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