid tissue absent (0). present (1)— 1, 56. Larval body width moderate (0), thin (1)— 1, 3. 57. Larval gut uniform (0), bipartite ( 1 )— 1 , 3. 58. Ossified distal pectoral radials (0), 1-7 (1)— 1. 59. CO, keel or ridge absent (0), present (1)— 1, *. group (the alepisauroids plus synodontoids) in his arrangement of the order. We do not have further evidence to present in favour of any of the above hypotheses (but do note the coiled gut of neoscopelid lai-vae resembles the condition found in higher groups). Generic Relationships Paxton (1972) analyzed features of the osteology and pho- tophore patterns of the Myctophidae and presented a taxonomy outlining his views of evolutionary relationships that included two subfamilies (Myctophinae and Lampanyctinae), six tribes (Myctophini, Gonichthyini, Notolychnini, Lampanyctini, Dia- phini and Gymnoscopelini), 28 genera and two subgenera. The Myctophinae was considered the more primitive of the subfam- ilies, while the monotypic Notolychnini was provisionally placed in the Lampanyctinae. In four papers Moser and Ahlstrom ( 1 970, 1972, 1974; Ahlstrom et al., 1976) detailed the larval charac- teristics of all but two genera of Myctophidae and translated their findings into a picture of evolutionary relationships. The relationships proposed by Paxton and Moser and Ahlstrom were strikingly similar overall and in many details. The larval studies supported the recognition of two subfamilies composed of the same genera indicated by the adult analysis, highlighted the enigmatic features of Notolychnus. and recognized three addi- tional tribes in the Lampanyctinae. Notable differences in the conclusions of the two studies included consideration of the Lampanyctinae as the most primitive subfamily by Moser and Ahlstrom, non-recognition of the tribe Gonichthyini ( Tarleton- beama. Loweina. Gonichthys, Ccntrohranchus) as a monophy- letic taxon in the larval study, inclusion of the genera Taan- ingichthys. Lampadena. Bolinchthys, Lepidophanes and Ceratoscopelus in the tribe Gymnoscopelini by Moser and Ahl- strom and the tribe Lampanyctini by Paxton, and recognition of the genera Metelectrona and Parvilux as valid genera on the basis of larval characters, which Paxton had synonymized with Electrona and Lampanyctus respectively on the basis of adult features. Neither study restricted characters to the derived state and the proposed phylogenies were based on overall similarities. The present work will attempt an analysis of derived character states and re-examine the proposed relationships within the family. We have used as character states (Table 63) features of adult osteology and photophore patterns as described by Paxton (1972), and features of larvae as described by Moser and Ahlstrom (1970, 1972, 1974) and Ahlstrom et al. (1976) summarized in Moser et al. (this volume). The distribution of the character states among the genera (we have not considered sub- genera in this analysis) is tabularized (Table 64). The criteria for determining apomorphic character states have been consid- ered by many, including Marx and Rabb (1972) and Zehren (1979:153). We have used three criteria, the numbers of which are listed after each character in Table 63: (1) Outgroup com- PAXTON ET AL.: MYCTOPHIDAE 241 Lampanyctini Diaphini Triphoturus Parvilux Lampanyctus Stenobrachius Lampadena Taaningichthys Bolinichthys Ceratoscopelus Lepidophanes Idiolychnus Lobianchia Diaphus Notoscopelus Lampichthys Scopelopsis Gymnoscopelus Hintonia Lampanyctodes Notolychnus Fig. 125. Phylogenetic diagram of the Myctophidae, subfamily Lam- panyctinae. Numbers refer to the apomorphic characters described in Table 63. Numbers in the middle of vertical lines (e.g., 4, 6) refer to characters for which the apomorphic state is unknown. Underlined numbers refer to apomorphic states unique to all members of a given lineage; bracketed numbers (e.g., 59) refer to apomorphic states that have secondanly reversed in at least one member of the lineage; non- bracketed, non-underlined numbers refer to character states found in all members of a given lineage but also by convergence in at least one other taxon in the family. parison. All previous workers have considered the Myctophidae and Neoscopelidae as sister groups; we have taken the character state in the Neoscopelidae to be the plesiomorphic condition for the Myctophidae. Paxton (1972:57) described the parallel evolutionary trends in the neoscopelids and myctophids, with SoliYonier similar to the Lampanyctinae and Neoscopelus sim- ilar to the Myctophinae. We have largely limited our analysis to those characters which display only one state in the Neosco- pelidae. Where variation occurs within the family, the character is discussed individually below. (2) Linear photophores. We have considered a photophore elevated out of linear series to be apomorphic. One line of support for this decision occurs in the ontogeny of those myctophid species with a larval PLO photophore, which develops on the pectoral base (where it pre- sumably has a different function from that of the adult) and moves dorsally during development (Ahlstrom et al., 1 976:Fig. 4). Also the photophores of Neoscopelus. the only luminous neoscopelid genus, are largely linear. However there is some question of the homology of Neoscopelus and myctophid pho- tophores. O'Day (1972:71) described the ultrastructure of myc- tophid photophores and ". . . confirm(s) Brauer's ( 1 908) original recognition of the close resemblance of photogenic tissue in the Neoscopelidae to that found in the Myctophidae." However Herring and Morin (1978:318) considered photophores of Neo- scopelus and the myctophids to be very different, on the basis Myctophini Gonichthyini 41,42,43 Notolychnus Krefftichthys Protomyctophum Electrona Metelectrona Symbolophorus Myctophum Benthosema s^ Diogenichthys Hygophum Loweina Tarletonbeania Gonichthys Centrobranchus 48,49 Fig. 126. PhylogeneticdiagramoftheMyctophidae, subfamily Myc- tophinae. Numbers are defined as in Fig. 125. of Kuwabara's (1954) description of Neoscopelus compared to that of Brauer (1908). As ventral photophores have evolved independently at least one other time in the stomiiform fishes (Fink and Weitzman 1982:71), the potential for such evolution in deeper water fishes is high enough that one cannot consider their mere existence a case for homology. A study of the ultra- structure of Neoscopelus photophores would be of value. (3) Generalized larvae. The larvae of neoscopelids are highly spe- cialized with a robust body, a large head and jaws with prom- inent teeth, a long gut that may be coiled and large pectoral fins. We do not think these features were present in the ancestors of the two families, and where they are present in the myctophids, consider they have evolved independently. We have used only one such feature, large pectoral fins (40, Table 63) in our anal- ysis. We consider the generalized larva of the myctophid ances- tor had the following characters, based on the distribution of larval features in myctophids and other teleosts: body moder- ately slender, gut slightly S-shaped, extending to about midbody, head moderate in size, eyes round or nearly so, without stalks or choroid tissue, small or moderate finfold and fins and Br, the only larval photophores present. We have used a total of 59 characters, far fewer than the total described in the previous studies. For many we were unable to determine a derived state, as they displayed two or more states or were absent in the neoscopelids. In the osteological descrip- tions small shape differences or classifications of a continuum were often found in both families and were not included. A number of the characters utilized require comment or expla- nation: (I) Jaws are long in Solhomer and short in Neoscopelus, and following our ground rules should not be utilized. However, they appear to be of such fundamental importance, affecting many correlated characters and appearing to represent a major subfamilial difference (Paxton, 1972), that they are included here. Paxton (1972:58) considered short jaws to be primitive, primarily because they occurred in Protomyctophum, thought 242 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Table 64. Character States in the Genera of Myctophidae. The 59 characters are described in Table 1.0 = plesiomorphic state, 1 = apomoi -phic state, 2 - = dif reren t or adva need apo mori )hic state 9 = unk now: 1 or 30th slates. 1 2 3 4 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 n 23 24 25 26 27 Krefftichthys 2 1 Protomyctophum 2 9 1 1 Electrona 2 1 1 1 Metelectrona 2 1 1 1 Benlhosema 2 9 1 9 1 1 Diogenichthys 2 1 1 1 1 Hygophum 2 1 1 Myctophum 2 1 9 1 9 Symbolophorus 2 1 9 1 Loweina 1 2 1 1 2 1 Tarlelonbeania 1 1 2 1 Gonichlhys 1 1 1 Centrobranchus : 1 1 Nololychnus 9 9 9 1 1 1 1 Lobianchia 9 1 1 1 1 1 1 1 1 Diaphus 9 1 1 1 1 1 9 9 1 1 Idiolychnus 9 1 1 1 1 1 9 9 1 9 Lampanyctodes 1 1 1 1 1 Gym noscopelus 2 1 1 1 1 Scopelopsis 2 1 9 1 1 1 1 Lampichlhys 2 1 1 1 Notoscopeliis 2 1 1 1 1 1 1 1 1 Hinloma 9 9 2 1 9 1 1 1 9 9 1 9 Lampadena 9 2 1 1 1 Taaningichthys 1 2 1 Ceratoscopelus 1 2 1 Lepidophanes 2 1 1 Bolinichlhys 2 2 1 1 9 1 Tripholurus 1 2 1 Stenobrachius 2 1 Parvilux 2 Lampanyctus 2 9 9 1 Solivomer 9 9 9 9 9 9 1 9 9 9 9 9 9 Neoscopelus -> 9 9 9 9 9 1 9 9 9 Scopeleng^'s 9 9 9 9 9 1 9 9 9 to represent the most primitive myctophid based on photophore pattern. However Myers (1958) has shown that short jaws have arisen from the long-jawed condition a number of times in teleost ^ olution, and discussed their adaptive advantages. We consider short jaws to be the apomorphic condition within both the Myctophidae and Neoscopelidae, and moderate jaws also to be derived from long jaws. (2) Extrascapulars are single in neoscopelids; therefore two extrascapulars in some myctophids should be the derived condition. However Paxton (1972:58) described how the neoscopelid extrascapular differs in position and shape from that of myctophids. Following Williston's Rule we consider a single extrascapular to be derived from the fusion of two elements, independently attained in each family. In Low- eina the single condition has arisen through the loss of the dorsal extrascapular. (4) With no outgroup with similar photophores for comparison, we are unable to determine whether 1-2 or 3- 9 Prcs is the apomorphic state. However the two character states follow subfamilial limits, and one of the states must be derived and definitive for its subfamily. (6) All myctophids have at least one of the orbital light organs, Dn and Vn, and most have both. We are not sure whether the presence or the absence of a Dn is apomorphic, but one of those stales defines a major line within the Lampanyctinae. (9) Although the Neoscopelidae have a su- pramaxillary, Paxton (1972:62) considered the supramaxillary of some Myctophidae to be an independently derived feature. due to a difference in shape and its required loss at least four times within the family if considered primitive. However, John- son (1974b:205, 1982:79) has shown the presence of supra- maxilla(e) to be primitive in other myctophiforms (sensu lato); the absence of a supramaxilla in myctophids is here considered a derived state through loss. (15) Although caudal luminous organs are not present in neoscopelids, they are present in all but three myctophid genera, where their loss is here considered derived. No other characters indicate that any of the three genera (Diaphus, Gymnoscopelus, Hintoma) are the most primitive in the family. (17) Two or three horizontal Pols are in a linear position and should be considered the plesiomorphic condition. However in those genera with horizontal Pols (Notoscopelus, Lampichlhys and Scopelopsis) the photophores are high, close to the lateral line. We consider the primitive myctophid state to be one with low photophores with none or one Pol (character 34). We therefore consider the horizontal position of Pols to be derived, while noting the state in Hintonia is intermediate between angled and horizontal. (19) Although Johnson (1982: 76) considered a higher number of vertebrae (42-62) plesio- morphic for iniomous fishes, lower numbers of vertebrae in neoscopelids and almost all myctophids indicate the higher number in Gymnoscopelus is a secondary specialization in these families. (23) The larval gut of some neoscopelids is long and coiled, clearly a specialization foreshadowing the condition of PAXTON ET AL.: MYCTOPHIDAE Table 64. Extended. 243 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 1 D 1 I 1 1 1 9 1 9 1 1 9 9 1 9 1 1 1 1 1 1 1 1 1 9 1 1 2 9 1 1 1 1 2 I 1 1 1 2 9 1 2 1 9 9 9 1 1 1 9 2 9 9 1 9 1 1 1 2 1 1 9 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 1 1 1 2 2 2 9 1 9 1 1 2 9 1 9 1 1 9 9 9 1 2 9 1 1 9 1 2 9 1 9 9 2 9 9 9 9 9 9 9 9 1 9 1 1 1 2 9 1 2 9 1 2 9 1 1 2 1 1 2 1 1 2 1 1 2 1 1 2 1 1 2 1 2 9 1 9 9 2 9 9 9 9 9 9 9 9 1 I 1 2 1 1 1 1 2 1 1 1 1 2 1 2 1 1 1 2 1 1 1 2 1 1 1 1 2 1 1 2 1 2 1 1 2 1 1 9 1 2 9 2 1 1 1 1 2 9 2 1 1 9 1 1 9 2 9 9 9 9 9 5 9 9 9 9 9 9 9 9 9 9 9 9 9 9 1 9 9 9 9 9 1 9 some acanthopterygians. Although it could be argued that the short gut that lengthens during development in a few forms of myctophids represents the primitive condition, we consider the primitive myctophid condition a moderate— iengthed gut, with different derived states, short and long. (26) Although the caudal luminous organs are sexually dimorphic in about half the genera, we assume the original caudal organs were not sexually dimor- phic. (27) No photophores are present on the described larvae of Neoscopelus. However the Br, develops in all larval mycto- phids except Taamngichthys and Notolychmis. and its univer- sality indicates it was present in the ancestral myctophid. Other larval photophores however are present in fewer than half of the genera and we consider their presence derived. (30) The strongly developed caudal luminous organs found in Lampa- dena and Taaningichthys are clearly a more specialized state than the relatively unstructured organs found in many other genera. (34) See the discussion of character 17. (36) Although a strongly angled set of SAOs represents a linear position for the first two photophores, we consider this condition developed by the SAO, rising from a lower position in the weakly angled, plesiomorphic position. (59) We consider the absence of a keel or ridge on the fifth circumorbital of Hintonia to be secondarily derived through loss. This is the only character state we have used which is not present in all examined members of the line it defines. We have thus attempted to determine polarity for 25 osteo- logical, 17 larval and 17 photophore characters. We initially attempted a phylogenetic analysis utilizing the distribution of 23 larval characters at the species level. The resulting diagram split some genera into as many as three unrelated lines. We remain convinced that the myctophid genera as currently de- fined by larval morph, photophore pattern and osteology rep- resent monophyletic lines (even though such genera as Diaphus, Lampanyctus, Myctophum and Hygophum may be formally di- vided as subgenera or genera by future work). These genera we use as the starting point in the present study. We have con- structed a phylogenetic tree (Figs. 1 25, 1 26) based on our knowl- edge of the family and used the apomorphic states of the 59 characters to define the various branching points, which is the basis of the following discussion. The subfamily Lampanyctinae is defined by two apomorphies restricted to all members of the subfamily (those characters found in all members of a lineage and nowhere else in the family are underlined in Figs. 125 and 126). the presence of a cleithral shelf and a single, fused extrascapular. The subfamily Mycto- phinae is defined by two apomorphies, short or moderate jaws and narrow larval eyes, but these features are also found in a few genera of the Lampanyctinae. The number of Pre photo- phores defines all members of one of the subfamilies (see dis- cussion of character 4 above). 244 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Notolychmis valdiviae. here considered a monotypic tribe, could not be placed with certainty in either subfamily. Moser and Ahlslrom ( 1 970: 1 38, 1 974:409) and Paxton ( 1 972:6 1 ) dis- cussed the characters and problems of this enigmatic species. With long jaws and the lack of a cleithral shelf both considered plesiomorphies, the apomorphic number of Pre photophores unknown, and the larval eyes variable and intermediate in shape, future work is required to resolve this trichotomy. We recognize three tribes in the subfamily Lampanyctmae (Fig. 125). The tribe Lampanyctini, with nine genera, is defined by the presence of a row of moderately to strongly hooked teeth in the posterior dentary; the only other genus with this feature is the myctophine Diogenichthys. These nine genera are also the only lampanyctines to lack a Dn orbital photophore, but we are unsure if this is a derived state (see discussion of character 6 above). Moser and Ahlstrom (1972) and Ahlstrom et al. (1976: 148) placed five of these genera (Lampadena. Taaningichthys. Bolinichthys. Lepidophanes, Ceratoscopelus) in the tribe Gym- noscopelini, based primarily on larval photophore pattern. Pho- tophores which appear in larvae of Lampanyctinae are essen- tially the same ones which develop in myctophine larvae (Moser et al., this volume) and, if they are adaptive as Moser (1981) has suggested, it is likely that they have appeared in these typical sites independently in a number of lineages. Moreover, these photophores develop at the end of the larval period, if at all, in Bolinichthys and no photophores develop in Taaningichthys larvae. Likewise, the larval pigment characters do not support the inclusion of these five genera in the Gymnoscopelini. In addition to the distribution of hooked dentary teeth and Dn photophores, other features influenced our decision about these five genera. The ischial ligament is medium or long in all Lampanyctini except Taaningichthys (and some species of Dia- phus). while the fifth circumorbital has a ridge or keel in all gymnoscopelines (but is lacking in some species of Diaphus) and no lampanyctines except Bolinichthys (thus the brackets around character 59 in Fig. 125). Finally all of the gymnoscopeline genera except Notoscopelus are restricted to the southern ocean (Moser et al., this volume: Table 59), while the Lampanyctini are found both north and south (except Stenobrachim) of the equator. Placement of the five genera in the Lampanyctini re- quires fewer character reversals and parallelisms. Within the Lampanyctini, the development of larval photo- phores in addition to Br, (character 27) unites the five genera discussed above. We recognize Dorsadena as a subgenus of Lampadena until specimens other than the types are available for osteological study and the larvae are discovered. We have not found an apomorphic character that defines the line in- cluding Stenobrachius. Triphoturus. Lampanyctus and Parvilux. We are recognizing Parvilux on the basis of a weakly angled SAO and larval shape and pigmentation. We consider the tribe Diaphini to be the sister group of the Gymnoscopelini. The relationships among the three genera of Diaphini are not clear. One of us (HGM) has re-examined the specimens on which the larval features of Idiolychmis urolampus were based (see Moser and Ahlstrom, 1974:405-406; Nafpak- titis and Paxton, 1978), and now thinks they could represent Lobianchia gemellari. with the larvae of Idiolychnus still un- known. Two characters shared by Lobianchia and Idiolychnus. the presence of caudal organs and the absence of a luminous patch above the pectoral fin, are considered plesiomorphic, while the absence of a Vn and differences of photophore positions are not clearly apomorphic. The most unequivocal derived state is the presence of a wide pubic plate, indicating Lobianchia and Diaphus are the sister group pair. Within the Gymnoscopelini the proposed generic relation- ships are based almost entirely on characters of the photophores and luminous tissue. No consistent osteological or larval fea- tures define generic groupings. Southern ocean larvae require more study. The larvae of Hintonia are unknown and not enough species of Gymnoscopelus have been studied to ascertain if the subgenus Nasolychnus can be defined by any larval characters. The species of Notoscopelus should also be studied to find sup- porting characters of the subgenus Parieophus. Within the subfamily Myctophinae (Fig. 126), we also rec- ognize three tribes, the Electronini, Myctophini and Gonichthy- ini. The Gonichthyini is clearly a derived lineage, with a num- ber of osteological, photophore and larval characters distinguishing the four genera from the rest of the subfamily. We think the larval specializations of eyes and pectoral fins arose after the split of the two generic pairs. Paxton (1972) was unable to find osteological characters to clearly separate the remaining genera of the Myctophinae into two lineages. We have utilized photophores to distinguish the Myctophini from the Electronini, while recognizing there is a mosaic of osteological and larval characters within these nine genera. We have little question of the sister group relationship of the generic pairs Krefftichthys—Protomyctophum. Mycto- phum — Symbolophorus and Benthosema — Diogenichthys. However two larval features, thin head and body and a bipartite gut, are shared by Metelectrona and some species of Hygophum. Since we think Hygophum is a monophyletic line, we consider these shared larval features parallelisms that do not indicate common ancestry. Paxton (1972) considered Metelectrona a synonym of Electrona. The description of a second species of Metelectrona (Hulley, 1981), coupled with its larval and pho- tophore characters, convinced us to recognize the genus. Of the 59 derived characters utilized in our analysis, only 20 are restricted to members of the lineage they define, and eight of these are autapomorphic at the generic level. The remaining 39 characters are not found in the apomorphic state in any member of the opposite lineage from the defined branching point, but are found in some members of other lineages within the family. This presumed homoplasy of larval, photophore and even osteological characters indicates that the proposed phy- logeny was arrived at with some difliculty. Ten of our proposed lineages are undefined by derived characters. We think that future work will support our proposed phylogeny, although some details may be modified, and that new, less plastic characters and better definitions of polarity will help resolve the problems. (J. R. P.) The Al'stralian Museum, 6-8 College Street, Sydney 2000, Australia; (H.G.M.) National Marine Fisheries Service, Southwest Fisheries Center, P.O. Box 271, La Jolla, California 92038. Scopelarchidae: Development and Relationships R. K. Johnson THE Scopelarchidae has traditionally been included with the primarily oceanic Alepisauroidei (Marshall, 1955; Gosline et al., 1966; Rosen, 1973; Johnson, 1974b, the most recent complete revision). Johnson (1982) excludes the scopelarchids from the alepisauroids, rejects putative sister-group relationship with the Evermannellidae, and provisionally allies the scope- larchids with the chlorophthalmoids. All scopelarchids are oceanic and meso- or bathypelagic. The majority of known adult specimens were taken in hauls to depths between 500 and 1 ,000 m. For most species there exists no evidence to suggest diel migration, however, Merrett et al. ( 1 973:39-40) present limited evidence for diel migration ("considerably dispersed vertically") in Benthalhella infans. Scopelarchids are relativedly large-bod- ied (to 302 mm SL; Iwami and Abe, 1980). All Scopelarchidae are tubular-eyed predators (see Munk, 1966; Locket, 1970; Muntz, 1976; Johnson, 1982) concentrating most frequently on fish, not capable of engorgement of enormously large food par- ticles (unlike evermannellids, Omosudis, Alepisaurus, Antop- terus and at least some paralepidids). Luminous tissue occurs in Benthalhella infans (Merren et al., 1973) and probably occurs in Scopelarchoides kreffti (Johnson, 1 974b). The family contains 1 7 species arranged in four genera and occurs throughout the world ocean except that no scopelarchid inhabits the Arctic Ocean or the Mediterranean Sea. Among iniomous fishes, the Scopelarchidae is distmguished by the following combmation of characters: ( 1 ) basihyal short to elongate but well-ossified; (2) lingual teeth strong, straight to strongly hooked, invariably pres- ent over basihyal, present or absent over basibranchials; (3) body and postorbital regions of head completely covered with cycloid scales; (4) lateral line scales large, differing distinctively in exact conformation between all species (Johnson, 1974b: Fig. 2); (5) parietal bones, when present, small, widely separated by frontals and supraoccipital; (6) coracoid broadly expanded; (7) two post- cleithra, widely separated in vertical dimension; (8) unossified gap (filled by tube-like structure of fibrous connective tissue) between skull and first vertebral centrum (see Merrett et al., 1973:17); (9) posttemporal unforked; (10) no basisphenoid, or- bitosphenoid, gill rakers, or free second ural centrum; (11) eyes tubular, directed straight upward (except in 3 species where directed dorsoanteriad); (12) larvae with 0, 1 or 3 peritoneal pigment sections. The genera and species are distinguished by gross morphological, meristic, morphometric, osteological, pig- ment and larval characters (Tables 65 and 66). Development Eggs of scopelarchids are unknown. Larvae are known for all species except Scopelarchoides kreffti and developmental series have been illustrated and described (Rosen, 1973; Merrett et al., 1973; Johnson, 1974b; Belyanina, 1981, 1982a;Moser, 1981). Except for limited information on Benthalhella infans in Merrett et al. ( 1 973), osteological description has been confined to adults. Except in Benthalhella. development is direct, adult characters are essentially acquired one by one, with completion of trans- formation at 30 to more than 80 mm SL depending upon the species. Larvae of Benthalhella undergo very rapid (i.e., small size increment) transformation after a prolonged period of growth while retaining larval form (see below). Larvae of most species are known from hauls within the top 100 m and the larvae of a number of species have been taken in the top 50 m. Con- trariwise the larvae of one species, Benthalhella dentata. have not been taken in hauls shallower than 150 m and most were taken in hauls to depths in excess of 500 m. Except possibly the cases oi Benthalhella elongata and B. macropmna (see Johnson, 1974b:228), the distributional ranges of larvae and adults are coextensive. There is no evidence (the data are quite incomplete) for seasonality in reproductive effort. Scopelarchids are syn- chronous hermaphrodites. The following paragraphs describe those characters most ev- ident in the early life history of scopelarchids, including those of value in distinguishing genera and species. Gross aspect (Fig. 127). — Larvae range from extremely elongate and shallow (Benthalhella) to quite short and deep (some species of Scopelarchus and Scopelarchoides). Small larvae are trans- lucent, scaleless, colorless (except for pentoneal pigment sec- tions, when present), with a characteristic "bowed down" an- terior dorsal profile. The body is deepest at the pectoral girdle and the trunk elongate. Anteriorly the hypaxial muscles do not embrace the abdominal cavity walls which are therefore highly translucent. Only the muscles of the pelvic girdle are visibly evident. The abdominal cavity is triangular, deep anteriorly. Peritoneal pigment appears early except in Benthalhella which lacks peritoneal pigment until transformation. The gut is mid- ventral. In larvae the anus is anterior (relative to distance be- tween pelvic fin insertion and anal fin origin) to position in adults, far anterior in some (Benthalhella). The head is very Table 65. Com PARISON OF Selected Meristic Characters among Scopelarchid Species. Lateral Dorsal Anal Pectoral line scales Vertebrae alalus 8-9 20-22 23-26 47-49 46-47 hubbsi 8-9 23-25 21-23 53 49 votucris 9-10 21-24 23-26 48-51 49-51 stephensi 8 20-22 18-20 41-44 42-43 michaelsarsi 7-9 18-21 18-21 40-44 40-44 anatis 7-9 21-26 18-22 45-50 44-49 guentheri 7-8 24-29 18-21 47-52 47-51 danae 6-9 24-27 20-22 50-52 48-50 nicholsi 6-7 20-23 20-23 46-50 45-48 kreffti 9 25-27 23-25 58-59 55-57 climax 7-8 25-27 25 53 49 signifer 9-10 26-29 22-25 49-52 48-49 macropmna 5-6 35-39 25-27 62-65 60-62 dentata 6-8 17-20 21-24 54-58 54-55 elongata 9-10 24-28 19-23 61-65 62-65 infans 8-9 20-26 25-28 55-59 55-58 linguidens 8-9 28-30 24-25 66 64 245 246 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM CAV Fig. 127. Larvae, juveniles and adult of Scopelarchidae. (A. B) Rosenblattichthys volucris. A = 14.5 mm SL, B = 26.0 mm SL, letters refer to pigment spots; (C, D) Scopelarchoides nicholsi, C = 1.5 mm SL, D = 23.0 mm SL, letters refer to larval pigment spots; (E, F) Benthalhella denlata. E = larva, 42.8 mm SL, F = transforming specimen, 53.0 mm SL, arrows indicate position of anus; (G) Scopelarchus guenlhen. juvenile, 48.5 mm SL, DS = dermal pigment stripes; (H) Scopelarchus analis. adult, 1 12.5 mm SL. large and massive, exceeding 30% of the SL in Rosenblattichthys, and large but not as large in other genera. The eye is elliptically narrowed, and initially small in comparison with the size of the bony orbit. The interorbital is initially broad and narrows during transformation. Development of the eyes is described for Ben- thalhella infans in Merrett et al. (1973). The snout is pointed. The mouth is large and low, with teeth appearing in very small larvae. The most striking changes take place during a period of transformation, which, as described below, can either be within a very short interval (ca. 10 mm in Benthalhella dentata) of growth (any statements implying time sequence are based solely on increments of length) as in Benthalhella, or over a long (20 mm) to very long (50 mm) interval. Meristic characters.— Counts of fin rays (Table 65) do not differ between larval and adult specimens. Most scopelarchid species can be uniquely distinguished from all other species on the basis ofmeristic characters alone (Johnson, 1974b: 14). Rosenhlattich- ihys is unique in precocious ossification of the pectoral fin rays, well in advance of the pelvic or median fins (except caudal). In all other scopelarchids the lowermost 5 or 6 pectoral fin rays are the last to be formed and the order of fin ray ossification is caudal > dorsal, anal, dorsal pectoral > pelvic > ventral pec- toral. As in all inioms the caudal is formed of 10 + 9 principle rays. In Scopelarchoides and Rosenhlattichthys the pelvic fins appear as buds on the midlateral abdominal cavity wall, well above the level of the intestine. In Benthalhella and Scopelar- chus the pelvic fin buds appear ventrolaterally, at or beneath the level of the intestine. In Benthalhella (except B. macropinna) the pelvic fin insertion in larvae is distinctly in advance of the dorsal fin origin. In other scopelarchid larvae the pelvic fin insertion is beneath or behind the dorsal fin base (but comes to be slightly in advance of dorsal fin origin in adult Rosenhlattich- thys and distinctly in advance of dorsal fin origin in all adult Benthalhella). The adipose fin develops within the dorsal finfold which extends between the dorsal and caudal fin in small larvae. In adults the adipose fin is inserted over the posterior one-third of the anal fin base (except B. dentata where inserted posterior to a vertical through base of last anal-fin ray). Ventral finfold extending from vent to anal-fin origin in smaller larvae, and is completely reabsorbed in early transformation. Peritoneal pigment sections. — \n all adult scopelarchids (except B. elongata) the gut is enclosed by a uniform tube of brown to jet-black pigment. In larvae this pigment appears in discrete sections (except in Benthalhella where peritoneal pigment is lacking prior to transformation) and in a conformation char- acteristic for each genus or group of apparently related species. All larvae larger than 20 to 22 mm possess peritoneal pigment (except in Benthalhella). One section only, unpaired, forming a saddle-like canopy over the gut, is present in Rosenhlattichthys, Scopelarchoides signifer, and S. clima.x (larvae of S. kreffii are unknown). Three sections, a single anterior section as above and two paired posterior sections are found in Scopelarchoides nich- JOHNSON: SCOPELARCHIDAE 247 olsi. S. danae. and Scopelarchus. However in S. nicholsi and 5. danae the posterior sections appear significantly "later" and appear above (S. danae) or anterior (S. nicholsi) to the pelvic fin bases. In Scopelarchus a.\\ 3 sections appear in near synchrony and the posterior sections appear well to the rear of the pelvic fin bases. In all cases the pigment section(s) expand during trans- formation and for all genera except Benihalhella the completion of transformation can be defined as acquisition of the adult state of a complete and unbroken tube of peritoneal pigmentation. In Benthalbella the first appearance of peritoneal pigment (not in discrete section but uniformly in mesentary dorsal to gut from between pectoral fin bases to behind pelvic fin bases) signals the onset of the period of "rapid" transformation. Other larval pigment.— Jht larvae of Scopelarchoides and Ro- senblattichthys are characterized by the presence of well-defined pigment spots or areas (accessory pigment of Johnson. 1974b; complementary pigment of Belyanina, 1982a) apparent in the smallest (6- 1 2 mm SL) known larvae. The presence and location of spots is uniquely diagnostic for each species possessing them. Pigment spots are present in all larvae of Scopelarchoides and Rosenblattichthys. absent in Benthalbella and Scopelarchus. In Scopelarchoides the middorsal spot, if present, and the mid- ventral spot are entirely behind the adipose base and anal fin base respectively. In Rosenblattichthys the middorsal and mid- ventral (where present) spots are entirely in advance of the bases of these fins. Transformation pigmentation.— Johnson (1974b:20) distin- guishes "dermal" vs "epidermal" pigmentation in scopelar- chids. Dermal pigmentation refers to the major pigment stripes present in some genera and species. These develop "early" dur- ing transformation and persist in the adult. In most cases the dermal pigment comes to be partially or completely overlain by the epidermal pigmentation associated primarily with the scale pockets. Dermal pigment is present in all 4 species of Scopelarchus and in certain Scopelarchoides and Rosenblatt- ichthys, it is absent in Benthalbella. The subequal pigment stripes oC Scopelarchus (Fig. 127), situated above and below the lateral line, are diagnostic for the genus. Gut morphology.— \n all scopelarchids the stomach is a heavily muscularized, greatly elongate blind pouch. In small larvae the stomach does not reach the pelvic fin base, but it expands pos- teriad during transformation, very "rapidly" so in Benthalbella. and in all adults extends to or nearly to a vertical through the anus (which in all is closely-adjacent to the anal fin origin). Johnson (1974b) and Wassersug and Johnson (1976) note that the tremendous expansion of the stomach allows ingestion of fairly large particles and hypothesize that the blind pouch ar- rangement is a device for maximal recovery of food energy. Transformation. — Larvae of Benthalbella undergo rapid trans- formation after a prolonged period of growth while retaining larval form. The onset of transformation (size of smallest known transforming specimen = 49.6 mm SL in B. dentata; 89. 1 mm SL in B. elongata: 55. 1 mm SL in B. infans; 65. 1 mm SL in B. macropinna; no transforming specimens of B. lingutdens are known, but the largest known larva is 85.5 mm SL) is signalized by appearance of a lens pad, appearance of peritoneal pigment, and invasion of the abdominal body wall by musculature. Other changes occurring during transformation include rapid elonga- tion of gut and stomach, "migration" of anus from just behind pelvic fin base to just anterior to anal fin origin, appearance of gonad, appearance of scales (especially lateral line scales), ap- pearance of head and body pigmentation, reabsorption of ven- tral adipose fin. great restriction of base of dorsal adipose fin, ossification of vertebral column, change (from dorsally convex to dorsally concave) in curvature in vertical plane of anterior portion of vertebral column (Merrett et al., 1973; Johnson, 1974b). The result is a miniature adult at the end of a trans- formation period covering as little as 1 mm of growth (Johnson, 1974b:68). In other scopelarchid genera these and other adult characters are acquired essentially one by one over an increment of growth ranging from 15 to 50 or more mm SL [in most transformation occurs over an actual size (SL) range of 1 5 mm to 40 or 50 mm]. Implications of changes in morphology during transformation in terms of activity, buoyancy, feeding and other aspects of biology are discussed for B. infans in Merrett et al. (1973). Relationships The scopelarchids were poorly known until the completion of Johnson's ( 1 974b) revision. Currently recognized are 1 7 species grouped in 4 genera. Phylogenetic analysis involving hypothe- sized derived states of 1 9 characters or character complexes (Table 66) supports allocation of species among 3 of the 4 genera. As will be shown, Scopelarchoides remains a problem. In the listing that follows characters are given a character number (de- rived state number). Documentation of character state catego- rization and hypothesized polarity are given in references listed in the key to Table 66. Of the 19 characters for which polarity is indicated, 6 involve larval features (Table 65: 18, 19, 20, 22, 23, 24). Of 13 adult characters, 5 represented noval autapo- morphies (Table 65: 1,4, 13, 14, 15), 3 occur in a sequence of 3 or more steps (Table 65: 5, 11, 16), and 5 represent reductive characters (Table 65: 6, 7, 8, 9, 12). Rosenblattichthys is dis- tinctive in having a greatly enlarged head in larvae 19 (19) and precocious development of the pectoral fins 20 (20). A single reductive character 8 (7) putati vely links the remaining 1 4 species of scopelarchids. Scopelarchus is specialized in having subequal dermal pigment stripes above and below the lateral line 4 (2), unique support of the first epibranchial 16(17); unique confor- mation of the three peritoneal pigment sections 22 (24), and in three reductive characters 9 (8), 12(11), and 23 (25). Scopelar- chus analis is linked with 5. michaelsarsi and 5. Stephens! by one reductive character 11 (10). Scopelarchus stephensi and S. michaelsarsi are linked by a reduced number of vertebrae 5 (3) and by early onset and completion of metamorphosis 24 (26). Benthalbella is specialized in having delayed but then extremely "rapid" metamorphosis 24 (27) and in three reductive char- acters 6 (5), 22 (21), and 23 (25). Linking Benthalbella dentata. B. infans, B. lingutdens and B. elongata is the unique presence of a hooklike process on the urohyal 15 (14) and two reductive characters 9 (8) and 1 1 (9). In dealing with the 5 species included by Johnson (1974b) in the genus Scopelarchoides the evidence available (Table 66, Fig. 1 28) suggests that this group is both unnatural and paraphyetic. Linking 5. nicholsi, S. danae and Scopelarchus are unique se- quential and fully correlated novel autapomorphies: support of the first epibranchial character 16 (states 15 - 16 - 17), and number and position of peritoneal pigment sections, character 22 (states 22 - 23 ^ 24). Further linking 5. nicholsi with S. danae and Scopelarchus are relative size of the opercle and 248 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Table 66. Characteristics of the Scopelarchidae. Characters and character states are defined and listed below. Positive integers indicate derived states, zeroes indicate primitive states, letters denote states of characters where polarity could not be determined. Rosenhlalltcblh ■< Scopelo rchus ■s copeiarchotdes Benlhalhella Char- volu- michael- guenih- nich- sigii:- mac ra- Imgui- acters alatus hubbsi cris stephensi sarsi a nail s en danae flist kreffn ■Imm.x fer pt una denlala ftongata inlans dens Gross morphology 1 ?1 1 2 a a a b b b b b b a b b a a a a a 3 c a a b b b b c c c c b c c c c ?c 4 2 2 2 2 Meristic characters 5 3 3 4 4 4 4 4 4 Osteological characters 6 ?0 ?0 ?0 ?0 ?0 5 5 5 5 ?5 7 ?0 ?0 ?6 6 6 6 6 6 ?0 ■'0 ?0 8 ?0 7 7 7 7 7 7 7 ?7 7 7 7 7 7 7 9 8 8 8 8 8 8 8 8 10 b b b a a a a a a b b a b b b b b 11 10 10 10 9 9 9 9 9 9 9 12 ?0 ?11 11 11 11 ?0 ?0 13 12 12 12 12 12 12 14 ?0 13 13 13 13 13 13 13 13 15 ?0 ?0 ?0 ?0 14 14 14 14 16 ?0 ?0 ?17 17 17 17 16 15 ?0 ?0 Developmental characters 17 b a a a a a a a a a b b b b b b b 18 18 18 18 18 18 ?0 19 19 19 19 ?0 20 20 20 20 ?0 21 a a a b b b b a a ?a a a b b b b b 22 24 24 24 24 23 22 ?0 21 21 21 21 21 23 25 25 25 25 ?0 25 25 25 25 25 24 26 26 ?0 27 27 27 27 27 KEY: Character state classification and hypothesized polanty based on detailed information presented in Johnson (1974b. 1982) and Iwami and Abe (1980). Both characters (boldface, in brackets) and character states (in parentheses) are numbered sequentially. Gross morphohgy —{\\ Luminous tissue is (0) absent. (1) present; |2| pelvic-fin insertion is (a) antenor to dorsal-fin ongin. (b) prasienor to dorsal-fin origin; |31 length of pectoral fin is (a) subequal to, (b) distinctly longer than, (c) distinctly shorter than length of pelvic fin; |4| dermal pigment stnpes as equal or subequal stripes above and below lateral line are (0) absent. (2) present. Merislic characiers —\5\ Modal number of vertebrae. Occurs within span of (3) 40 to 44. (0) 45 to 51, (4) 54 lo 65, hypothesized character slate sequence: 3-0-4. Oiieologica/ characlers —\(t\ An anterovcntrally directed prong from opisthotic reaching or nearly reaching border of proolic is (0) present. (5) absent; |7| panental bones are (0) present, (6) absent; |8] supraorbital bones are (0) present. (7) absent; |9| antorbital bones are (0) present. (8) absent; |10| Ethmoid process on first infraorbital bone is (a) present, (b) absent. |ll| Supramaxillary bones are (0) large, one-third to one-fourth the maxillary length; (9) splinllike. less than one-ninth of maxillary length, (10) absent: hypothesized character state sequence 0-9-10, [12] Discrete postenor arm of hyomandibular bone which articulates with opercle is (0) present, (11) absent, represented only by a rounded ridge. 113) Opercle— (0) subequal to or less than, (12) distinctly great than— subopercle in size. 1I4| (0) basibranchial teeth present, basihyal short, (13) basibranchial teeth absent, basihyal long. |I5| Hook-like process on anterodorsal surface of urohyal is (0) absent. (14) present, |16| (0) suspensory phar>ngobranchial (PBl) present, uncinate process (UP) of first epibranchial (EBl) and second pharyngobrancial (PB2) connected by a ligament; (15) FBI lacking, support of EBl near proximal end of PB2 — UP of EBl and PB2 connected by a ligament; (16) PBl lacking, support of EBI near middle of PB2. no UP. (17) PBl lacking, support of EBl at point of articulation between PB2 and EB2. no UP. Hypothesized character state sequence: 0-15-16-17 Developmental characiers — |I7| Dermal pigmentation as defined in text is (a) present, (b) absent, [181 Adipose fin (0) remains elongate (extending antenad to over antenor anal-fin base) throughout transformation penod, (18) is reabsorbed early in transformation, exhibiting adult proportions in specimens 20 to 22 mm SL and larger, II9| Head length in larvae (=28 mm SL) (0) not exceeding 30% SL. (19) exceeding 30% SL. |20| Pectoral fin (0) not precocious, all other fins with completely differentiated rays pnor to ossification of ventralmost rays {at least) of pectoral fin, (20) precocious, all rays completely differentiated pnor to formation of complete complement of rays of all other fins (except caudal fin), (211 Pelvic fin buds (a) form midlaterally, well above level of intestine, (b) form ventrolaterally, at or below level of intestine, |221 Number of pentoncal pigment sections in larvae (2 1 ) = 0. (0) = I , (22) = 3, the postenor paired sections appeanng much later in development than the single antenor section, and appeanng entirely antenor to the pelvic-fin bases. (23) = 3. the postenor paired sections appeanng much later in development than the single antenor section, and appeanng over the pelvic-fin bases. (24) = 3. the postenor paired sections app)eanng in near synchrony with the single antenor section and appeanng entirely posterior to the pelvic-fin bases. Hypothesized character state sequence: 21 - - 22 - 23 - 24, |23| Other pigment spots or areas (as defined in text) are (0) present. (25) absent, |24| Transformation is (26) gradual, onset at 12-14 mm SL or smaller, completion at 30-35 mm SL or smaller. (0) gradual, onset at 16-22 mm SL or larger, completion at 40-60 mm SL (most species, R alaius is extreme with onset at 9-10 mm SL and not yet complete in 6 (39,9-80. 1 mm SL) juveniles examined by Johnson { 1 974b)], (27) abrupt; onset at 49 6-89 I mm SL or larger, completion at 68.3-98.6 mm SL or larger (size for both onset and completion of metamorphosis vanes among the 5 species of Benlhalhella). Hypothesized character states sequence: 26 - - 27. subopercle 13 (12) and two reductive characters 7 (6) and 1 1 (9). Further linking S. danae 'wiih Scopelarchus is a unique early restriction of the base of the dorsal adipose fin 18 (18). I am convinced that the characters previously detailed warrant ge- neric level recognition for the group of 4 species assigned to Scopelarchus. Thus Scopelarchoides (type-species S. nicholsi) should be restricted to S. nicholsi and S. danae. This leaves the three species currently assigned to Scopelar- choides, viz. S. signifer, S. climax, and 5". kreffti. These three share no known derived character unique to just this group. They share a single presumably derivative character— loss of basibranchial teeth, extension of length of basihyal tooth row 14 (13)— with Benlhalhella but as noted by Johnson (1974b: 204) this may represent adult retention of a larval character state common to all scopelarchids. Scopelarchoides kreffti. a subtropical convergence species, shares with Benlhalhella an increase in the number of vertebrae 5 (4) and probably shares with B. infans the presence of luminous tissue 1(1). Most os- teological characters arc unknown for 5. climax and S. kreffti (as a resuU of paucity of available material) and the larvae of JOHNSON: SCOPELARCHIDAE 249 -26 - 3 —25 — 24 — 17 — 11 8 — 2 — 14 9 - 8 23 —18 —16 -22 -15 -12 - 9 - 6 -20 -19 -27 -25 -21 - 5 Fig. 128. Proposed relationships among scopelarchid species based on adult and larval characters. Integers indicate derived character states, listed in Table 66, possessed by taxa above indicated point in dendrogram. S. kreffti are unknown. I would argue that the specializations oi Benthalhella, especially in larval characters relating to a unique, rapid pattern of transformation preclude addition of 5. signifer, S. climax, and presumably 5. kreffti to Benthalhella. But with S. climax and S. kreffti very poorly known and with the only "character" uniting this "group" of three being that they are "left over," I remain with my 1974b (p. 217) compromise. Uniting all 5 species of "Scopelarchoides" and diagnostically separating them from Scopelarchus and Benthalhella are de- velopment and conformation of accessory pigment spots char- acter 23, and lateral appearance of the pelvic fin bud. character 21. It is possible that the state exhibited by Scopelarchoides larvae is primitive in both cases (I doubt that lateral appearance of the pelvic fin buds is primitive) but until this can be shown through adequate outgroup comparison and until S. climax and S. kreffti are better known, I refram from attempting the de- scription of an additional genus. Thus, for now, the possibly paraphyletic genus "Scopelarchoides" is retained. A summary of the contribution of 6 ontogenetic characters to this analysis is presented below. Dermal pigmentation (character #/7j. — Dermal pigmentation and/or dermal pigment stripes are found in all scopelarchid genera except Benthalhella. however, the fixation of such pig- ment into subequal stripes above and below the lateral line is diagnostic of and unique to the four species of Scopelarchus. This fixation is regarded as autapomorphous for this genus. .Adipose fin (character #18).— Scopelarchoides danae shares with Scopelarchus an early reabsorption of most of the adipose (fin, resulting in restriction to essentially adult proportions of the base of this fin in specimens 20-22 mm SL. In other Scopelar- choides as in Benthalhella and Rosenhlattichthys the adipose fin remains elongate, to over the anterior anal fin rays, throughout transformation, assuming adult proportions in specimens >30 mm SL. In combination with other characters uniting .S. danae with Scopelarchus (Fig. 128) fixation of early restriction of the dorsal adipose base is regarded as apomorphous for this group. Head length (character #19}. — The head in larval Rosenhlatt- ichthys is unusually large, deep and massive, the head length exceeding 30% of the SL. The head length in other scopelarchid larvae does not exceed 30% of the SL and this is apparently the caseinchlorophthalmoids(Taning, 1918;Okiyama, 1972, 1974b, 1981) and most alepisauroids (Rofen, 1966a: Johnson, 1982). Larvae of Omosudis and .Alepisaiirus do exhibit exceptionally large heads (Rofen, 1966b). The fixation of this character in Rosenhlattichthys alone among scopelarchids is presumed to be apomorphous. Pectoral fin development (character #20). — The order of fin ray differentiation varies within and between iniomous families. Precocious pectoral fin development is unique to Rosenhlattich- thys among scopelarchids. It is also found in ipnopids (Okiyama, 1972, 1981) and myctophids (Moser and Ahlstrom, 1970) but 250 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM not evermannellids, Omosudis. or chlorophthalmids (Tuning, 1918; Rofen, 1966b; Johnson, 1982). It is presumed that pre- cocious pectoral fin development in Rosenblattichthys is the derived state. Peritoneal pigment sections (character #22). — For an overview of the distribution of peritoneal pigment sections in inioms see Johnson (1982) and the account of the Evermannellidae in the present work. The single, transverse section seen in Rosenblatt- ichthys. Scopelarchoides climax. S. signifer and presumably 5. kreffti is here considered the primitive state. Loss of peritoneal pigment in the larvae of Benthalbella is clearly apomorphous. The single and paired conformation of the 3 sections in Sco- pelarchoides nicholsi. S. danae and Scopelarchus is unique to this lineage among inioms. The seemingly sequential progres- sion of states 22 - 23 - 24 (Table 66: character 22) and the correlation of these states with states 15-16-17 of character 16 strongly reinforce the concept of monophyly for this lineage. Larval pigment spots (character #23). — Deep-lying pigment spots or areas occur widely among iniomous fishes (TSning, 1918; Gibbs, 1959; Anderson et al., 1966; Rofen, 1966a; Moser and Ahlstrom, 1970; Johnson, 1982) and their presence is here pre- sumed to be primitive. As noted above, the position and relative size of the spots differs between and is diagnostic of Scopelar- choides (all 5 species) vs Rosenblattichthys. Transformation (character #24).— Larvae of Benthalbella are unique among scopelarchids and possibly among inioms in achieving very large size— 50 to 100 mm or more (varying by species) while retaining a purely larval form and then exhibiting a very "rapid" (based on size increment relative to total size) transformation. This pattern is regarded as autapomorphous for this genus. Larvae of two central- water species of Scopelarchus. S. stephensi and S. michaelsarsi. exhibit a gradual transfor- mation typical for most inioms, but, relative to other scopelar- chids, exhibit onset and completion of transformation at sub- stantially smaller sizes. This is regarded as an apomorphous feature linking these two species (as does the possibly redundant character 5, reduction in number of vertebrae). Johnson ( 1 982:62-10 1 ) reviews some 49 characters seemingly related to the question of sister-group relationship of the sco- pelarchids and evermannellids. Found were derived states in eight characters— multiple peritoneal pigment sections, lateral attachment of dermosphenotic, restricted insertion of RAB (Ro- sen, 1973) muscle, reduction in number of supraneurals, and loss of the following: sclerotic bones, antorbital bones, tooth- plate of second pharyngobranchial and basibranchial denti- tion—characteristic of all alepisauroids (Alepisauridae, Ano- topteridae, Evermannellidae, Omosudidae, Paralepididae) but not the Scopelarchidae (at least primitively). Also found were 5 derived states characteristic of the Evermannellidae + Alep- isauridae + Omosudidae but not the Scopelarchidae. viz. pos- session of eight infraorbital bones, reduction in number of ep- urals and loss of the following: body scales, lateral line scales, suspensory pharyngobranchial. Admittedly many of the features listed are "loss" characters and thus potentially worrisome, but why should they uniformly be absent in the groups indicated and not in the Scopelarchidae if their correlated loss is not indicative of relationship? On the basis of the large number of derived states shared among alepisauroids but not shared by scopelarchids Johnson (1982) excludes the scopelarchids from the alepisauroids and links them (tentatively) with chloroph- thalmoids. Only a single derived state— gap in ossification between first centrum and the skull— links the scopelarchids with chlorophthalmoids, but this feature is found in no alepi- sauroid. It should be reemphasized that the characters discussed in Johnson (1982) were specifically chosen to explore the hy- pothesis of sister-group relationship of evermannellids and sco- pelarchids— a notion rejected. Many additional characters need to be studied for any rigorous analysis of iniom relationships. It is clear that the contribution of larval characters to this anal- ysis will be great. Field Museum of Natural History, Roosevelt Road at Lake Shore Drive, Chicago, Illinois 60605. Evermannellidae: Development and Relationships R. K. Johnson THE Evermannellidae is one of five families included by Johnson (1982, the most recent revision) in the primarily oceanic Alepisauroidei. Excluded from this group are the Sco- pelarchidae, long the supposed sister group of the evermannel- lids, but tentatively allied by Johnson with the chlorophthal- moids. All evermannellids are oceanic and mesopelagic, occupying (as juveniles and adults) a wide vertical range in the upper 1,000 m, and are not known to exhibit diel vertical mi- gration. Evermannellids are relatively large-bodied (to 184.5 mm SL) predators, capable of engorging large food particles, and concentrating most frequently on fish although Coccorella may more frequently prey on squid. The family contains 7 species arranged in 3 genera. Evermannellids are distinguished among other alepisauroids by the following combination of characters: ( 1) an externally visible tripartite division of the tail musculature with the epaxial and hypaxial muscles separated by a midlateral band of muscle tissue, the lateralis superficialis; (2) lack of scales; (3) greatly reduced, edentate basihyal; (4) restriction of gill teeth to ceratobranchial of second arch; (5) presence of tubular or semitubular eyes in 6 of 7 species; (6) lack of external keels on body. The genera and species are distinguished by gross mor- phological (eye, laterosensory pores, gut morphology, luminous tissue), meristic, morphometric, osteological, pigment and lar- val characters (Table 67). JOHNSON: EVERMANNELLIDAE 251 Fig. 129. Larvae and juveniles and Evermannellidae. (A) E. balbo. showing larval phase pigmentation, D 3553 II, 8-10 mm SL; (B) E. indica. showing juvenile phase pigmentation, ORSTOM CY III-5, 28.0 mm SL; (C) O. normalops. illustrating larval phase pigmentation and multiple peritoneal pigment sections (shown in solid black), UH 73/8/38, 10.5 mm SL; (D) C. allantica. showing juvenile phase pigmentation, RHB 2960, 6.3 mm SL; (E) C. allantica, arrow shows location of cephalic extension of pyloric caecum, ACRE I2-18A, 25.2 mm SL (peritoneal pigment sections not shown). Development Eggs of evermannellids are unknown. Larvae are known for all species and developmental series have been partly illustrated and described (Schmidt, 1918; Rofen, I966d; Wassersug and Johnson, 1976; Johnson, 1982). Osteological examination has been confined to adults. Development is direct, transformation gradual, adult characteristics are acquired essentially one by one but for the most part such acquisition is complete in specimens exceeding 30 mm SL. For all species the great majority of larval specimens has been taken in the upper 100 m but only the larvae of three species (Evermannella balbo, E. indica, Odontostomops normalops) have been commonly taken in hauls to 50 m or less. The distributional ranges of larvae and adults are coextensive and there is no evidence (the data are very incomplete) for seasonality in re- productive effort. Evermannellids are synchronous hermaph- rodites. The following paragraphs describe those characters most ev- ident in the early life history of evermannellids including those of value in distinguishing genera and species. Gross aspect (Fig. /29A — Larvae and smaller juveniles of all three genera are similar in general proportions and in having a relatively smaller eye, smaller lens, broader interorbital, and larger snout than larger juveniles and adults. The body is deepest just behind the pectoral fin base. The anterior dorsal profile descends gradually and is not bowed down. The eye in larvae of Evermannella and Coccorella but not Odontostomops is el- liptically narrowed, broader dorsoventrally than antero-poste- riorly. The gut cavity is essentially triangular and quite deep anteriorly. The snout is pointed, the mouth large, and teeth appear in very small larvae. The most striking changes in body proportions, in all evermannellid larvae, are correlated with the transition from individuals with a "larval phase" pigment pat- tern to those with a "juvenile phase" pigment pattern (see pig- mentation, below), with the result that individuals exceeding ca. 25 mm in the latter category are essentially miniature adults. Meristic characters.— Counts of fin rays (Table 67) do not differ between larval and adult specimens. The caudal is the first fin to form, it develops 10 + 9 principal rays, as in all Aulopiformes and Myctophiformes (sensu Rosen, 1973). Next to form, in order, are the dorsal, pelvic, anal and pectoral fins. The pelvic fins do not greatly change position during ontogeny, they appear ventrolaterally beneath the posterior half of the dorsal fin and are inserted beneath the anterior half of the dorsal fin in adults. An adipose fin connects the incipient dorsal fin with the caudal fin in small larvae but loses this connection and shrinks in extent with growth of the individual, inserted over posterior one-third of anal fin base in adults. There is apparently no variation in the above-described features among evermannellid larvae. Peritoneal pigment sections (Fig. 129). — In all adult everman- nellids the gut is completely enclosed by a uniform lube of dark brown to black peritoneal pigment. In larvae, this peritoneal pigment appears in discrete sections. In Odontostomops there are 12 or more peritoneal pigment sections, typically 13 to 15. In Evermannella and Coccorella there are invariably 3 sections, one centered over and medial to the pectoral fin insertion, one centered (or nearly so) under the dorsal fin insertion, and one (roughly) centered between the posteriormost pelvic fin ray base and the anal fin origin. In all cases the sections are unpaired and are connected broadly over the dorsal surface of the stom- ach. In small larvae the sections form canopy-like continuous 252 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Table 67. Characteristics of the Evermannellidae. In the list that follows only characters useful in distinguishing evermannellid taxa are included. Those characters also included in phylogenetic analysis are numbered; presumed primitive states denoted by 0. presumed derivative states by integers. Coccorella allantica Coccorella Everntannella Evermannelta Evermanneita Evermannella Odonloscomops atrata ahblromi halbo tndtca megatops normatops Gross morphology (1) Eye (each state includes a suite of presumably cor- related features listed in Johnson, 1982, p. 68): (0) nontubular, (I) semitubular, (2) tubular (2) Pyloric caecum with cephalic extension: (0) absent, (3) present (3) Luminous tissue, associated with ventral wall of intestine and pyloric caecum; (0) absent, (4) pres- ent (4) Medial snout-pad pore (Johnson, 1982, p. 8) is: (0) present, (5) absent Meristic characters (5) Dorsal fin, modal number of rays; (0) 12 or 13, (6) 10 or II (6) Number of lateral line segments: (0) S43, (7) £34,(8) SI8 —Anal fin rays — Vertebrae Morphometric characters (as thousandths of SL) —Body depth at anal-fin origin —Horizontal diameter of eye — Vertical diameter of eye — Interorbital width — Length of longest palatine tooth Osteological characters (7) Basisphenoid; (0) present, (9) absent (8) Ethmoid cartilage: (0) not forming orbital septum, (10) considerably expanded posteriorly forming an orbital septum (9) Supraorbitals; (0) present, (II) absent (10) Vertically elongate fossa centered at dentary sym- physis; (0) absent, (12) present (11) Jaw and palatine teeth; (0) dentary teeth in two se- ries, at least some dentary and palatine fangs barbed, (13) dentary teeth uniserial, all fangs un- barbed (12) Basihyal toothplate; (0) covers dorsal and dorsolat- eral surface of basihyal, (14) covers only posterior 2/3 of dorsum of basihyal, (15) absent (13) Toothplate of fourth pharyngobranchial: (0) bears teeth, ( 1 6) edentate (14) Toothplate of fifth ceratobranchial: (0) bears teeth, (17) edentate Developmental characters ( 1 5) Number of peritoneal pigment sections; (0) three, (18) twelve or more (16) Juvenile phase pigmentation; (19) characterized by development of three distinct rows of very large melanophores, each row associated with one of three main divisions of tail musculature, (0) juve- nile phase pigmentation not as above, with many more melanophores and no distinct trilateral pat- tern 1 1 2 2 2 2 3 3 4 4 5 7 7 S 8 8 8 26-30 27-29 29-32 33-37 27-31 29-31 30-35 48-50 45-47 47-49 52-54 48-52 48-50 48-52 144-191 171-210 173-200 145-181 136-173 148-162 135-170 40-65 47-65 67-81 52-72 49-93 74-85 27-42 42-76 54-70 69-87 59-81 60-97 86-110 28-40 32-47 47-61 17-26 9-19 8-20 4-17 36-52 71-96 80-100 46-69 54-69 48-73 61-69 53-69 10 10 11 13 13 12 12 12 12 14 15 16 17 17 17 17 17 18 19 19 19 19 JOHNSON: EVERMANNELLIDAE 253 sheets over the dorsal and dorsolateral margins of the gut and these sections expand ventrad as well as longitudinally with growth. In specimens larger than 35 to 45 mm SL the peritoneal pigment sections coalesce to form the complete gut-enclosing pigment tube characteristic of adults. Other pigmentation (Fig. 129).— The major pattern of body pig- mentation in evermannellid larvae occurs in two phases, a larval phase and a juvenile phase, with a gradual transition between the phases. In smaller larvae (less than 12-15 mm SL) the most prominent body pigmentation consists of a pattern of pigment bands arranged along the myosepta. Typically these bands are arranged in groups (symmetrically distributed in epaxial and hypaxial myotomal bands in the tail region, nonsymmetrical and predominantly epaxial in the trunk region), resulting in a characteristic barred appearance. In larvae larger than 1 2 to 15 mm SL the body pigmentation characteristic of adults begins to appear. In Odontostomops juvenile phase pigmentation is characterized by the development of numerous highly punctate melanophores generally distributed over the head and body. In Evermannella the juvenile phase is typically characterized by the development of three rows of very large melanophores, each row associated with one of the 3 main divisions of the trunk/ tail musculature. The median row, that associated with the lat- eralis superficialis, is limited to the tail. Both of the other rows, epaxial and hypaxial, extend the length of the body, from the posterior border of the head (or nearly so) to the caudal peduncle. In Coccorella the juvenile phase pigmentation tends to be in- termediate in state between that of Odontostomops and Ever- mannella. the developing melanophores tend to be larger and more prominent than in Odontostomops. but much more nu- merous and not arranged in rows as in Evermannella. Body pigmentation in juveniles larger than 25 to 30 mm SL is similar to that in adults. Development of adult pigmentation in ever- mannclid larvae is associated with gradual (all statements im- plying time course are based solely on size increments) disap- pearance of the larval myoseptal pigment bands. Four of the seven evermannellid species (Coccorella atlantica. C. atrata. Evermannella megalops. Odontostomops normalops) are highly melanistic as adults. In Evermannella balho. E. indica. and es- pecially E. ahlstromi the pigmentation in adults tends to be much more mottled, with numerous, variably-sized melano- phores (some very large) on a light brown (in alcohol) ground color. Obscured in adults is the longitudinal tnlateral melano- phore pattern characteristic of juveniles. Gut morphology (Fig. 129). — \n all evermannellids the stomach is a heavily muscularized blind sac. The stomach expands pos- teriad with larval growth reaching its full extension (to a vertical just behind the pelvic fin base) in specimens exceeding 20-25 mm SL. Larvae of Coccorella are distinguished by the unique possession of a pyloric caecum that expands anteriad with growth and enters the head in larger larvae, juveniles and adults (Fig. 129E). The caecum is visible as a short, blind, bud-like sac on the ventro-anterior margin of the intestine in the smallest known larvae of Coccorella. Wassersug and Johnson (1976) describe in detail the structure and development of this remarkable or- gan. Neither Evermannella nor Odontostomops nor (as far as is known) any other alepisauroid possess a pyloric caecum. rra«s/orwa?/o/i. — Development of juvenile phase pigmentation signals the onset of transformation in all evermannellid larvae. 2,8,12,19 Fig. 1 30. Proposed relationships among evermannellid species based on adult and larval characters. Integers indicate derived character states, listed in Table 67, possessed by taxa above indicated point in dendro- gram. Transformation in Evermannellidae is gradual, adult characters are essentially acquired one by one, and there are no abrupt and radical changes in morphology. In all evermannellid species, individuals larger than 25 to 30 mm SL are (except for final fusion of pentoneal pigment) essentially miniature adults and can be distinguished readily on the basis of adult characters (e.g., eye morphology, presence or absence of dentary fossa, posterior extent of lateral line, arrangement of cephalic latero- sensory pores, dentition, pigmentation, meristic and morpho- metric characters). Final fusion of the peritoneal pigment sec- tions occurs by about 35 mm SL (Coccorella. Evermannella) or by about 45 mm SL (Odontostomops). Relationships The evermannellids were poorly known until the completion of Johnson's (1982) revision. Currently recognized are 7 species in 3 genera (Fig. 130). Phylogenetic analysis involving presum- ably derived states of 1 6 characters or character complexes sup- ported previous allocation of species among the 3 genera. In the listing that follows characters are given as character number (derived state number). Of the 1 6 characters, 2 involved larval features (Table 67: 15, 16). Of the 14 adult characters, 5 rep- resented novel autapomorphies (Table 67: 2. 3, 8, 10, II), 3 exhibited a sequence of 3 steps (Table 67: 1,6, 12) and 6 rep- resent reductive characters (Table 67: 4, 5, 7, 9, 13, 14). Odon- tostomops is specialized in having 12 or more serially arranged pentoneal pigment sections 15 (18) and in two reductive char- acters 7 (9) and 9 (11). Coccorella exhibits autapomorphies in four characters: cephalic extension of pyloric caecum 2(3), pres- ence of luminous tissue 3 (4), posterior expansion of ethmoid cartilage 8 (10), arrangement and morphology of dentary and palatine teeth II (13) and is apomorphous in two additional reductive characters 12 (14) and 14 (17). Coccorella atrata is apomorphous in two reductive characters, 12(15) and 13 (16). Linking Coccorella and Evermannella are intermediate states in the two 3-step characters 1(1) and 6 (7). Evermannella shows 254 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM autapomorphies in three characters: unique pattern of juvenile phase pigmentation 16 (19) and presence of vertically elongate fossa at dentary symphysis 10 (12), presence of fully tubular eye, 1 (2), unique to them among evermannellids, and show further reduction in the number of lateral line segments 6 (8). A single reductive character 14(17) also shared with Coccorella links E. indica and E. ahlstromi and E. megalops. A final, ques- tionable character 5 (6) links the latter two. In each case well- defined autapomorphous features support the hypothesis of monophyly of each genus and the information available appears to adequately support most of the proposed scheme. Details concerning the contribution of two larval characters to this analysis are discussed below. Peritoneal pigment ^frt/0/15. — Discrete peritoneal pigment sec- tions are striking features of most aulopiform but not mycto- phiform fishes (Johnson, 1974b, 1982; Okiyama, 1974b, this volume). A single dorsomedial section characterizes the larvae of all Aulopus (Okiyama, this volume), chlorophthalmoids and (primitively) scopelarchids. Multiple (3 or more, serially ar- ranged, paired or unpaired) sections occur in ipnopids (Bathy- pterois), bathysaurids, synodontids, harpadontids, paralepidids, Oinosudis and evermannellids. Peritoneal pigment sections are paired, left and right, in synodontoids (sensu Johnson, 1982) but single and connected dorsomedially over the gut in alepi- sauroids. Peritoneal pigment sections are apparently lacking in notosudids, some ipnopids, Alepisaurus, neoscopelids (perito- neal pigment present but not in a discrete section, see Okiyama, this volume) and myctophids. Johnson (1982) concludes that a single dorsomedial section is primitive for aulopiform fishes. Three unpaired sections are found in larvae of Coccorella, Ev- ermannella, Omosiidis and the paralepidine barracudina Par- alepis atlantica (said by Rofen, 1966a:238, to be ". . . the most primitive species in the Paralepididae."). Larvae of Odontosto- niops norinalops exhibit 1 2 or more unpaired peritoneal pig- ment sections, unique in the order, and a feature regarded as autapomorphous. Juvenile phase pigmentation— ]o\\nson (1982) regarded fixa- tion of the trilateral longitudinal pattern of juvenile phase pig- mentation, as described above, as autapomorphous for the genus Evermannella. It has long been supposed (Gregory and Conrad. 1936; Mar- shall, 1955; Gosline et al., 1966) that the Scopelarchidae and Everrriannellidae are closely related. This supposition was based mainly on the occurrence of tubular eyes in both groups. John- son (1982) argues against this notion, rejecting any close rela- tionship of the Evermannellidae and Scopelarchidae, placing the latter (tentatively) among the chlorophthalmoids, and plac- ing the Evermannellidae as the sister group of the Omosudidae plus Alepisauridae. The evidence for these conclusions is pre- sented in Johnson ( 1 982) and briefly summarized in the account of the Scopelarchidae in the present work. Field Museum of Natural History, Roosevelt Road at Lake Shore Drive, Chicago, Illinois 60605. Myctophiformes: Relationships M. Okjyama IN the traditional concept, the order Myctophiformes is con- sidered to be a monophyletic assemblage with taxa having much the same levels of organization, even though they have undergone considerable adaptive radiation including some ex- tremely specialized forms for particular habitats (Goody, 1969; Marshall and Staiger, 1975; Johnson, 1982). Modem definition of the order including 16 families was first established by Gosline et al. (1966). They recognized the fol- lowing two suborders: Myctophoidei: Aulopidae, Synodontidae, Bathysauridae, Harpadontidae, Bathypteroidae, Ipnopidae, Chlorophthal- midae, Myctophidae and Neoscopelidae. Alepisauroidei: Notosudidae (=Scopelosauridae), Paralepidi- dae, Omosudidae, Alepisauridae, Anotopteridae, Ever- mannellidae and Scopelarchidae. This dichotomous system has been generally followed by re- cent workers (Rosen and Patterson, 1969; Marshall and Staiger, 1975; Sulak, 1977), despite some minor changes or disagree- ments in the definition of family limits. On the other hand, Gosline (1971) proposed the idea of splitting the order into four groups (!) without giving rigorous evidence. Rosen (1973) reevaluated the relationships among the Myc- tophiformes and produced a very different provisional classi- fication based on a cladistic analysis of the group, where all of the myctophiform fishes (except Myctophidae and Neoscopeli- dae) form a monophyletic group, and likewise all the alepisau- roid families (except Giganturidae) form a monophyletic assem- blage. His phyletic hypothesis is radically different from those of Gosline et al. (1966) and Johnson (1982). Notosudidae was later transferred from Alepisauroidei to Myctophoidei (Bertelsen et al., 1976), and furthermore, Sco- pelarchidae was removed from Alepisauroidei (sensu lato) in the recent study of Johnson (1982) who further subdivided the order into five possible major groups in three perceived lineages (Fig. 131). Among these studies, Johnson (1982) is unique in carefully evaluating larval characters such as the peritoneal pigment sec- tions and the stomach pigmentation in juveniles, in considering myctophiform phylogeny with special references to Scopelar- chidae and Evermannellidae. As finely reviewed by Kendall ( 1 982), myctophiforms provide an excellent example for elucidating systematic relationships among fishes using larval characters, because larvae are known for representatives of most of the families and in some cases nearly all of the species within the families. Potential usefulness of the larval groups in this connection has been well documented for several families such as Myctophidae (Moser and Ahlstrom, OKIYAMA: MYCTOPHIFORMES 255 1 AULOPIDAE < w w — O t^ -• TTtH myctophidae ttH neoscopelidae y ttttHNOTOSUDIDAE I nil I SCOPELARCHIDAE "pCHLOROPHTHALMIDAE ttH IPNOPIDAE — 3 TTT TTT tH synodontidae ^ harpadontidae TT T-i bathysauridae ^ TTTTTTTnr paralepididae N II i iMiii I I I ANOTOPTERIDAE Mill -tH EVERMANNELLIDAE St TTTTTrr tHOMOSUDIDAE ttH ALEPISAURIDAE J Fig. 131. Possible interrelationships among myctophiform fishes (Johnson, 1982). 1972, 1974), Scopelarchidae (Johnson, 1974b, 1982), Notosu- didae (Bertelsen et a!., 1976) and Evermannellidae (Johnson, 1982). At higher taxonomic levels. Okiyama (1974b, 1979b, 1981) considered the relationships among families with partic- ular reference to the peritoneal pigment sections in association with the meristic features of the axial skeleton, notably precau- dal and caudal vertebrae. Larval characters of possible system- atic importance among Myctophoidei in Okiyama ( 1 979b) have been closely analyzed by Kendall (1982) in establishing familial interrelationships on the basis of the cladistic method, although several larval stages critical to this were not available at that time. Since current knowledge reveals slightly different conclusions for larval characters of potential phylogenetic importance from those employed in Okiyama ( 1 979b), some comments are given below for a revised character catalogue with a discussion of possible evolutionary direction. The determination of this di- rectional change is generally based on the assumption that the family Aulopidae, as presently considered, represents the prim- itive character state. In the following discussion, the character states believed to be primitive are all identified with a "0," and those believed to be derived are designated by a positive integer. Peritoneal pigment sections {1).—The development of the dis- crete peritoneal pigment sections is a remarkable feature of lar- val myctophiform fishes. Nothing is known of their function, but the systematic importance of this unique structure has been 256 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Table 68. Distribution of Larval Character States Among Myctophiform Families. Characters Family 1 T 3 4 5 6 7 8 D* Aulopidae (Au) Myctophidae (My) 3 1 1 3 Neoscopelidae (Ne) 3 1 1 1 2 5 Chlorophthalmidae (Ch) Ipnopidae (Ip) 1 1 2 Notosudidae (No) 3 1 1 3 Scopelarchidae (Sc) 1 1 -> Bathysauridae (Ba) 1 1 t 0? 1 1 5 Harpadontidae (Ha) 2 1 1 3 Synodonlidae (Sy) 2 1 1 3 Alepisauridae (Al) 3 1 1 1 2 6 Anotoptendae (An) 3 1 3 Evermannellidae (Ev) 1 1 1 -> 5 Omosudidae (Om) 1 1 2 4 Paralepididae (Pa) 1 1 3 * Number of denved character slates. repeatedly emphasized (i.e., Okiyama, 1974b, 1979b, 1981. Johnson, 1974b, 1982). Contrary to earlier understanding (Oki- yama, 1974b), much diversity of this character has been re- vealed. Based on the number and shape of the sections, a pro- visional classification is as follows: (A) Dorsomedial pigment sections; (A-1) Single patch— Aulopidae, Chlorophthalnms. Bathytyphlops. Rosenblattichthys. Scopelarchoides (in part); (A- 2) Many (three or more)— Bathysauridae, Bathypterois (in part), Sudis. Omosudidae, Evermannellidae; (A-3) Single to many patches with growth— Paralepididae (except Sudis). (B) Paired pigment sections— Harpadontidae, Synodontidae. (C) Dorso- medial and paired pigment sections— Scopelarchoides (in part), Scopelarchus. (D) No pigment sections— Neoscopelidae (except Solivomer), Myctophidae, Ipiiops. Bathymicrops. Bathypterois (in part), Benthalhella. Notosudidae, Alepisauridae, Anotop- teridae. Rare exceptions are also known for several of these types. The only known exception to the presence of the A-3 type in paralepidids is in Notolepis coatsi with a single pigment section throughout all stages (Efremenko, 1978, 1983a). However, the ontogenetic development of this section into the extensive per- itoneal pigment tube around the gut as in other paralepidids reveals little phylogenetic difference for this exception. Among those having B-type, some Synodus reportedly lack the peri- toneal pigment sections and may represent an extremely spe- cialized character state (Cressey, 1981). On the contrary, a my- tophid, Protomyctophuin anderssoni. is known to develop the serially arranged paired pigment patches similar to those of B- type(MoserandAhlstrom, 1974; Efremenko, 1976). Their over- all resemblance including this pigmentation may be a result of a simple convergence. As is clear from the above classification, character states are remarkably diverse in the Scopelarchidae and Ipnopidae. C- type, peculiar to the former, is of particular significance in sug- gesting the possible direction differentiating the paired and un- paired character states (Johnson, 1974b). Unclear limits of the family are partly responsible for the confusion in Ipnopidae. It is generally agreed that the presence of a single, dorsomedial peritoneal pigment section (A-1 type) represents the primitive state. Since A-3 and C types are referable to the ranges of either A-1 or 2, four states are recognized as in Johnson ( 1 982); (0) = A single, dorsomedial peritoneal pigment section. (1) = Multi- ple (3 or more), serially arranged, unpaired peritoneal pigment sections. (2) = Multiple (3 or more), serially arranged, paired peritoneal pigment sections. (3) = Peritoneal pigment section absent. Position of anus f2A— Contrary to the usual pattern of the anus location immediately anterior to the anal fin origin in much teleosts, a more or less wide preanal interspace is commonly shared by many taxa of this order. This character can be of much use in distinguishing the groups of Myctophiformes (Rosen, 1 97 1 ; Okiyama, 1 979b). The character states of the diverse anus location relative to the pelvic fins are not recognized herewith due to the unclear patterns of occurrence. The ontogenetic rearward shift of the anus is restricted to some speciose families such as Scopelarchidae, Paralepididae (except Sudis), Notosudidae, and Myctophidae (in part). There is, however, a sharp contrast in the final condition among them: no preanal interspace in Myctophidae and Scopelarchidae vs a distinct space in the remaining two. As in Kendall ( 1982), who employed this character in the first step of branching, two char- acter states are recognized. (0) = Anus with interspace from the origin of the anal fin. (1) = Anus just in front of the ongin of the anal fin. Fin features (3). — Except Bathysauridae with magnificently en- larged fins, the elongated pectoral fins are the pronounced larval character found in many representatives of this order. The Ip- nopidae displays the most diverse pattern of specialized pectoral form in being bifid, large and fan-like, or extremely elongated. Parallel features are known to occur sporadically in some spe- cialized myctophidae (Moser and Ahlstrom, 1970, 1974). Sco- pelarchidae is another member of less cohesiveness in this char- acter; prominent pectoral fins are peculiar to Rosenblattichthys, the most specialized genus in this family (Johnson, 1974b). Likewise, only the aberrant genus Sudis has elongated pectoral fins in Paralepididae. The character states recognized are: (0) = All fins short. (1) = Only pectoral fins elongated. (2) = All fins elongated. Sequence of fin formation i^-^A- Although current knowledge is far from complete, dichotomous patterns can be recognized in the sequence of fin formation, especially in the pectoral fins. Johnson (1982) defined the derived character state of Rosen- blattichthys by the development of pectorals prior to all fins except caudal. The precocious nature of this fin apparently rep- resents the derived state. (0) = Pectoral fins not precocious. (1) = Pectoral fins precocious. Eye shape (5). — Moser and Ahlstrom (1974) showed that two types of eyes, i.e., round and narrow, reflect the major two lineages of Myctophidae with several exceptions. These patterns are commonly duplicated at familial levels in this order. In view of the specialized morphology of the narrow eyes including the peculiar choroid tissue and following the suggested phylogeny of Myctophidae (Moser and Ahlstrom, 1974), round or nearly round eyes are regarded primitive. The states recognized are: (0) = Eyes rounded or nearly rounded. (1) = Eyes narrowed. Head armature (6).— The development of head spines is rare or rather exceptional particularly in the adult myctophiform fishes. However, larvae of at least five families have head ar- mature. These include preopercular spines and supraorbital and/ OKIYAMA: MYCTOPHIFORMES 257 Table 69. Similarity Matrix of 1 5 Families of Mvctophiformes. Based on the total number of characters shared in the same state regardless of the primitive or denved (below the diagonal) and that of the shared derived states (above the diagonal, with similarity index in parentheses). Subordinal groups are indicated by enclosure. Similarity index is calculated by the following formula: P„ = (C,/\/S,S,) x 100, where S, and S, are number of derived characters in families i and j, and C„ is number of the shared derived states between the same set of families. My Ip Sy Au - My 5 - |3(77)| 1 (40) 1 (33) 1 (40) 1 (26) 1(33) 1 (33) 2 (47) 1(33) 1(26) Ne 3 g [6] 5 3 1(22 ) 1 (26) 1 (32) 1(20) 1(26) 1 (26) 1 (55) 1(26) 2(40) 2(45) Ch Ip 6 5 3 6^^ \^"-^ 1(32) 2 (58) No 5 6 3 4 5 3 3 3 2 5 6 3\ 4 ^ 1(45) 1(20) 1(33) 1(41) 1 (33) 1 24) 2(67) 1(26) 1(26) 2(64) 1(20) 1 (33) So - 1(32) 1(41) Ba 3 2 2 2 [2^2) 2(52) 1 (26) Ha 5 5 2 5 5 4 4 4 5 5 5 2 3 3 4 5 5 2 5 5 5 "^ 3(100) 1(33) 1(33) 1(26) 1(26) 1(33) Sy - 1(33) A! 2 1 1 12^7) 2(37) 3(61) 1 (24) An 5 4 3 5 3 6 3 5 5 5 4 ~~"\ ,^^^^ ^26) 1(29) 2(67) Ev 3 3 3 3 2 4 5 5 5 5 2 4"^ ,^^^^^. 3(67) 2(58) Om 4 5 2 3 4 2 4 5 2 3 3 4 2 4 3 5 3 5 3_ 5 7 5 "^ 2(52) Pa 3 6 5 - or frontal ridges. Development of head armature generally oc- curs in the forms with a massive head more than 30% of body length, thus suggesting the specialized condition of this char- acter. Several myctophid species {Lampanyctus) having pre- opercular spines provide a fine example of this trend, while this is not the case in Scopelarchidae. According to Nafpaktitis(1977), the character state of Neoscopelus is assigned to the Neosco- pelidae. The states recognized are: (0) = Head armature absent. (1) = Head armature present. Body shape f/A — The general body shape can range from ex- tremely slender and elongate to stubby and deep. These are tentatively grouped into three character states with possible evo- lutionary trends towards the opposing directions from the mod- erately slender body shape shared by primitive groups such as Aulopidae and Chlorophthalmidae. The character states rec- ognized are: (0) = Body moderately elongate. (1) = Body ex- tremely slender and elongate. (2) = Body stubby and deep. Pigment spots or area ("(SA— Johnson (1982) suggested the po- tential importance of pigmentation other than the peritoneal sections in the systematics of the Myctophiformes, even at high taxonomic levels. A difficulty in this regard is how to recognize the meaningful character states. Based on the various pigmen- tation patterns in the tails of larvae (posterior to the anus except for the caudal fin) such as (a) absent, (b) present along only the ventral midline, and (c) present along lateral or dorsal surfaces of body sometimes forming clear bands, formal recognition of this character is undertaken. Since patterns (a) and (b) are shared commonly during the ontogeny of the same species, two char- acter states are recognized with the assumption that (c) repre- sents the derived state. (0) = Pigment spots or areas in tail absent or present along only the ventral midline. (1) = Pigment spots or areas in tail present along lateral or dorsal surface. The primitive or derived states for these eight characters are summarized in Table 68. Family level designation of character states is mostly based on the assumption of Johnson (1982) that "possession by one or more representatives of a particular OTU of a state considered primitive indicates (except where contrary evidence can be cited) the primitiveness of that state for that OTU." A similarity matrix based on the total numbers of characters shared in the same state, regardless of whether the states are primitive or derived, is given below the diagonal in Table 69. Above the diagonal are shown the numbers of derived characters shared in the same state and the similarity index calculated on the same data. These two sets of figures are expected to reveal certain clues to clarify the interfamilial associations of this order from the larval standpoint. AiiLOPOiDEi: Aulopidae So far as the selected larval characters are concerned, the Aulopidae can not be separated from the Chlorophthalmidae. This unclear distinction is due to the limited numbers of char- acters selected, because other larval and adult features shown in Table 70 reveal the trenchant differences between them. Of these, the possession of maxillary teeth and fulcral scales, and the earlier differentiation of the peritoneal pigment spots well justify the distinct and less specialized systematic status of the Aulopidae. Other aspects of sharp contrast such as in the den- tition, particularly of the basihyal, and gut morphology sub- stantiate the above conclusion. Although the diversity within the Aulopidae once suggested on the basis of larval characters (Okiyama, 1974b) has proved to be unacceptable, there still remain problems concerning the monotypic nature of this family. As mentioned elsewhere (Oki- yama, 1979b), it is likely that the Myctophiformes evolved along several lines, one of the major trends being the elongation of the body shape accompanying an increase in vertebral number. Obviously, aulopids lie near the base of this trend with clear orientation toward an increase in the number of abdominal components. The uniquely elongated larval oesophagus in A. 258 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Table 70. Anatomical Differences of Early Stages Betwfen AuLOPvs and Chlorophthalmus. Table 71. Distribution of Larval Character States among Four Genera of the Ipnopidae. Autopus Chtorophlhatmits (afler Rosen (1971) and Sulak(1977)] D« Maxillary teeth Vomerine teeth Basihyal Fulcral scale Gut morphology Peritoneal pig- ment sections Present Only two widely sepa- rated at opposing anterolateral comer Ovoidal with slightly indented tip; teeth absent Present Moderately elongated, straight; intestine slightly fat Single; distinct at less than 3.5 mm SL Absent Transverse row of six teeth divided into two rows of three each Triangular with similar anterior indentation; a transverse row of six teeth divided into two series Present(?) Short, compact with slender stomach; in- testine fat Single; distinct at more than 5 mm SL japonicus is a probable indication of this evolutionary trend (Okiyama, 1974b). Among recent congeners, A. damasi may be the most generalized species in view of its smallest number of vertebrae (20 + 16) similar to the known counts in the fossil aulopids (Goody, 1969; Rosen and Patterson, 1969). Further- more, this species is clearly separable from congeners by the mode of direct association between the first haemal spine and anal pterygiophores (Okiyama, 1979b). A look at the larvae of A. damasi would be enlightening in clarifying the problem in question. Myctophoidei: Neoscopelidae, Myctophidae The two families of this suborder are readily discriminated from the others by the greatest similarity index value based on a suite of derived characters (1 and 2) not shared by any other families. The smaller sizes at metamorphosis are also peculiar to these families. These larval evidences offer strong support for the views of Moser and Ahlstrom (1974) and Johnson (1982), warranting a distinct subordinal ranking. My observation of the vertebrae of Solivomer (see Table 57 in my Myctophiformes: Development, this volume) also disclosed their closer linkage than assumed by Johnson (1982). The similarity matrix in Table 69 would offer little support for Rosen's scheme to transfer these families to a different order. Chlorophthalmoidei: Notosudidae, Scopelarchidae, Chlorophthalmidae, Ipnopidae The larval character states indexed in Table 7 1 are less prom- ising in support of this familial assemblage, because only the Notosudidae and Scopelarchidae share a single derived char- acter state (narrow eye). It seems that this ambiguity is also associated with the inadequate numbers of characters in ques- tion. Although the admitted cohesiveness of larval characters of Chlorophthalmidae may be altered by the discovery of larval Bathysauropsis or Parasudis. larval characters support the tra- ditional view that it is one of the basal stocks of this order, lying Bathytyphlops Ipnops Bathymicrops Bathypterois 10 2 10 10 4 10 10 4 10 15 ' Number of denved character slates. at a somewhat advanced place along a line different from the Aulopidae. Trenchant characters in this connection such as the dentition and the mode of anal fin support are shared with the Ipnopidae. Members of the Notosudidae, the most cohesive family in this suborder, have the greatest numbers of derived characters of the group. Marshall ( 1 966a) and Bertelsen et al. ( 1976) stated that it seems most closely related to Chlorophthalmidae. The superficial resemblance of larval stages between this and the Paralepididae was also suggested (Ahlstrom, 1972a). On the other hand, the similarity matrix indicates its affinity with An- otopteridae, along with Scopelarchidae. Of these associations, the last grouping based on a single derived state in character 5 (narrow eye) appears less arguable. Other features such as the maxillary teeth and the uncommon morphology of the corpus cerebelli suggest the aberrant systematic status of this family. Since Table 68 provides few clues to discuss the confused family limits of the Ipnopidae, the same coding of the character states is applied to the four genera of this family (Table 71). Except for the distinct larval status of Bathypterois. derived characters shared among the remaining three genera do not reveal the generic linkages suggested by Sulak (1977). By the same reasoning as discussed before concerning the relationships between Aulopidae and Chlorophthalmidae, the derived state in character 1 (peritoneal pigment sections) shared by Ipnops and Bathymicrops includes the different states of gut morphol- ogy. It seems these genera form a loose but distinct assemblage warranting family rank. Besides the shared dentition mentioned before, the close fit of general larval morphology between Bathy- typhlops and Chlorophthalmus may suggest their relationship. The diverse larval characters of Scopelarchidae were elabo- rately enalyzed in the light of adult systematics (Johnson, 1 974b). It is remarkable that this family has no phenetic similarity with Alepisauridae in terms of catalogued characters. On the other hand, two derived states in character 2 (anus location) and 5 (eye shape) shared with Evermannellidae give the greatest sim- ilarity index value. Johnson (1982) suggested the independent occurrence of the tubular eyes in adults of both families, but traditional concepts of their close association should be reevalu- ated using larval evidence. Synodontoidei: Bathysauridae, Harpadonti[5ae, Synodontidae Accepted linkage between Synodontidae and Harpadontidae is clearly substantiated by the larval characters, while familial allocation of Saurida remains to be solved. Synodus lucioceps, having the intermediate state of larval characteristics between these families, may be important here. The relationships among four genera are thus indistinct from the standpoint of the larvae, but Saurida appears to be the most generalized. Possible phy- OKJYAMA: MYCTOPHIFORMES 259 logenetic association between Aulopidae and these families has been suggested on the basis of larval characters and the similar mode of anal fin support (Okiyama, 1974b, 1979b). To these can be added the peculiar structures on the chorion surface of the extremely transparent eggs, the pigmentation patterns in the newly hatched larvae, and the mode of reproduction shared by these families, characters which favor their close association. Bathysauridae is distinguished from other families of this suborder by some trenchant differences in the peritoneal pig- ment sections and the mode of reproduction, while two derived states are shared by all families. The phylogenetic relationship of these families depends on whether the above mentioned dif- ferences are due to divergence. Larval stages of Bathysauridae are surely highly specialized, adapting to a prolonged pelagic life, but larval dentition described in detail by Rosen (1971) and Johnson (1974) and the character state of the axial skeleton, including the mode of anal fin support (Okiyama, 1976b) are of particular interest in showing the pattern common to Ipno- pidae. Alepisauroidei: Paralepididae, Anotopteridae, Evermannellidae, Omosudidae, Alepisauridae The similarity matrix provides certain indication of the co- hesiveness of this suborder. Most remarkable is their common sharing of the derived state of character 8. Regarding the per- itoneal pigment sections dividing five families into two groups, some comments are warranted for Alepisauridae. As discussed by Johnson ( 1 982), this character state is very tentatively defined due to the inadequate state of available material. Even so, a distinct family pair of Alepisauridae and Omosudidae can be readily separated from the remaining families by the many de- rived character states shared by them. Although the possibility of their convergence cannot be fully rejected in view of the clear contrast in the ontogenetic aspects of the pectoral fins, the close similarity between Alepisaurus ferox and Omosudis lowei (trop- ical western Pacific specimen) (see my Myctophiformes: De- velopment, Fig. 1 1 2B, E, F, this volume), in head armature and pigment pattern is extremely striking. An association between the Anotopteridae and Paralepididae, particularly the more elongated paralepidids such as Stemo- nosudis and Macroparalepis (Rofen, 1 966a, c), can be seen from the larval standpoint. In addition to their shared derived char- acter states (character 7 and 8), a fleshy projection on the lower jaw tip peculiar to Anotopteridae and Stemonosudis macrura. and the similar larval dentition (huge canines) may substantiate the above association. Their disagreement in the character of the peritoneal pigment sections is probably associated with the odd systematic position of Anotopteridae lying at "an extreme specialized end-point of the paralepidid line" (Rofen, 1966a, c). On the basis of the larval characters, two subfamilies of Par- alepididae are well separated. As compared with the relative constancy of conservative characters in larval Paralepidiinae, the many derived character states of larval Sudinae are too specialized to be consistent with the accepted subfamilial level. The latter may be an earlier offshoot preceding the remarkable paralepidine radiation. The complete lack of intermediate forms between them offer strong support for this suggestion. As in Scopelarchidae (Johnson, 1974b), the systematics of Evermannelidae were studied in detail using a large character suite, including larval aspects (Johnson, 1982). So far as the present analysis is concerned, this family seems variously as- sociated with families of Alepisauroidei such as Paralepididae, Alepisauridae and Omosudidae, besides Scopelarchidae. It is of interest that limited character states shared by Evermannellidae and Alepisauridae are restricted to derived ones, probably sug- gesting their close association. Perhaps, an Evermannellidae and Scopelarchidae linkage is much more loose, if valid. Concerning the possible three main lineages in this order, the larval evidence is less promising. However, additional larval evidence regarding developmental sequences, including osteol- ogy as well as internal morphology, would provide much more fruitful information for elucidating the phylogeny of this inter- esting group. Ocean Research Institlite, University of Tokyo, 1-15-1, MiNAMiDAi, Nakano-ku, Tokvo 164, Japan. Gadiformes: Overview D. M. Cohen GADIFORMES is a particularly interesting order with which to work because it encompasses a high degree of diversity that suggests the existence of several lineages, apparent conver- gence and reductive trends to trap the unwary, a useful fossil record that allows a consideration of the distribution in time of some important taxa and character states, and new suites of characters based on the study of ELH stages. Although study of the classification of gadiforms dates from pre-Linnean times, there is still insufficient properly evaluated data available to derive a phyletic classification. In fact, there is not even agreement as to what should be included. Berg (1947) restricted the order to the muraenolepids, bregmacerotids, mor- ids, and gadids (including merlucciids) and excluded the mac- rourids. He noted primitive and advanced characters in his gadiforms and suggested derivation from primitive fishes. Rosen and Patterson (1969) revived an expanded Gadiformes dating at least from the time of Gill, which included not only gadoids and macrouroids but also ophidioids and zoarcoids, and which they placed in a supraorder Paracanthopterygii, postulated as being, "in many ways more primitive than the acanthoptery- gians" and representing "a spiny-finned radiation more or less comparable morphologically with that of the Acanthopterygii" 260 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM MERLUCCIUS BRECMACEROS EUCLICHTHYS CADINAE (2) LOTINAE (2) EUCLICHTHYS ( u) MURAENOLEPIS (2) PHYCINAE 12) MORIDAE («-5) MELANONUS ( if- 5) MERLUCCIUS (21 BRECMACEROS (2) MELANONUS HYPURAL RAYS Fig. 133. Numbers of hypural bones (in parentheses) and fin rays supported by hypural bones in nine groups of gadiform fishes. Data from Fahay and Markle (this volume) and original. TOTAL CAUDAL RAYS Fig. 132. Total caudal rays in eight groups of gadiform fishes. Data from Fahay and Markle (this volume) and original. and including in addition to their gadiforms the polymixoids, percopsiforms. batrachoids. iophiiforms, and gobiesocoids. Gosline (1968) analyzed the characters used in defining the ex- panded Gadiformes and concluded that ophidioids and zoar- coids are perciform derivatives, while gadoids are widely sep- arate and probably close to the percopsiforms (Gosline, 1963a). Marshall and Cohen (1973), whom I follow for present purposes, restricted the Gadiformes to the gadoids and macruroids but did not consider the question of relationships. In the following brief preliminary consideration of the order, I discuss several characters, mention the groups that I think must be considered, and outline some of my ideas about the course of evolution in the gadiforms. Characters Several character complexes that require consideration are discussed below. Others are noted later under groups in which they occur. Additional relevant information is presented by Fa- hay and Markle and Dunn and Matarese in subsequent sections of this volume. Caudal fin.— Considering the fact that well over half the known species of gadiform fishes lack the slightest vestige of a caudal fin, it is a little astonishing how much importance has been attached to the origin and homologies of the various skeletal supports and of the fin rays themselves. There is no denying, however, that when present the gadiform caudal complex is unique in several respects. Most fish groups may be character- ized by a set number of branched caudal rays. Furthermore, the branched rays are generally supported by only hypurals. In gad- iforms with tail fins, the number of branched caudal rays is highly variable, as is their skeletal support. Bregmaceros may have as few as 1 2 branched caudal rays, most of which are supported by hypurals, while at the upper end of the range, the lotine Brosmc may have as many as 43 branched rays, which are supported by hypurals. epurals, and haemal and neural spines. This high degree of variation in an otherwise conservative an- atomical complex lends credence to the idea of Boulenger(1902) and Regan ( 1 903b) that the caudal fin of gadiforms is a structure newly evolved from an essentially tailless condition such as that of the macrourids or of some merlucciids. It was partly to test Regan's hypothesis that Barrington (1937) compared the de- velopment of the caudal fin of Gadus with that of Pleuwnectes and concluded that, although the tail of Gadus was unique in several respects, it could have been derived from an ordinary homocercal tail that was less specialized than that of Pleuw- nectes. I agree with Barrington. Barrington commented also on the presence in gadids of a high number of procurrent caudal rays, which he interpreted as being far posterior dorsal and anal rays, so that the functional caudal of a cod is composed of elements of three fins, dorsal, anal, and caudal proper. This interpretation has been neither falsified nor verified by the study of early life history stages. Barrington coined the term pseu- docaudal for what he took to be this kind of fin. In his lectures and during conversations with me. Ahlstrom disagreed with Barrington's explanation and its acceptance by Marshall and Cohen (1973) because procurrent rays lack pterygiophores. It is instructive to note in this respect the caudal fin structure of Muraenolepis (see Fig. 1 43 of Fahay and Markle in this volume), which has confluent vertical fins and in which the distinctive, elongate pterygiophores grade into hypurals. It is, in fact, im- possible to distinguish between the last anal pterygiophore and the first hypural or parhypural. But see Fahay and Markle later in this volume. A variety of controversial interpretations (Gosline, 1963a; Monod, 1968; Rosen and Patterson, 1969) have been advanced concerning supposed sequences effusions and deletions of bony elements in gadiform tails. This particular use of caudal fin structure in phylogeny has yet to be proven, as few hypotheses have been verified or falsified. For purposes of classification within the order, at least four COHEN: GADIFORMES 261 BREGMACEROTIDAE EUCLICHTHYS MERLUCCIUS 10 20 30 40 50 BRANCHED CAUDAL RAYS Fig. 134. Branched caudal rays in seven groups of gadiform fishes. Data from Fahay and Markle (this volume) and onginal. caudal fin characters require comment. They are: 1) presence or absence of a caudal fin; 2) number of hypurals; 3) relationship between branched caudal rays, hypurals, and procurrent caudal rays; 4) presence or absence of X- Y bones. Although vestiges of a caudal fin are sometimes found in a few macrourid species, it is essentially absent from all of them. The same is true of the merlucciid genus Lyconus and also Steindachneria. Loss of a caudal fin has certainly occurred two times and perhaps more. The number of hypurals is a useful systematic character. There are almost always 4 or 5 in morids and Melanonus, and almost always 2 in gadids, Merliiccius. Bregmaceros, and Muraenolepis: Euclichthys has 4, nearly fused to 2. 1 consider the lower number to be an advanced character; the study of developmental series has verified this interpretation for Raniceps at least (Dunn and Matarese, this volume). Certainly the loss of hypurals, whether through deletion or fusion has occurred several times in gadi- forms. The evolutionary complexity of the caudal fin in gadiforms is particularly apparent when considering the numbers of dif- ferent kinds of caudal fin rays (Figs. 132-134 and Fahay and Markle, this volume. Table 76). Morids in general have caudal fins that are small and probably of reduced importance in pro- pulsion, and which 1 interpret as a derived state; they also have generally fewer total rays, which Fahay and Markle (this volume) consider an ancestral state, and unbranched rays that tend to be short and contribute little to overall caudal fin size; yet, morids have 4-5 hypurals. Melanonus also has a weakly de- veloped caudal fin but has 4-5 hypurals and many rays. Gadine fishes on the other hand, have well-developed caudal fins with many rays, both branched and unbranched, but have only 2 hypurals. Gadines are in general good swimmers, and one of the most active of all, Pollachius vtrens, has the most total caudal fin rays (70 in one specimen) of any gadiform fish. (Sluggish fishes like the lotines, Brosme and Lota, also have numerous caudal fin rays but have rounded caudal fins and must swim in a very different way, probably using the caudal fin as an exten- sion of the body rather than as an oar.) Although numbers of different kinds of fin rays may prove useful in taxonomy, the relationship of branched to unbranched or total caudal fin rays is variable and has limited apparent value in the present context. Many gadiform fishes have in their caudal fin skeletons a pair of bone splints resembling neural and haemal spines. These structures have been mentioned in the literature as accessory bones or X and Y bones and have been interpreted as modified relict pterygiophores or detached neural and haemal spines whose centra have been lost (Rosen and Patterson, 1969). 1 agree with Markle (1982) that the absence in any gadiform of X and Y bones is a derived character. Dorsal and anal fins.— Gadiform fishes have 1, 2, or 3 external dorsal fins and 1 or 2 external anal fins. The number, size, and location of these fins have been used for hundreds of years to characterize groups of species. Prior to the recognition of Mor- idaeasa distinct family (Svetovidov, 1937), convergence in this character was not recognized; most ichthyologists lumped gadids and morids with similar fin patterns. Svetovidov ( 1948) assumed on functional grounds that a sin- gle dorsal and single anal is the primitive condition and arranged the gadid genera in a transition series based on increasing num- ber of fins and the distance of their separation from each other. His hypothesis is supported by the presence in all gadiforms of a single, continuous, postanal series of pterygiophores, present even over areas that lack fin rays. Complete or partial division of the exterior fin has occurred several times, for example in the gadines, Euclichthys. Merluccius. and in the morid genera Mora, Halargyreus. Lepidion, Laemonema. and Tripterophycis. Although only a few gadiforms have a single dorsal fin, the condition has a broad taxonomic distribution; examples are the gadid Brosme. the merlucciid Lyconus. Melanonus, and the ma- crouroidine rattails. Nearly all gadiforms have 2 or 3 dorsals, but even in those with 3, there are only two series of pteryg- iophores. From fewer to more dorsals would seem to be a rea- sonable transition series. But it certainly has occurred more than once, even within Gadidae, as Markle (1982) has demonstrated. Pectoral radials. — Mosx gadiforms have five pectoral radials. Muraenolepis has more; Bregmaceros has fewer; both are in- terpreted as derived conditions. First neural spine. — Many gadiforms have the first neural spine closely adpressed to the occipital crest. I take this as a derived character. Muraenolepis has a free spine, but it is modified by the presence of a prominent wing-shaped enlargement extending on either side of the occipital crest. Olfactory lobes. — In his classical monograph on the Gadidae, Svetovidov (1948) discussed the position of the olfactory lobes 262 ONTOGENY AND SYSTEMATICS OF FISHES-AHLSTROM SYMPOSIUM Fig. 135. Dorsal view of cranium in three genera of gadiform fishes; left. Rhinocephalus planiceps: center, Palaeogadus intergerinus; right, Merluccius merluccius. From Fedotov ( 1 976). of the brain and used their advanced position, adjacent to the nasal capsule, as his primary character for defining the Gadi- formes. This is a derived character, which has been found also in cyprinids, galaxiids, and mormyrids. Svetovidov noted that the olfactory lobe is located in an intermediate position in the gadid Raniceps. A posterior location of the lobe was subse- quently recorded in Melanonus and several macrourids and an intermediate location in merlucciids, Steindachneria, the gadid Raniceps, and two macrourids (Marshall and Cohen, 1973). Svetovidov ( 1 969) pointed out the size dependent nature of this character, especially in Merluccius (which I have verified in M. bilinearis and M. productus). Further investigation is required, especially in species that mature at small sizes. V-shaped crest on skull.— As long ago as 1903b Regan noted the shared presence in Merluccius and Macruronus of prominent V-shaped ridges on the frontals, which converge on the supra- occipital crest. These structures have subsequently been found in the extinct genera Rhinocephalus and Palaeogadus (Fig. 135) as well as in some fossil percopsiforms (Rosen and Patterson, 1969) and are present in varying degrees in Lyconus and Stein- dachneria. Groups and Their Relationships In this section I briefly discuss those taxonomic units that I think require consideration and explain as best possible the reasons for their placement on Fig. 1 36. "Protocodus" is an unnamed species' from the Paleocene of Greenland (discussed by Rosen and Patterson, 1969 and Fe- ' The name "Protocodus" is used as a designation of convenience and does not have formal, nomenclatural significance. dotov, 1 976; I too have examined it), which is the oldest known non-otolith gadiform. It has a number of characters that may be interpreted as primitive for the group, including five, slender, well-separated hypurals, X-Y bones, numerous procurrent rays, and a V-shaped ridge on the frontals. It has a dorsal and anal fin configuration much like that of Merluccius (Rosen and Pat- terson, 1969). Muraenolepis is a highly distinctive genus with four or more species. It has such primitive characters as a single anal and long-based second dorsal fin, a dermal basibranchial plate (Ro- sen and Patterson, 1969), the similarity of the lower hypurals to pterygiophores and to caudal fin elements, and a free first neural spine. Derived characters include 12-14 pectoral radials, a single epural, first dorsal fin a single-rayed anteriorly placed filament, vertical fins confluent around the tail, an oblique pat- tern of squamation, and modifications of the first neural spine. Muraenolepis is not obviously related to any other gadiform and appears to represent an ancient lineage. Bregmaceros is another distinctive genus with no obvious close relatives. Like Muraenolepis it retains a dermal basi- branchial plate, but this is a primitive character, as is possession of a uroneural and a set of X-Y bones in the tail. Derived characters include the conjunction of the first neural spine with the occipital crest, a large consolidated hypural plate supporting many branched rays, a unique lateral line system, only two pectoral radials, and a long dorsal ray on top of the head. The tropical pelagic habitat of these fishes is also different from that of any other gadiform. If fusion of the first neural spine with the occipital crest has occurred only a single time, then Breg- maceros must have originated after Rhinocephalus. Rhinocephalus is an Eocene fossil, the skull of which has been described in some detail and compared with other gadiforms by Rosen and Patterson ( 1 969). They mention and illustrate a COHEN: GADIFORMES 263 RECENT. PLEISTOCENE. PLIOCENE. MIOCENE. OLIGOCENE. EOCENE. PALEOCENE. "PROTOCODUS" Fig. 136. Phylogenetic bush showing hypothetical inter-relationships among gadirorm fishes. Beginning of soHd Unes based on fossils, not including otoliths or scales. V-shaped indge on the frontals and also lateral flanges on the rear of the skull that characterize gadines and at least some morids. They write, "The skull roof of Rhinocephalns shows many features common to morids, merlucciids, gadids. and macrourids . . . ." In addition, the first neural spine is free from the supraoccipital crest. Eucltchthys (Fig. 137), represented by a single South Austra- lian and New Zealand species, was incorrectly placed in Moridae but removed by Svetovidov (1969), who pointed out some sim- ilarities to Macrouridae. Enclichlhys can not be placed in any currently recognized family. It has a free first neural spine, which may indicate an origin prior to Palaeogadus. lacks an otophysic connection, has four hypurals nearly fused to two, and in two specimens has only one of the X-Y bones. As in morids, which are more specialized than macrourids and could not have given rise to them, Eitclichthys has an asymmetrical, rather reduced caudal fin. Perhaps this curious fish is a modem representative of a macrourid progenitor. Macrouroidinae is represented by two small genera and has been treated both as a subfamily of Macrouridae (Marshall, 1973) and a separate family (Okamura, 1970a). It has single dorsal and anal fins and a number of distinctive features in the head skeleton and may represent the most primitive tail-less macruroid. Macrourinae-Trachyrincinae, which may well constitute two quite separate groups, has 20-25 genera and contains more than half of all gadiform species (Okamura, 1970a; Marshall, 1973). The caudal fin is absent in most, vestigial in a few; the first neural spine is free, and there is no V-shaped ridge. Eggs of the few species for which information is available have a distinctive hexagonal pattern; many species have light organs. Bathygadinae, with two genera, differs from other macrourids in having a large, terminal mouth, dorsal rays longer than anal ones, and in a variety of other ways summarized by Okamura (1970a), who interprets most of the bathygadine characters as primitive ones. Differences in functional morphology between bathygadines as pelagic feeders and macrourines as benthic to benthopelagic feeders have been described by McLellan (1977). Melanonus has two meso-to-bathypelagic species formerly placed in Moridae, where they do not belong as they lack an otophysic connection, have a single dorsal fin, and have lost the X-Y bones. Otherwise, they seem similar to Moridae. The first neural spine is joined to the occipital crest, suggesting an origin after Rhinocephalus. A separate family was proposed by Mar- shall (1965). Moridae consists of 12-15 genera, some highly diverse, and all characterized by possession of an otophysic connection, 4 or 5 hypurals, X-Y bones, a joined first neural spine, and distinctive otoliths; many species have light organs. Morids probably di- verged from the main Rhinocephalus-Palaeogadus-Merluccius 264 ONTOGENY AND SYSTEMATICS OF FISHES- AHLSTROM SYMPOSIUM Fig. 137. Euclichlhys polynemus, ho