UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The evolution of branchiostegal rays in teleostome fishes McAllister, Don Evan 1964

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-UBC_1964_A1 M2.pdf [ 25.34MB ]
JSON: 831-1.0105258.json
JSON-LD: 831-1.0105258-ld.json
RDF/XML (Pretty): 831-1.0105258-rdf.xml
RDF/JSON: 831-1.0105258-rdf.json
Turtle: 831-1.0105258-turtle.txt
N-Triples: 831-1.0105258-rdf-ntriples.txt
Original Record: 831-1.0105258-source.json
Full Text

Full Text

THE EVOLUTION OF BRANCHIOSTEGjrAL RAYS IN TELEOSTOME FISHES < B Donald Evan McAllister B. A., University of British Columbia, 1955 M. A. , University of British Columbia, 1957 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF . DOCTOR OF PHILOSOPHY in the Department of Zoology We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA March, 1964 I n p r e s e n t i n g t h i s t h e s i s i n p a r t i a l f u l f i l m e n t of the requirements f o r an advanced degree at the U n i v e r s i t y of B r i t i s h Columbia, I agree that the L i b r a r y s h a l l make i t f r e e l y a v a i l a b l e f o r reference and study. I f u r t h e r agree that per-m i s s i o n f o r extensive copying of t h i s t h e s i s f o r s c h o l a r l y purposes may be granted by the Head of my Department or by h i s representatives,, I t i s understood that copying or p u b l i -c a t i o n of t h i s t h e s i s f o r f i n a n c i a l gain s h a l l not be allowed without my w r i t t e n p e r m i s s i o n . Department of The U n i v e r s i t y of B r i t i s h Columbia, Vancouver 8, Canada. Date The U n i v e r s i t y of B r i t i s h Columbia FACULTY OF GRADUATE STUDIES PROGRAMME OF THE FINAL ORAL EXAMINATION FOR THE DEGREE OF DOCTOR OF PHILOSOPHY B.A., The Univ e r s i t y of B r i t i s h Columbia, 1955 M.A., The University of B r i t i s h Columbia, 1957 MONDAY, JUNE 29, 1964 AT 10 A.M. IN ROOM 3332, BIOLOGICAL SCIENCES BUILDING of DONALD EVAN McALLISTER COMMITTEE IN CHARGE Chairman: F. H. Soward J. R. Adams R. V. Best I. McT. Cowan P. A. Dehnel H. D. Fisher W. S. Hoar P. A. Larkin C. C. Lindsey External Examiner: S. H. Weitzman United States National Museum Washington, D.C. THE EVOLUTION OF BRANCHIOSTEGAL RAYS IN TELEOSTOME FISHES ABSTRACT The o r i g i n , function and evolution of the branchio-stegal rays, the r e l a t e d opercular and gular bones and associated hyoid elements were investigated i n teleo-stome f i s h e s . A l i z a r i n , s k e l e t a l or a l c o h o l i c speci-mens of over h a l f the l i v i n g f a m i l i e s (over 240) and a l l the l i v i n g orders of teleostome fishes with branchio-stegals were examined. L i t e r a t u r e provided data for most of line remaining l i v i n g and f o s s i l f a milies and orders. Several evolutionary trends became apparent; a ten-dency for number of branchibstegals to decrease, follow-ing W i l l i s t o n ' s Law; increasing separation of mandibular and hyoid arches; and an increase i n number and complexity of hyoid elements. In the development of hyoid elements, but not of branchiostegal rays, the ontogenetic sequence p a r a l l e l e d the phylogenetLc sequence. Examination of the unusual adult hyoid r e l a t i o n s h i p s i n the neotenic S c h i n d l e r i a showed i t to resemble the l a r v a l condition of normal f i s h e s . The condition i n the t i n y goby, Mistichthys, i s s i m i l a r . The structure of the branchiostegal series and hyoid elements proved valuable i n tracing the r e l a t i o n s h i p s of f i s h e s . Major findings include: Hiodontidae were found not be be c l o s e l y related to the Notopteridae; the Notopteridae and osteoglossoid families to be r e l a t e d to the mormyriforms; the Neoscopelidae and Myctophidae to d i f f e r from other myctophiforms; the ophidioids to require ordinal separation from the Perciformes and-placement near the Gadiformes and Ateleopiformes; .-the Amblyopsidae to belong i n the Percopsiformes; the Anabantoidei and Ophicephaloidei to be c l o s e l y r e l a t e d suborders of common ancestry deserving placement i n the same order; the Beloniformes to d i f f e r from most other orders i n the loss of the interhyal and upper hypohyal; and the Echeneiformes to d i f f e r from most Perciformes i n the possession of 8-11 branchiostegals. The number of branchiostegals was found to be influenced by posterior extension of the jaws, small body length, feeding habits, g i l l membrane attachment and deepsea existence. GRADUATE STUDIES F i e l d of Study: Zoology Ichthyology Marine Zoogeography Systematics Evolutionary Mechanisms Developmental Genetics Other Studies: Contemporary L i t e r a t u r e Philosophical Problems J . C. Briggs J . C. Briggs C. C. Lindsey R. M. Bailey (Michigan) R. R. M i l l e r (Michigan) H. C. Lewis B, Savery PUBLICATIONS M c A l l i s t e r , D.E. 1960. Keys to the marine f i s h e s of A r c t i c Canada. Nat. Mus. Canada, Natural H i s t . Pap. 5: 1-21. M c A l l i s t e r , D.E. 1960. Le Gasterosteus wheatlandi, nouv e l l e espece de poisson pour l a province de Quebec. Le N a t u r a l i s t e Canadien, 87(5):117-118. M c A l l i s t e r , D.E. 1961. The o r i g i n and status of the deepwater s c u l p i n , Myoxocephalus thompsonii, a n e a r c t i c g l a c i a l r e l i c t . Nat. Mus. Canada, B u l l . 172: 44-65. M c A l l i s t e r , D.E. and C C . Lindsey. 1961. Systematics of the freshwater s c u l p i n s (Cottus) of B r i t i s h Columbia. Nat. Mus. Canada, B u l l . 172: 66-89. M c A l l i s t e r , D.E. 1962. F i s h remains from Ontario Indian S i t e s 700 to 2500 years o l d . Nat. Mus. Canada, N a t u r a l H i s t . Pap. 17: 2-6. M c A l l i s t e r , D.E. 1963. A r e v i s i o n of the smelt f a m i l y , Osmeridae. Nat. Mus. Canada, B u l l . 191: 1-53. THE EVOLUTION OF BRANCHIOSTEGAL RAYS IN TELEOSTOME FISHES ABSTRACT The origin, function and evolution of the branchiostegal rays, the related opercular and gular bones and associated hyoid elements were investigated in teleostome fishes. Alizarin, skeletal or alcoholic specimens of over half the l iving families (over 2/+0) and a l l the living orders of teleostome.fishes with branchiostegals were examined. Literature provided data for most of the remaining living and fossil families and orders. Several evolutionary trends became apparent; a tendency for number of branchiostegals to decrease, following WiHiston fs Law; increasing separation of mandibular and hyoid arches; and an increase in number and complexity of hyoid elements. In the development of hyoid elements, but not of branchiostegal rays, the ontogenetic sequence paralleled the phylogenetic sequence. Examination of the unusual adult hyoid relationships in the neotenic Schindleria showed i t to resemble the larval condition of normal fishes. The condition in the tiny goby, Mistichthys, is similar. The structure of the branchiostegal series and hyoid elements proved valuable in tracing the relationships of fishes. Major findings include: Hiodontidae were found not to be closely related to the Motopteridae; the Notopteridae and osteoglossoid families to be related to the mormyriforms; the Neoscopelidae and Myctophidae to differ from other myctophiforms; the ophidioids to require ordinal separation from the Perciformes and placement near the Gadiformes and Ateleopiformes; the Amblyopsidae to belong in the Percopsiformes; the Anabantoidei and i i i Ophicephaloidei to be closely related suborders of common ancestry deserving placement in the same order; the Beloniformes to dif f e r from most other orders i n the loss of the interhyal and upper hypohyal; and the Echeneiformes to dif f e r from most Perciformes in the possession of 8-11 branchiostegals. The number of branchiostegals was found to be influenced by posterior extension of the jaws, small body length, feeding habits, g i l l membrane attachment and deepsea existence. iv NEW TAXA Hiodontoidei, new suborder P» 151 Ichthyotringidae, new name p. 197 Barbourisioidei, new suborder p. 202 Gymnarchoidea, new superfamily.... p. 224 Ophidiiformes, new order. . . . p. 311 V TABLE OF CONTENTS INTRODUCTION p. 1 Literature p. 3 Definitions p. 4 Classification p. 11 METHODS p. 11 EMBRYOLOGY p. 13 FUNCTION p. 16 Respiration p» 17 Feeding p. 18 Behavior p. 19 Protection of Gil ls p. 21 Defense p. 21 THE ORIGIN OF THE BRANCHIOSTEGAL SERIES p. 23 The branchiostegals p. 23 The operculum and suboperculum p. 26 The interoperculum p. 27 The gular plates p. 28 PHYLOGENETIC OR VERTICAL EVOLUTION p. 31 CLASS TELEOSTOMI p. 32 Key to the classes of Gnathostomata p. 33 Key to the subclasses of Teleostomi p. 34 Key to the orders of Crossopterygii p. 34 Key to the orders of Dipneusti p. 35 Key to the orders of Actinopterygii - p. 36 SUBCLASS CROSSOPTERYGII p. 41 /Order Hoploptychiformes p. 41 /Order Osteolepifonnes p, 43 Order Coelacanthiformes p. 48 SUBCLASS DIPNEUSTI p. 53 /Order Dipteriformes p. 55 /Order Phaneropleuriformes p. 58 /Order Uronemiformes p. 60 /Order Ctenodontiformes p. 61 vi SUBCLASS DIPMEUSTI (Cont'd) Order Ceratodiformes P» 63 Order Lepidosireniformes P» 64 SUBCLASS BRACHIOPTERYGII P« 6 ? Order Polypteriformes P» 67 SUBCLASS ACTINOPTERYGII P« 6 9 GROUP I. CHONDROSTEI P' 7 1 /Order Palaeonisciformes P* 71 /Order Tarrasiiformes P" 7 9 /Order Phanerorhynchiformes P* 8 0 /Order Haplolepiformes ^* f /Order Redfieldiiformes P. 83 /Order Perleidiformes P» 85 /Order Dorypteriformes P* 87 /Order Bobasatraniifonnes P» 8 8 /Order Pycnodontifornies P* 89 /Order Ptycholepiformes P» 91 /Order Pholidopleuriformes P» 92 /Order Cephaloxenifoi-mes P» 94 /Order Aethodontiformes , P* 95 /Order Luganoiiformes P» 96 /Order Peltopleuriformes p« 96 /Order Platysiagiformes P» 98 /Order Chondrosteiformes P« 99 /Order Saurichthyiformes P» 1^ 1 Order Acipenseriformes P» l ^ 2 GROUP II. HOLOSTEI P» 1 0 6 a /Order Ospiiformes P» l°6a Order Amiiformes ' P» ^ 9 Order "Lepisosteiform.es P* -L19 /Order Aspidorhynchiformes P* /Order Pachycormiforraes P # -*-23 /Order Pholidophoriformes P* ^-24 GROUP III. TELEOSTEI P* 1 2 8 MALACOPTERYGII P" 1 2 8 Order Clupeiformes P« 1 29 Order Myctophifomies P» 184 Order Motacanthiformes P* 2^6 Order Giganturiformes P» 2 ^ 2 Order Saccopharyngiformes P« 214 v i i SUBCLASS ACTINOPTERYGII . (Cont»d) Order Mormyriformes P« 216 Order Cyprinif onnes P« 226 Order Anguillifonnes P» 251 Order Beloniformes P» 273 ACANTHOPTERYGII P- 280 Order Berycifonnes P» 281 Order Lampridifonnes P» 291 Order Zeiformes P» 296 Order Bathyclupeiformes P» 301 Order Syngnathifonnes p. 303 Order Ophidiiformes P» 311 Order Ateleopifonnes . P» 319 Order Gadifonnes P» 322 Order Percopsiformes P« 329 Order Cyprinodontifonnes P» 335 Order Pleuronectiformes P« 340 Order Perciformes p. 346 Order Gasterosteifonnes P» 373 Order Icosteiformes p. 376 Order Echeneiformes P» 378 Order Tetraodontifonnes P* 381 Order Mastacembeliformes P» 389 Order Synbranchiformes P» 393 Order Lophiiformes P» 395 Order Batrachoidifonnes p. 407 Order Gobiesocifonnes P« 408 Order Pegasifonnes P» 413 DISCUSSION OF RESULTS P» 417 ADAPTIVE EVOLUTION P« 422 Accessory respirator;'- organs P» 422 Filter feeding apparatus P» ^2° Size P« 427 Jaw length P« 431 G i l l membrane attachment P» 433 Deep sea P» 434 SUMMARY AND CONCLUSIONS p. 436 LITERATURE CITED p. 439 v i i i LIST OF TABLES Table 1. Summary of data oh branchiostegal series and hyoid arch in teleostome fishes (page 419). Table 2 . Relation between aerial respiratory organs and branchiostegal number and g i l l membranes (page 424). Table 3. Number of branchiostegals and g i l l membrane attachment in some giant fishes (ten feet or longer) (page 428). Table 4. Branchiostegal number and g i l l membrane attachment in miniature fish (groups having many species of three or less inches) (page 429). ILLUSTRATIONS Fig. 1. General features and relative positions of branchiostegals, operculars, gulars, hyoid arch and g i l l membranes. (Follows page 4)* Fig. 2. Percopsid projections (arrow) in a percopsiform, gadiform and ophidiiform (top to bottom). (Follows page 3H). Fig . 3» Opercular spines in Anabas and Holocentrus. Note the similarity, although the suboperculum of Holocentrus differs in occluding the posterior border of the operculum. (Follows page 356). The following plates at end of thesis. P l . I . Branchiostegal series in the Crossopterygii. P l . II. Branchiostegal series in the Dipnoi. P l . III. Branchiostegal series in the Actinopterygii (chondrostei). P l . IV. Branchiostegal series in the Actinopterygii (Chondrostei and Holostei) and Brachiopterygii. P l . V. Branchiostegal series in the Actinopterygii (Chondrostei and Holostei). P l . VI. Branchiostegal series and hyoid arch in the Actinopterygii (Mai acopterygii ) • P l . VII. Branchiostegal series and hyoid arch in the Actinopterjrgii (Holostei and Teleostei). (In these and the following photographic plates retouching was used to improve definition). PI. VIII. Branchiostegal series and hyoid arch in the Actinopterygii (Maiacopterygii). PI. IX. Branchiostegal series and hyoid arch in the Actinopterygii (Maiacopterygii). PI. X. Branchiostegal series and hyoid arch in the Actinopterygii (Malacopterygii and Acanthopterygii). PI. XI. Branchiostegal series and hyoid arch in the Actinopterygii (Acanthopterygii). PI. XII. Branchiostegal series and hyoid arch in the Actinopterygii (Ac anthopterygii)• PI. XIII. Branchiostegal series and hyoid arch in the Actinopterygii (Acanthopterygii). PI. XIV. Branchiostegal series and hyoid arch in the Actinopterygii (Acanthopterygii). PI. XV. Branchiostegal series and hyoid arch in the Actinopterygii (Acanthopterygii). PI. XVI. Branchiostegal series and hyoid arch in the Actinopterygii (Acanthopterygii). P l . XVII. Branchiostegal series and hyoid arch in the Actinopterygii (Ac anthopterygii). P l . XVIII. Evolutionary relationships of the teleostome fishes. Based on the study of the branchiostegal series and hyoid arch and on other characters. ACKNOWLEDGEMENTS My deepest gratitude is due Dr. J . C. Briggs under whom this study began and Dr. R. H. Rosenblatt under whom the final drafts were written. Drs. N. J . Wilimovsky, C. C. Lindsey, V. J . Krajina, P. A. Dehnel and P. A. Larkin also criticized the manuscript. Many generous people donated, loaned or exchanged rare groups of fishes, greatly adding to the scope of the study. These include Dr. W. Aron, then at the University of Washington, Dr. R. M. Bailey of the University of Michigan, Mr. G. Bell-Cross of the Northern Rhodesia Department of Game and Fisheries, Dr. E . J . Crossman of Royal Ontario Museum, Dr. W. A. Gosline of the University of Hawaii, Mr. R. Kanazawa of U.S. National Museum, Mr. E . D. Lane then of the Fisheries Laboratory, Wellington, New Zealand, Dr. A. G. K. Menon of the Zoological Survey of India, Dr. G. S. Myers of Stanford University, Dr. R. H. Rosenblatt of Scripps Institution of Oceanography, Dr. L . P. Schultz of U.S. National Museum, Dr. W. B. Scott of Royal Ontario Museum, Dr. F . H. Talbot of the South African Museum, Dr. W. Templeman of the Biological Station, St. John's, Newfoundland, Dr. J . Thiemmeda of Uasetsart University, Thailand, Dr. A. D. Welander of the University of Washington, Mr. P. J . Whitehead of the British Museum (Natural History). Miss Audrey Dawe, librarian of the National Museum of Canada, was untiring in seeking out rare journals. Mr. 5. Gorham of the National Museum of Canada prepared a series of excellent skeletons. The author is grateful to Dr. L.S. Russell and Dr. A. W. F . Banfield for their continued support of this project. To a l l these persons the author extends his most sincere thanks. INTRODUCTION This is a study of the evolution of branchiostegals in teleostome fishes. The study considers the vertical evolution of branchiostegals with special regard to ordinal phylogeny of teleostome fishes. It also considers the horizontal evolution of branchiostegals in regard to the effect of internal and environmental factors. The homologous structures, the opercular bones and gular plates, and the elements of the hyoid arch (except the hyomandibular, urohyal, and the glossohyal)"*" upon \\rhich the branchiostegals insert, are also studied, although in less detail. The origin, development, and function of the branchiostegals are also dealt with. The central problem considered is to what extent the branchiostegals reflect the phylogeny of the teleostome fishes. Lh the evolution of a group, a morphological structure has three courses. It may remain the same or almost the same. It may do this even i f other structures are evolving rapidly. Secondly, i t may advance through modification of form, through complication or addition. Thirdly, i t may degenerate through simplication, loss or disappearance. Al l of these courses are of value to the student of phylogeny. Possession of a structure nearly similar by two groups, supposing the similarity is not due to parallelism, suggests relationship and common origin. Differences wrought through evolutionary change towards complexity or simplicity suggest placement in different phylogenetic lines or in different taxa, the closeness of relationship depending on the degree of difference. Through this method a sequence of relationships may be built up, and These bones can be important systematically but are not directly associated with the branchiostegals. 2 through examination of fossils and determination of primitive characters the sequences can be transformed into a phylogenetic tree or dendrogram. A clear picture of phylogeny cannot be gained through examination of a single character. Because of parallel evolution and because of different rates of evolution in different structures a single character may lead one astray in tracing phylogeny. A clear picture of phylogeny is based on as many firm taxonomic characters as possible. This reduces the probability of confusing homologous and analagous characters and presents a picture of the evolution of the whole animal and not just one port. Therefore, in this study the relationships suggested by the branchiostegal series (operculars, branchiostegals, gulars) and the hyoid arch, are checked with other sound taxonomic characters (from literature or original observations). The concentration which the study of a single characters complex enables has advantages over the studying of many characters simultaneously. Parallelisms are more readily noted. Smaller differences are less l ikely to be ignored. Some valuable single character complex studies may be noted: Hubbs (1920) on the branchiostegals, Starks (1930) on the bones of the shoulder girdle and (1926) on the ethmoid bones, Burne (1909) and Derscheid (1924) on the olfactory organ, Whitehouse (1910) and Gosline (i960), (1961) on the caudal skeleton, Emelianov (1935) on the ribs, Frost (1925), (1926), (1927), etc. on the otloliths, Stensio (1947) on the sensory canals of the head, Lindsey (1956) on the vertical fins. Holstvoogd (1963) on the retractores arcuum branchialium 3 and Freihofer on the ramus lateralis accessorius.^ None of these studies ha3 included a l l the orders and only one more than 100 families. It is from the combination of detailed studies such as these and others yet to be done that the ultimate picture of the phylogeny of the Teleostomi wi l l be constructed. Towards this end the present paper hopes to contribute. Literature Branchiostegal counts have been used in fish taxonomy even before the time of Linnaeus. The more careful ichthyologists such as Day, Regan, Smitt and Jordan have included branchiostegals in their descriptions and analyses of fishes, as have G i l l , Ryder and Starks in their osteological studies. There have been few studies of branchiostegals on the broad scale however. Bertelsen and Marshall (1956) discuss the number and arrangement of branchiostegals in some of the malacopterygian orders while placing the Miripinnati. The only other study, and the most important one, is that of Hubbs (1920). This concise study enumerates many of the important evolutionary changes in the branchiostegals of living fishes. Hubbs noted the tendency for decrease in branchiostegals during evolution (noted previously as early as 1904), and the differences in shape and arrangement of the branchiostegals of malacopterygians and acanthopterygians. He noted the malacopterygian nature of the branchiostegals in the Synentognathi, and the acanthopterygian nature of the branchiostegals in the Microcyprini, Symbranchia and Opisthomi. Two further valuable studies in progress may be noted, that of Quentin Bone, Marine Laboratory, Plymouth on lateral muscle innervation and G. Nelson, University of Hawaii on the branchialarches. 4 He based his conclusions on the examination of about 140 families distributed in less than twenty orders, but gave no figures, and described only a few in detail. The hyoid arch, aside from papers on the osteology of single species, has received little.attention from the broad comparative point of view in fishes. Corsy (1933) studied the evolution of the hyoid arch of vertebrates but only a small portion of this study was devoted to teleostome fishes. Khanna (1961) described the hyobranchial skeleton of some Indian fishes. AUis (1915) and (1928) and Edgeworth (1926) and (1931) comment on the lox\rer elements in discussions upon the homologies of the hyomandibular. This paper tries to extend these and other comparative studies by the inclusion of fossil groups, the examination of every order of fishes with branchiostegals and as many families as possible (over half of the . ( l iving families),"'" the presentation of data in a consistent style, the illustration of many forms, and in the interpretation of the data in terms of phylogeny, adaption and parallelism. Definitions To introduce the subject and to clarify the terminology, the technical terms pertinent to the study are here defined, synonyms listed and structures are illustrated (fig. l ) . The terminology of Harrington 1 ' — — — — — Over 240 of the 402 l iving families and over 330 species and 700 specimens examined. Counts lacking on only 16 living and 42 fossil families ( i .e . data (own & literature) available for 452 of total of 510 living and fossil teleostom.3 families). A l l l iving orders with branchiostegals were examined. 5 (1955) is used, as far as i t applies. Starks (1901) may be referred to for the older synonyms. Two terms, spathiform and acinaciform are introduced for the f irst time. Branchiostegals or branchiostegal rays; These are the struts of dermal bone (sometimes cartilage) below the operculars, that insert on the hyoid arch and support the g i l l membrane of Teleostomi. The term lateral gulars is sometimes misapplied to the broad branchiostegals of Chondrostei. The branchiostegals are here believed to be homologous with the hyoid rays of Acanthodii and ELasraobranchii (fused into the "opercular plate" of chimaeras). Branchioperculum: This is the enlarged uppermost branchiostegal in Amia. Jugostegelia: These are the branchiostegals free from the hyoid arch and overlapping, found in the anguilliform families, Echelidae, Ophichthidae and Neenchelyidae. Spathiform: Applies to the broad, laminar, paddle-shaped branchiostegals found in some malacopterygian and most of the more primitive teleostomes. From the Latin spatha or spatula. E.g. branchiostegals of Amia. Acinaciform: Applies to the slender, non-laminar, sword or scimitar-like branchiostegals found in some malacopterygians and most actinopterygians. From the Latin acinaces or scimitar. E.g. branchiostegals of Perca. Operculum: The large dermal bony (sometimes cartilaginous) upper element of the g i l l cover behind the preoperculum and inserting on the hyomandibular in teleostomes. Believed to be an enlarged branchiostegal. Ventral View GEKERAL VICt£ CP BRANCH 06 TEGALS (ID Amia calva) BRANCH!OSTEGAL AND HTOID ARCH TERMINOLOGY GUI Membranes Separate G i l l Membranes United G i l l Membranes Joined G i l l Membranes Joined G i l l Opening G i l l Membrane Joined and Free From Isthmus to Isthmus - Narrowly to Isthmus - Broadly Restricted to Isthmus and Forming a Free Fold Over I t GILL MEMBRANE ATTACHMENT Fig. 1. General features and relative positions of branchiostegals, operculars, gulars, hyoid arch and g i l l membranes. 7 Suboperculum; The large, dermal, bony (sometimes cartilaginous) g i l l cover element below the operculum in teleostomes. Believed to be an enlarged branchiostegal. Interoperculum; The large, dermal, bony (sometimes cartilaginous) element below the lower arm of the preoperculum and anterior to or under the front half of the suboperculum. It typically is connected to the mandible, epihyal and suboperculum. Believed here to be an enlarged branchiostegal. Found only in higher teleostomes (Group II and above). Operculars; Collective terra for the operculum, suboperculum and, i f present, the interoperculum. G i l l membrane or branchiostegal membrane: The membrane lying between the operculars and the isthmus which is supported by the branchiostegals. The g i l l membranes may be variously connected to the isthmus. If the g i l l membranes are not connected to one another or the isthmus and overlap anteriorly they are termed separate (e.g. Salmo, Sphyraena). If the g i l l membranes are attached to one another, yet not joined to the isthmus (at least posteriorly), then they are said to be united and free from the isthmus (e.g. Polyodontidae, Notopteridae); they may in this case have a small or large free border posteriorly (whether the posterior edge is free may be determined by running a needle under i t ) . In the Synbranchifonnes the g i l l openings are united and free from the isthmus but dorsally attached to the body before the pectoral f in; this gives the appearance of a single ventral g i l l opening but is really only a special case of being united and free. Lastly, the g i l l membranes may be joined to the isthmus (e.g. most Cyprinidae, Gobiidae). They may be narrowly joined to the isthmus (e.g. Gasterosteus) 8 or joined to the isthmus anteriorly in which case the g i l l opening i3 wide, or they may be broadly joined to the isthmus (e.g. most Cyprinidae) in which case the g i l l opening is narrow and the space between the g i l l openings is wide, or the g i l l opening may be restricted (e.g. Anguillidae) in which case the g i l l opening is reduced to a small aperture on the side of the head. When the g i l l membranes are joined to the isthmus a fold may form across the isthmus between the g i l l s l its (e.g. Myoxocephalus polyacanthocephalus) which secondarily resembles the g i l l membranes united and free condition. The preceding terms have not been used consistently by some authors but the definitions and illustrations given here should make the distinctions clear and should help standarize the terminology. Median gular; A median bony plate extending backwards from the symphysis between the mandibles. It is bordered posteriorly by the lateral gulars or branchiostegals, i f present. Primitively i t bears a v-shaped pit l ine. It may be homologous with the branchiostegals. E.g. Amiaf Elops. A second median gular, posterior to the normal median gular is found in some Dipneusti. Lateral gulars; Pair(s) of bony plates, larger than the branchiostegals, lying between the median gular and branchiostegals (when present), and inserting on the hyoid arch medial to the mandibles. It is believed here to be homologous with the branchiostegals. Primitively bears a transverse pit l ine. E.g. Polypterus, Calamoichthys, Latimeria. In some Dipneusti there may be two pairs, an anterior and a posterior pair. 9 Gulars: Collective term for median and lateral gulars. Branchiostegal series: Collective term for the operculars, branchiostegals and gulars (al l branchiostegal derivatives). Interhyal: Endochondral bony or cartilaginous, usually cylindrical element in teleostomes connecting the lower portion of the hyoid arch to the hyomandibular. It acts as a pivot. It may or may not be homologous \-rith the tetrapod stylohyal, which term has been used for i t . The interhyal typically inserts on the lower tip of the hyomandibular. Paired. Epihyal: Endochondral bony or cartilaginous usually triangular element in higher teleostomes (Group II and III) lying between the ceratohyal and interhyal. Probably derived from the ceratohyal. Paired. Ceratohyal: Endochondral bony or cartilaginous usually hourglass-shaped element in teleostomes lying between the epihyal, i f present, or interhyal and the hypohyal(s). Paired. Found in Acanthodii, Elasmobranchii and Teleostomi. Kypohyal(s): Endochondral bony or cartilaginous element(s) lying below the ceratohyal and lateral to the glossohyal, in Acanthodii and Teleostomi. In the Crossopterygii, Dipneusti and lower Actinopterygii (Group IIB and lower) the hypohyal is single; in the higher Actinopterygii (except where secondarily degenerate) there is a lower (hypohyal l ) and an upper (hypohyal 2) hypohyal. Called basihyals by some authors. Hyoid arch: Restricted in this study to include only the interhyal, epihyal, certaohyal and hypohyals (since the other hyoid elements, the hyomandibular and basihyal (= glossohyal) were not included in this studjO. 10 Sutured: The epihyal and ceratohyal are termed sutured i f they are joined by interdigitating prongs emitted by each bone, while they are termed separate i f not so joined. Clupeoid projection: This is a bulge on the anterior edge of the base of branchiostegals in clupeoids and their derivatives. Percopsoid  projection is an angulation on the anterior branchiostegal base (see f ig . 3). Beryciform foramen: This is a perforation above the midsection of the centre of the ceratohyal found in beryciform fishes and some of their derivatives. In some fishes the roof of the foramen is lost and only a notch appears on the dorsal edge of the ceratohyal. The foramen perforates the groove along which the hyoid artery runs on the outer face of the ceratohyal. Positional terms: Since the hyoid arch may be almost horizontal or almost vertical the branchiostegals towards the interhyal end of the arch may either be called dorsal or posterior branchiostegals. The branchiostegals toward the hypohyal end of the arch may similarly be called ventral or anterior branchiostegals. In numbering the branchiostegals the uppermost (or posteriormost) provided the starting point since the lower (anterior) branchiostegals are the most variable and do not provide as stable a point of enumeration. The portion of the branchiostegal inserting on the hyoid arch is termed basal, the opposite end the distal t ip . The two long edges may be called the ventral (anterior) edge and the dorsal (posterior) edge. According to their insertion branchiostegals are divided into epihyal and ceratohyal 11 (sometimes interhyal and hypohyal), external and ventral or internal branchiostegals. When a branchiostegal straddles the epihyal-ceratohyal or a ceratohyal-hypohyal joint a half a branchiostegal is awarded to each (recorded in descriptions and tables as 2 ) . Classification The basis of the classification used herein is that of Berg (1947; 1955)* Modifications of this classification were made from the later literature. Changes, sometimes considerable, were also made in the classification of living fishes, mainly in the rearrangement of orders and status of certain groups, as suggested by this study and data from other studies. The uniform -iformes ending was adopted for orders, -oidei for suborders, -idae for families and -inea for subfamilies. METHODS The branchiostegal series and hyoid bones were examined by several methods. The principal method was by clearing and staining xvith alizarin following the method of Hollister (1934). Specimens preserved in alcohol usually from three to six inches but sometimes a3 short as one inch (Phallostethidae) or as long as sixteen inches were employed for staining. The stained hyoid arch with the branchiostegals was usually removed and examined. Some were photographed under a binocular microscope. In quite a few cases dermestid-cleaned skeletons were examined; here caution must be employed since branchiostegals may be lost during skeletonizing. Alcoholic specimens were dissected for examination of the arch and branchiostegals. From other alcoholic specimens branchiostegal counts were taken without dissection. In the 12 latter method one mu3t check to see i f the uppermost branchiostegal is hidden under the suboperculum or whether the lowest branchiostegal, which may be quite small, is not obscured by the skin. An attempt was made to examine at least one specimen from as many families as possible. Representatives were chosen by availability except that as many families and suborders as possible were examined. Branchiostegals were counted on the left . When the branchiostegals were abnormal (see Crossman, I960, for examples) the counts were not recorded. E.g. fused, bent, or irregularly placed branchiostegals. For families not examined and to supplement families examined, counts were obtained from the literature. Information on fossils was wholly obtained from the literature. These sources are included in References under the family. Synonyms of taxa follow enclosed in parentheses. Fossil groups are indicated by the sign Observations were made at the Vancouver Public Aquarium to gain an understanding of the movements of the elements in a living fish. The principal sources of material were the National Museum of Canada, Ottawa (NMC), and the museum of the Institute of Fisheries, Universitj' of British Columbia (BC). Other material was borrowed from Stanford University (SU), and Roj^ al Ontario Museum (ROM), British Museum (BM), Scripps Institution of Oceanography (SIO), and University of Michigan Museum of Zoology (UMMZ), or examined at the United States National Museum (USNI-l). Acknowledgements note other sources. 13 EMBRYOLOGY The hyoid arch develops from the second visceral arch, the f irst becoming the jaws, the more posterior ones supporting g i l l 3 . The following, from de Beer (1937)* Wade (1962) and original observations on Amia appear to be the usual order of development: l ) ventral extension of a cartilaginous hyomandibular-symplectic rod from the auditory capsule and appearance of a ceratohyal cartilage below this, 2) separation of the hyomandibular-symplectic cartilage from the auditory capsule, 3) appearance of the interhyal half or three quarters of the way up the lower side of the hyomandibular-symplectic cartilage and appearance of a hypohyal, /+) appearance of the upper, then the lower branchiostegals, appearance of the median gular, operculum, suboperculum and interoperculum, 5) ossification of the ceratohyal in the anterior and epihyal in the posterior end of the ceratohyal cartilage, and of the symplectic and the hyomandibular in their cartilage, 6) ossification of the upper and lower hypohyal in the hypohyal cartilage, 7) suturing of epihyal and ceratohyal (delayed t i l l adult in Anarhichas). In different groups the exact sequence may vary and of course not a l l of these elements are found in a l l teleostomes. It is interesting to compare the ontogenetic and phylogenetic appearance of the hyoid elements. This is done below: Ontogeny Phylogeny l ) ceratohyal cartilage hypohyal and ceratohyal 2) a hypohyal and interhyal cartilage interhyal 3) epihyal and ceratohyal ossification epihyal 14 4) upper and lower hypohyal ossification upper and lower hypohyal 5) suturing of epihyal and ceratohyal suturing of epihyal and ceratohyal In comparing the sequence of appearance of the hyoid elements in the table above i t may be seen that the embryonic sequence of appearance of every element follows the phylogenetic sequence of origin in every case except that of the hypohyal which is delayed to the next stage in embryonic development. The embryonic development of the hyoid arch is also valuable because i t suggests the two hypohyals develop from a single precursor, the cartilaginous hypohyal, a point which the positional relationships of the two adult hypohyals would tend to confirm. The embryonic development of the epihyal from the ceratohyal would also appear to be a morphologically and phylogenetically plausable development. The development of the branchiostegals does not appear to recapitulate phylogeny. In the higher teleosts numerous branchiostegals do not appear and then secondarily diminish to a reduced number, perhaps because of selection at the embryonic stage. Nor do the embryonic branchiostegals of higher teleosts commence embryonically as spathiform and then change to acinaciform shape: instead, they commence as acinaciform. Embryology enables one to explain one of the unusual characters of the Schindlerioidei, Gosline (1959) states the Schindlerliidae were, as far as he knew, unique in having the epihyal inserting on the upper head of the hyomandibular. However, as noted above, the interhyal 15 commonly inserts high up on the lower side of the hyomandibular-symplectic cartilage embryonically; the point at which i t inserts marks the later point of division of the lower end of the hyomandibular and the upper end of the symplectic. Development in the neotenic Schindleria appears to have, ceased at a point before the cartilage differentiated into the hyomandibular and symplectic; the epihyal therefore appears to insert on the upper head of the "hyomandibular" (= hyomandibular-symplectic cartilage). Thus, the position of the hyoid arch and "hyomandibular" is unique only that i t is found in the adult, and this is not surprising since one expects to find larval conditions in a neotenic fish. An unusual condition is that a hypohyal is not present, although the epihyal has made its appearance. A similar condition is found in the minute goby, Mystichthys luzonensis where the interhyal inserts above the ventral end of the hyomandibular. Another unusual condition that embryology elucidates is the origin of the mental barbels in the Mullidae. Here there are only 3 normal branchiostegals on the external face of the hyoid arch. At the anteriormost end of the ceratohyal close to the symphysis in the adult is a slightly twisted osseous ray which becomes attenuate and cartilaginous distally. This ray has been suggested to be a branchiostegal (Starks, 1 9 0 4 ) , thus accounting for the missing fourth branchiostegal one expects on the external face of the hyoid arch. Lo Bianco ( 1 9 0 ? ) and Caldwell ( 1 9 6 2 ) investigated the development of mullids. At 6 - 8 mm. 4 slender branchiostegals were found in the normal position, at 1 1 - 1 5 ram. the fourth branchiostegal thickened, at 2 2 inm. the branchiostegal began moving anteriorly and medially, to achieve at 3 0 mm. 16 nearly the adult mental condition. Here the study of development confirms the theory that the mental hyoid ray is indeed a branchiostegal. It would be interesting to follow the development of Polymixia, an unrelated form which has a hyoid barbel believed to be supported by branchiostegals. Another worthy problem would be the close following of the development of the median gular to compare i t with branchiostegals development for evidence on whether or not they are homologous. A series of young Amia calva were examined from two Ontario collections (NMC53-192-3, Pt. Pelee, Lake Erie, 20.6-26.7 mm. standard length; NMC58-209, Georgian Bay, Lake Huron, 44.5-50.1 mm. length). The following relationship between standard length and number of branchiostegals was found: Standard Length (mm.) 21 23 25 .26 27 44 46 47 48 49 50 No. Branch-iostegals 5 7 7-8 8-9 9 8-13 12 13 12 12 10-12 These data would suggest that by 46 mm. standard length the adult number of 10-13 branchiostegals i s attained. Gasterosteus at 25 mm. have attained the adult number of 6 (Runyan, 1961); Neostethus of 25 mm. had attained the adult complement of 5 (own observation). It would seem advisable not to take branchiostegal counts as representative of the adult condition from very young specimens. FUNCTION OF TIDS BRANCHIOSTEGAL SERIES The bones of the branchiostegal series may function in five ways, in respiration, in feeding, i n sensing, in protection of the g i l l s and in protection or defense of the f i s h . These functions are discussed below. 17 Respiration Respiration in teleosts is accomplished by maintaining an almost continuous flow of water over the g i l l filaments. The buccal cavity creates a positive pressure before the gi l ls and the operculum and branchiostegals create a negative pressure behind the g i l l s . The cycle is summarized (from Hughes and Shelton, 1958) in four phases: l ) Water is drawn into the mouth past the open buccal valves by negative pressure created by.dropping the lower jaw, which then starts to close. As this happens the operculum is abducting \n.th the opercular cavity closed by the branchiostegal membrane (opercular valve), thus creating a negative pressure and drawing water through the gi l l s from the buccal cavity. 2) As the mouth closes the buccal cavity is reduced; water leaves the mouth until the buccal valves close, causing a rise in pressure. As this happens pressure in the opercular cavity is less negative because of flow through the g i l l s . A3 the operculum reaches the end of its abduction the pressure within the opercular cavity equals the external pressure and the branchiostegal membrane opens. 3) The operculum begins to abduct with a gap between i t and the flank, through which water exits. The mouth closes increasing buccal cavity pressure, forcing \tfater over the g i l l s with l i t t l e loss out the mouth as the l ips are close and the buccal valves effective, k) The mouth begins to open increasing the volume of the buccal cavity and opening the buccal valves, and dropping the pressure in the buccal cavity. The operculum continues to abduct raising the opercular cavity pressure as the gap between the operculum and flank decreases. At this point there is a tendency for reversal of 18 water flow across the g i l l s . From this resume i t may be seen that the branchiostegal membrane functions as a valve to seal the g i l l opening during abduction of the operculum. Secondly, i t acts as an expansable wall (comparable to the rib cage) which permits the volume of the opercular cavity to enlarge and allo>?s a grea.ter volume of water to be drawn through the g i l l s . This is accomplished by spreading the branchiostegals (much like unfolding a fan). Intermittently the branchiostegals take part in a coughing reflex. Mere water flow is reversed through the gi l l s to clean them of debris. It is possible that branchiostegals function also in aerial respiration. Under conditions of low oxygen, surface dwelling physostomous fishes may inhale air bubbles. It is possible that abduction of the branchiostegals may assist in this process. It is notable that the branchiostegals of surface dwelling physostomous fish are broad imbricating structures while those of deepwater or physoclistous fish are slender and non-imbricating. It is possible that reflexing the spathiform branchiostegals assists in inhaling bubbles of air. Depression of gular bones may assist in inhaling air also. Feeding Many fish feed by inhaling the food particles into the mouth. In macrophagus fish this probably takes place mainly by dropping the floor of the buccal cavity, although the creation of negative pressure in the opercular cavity by the operculum and branchiostegal membrane may take some part in this. It is in the microphagus or plankton feeders that these latter actions are more important. Here water is drawn through 19 the sieving apparatus of the g i l l s , the g i l l rakers and out the long g i l l s l i t . A long g i l l membrane with numerous, branchiostegals is necessary to open and close the long g i l l s l i t during feeding and respiration. Curry-Lindahl (1956) reports that the lungfish, Protopterus, sucks its food into the mouth. He quotes an author saying this is accomplished by depression of the hyoid bone. It may be that this behavior is a holdover from ancestors which had gular plates. Gulars would help in sudden depression of mouth floor and hence sucking in of prey. Behavior The branchiostegals and their membranes have been shown to play a part in behavior of fishes by modern ethologists. The branchiostegal membranes are commonly employed in agonistic displays by fishes. Here the branchiostegals are spread and thrust laterally. For example in cichlids, "fighting begins with lateral display in which the fishes, in breeding colors, hold themselves parallel to each other, with the median fins and eventually the branchiostegal membranes erected". (Baerends, 1957). This aggressive component, raising the branchiostegals, may be employed in territorial behavior, in establishing social hierarchies (Allee, 1952) and in courtship (Morris, 1954). It may be noted that the erecting of the operculum and branchiostegal membrane has the effect of enlarging the head. This type of display is known in such fishes as Gasterosteidae, Cottidae, Cichlidae, and Anabantidae. Another behavioral function of the branchiostegal membrane can be suggested, although not yet proven. Differences exist in coloration of 2 0 related forms. Red slashes occur in the hyoid cleft on the lower side of the head in Salmo c larki i . In other species of Salmo this cleft is light colored. In Thymallus arcticus on the other hand i t i3 black. In other groups there are differences in photophore pattern on the branchiostegal membrane. .In Porichthys species may have U-shaped or V-shaped patterns of photophores (Hubbs and Schultz, 1939). Differences in the number of photophores on the branchiostegals of sternoptychids are given by Schultz (1961): Argyropelecus and Polyipnus 6, Sternoptyx 3. It is possible that the role of color is important only in agonistic behavior. However, the fact that the membrane is used in courtship and that there are interspecific differences suggests that these forms use the patterns in species recognition. By analogy i t is suggested that photophores have a function parallel to that of color. A further function is suggested by Tavolga (1958). In Bathygobius soporator the males make low-pitched grunting sounds to attract the female, apparently by forcible ejection of water through the g i l l openings (in which the branchiostegals would take part). Sensory In some fishes modified branchiostegals have a sensory function. In both Mullidae and Polymixiidae the anterior branchiostegal becomes free from the branchiostegal membrane and forms a long barbel-like structure. Of the Mullidae, Herald (1961) says "The long, tactile barbels under the chin, constantly working in the same way as a mine detector as they are dragged over the bottom, enable the goatfishes to locate small items of food that might otherwise be missed. These barbels 21 are highly flexible, often moving back and forth even when the goatfish is at rest. When not in use, the barbels can be pulled under the throat, where they are fairly inconspicuous." According to Andriashev (1944) the barbels in Hullus are also employed in digging and chemoreception. As the polymixiids are deepwater forms, l iving between 600 and 1200 feet, their habits are not well known. Through analogy with the barbels of Mullidae i t is possible to suggest that they also have a sensory function. It is difficult to conceive of any other function. In Linophryne coronata the hyoid barbel "was found to be formed of a nerve issuing from the hind corner of the mandible and of a strand of the interhyoideus muscle. Protection of Gil ls The branchiostegal series serve to protect the g i l l s . In some forms lacking branchiostegals, scales have assumed the protective function (Mesturus), in others the lateral gulars have expanded to replace them (Polypterus). Defense and Protection In certain eleotrids such as BelobranchU3, the base of one or more branchiostegals is pointed and projects from the skin. By analogy i t is conceivable that this spine or spines is used as a deterrent defensive mechanism (similar to the suborbital spines in Cobitidae, maxillary spines of Notacanthidae, caudal peduncle spines of Acanthuridae). Many acanthopterygian fishes have spines on the opercular bones, probably of similar function. Other fishes, such as Denticipitidae have small 22 spinules on the branchiostegals. The function of these is more difficult to understand. Morris (1955) has experimentally demonstrated the protective value of dorsal spines in sticklebacks. A pike which had been pricked in attempting to swallow a stickleback was less likely to make the attempt again. THE ORIGIN OF THE BRANCHIOSTEGAL SERIES The Branchiostegals In exploring the origin of branchiostegal rays i t is necessary f irst to examine the earliest teleostome fossils. Although fragmentary fossil teleostomes are found in the Lower Devonian deposits, i t is not until the Middle Devonian that adequate specimens are known. In the Middle Devonian the three major early teleostome subclasses, Actinopterygii, Crossopterygii and Dipneusti are already clearly distinguished (the subclass Brachiopterygii not being known until the Eocene). In a l l of these we find broad, flat, spathiform branchiostegals, a median gular and a pair of lateral gulars and operculum and a sub-operculum. The branchiostegal series of these subclasses are much more similar than in modern representatives of the groups, as the later Actinopterygii tend to lose the gulars while the Dipneusti and Crossopterygians lose the branchiostegals. It is also notable that the opercle and particularly the subopercle are more similar to the branchiostegals, than are the opercle and subopercle in more modern teleosts. However, i t is evident that the branchiostegal series is already quite well developed in the earliest teleostome remains and that i t is therefore necessary to examine the forerunners of teleostomes to determine the origin of the branchiostegal series. The forerunners of the teleostomes are not known with certainty. However, most evidence points towards origin from the Acanthodii or a closely related group. The Acanthodii are sufficiently ancient to be ancestral to the teleostomes (unlike the Elasmobranchii). The Acanthodii agree with the Teleostomi in possession of true bone in the skeleton, jaws, circumorbital 24 bones, ganoid type scales, similar neurocrania (Watson, 1938), shaped branchial arches, small olfactory organs and large anterior orbits. Berg (1947) indicated the two otoliths in Acanthodes are similar to palaeoniscid otoliths. Of the similarity of the scales of Acanthodii and Actinopterygii Aldinger (1937, translation) states "In contrast (to important differences between Acanthodii and Rhipidistia) the scales of Acanthodii are b u i l t after the same plan as those of the oldest Elonichthyiformes and of Cheirolepis". Both the Acanthodii and early Teleostomi are found in freshwater deposits. Arambourg (1958a), Romer (1955), H i l l s (1943) agree that the Acanthodii or forms close to i t gave rise to the Teleostomi. Berg (1947) considers ... "that the Acanthodii are a l l i e d to the Teleostomi." Watson (1938), in a revision of the Acanthodii, considered that the Acanthodians had no close relationship with the bony f i s h , xi/hile admitting, "Nevertheless there i s a most curious set of qualities i n which the members of two great groups agree." But he did regard the Teleostomi as derived from the Class Aphetohyoidea in which he included the Acanthodii. In summary then i t i s quite probable that acanthodians or close gnathostome relatives were ancestral to the Teleostomi, and i t i s thus then i n the Acanthodii that the origins of branchiostegals rays are here sought. From the visceral arches in the Acanthodii extend rows of bony splints. It i s possible that from these the branchiostegal rays . developed, as has been suggested by Gregory (1951). The anteriormost of these rows bony splints i s well developed, the rays being large and rod-like dr paddle-like and forming an operculum. Watson (1938) believed 25 that this row of rays w a 3 carried by the mandible, (the rays being retained from a time when the jaws were g i l l bars) and that a f u l l g i l l s l i t separated the mandibular and hyoid arches. He supported this theory by showing the hyoid arch had a set of g i l l rakers along its whole length and suggesting that these guarded a f u l l g i l l s l i t before the hyoid arch. If there was a f u l l s l i t between the hyoid arch and mandibular arch and not just a small or large dorsal spiracle, then the well developed row of visceral rays must have proceeded from the mandibles. It is not, however, definite that this operculum was mandibular. In certain of Watson's photographic plates (plate 5 of Climatius and plate 7 of Euacanthus) the hyoid arch has been displaced posteriorly and has carried with i t the operculum. In Davis (16*94> pi. 27, f i g . 1 and la of Acanthodes) the hyoid rays are seen to insert on the ventral face of the hyoid arch. If the rays were closely applied to the mandible i t would appear they would interfere with opening of the jaws as opening the jaws meet with resistance as the rays wore forced upon one another. The branchiostegals of palaeoniscids appear in lateral view to insert upon the mandible while they are actually inserting on the underlying hyoid arch. Stensio (1947) doubted that there was a f u l l hyoid g i l l s l i t and that the principal operculum was mandibular. He also mentions that Holmgren has come to a similar conclusion.1 A further paper has just been published which also supports this view. Westoll, T.S. 1963. The hyomandibular problem in placoderm fishes. Proc. XVI Int. Zool. Congress 1: 176. 26 From the above arguments three hypotheses may be made. If the principal operculum is mandibular the branchiostegals may be derived from: i) the smaller hyoid rays behind the principal operculum, the principal operculum later being lost, or i i ) the rays of the principal operculum which became attached to the hyoid arch on closure of the spiracular s l i t . But i f the rays of the principal operculum were indeed hyoidal i t can be suggested: i i i ) that they remained and developed into the branchiostegals. To the author the third hypothesis seems most economical and preferable. However, through sectioning of the hyoid-mandibular region of acanthodian fossils i t would be readily determinable which theory is true. The operculum and suboperculum The operculum and suboperculum, found in the f irst known teleostomes, are probably expansions of branchiostegals or their hyoid ray homologues, as has been suggested by Traquair (in Ridewood, 1904) and by Hubbs (1920). Little difference but size may be seen between the opercular, subopercular and branchiostegals of primitive teleostomes. The embryological development of the operculum and suboperculum is very similar. One can see in some acanthodians (e.g. Euthacanthus) a tendency towards expansion of certain opercular rays. It does not seem unreasonable, therefore, to suggest derivation of the operculum and suboperculum from branchiostegals or hyoid rays. With the expansion of branchiostegals into opercular bones larger muscles might insert upon them and a stronger branchial pump develop. A stronger branchial pump may have developed purely for the sake of 27 efficiency or because of greater respiratory demands. Greater activity would require more oxygen. The acanthodians are known, through a fortunate fossi3.iza.tion (Watson, 1959) to have cyclootome type myomeres with a low central vertex and no horizontal septum. It might be suggested that this type of myomere indicates a lower activity level than would the more highly developed piscine myomere type (Nursall, 1956) found in teleostomes. It might alternately be suggested that during teleostome evolution, conditions of lower oxygenation due to warmer climate produced selection for a more efficient branchial pump. Either of these two factors might explain the evolution of opercular bones from branchiostegals. The Interoperculum The interoperculum is absent from the Dipnoi, Crossopterygii, Brachiopterygii and lower Actinopterygii. It is f irst found in the Ospiiformes of the Lower Triassic. Regan (1929) siiggested that the forward migration of the mandible and quadrate dragged the suboperculum down, perhaps causing the lower end of the suboperculum to be fractured and pulled forward to form the interoperculum. This explanation appears rather Lamarkian. The following hypothesis would appear to be preferable. Several authors have suggested that the interoperculum developed from a forwardly displaced branchiostegal (Cole and Johnstone 1902 in Ridewood, 1904; Westoll, 1944; Saint-Seine, 1955). The development of an interoperculum seems to have been a further development in the separation of the mandibles and operculum. The preoperculum developed in the teleostome ancestors, partially separating 28 the jaw and opercular movements. With the freeing of the maxillary in the Ilolostei the need again arose to increase separation of the jaws and operculum to make their movements more independent from one another. The evolutionary solution was the interpolation of the upper.nost branchiostegal between the jaws and the suboperculum; thus, the uppermost branchiostegal became the interoperculum. Schaeffer (1956) has shown there is a high degree of correlation between the freeing of the maxillary and the presence of an interoperculum. An element in the chondrostean Platysiagum has been termed by Brough (1939) a modified branchiostegal or an incipient interoperculum. This element l ies in a slight concavity under the front two thirds of the suboperculum. The element cannot be identified as an interoperculum since i t does not separate the jaws from the suboperculum, nor is i t broadly bordered by the preoperculum. It is very similar to the upper-most branchiostegal of the chondrosteans Ptycholepis and Brookvalia which l i e in a slight emargination of the suboperculum. It therefore seems inadvisable to call this element an incipient interoperculum. The Gular Plates Both median and lateral gular plate3 are found in the earliest of the actinopterygian, crossopterygian and dipnoan teleosts. The lateral gulars are hardly distinguishable (only by being slightly larger and bearing a pit line) from the adjacent branchiostegals in the earliest teleost fossils (Cheirolepidae). Indeed there is no reason to suppose that the lateral gulars are not simply expanded branchiostegal rays. The lateral gulars are already characteristically large in the earliest 29 crossopterygians. As the crossopterygians evolve the lateral gulars eventually supplant both the median gular plate and the branchiostegals. The branchiostegals and median gular are also lost during the evolution of the Dipnoi. The Brachiopterygii also lack a median gular and branchiostegals, the large lateral gulars have supplanted them. The pit lines are of great value in tracing the gular elements. The median gular of Crossopterygians and Actinopterygians may be identified by a V-shaped pit l ine. The lateral gulars of a l l four sub-classes, at least primitively bear a transverse pit l ine. In the Crossopterygians (osteolepids) one or two of the branchiostegals under the end of the mandible may bear a short pit l ine. These appear to have been retained in the dipnoans where the anterior of the two marked branchiostegals has moved to the medial edge of the posterior branchiostegals. In the Dipnoi there are primitively two pairs of lateral gulars, the posterior pair of which bears the pit line and which is thus identifiable with the lateral gulars of the other suborders. The second pair of gulars may have formed by division of the single crossopterygian pair, or by expansion of one of the anterior branchiostegals. The f irst suggestion is favoured since the gular pit lines s t i l l retain a position (at the anterior end of the posterior gulars) that would be in the centre of the single gular, although i t is now divided into two. If the anterior gulars were formed from expanded branchiostegals one might expect the pit lines on the posterior gulars to be in the middle of the gular rather than at the anterior end. The origin of the median gular is less certain. There is no bone from which i t can be derived. However, i t is possible that i t arose 30 from a branchiostegal close to the midline. This is difficult to conceive as the median gular overlies the branchiostegals and is not in the same plane. It is possible that i t arose de nouveau from dermal tissue. Or perhaps i t evolved from the hyoid rays of the Acanthodii. The hyoid rays are continuously distributed around the hyoid arches in the Acanthodii, not restricted bilaterally as in the Teleostomi. Possibly the median hyoid ray of the Acanthodii gave rise to the median gular. 31 PHYLOGENETIC OR VERTICAL EVOLUTION The phylogenetic or vertical evolution of branchiostegals as opposed to adaptive or horizontal evolution is traced in this section. The phylogeny of the operculars, gulars and hyoid arch elements is also traced but in less detail. The groups are treated in phylogenetic order, as closely as is possible in a linear series. For the best picture of phylogeny see plate XVIII. Evolution is followed down to family level, although occasionally comments may be made on lessor taxa. An attempt is made to illustrate at least one example of every order, (see plates I, II, III, etc.). 32 CLASS TELEOSTOMI (OSTEICHTHYES INCLUDING DIPNOI) Branchiostegals (0-2)3-20(21-50). One (or two) median gulars, one or two pairs of lateral gulars present or absent. Operculum, suboperculum and (in higher groups) iivteroperculum usually present, but sometimes absent. Interhyal, epihyal (in higher groups), ceratohyal and one or two (in higher groups) hypohyals usually present but sometimes lost. Lower Devonian to present. Four subclasses and 66 orders, 38 of which are l iving. Although the subclass Actinopterygii is considered most primitive, then the Crossopterygii, Dipneusti and Brachiopterygii, the latter are placed f irst because they are more primitive than the higher Actinopterygii. The Brachiopterygii might best be placed as a chondrostean order of the Actinopterygii. But until i ts affinities are known with certainty i t would appear preferable to leave i t in a separate subclass. The Dipneusti are often included in the same class as the other bony fishes, but Berg isolates them in their own class. Since the Dipneusti are derivable from the Crossopterygii and since they are not distinguished to a higher degree than the other subclasses of teleostomes they are included in the Teleostomi, not in a separate class. KEY TO TELEOSTOMI BASED ON THE BRANCHIOSTEGAL SERIES AND HYOID ARCH The following key outlines the major events in the evolution of the Teleostomi. While i t follows the phylogeny rather closely, for the best picture of phylogeny of the Teleostomi see the evolutionary dendrogram. For the keying out of groups and determining of relationships one cannot depend on one set of characters, so that some orders cannot be 33 separated in this key. The key demonstrates that at many points in evolution that the branchiostegals and hyoid arch provide sharp division between related groups. For example the Beloniformes are separated from a l l other malacopterygians by the lack of an interhyal and non-filiform branchiostegals, the Cyprinodontiformes and Gobiesociformes differ from a l l other ?.canthopterygians (except a genus of Gobiidae and Syngnathidae) in having lost one hypohyal, a different one in each case. The Tarasiiformes, Chondrosteiformes and Peltopleuriformes are neatly separated by branchiostegals. Just as important are the similarities shown among orders which cannot be conveniently separated on the basis of branchiostegals. Thus, is indicated the closeness of the great proportion of the acanthopterygians. Key to the Classes of Gnathostoraes A With jaws, laterally paired nostrils, 3 semicircular canals. B Without rays on the hyoid arch . . . Pterichthyes, Coccostei. BB With rays on the hyoid arch. C Lacking gular plates, an interhyal and opercular bones. ("Opercular plate" of holocephalans composed of fused hyoid rays). D Hypohyal and ceratohyal present. Class Acanthodii. DD Only ceratohyal present. Class Elasmobranchii (including Holocephali). ; . CC With or without gular plates, with an interhyal (lost in some orders), with opercular bones (completely lost only in Giganturiformes and Saccopharyngiformes). Class Teleostomi. 34 Key to Subclasses of Teleostomi The subclasses are difficult to separate in a convenient manner because the primitive members are so similar and because of the convergence between the Coelacanthiformes and Brachiopterygii. A With one large pair of gulars longer than 2/3 of the mandible or with 2 pairs of gulars (except gulars absent in Ceratodiformes and Lepidosirenformes these recognizable because of cartilaginous disk under tip of operculum). B With or without median gular and branchiostegals. Medium-sized subquadrate or subtriangular operculum with a broad or triangular suboperculum. Where suboperculum is triangular opercular fold passes behind i t . . . Crossopterygii. BB With or without a median gular and branchiostegals. Operculum large and pentagonal (or reduced to a slender ray with a cartilaginous disk under its t ip) . Narrow elongate suboperculum below . . . Dipneusti. BBB Without median gular or branchiostegals. Operculum and sub-operculum medium-sized and subtriangular; opercular fold in front of suboperculum . . . Brachiopterygii (contains a single order Polypteriformes). AA With one pair of gulars shorter than g of length of mandible or without gulars . . . Actinopterygii Key to Orders of Crossopterygii A Branchiostegals and sometimes median gular present. Lateral gulars taper along whole length towards anterior t ip . Suboperculum 35 completely ventral to operculum. Supraorder 0steolepide3. B Branchiostegals 1 0 . Suboperculum xdth ventral edge on the dorsal edge of the lowest preoperculum (third) . . . Hoploptychiforme s. BB Branchiostegals 4-8. Suboperculum with ventral edge not on dorsal edge of lowest preoperculum . . . Osteolepiformes. AA Branchiostegals and median gular absent. Lateral gulars of even breadth, do not taper through whole length towards anterior t ip. At least dorsal portion of suboperculum opposite front of lower portion of operculum. Supraorder Coelacanthi . . . Coelacanthiformes. Key to Orders of Dipneusti The following is a tentative key based on the assumption that the Uronemiformes have gular plates and that they and the Ctenodontiformes have less than three branchiostegals. A Gular plates and usually branchiostegal(s) present. Operculum large and circular or pentagonal in shape. Supraorder Dipteri. B Branchiostegals 3 . . . Dipteriformes. BB Branchiostegals less than 3 . C Suboperculum 1/3 to ^ of size of operculum . . . Ph oneropleuriforme s. CC Suboperculum smaller. D Operculum oval . . . Uronemiformes. DD Operculum scallop-edged, roughly circular . . . Ctenodontiformes. AA Gular plates and branchiostegals absent. Operculum small and 36 elongate. Supraorder Ceratodi. E Operculum arrowhead-shaped; suboperculum a slender ray . . . C er atodiforme s. EE Operculum a slender ray; suboperculum chevron-shaped . . . Lepido sireniformes. Key to Orders of Actinopterygii Due to paucity of information, the orders Cephaloxeniformes, Aethodontiformes, Luganoiformes, and Ptycholepiformes must be omitted from this key. A Interoperculum absent. Lateral gulars may be present. Group I (Chondrostei). (Maxillary not free from cheek, except Dorypteriformes). B With a pair of lateral gulars and usually with a median gular. Always with a suboperculum. C With 4 or more branchiostegals D With pit line on lateral gulars . . . Palaeonisciformes, Perleidiformes. DD Without pit line on lateral gulars . . . Pholidopleuriformes. CC With 1-3 branchiostegals . . . Haplolepiformes. BB Without lateral and median gulars. With or without suboperculum. E With suboperculum (except Polydontoidei). F Branchiostegals 3 or more. Tarasiiforme3 15 branchiostegals. Platysiagiformes 12 branchiostegals. Phanerorhynchiformes "series" of branchiostegals. . . . . . Chondrosteiformes 9 - 1 0 branchiostegals. . . . . . Peltopleuriformes 6 - 7 branchiostegals. FF Branchiostegals 0 - 1 . G Operculum smooth . . . Redfieldiiformes. GG Operculum grooved or incised . . . Acipenseriformes. EE Without suboperculum H Operculum deeper than long . . . Dorypteriformes, Bob as atr aniiforme s, Pycnodontiforme s. H H Operculum as long as deep . . . Saurichthyiformes. Interoperculum present (secondarily absent in Lepidosteiformes, some Mormyriformes, Giganturiformes, Saccopharyngiformes, some Anguilliformes). Lateral gulars never present. (Maxillary free from cheek). I A single hypohyal present, a median gular often present. Group II (Holostei). J Median gular present. K Branchiostegals 3 0 or fewer . . . Ospiiformes, Amiiformes, Pholidophoriformes. KK Branchiostegals 3 0 - 5 0 . . . Pachycormiformes, JJ Median gular absent, Branchiostegals 3 . . . Lepisosteiformes. Branchiostegals 1 2 - 1 3 . . . Aspidorhynchiforra.es. II Two hypohyals present (except Lycopteridae, Phractolaemidae, Mormyriformes, Amphiliidae, Chacidae, Anguilliformes, Stylephoridae, Beloniformes, some Syngnathidae, Cyprinodontiformes Gobiesociformes). Median gular only in Elopoidei and Albuloidei. Group III (Teleostei). branchiostegals often spathiform (except Stomiatoidei, Myctophiformes, Mormyriformes, some Cypriniformes, Notacanthif ormes, Anguilliformes). Epihyal and ceratohyal sutured together only in Siluroidei and Beloniformes. Never spines on opercular bones. Branchiostegals usually inserting on external face of hyoid arch with a minority below on the ventral or internal, sometimes a l l on ventral edge; only sporadically in Clupeiformes (Hiodontidae, Chanidae, Gonorhynchidae, Osmeridae, Argentinidae) and commonly in the Myctophiformes is the acanthopterygian pattern of the 4 upper branchiostegals on the external face and 0-5 lower ones on the ventral (or internal) face of the hyoid arch found. Group IIIA Malacopterygi. M With branchiostegals, hyoid arch and operculum. N Hypohyals 2 (except Phractolaemidae, L Branchiostegals 0-36 , one or more of upper Araphiliidae and Chacidae). 0 One or more upper branchiostegals spathiform, at least distally ... Clupeiformes, Cypriniformes 00 Branchiostegals not spathiform. P Branchiostegals not straight, curve up behind g i l l cover . . . Myctophiformes. PP Branchiostegals nearly straight, do not curl up behind g i l l cover . . . Notacanthiformes. NN Hypohyals 0-1 Q Branchiostegals not filiform and not curled up around dorsal edge of operculum. R With interhyal . . . Mormyriformes. RR Without interhyal . . . Beloniformes. QQ Branchiostegals filiform and curled around operculum . . . Anguillifonnes. MM Without branchiostegals or operculum . . . Giganturiformes, Sac coph aryngif orme s. LL Branchiostegals 1-9 (10) (except 8-11 in Echeneiformes). Branchiostegals acinaciform, sometimes filiform, never spathiform (except Molidae). Epihyal and ceratohyal usually sutured together (but suture secondarily lost in several groups). Often spines on operculum, sometimes on suboperculum and interoperculum. The upper 4 branchiostegals on the external face of the hyoid arch, the other 0-7 (usually 2-3) on the ventral or internal face. Group IIIB Acanthopterygi. S Hypohyal3 2 (except in one genus of Syngnathidae, Gobiidae and Stylephoridae and a l l the Phallostethidae). T Branchiostegals modally 7 or more. G i l l membranes usually separate (except some Gadiformes) . . . Beryciformes, Zeiformes, Lampridiformes, Bathyclupeiformes, Gphidiiformes, Gadiformes, Ateleopiformes, Echeneiformes. TT Branchiostegals rnodally 6 or less. Gi l l membranes separate, united and free from isthmus, or joined to isthmus . . . Percopsiformes, Syiignathiformes, Perciformes, Pleuronectiformes, Tetraodontiformes, Icosteiformes, Gasterosteiform.es, Mastacembeliform.es Synbranchiiformes, Batrachoidiformes, Lophiiformes, Pegasiformes. Hypohyals 1 U Lower hypohyal present. With epihyal-ceratohyal suture . . . Cyprinodontiformes. UU Upper hypohyal present. Without epihyal-ceratohyal suture . . . Gobiesociformes. 41 SUBCLASS CROSSOPTERYGII Branchiostegals 10 or fewer; median gular present or absent, primitively with a V-shaped pit line; a single pair of large gulars at least 2/3 the length of the mandible; operculum present; usually sub-operculum present; hypohyal, ceratohyal and interhyal present. Lower Devonian to present. Two superorders with three orders and 12 families. SUPERORDER OSTEOLEPIDES Branchiostegals and median gular present (median gular in Hoploptychiformes?). The lateral gulars taper along their whole length towards the anterior t ip. Suboperculum ventral to operculum and taking normal part in movement of g i l l cover. Lower Devonian to Upper Carboniferous. Two orders. Thomson (1962) would reduce the status of these two orders. / ORDER HOPLOPTYCHIFORMES Branchiostegals 10; median gular unknown; a pair of large lateral gulars with an arc-shaped pit line; operculum and suboperculum present lower edge of suboperculum on upper edge of third preoperculum. Lower to Upper Devonian. TITO families. ^ Porolepidae Branchiostegals: No data available. ^ Hoploptychidae Branchiostepials; In Hoploptychius flemingi about 10. The f irst six are elongate and situated in a series below the suboperculum. The remaining four are very short and l ie between the gulars and the 42 mandibles; the f irst of these bears a short vertical pit l ine. At least two of these short branchiostegals are known in Glyptolepis. Gulars: One of Woodward's specimens of Hoploptychius flemingi shows the lateral gulars but no median gular nor does his text report a median gular in Glyptolepis* The lateral gulars in Hoploptychius are large, about the same length as the mandible. The posterior edge of the lateral gulars curve inward and anteriorly, leaving a V-shaped gap. At the centre of each gular is a short arc-shaped pit line with apex anterior-most. Operculars: The operculars of Hoploptychius are rather different from those i n 0 s t e o l e p i 3 since they insert on the diagonal base of the upper preopercular (squamosal), rather than on the vertical base of the lower preoperculum as in Osteolepis. Thu3 the opercular bones are more dorsal in Hoploptychius and their bases diagonal. The operculum is larger than the suboperculum. The jaw of Hoploptychius is shorter than in Osteolepis and the triangular lowest preoperculum acts an interoperculum of the holostean type, a piviting point for the opercular bones. Relationships: The more numerous branchiostegals would indicate this order to be more primitive than the Osteolepiformes. The opercular bones and number of branchiostegals separate the two orders. Yet the two orders share features which separate them from the Coelacanthi: their gular plates taper anteriorly, they possess branchiostegals and the suboperculum i 3 below rather than anterior to the operculum. References: Jarvik (1948), Woodward (1891), Stensio (1947). 43 / ORDER OSTEOLISPIFORMES (RIHZ0D0NTIF0RME3) Branchiostegals 4-8; visually a median gular with p i t line; one pair of large lateral gulars with p i t line; suboperculum not with lower edge on upper edge of third preoperculum; operculum and suboperculum present* Six families. Middle Devonian to Lower Permian. Taxonomy! The operculars, gulars, branchiostegals and other skull bones of Rhizodontiformes are so similar in form and arrangement I feel constrained to return this order to the Osteolepiformes, as Arambourg (1958) has done. ^ Gyroptychiidae Branchiostegals; In Gyroptychius branchiostegals 6-8. The f i r 3 t i s almost quadrangular but expands d i s t a l l y . The remainder are short and broad and bridge the gap between the gulars and the mandible. The second i s characterized by a diagonal p i t line somewhat anterior to the center. The last two branchiostegals may fuse into a single large plate. Gulars: The diamond-shaped median gular bears a broad V-shaped pi t l i n e . The l a t e r a l gular3 are large, narrow anteriorly to a point and bear a short arc-shaped p i t line at the centre of the side next to the mandibles. The posterior edges of the plates curve anteriorly and medially, leaving a wedge-shaped gap between them. The gulars are about % the length of the mandibles. The anterolateral portion of the gulars contacts the mandibles (a primitive character). Operculars: Tne operculum and suboperculum are small and rectangular. The operculum Is slightly larger. The opercular bones are slightly 44 shorter than in Osteolepis. Taxonomy: This family erected by Berg (1955)• Relationships: The branchiostegal series speak for a very close relationship to the Osteolepidae. References: Berg (1955), Jarvik (1948). ^ Osteolepidae PI. I Branchiostegals: In Osteolepis macrolepidotus about 7 spathiform branchiostegals. The f i r s t , broad but elongate, i s under the suboperculum, which i t much resembles. The remainder of the branchiostegals are short and wide; they span the gap between the gulars and the mandible. The second and sometimes the third branchiostegal bear a small arc-shaped p i t l i n e . These two p i t lines are important because they act as labels. They provide evidence that these two branchiostegals are homologous to the two pit-line-bearing branchiostegals between the gulars and the mandible i n Dipterus and further that the bone called suboperculum in Dipterus i s indeed that bone. The pit line on the third branchiostegal °^ i l l macrolepidotus i s apparently in the process of being lost, since i t i s often not present. In Osteolepis panderi there are only 4 branchiostegals, the anterior ones being displaced by the later a l gulars; in Thursius there are 5-6 much as i n Osteolepis. the second with a p i t l i n e . The branchiostegals narrow anteriorly as the later a l gular approaches the jaw, u n t i l the f i r s t branchiostegal becomes pointed. 45 Gulars: In Osteolepls a small diamond-shaped median gular is located in the fork of the mandibular rami. It bears a V-shaped pit line, the arms of which curve slightly outward. The very large lateral gulars nearly equal the length of the mandibles. They taper only slightly anteriorly. The posterior edge curves inward and anteriorly. In the middle of each lateral gular is a short arc-shaped pit line whose apex faces anteriorly. The width of the lateral gulars enters about 3 times in their length. In Thursius the gulars are pointed anteriorly and about .9 of the mandibular length. They bear a pit line and their posterior edges curve medially and anteriorly. Operculars: In Osteolepis the operculum is slightly larger than the suboperculum; both are elongate in horizontal direction and very similar in shape. The operculum of Thursius is said to be deeper than broad and the suboperculum smaller and to be broader than deep. References; Woodward (1891), Berg (1947, 1955), Jarvik (1948). ^ Euathenopteridae P l . I . Branchiostegals: In Eusthenopteron foordi 8 branchiostegals. The f irst is elongate and l ies below and slightly anterior to the suboperculum. The remainder are very short and are situated between the gulars and the mandible; the f irst of these bears a vertical pit line; the last is triangular. In Eusthenodon 8. Gulars; A small median gular with an arc-shaped pit line with apex anteriormost is present, an unusual shape for this pit l ine. The lateral 46 gulars are relatively small—about 2/3 of the length of the mandible. They are narrow (their width entering 4*5 times in the length) and f a i l to touch the mandible anteriorly. Slightly anterior to their centre is a J-shaped pit line, instead of the usual arc-shaped or straight l ine. Operculars; The operculum and suboperculum are of about equal size and trapezoidal in shape. Taxonomy; Berg (1955) erected this family. Relationships: This family differs from other Osteolepiformes by the shorter and narrower lateral gulars. It i s closest to Osteolepidae in this respect. References: Berg (1955). Jarvik (1944, 1952), Bryant (1919). f Rhizondontidae Branchiostegals; In Rhizodppsis sguroides there are 5 branchiostegals. The f irst of these is elongate and situated under the suboperculum0 The remaining 4 l i o between the lateral gulars and the mandibles; they narrow anteriorly until the f irst i3 triangular. Gulars: In Rhizodopsis a small anterior median gular is followed by a large pair of gulars. The median gular, situated in the fork of the mandibles, is egg shaped with apex anteriormost. Behind i t are the pair of large lateral gulars whose length almost equals that of the mandible. Their posterior edge curves inward and anteriorly. The left is shown to overlap the right in Woodward's f ig . 57. The anterior tip of the lateral gular touches the mandible. The length of the gular is about 2^ times its width. 47 Operculars: The operculum in Rhizodopsis ornatus is scallop-shaped (with hinge uppermost) and heavily decorated. The operculum is large and pentagonal in Rhizodopsis. The smaller suboperculum is approximately trapezoidal. Relationships: The operculars, gulars and branchiostegals are very similar to those in other Osteolepiformes. References: Woodward (1891), Traquair (1883). J. ' Parabatrachidae Branchiostegals: No data available. Gulars: A pair of lateral gulars, each about Z\ times as long as wide, abruptly truncated posteriorly or rounded in Parabatrachus (= Megalichthys Agassiz). Operculars: Operculum nearly as broad as deep. References: Woodward (1891). 48 SUPRAORDER COELACANTHI (ACTINISTIA) Branchiostegals and median gular absent. Lateral gulars of even breadth, tapering only at tips. Suboperculum at least partially opposite lower portion of operculum, and apparently not taking part in movement of g i l l cover. Middle Devonian to present. One order. ORDER COELACANTHIFORMES A large pair of lateral gulars of even breadth bearing an arc-shaped pit line. Suboperculum at least partially opposite lower portion of operculum. Suboperculum apparently not taking part in movement of the g i l l cover. Interhyal and ceratohyal present. Middle Devonian to present. Three suborders, four families. Taxonomy: There are no notable differences in the branchiostegal series between the three suborders of Coelacanthiformes, but they may be separable on other grounds. Arambourg (1958) does not employ the sub-orders in his classification. Relationship: The Coelacanthiformes are more similar to the Osteolepiformes than to the Hoploptychiformes in that the suboperculum is next to the third preoperculum, in that the branchiostegals are fewer in Osteolepiformes and in that the sensory canals of the head, as portrayed by Stensio (1947), are of a more similar pattern. f SUBORDER DIPLOCERCIDOIDEI Middle Devonian to Upper Devonian. A single family. 49 f Diplocercidae Gulars: In Nesides schmidti a pair of large lateral gulars, length about 5/6 of the mandible. Posterior end protruding slightly beyond mandible. Operculars: Operculum large and triangular, concave where i t meets the upper preoperculum. Between the lower portion of the operculum and the preoperculo-quadrajugal is a small presumably triangular suboperculum. The suboperculum is anterior and not ventral to the operculum. Taxonomy: Romer (1955) is followed as to the limits of this family. References: Stensio (1947). ^ SUBORDER LAUGIOTJJEI Lower Triassic. A single family. / Laugiidae Gulars: A pair of large lateral gulars 7/8 of the length of the mandibles. Breadth even, about 3z times their length. Litt le gap posteriorly between the gulars. Operculars: Operculum medium sized and triangular. Antero-dorsal corner pointed, without notch. Suboperculum not known, presumably absent. Relationships: The opercular bones are l i t t l e different from those in the Coelacanthoidei. References: Stensio (1932). 50 SUBORDER COELAGANTHOIDEI Lower Carboniferous to present. Two families. Coelacanthidae PI. I Gulars: There are a pair of large lateral gular plates in members of this family. In Rhabdoderma elegans the gulars are about 4/5 the length of the mandible and do not protrude posteriorly beyond the end of the mandible. In R. aldingeri the width is hh times the length and the main body of the plate is of even breadth, tapering only at the tips. In Holophagus (=Undina) the gulars are of even breadth—about 3 times in length; they are about equal to the length of the mandible and just posteriorly from i t . In Macropoma the gulars are almost equal to the length of the mandible and bear arc-shaped pit lines centrally. In Diplurus the gulars are long and narrow, the width kk i n length. In Whiteia the gulars occupy only about 2/3 of the mandibular length; they are of even breadth. They bear a short straight pit line centrally. Their posterior ends are rounded but they soon meet on the midline so there is not a V-shaped gap between them. Operculars: In Rhabdoderma the operculum is of middling size and triangular; i t has a notch where the antero-dorsal corner meets the spiracular bone. The suboperculum is small and triangular and the upper half is opposite the lower end of the operculum. In Holophagus (=Undina) the operculum is longer, extending further ventrally than in Rhabdoderma and the suboperculum is a long or short triangle, 2/3 or more of which 51 i s below the operculum. The operculum i 3 also larger in Wimania ( h a 3 an antero-dorsal notch) and Axelia (lacks notch). In Macropoma the operculum is of middling size and has a small concavity on its dorso-anterior edge; a small rectangular suboperculum i s present. I n Mylacanthus the large operculum has a lobate or spinous posterior margin. In Whiteia the operculum is medium sized, triangular, and has a slight concavity on its dorso-anterior corner where i t contacts the postspiracular. A small suboperculum is present. In Diplurus the operculum is medium sized, triangular, and without an antero-dorsal notch; the dorso-posterior edge is crenulate; the suboperculum unknown. Hyoid arch: In Macropomus ceratohyal with postero-ventral projection; cylindrical interhyal. References: Moy-Thomas (1937), Zittel (1887), Stensio (1921, 1932), Lehman (1952), Schaeffer (1952), Goodrich (1909), Gardiner (I960). Latimeriidae Gulars: In Latimeria chalumnae a pair of large lateral gulars is present. These are only slightly less than the length of the mandible. They are of even breadth, and taper at the ends to a point. Width is a l i t t l e more than 3 times in length. An arc-shaped pit line is present in the centre of each gular; as usual the apex of the arc is anterior-most. Operculars: Operculum middle sized, a rounded triangle without antero-dorsal notches. Suboperculum small and forming a narrow triangle the apex uppermost. The suboperculum l ies entirely in front of the lower 52 half of the operculum. The opercular fold in Latimeria, unlike in other orders of fishes, passes between the operculum and the suboperculum. The close attachment of the suboperculum to the dorsal operculum and the passage of the opercular fold behind, rather than in front of the suboperculum, would seem to forbid the suboperculums taking part in the movement of the g i l l cover. From the similar arrangement of the opercular bones in other families of the order Coelacanthiformes one would conclude that the rigidity of the suboperculum also pertains to them. In this character the Coelacanthiformes differ from other crossopterjrgians and other teleostomes. J . L . B. Smith, from a poorly preserved specimen, reported an interoperculum. Thi3 bone is known only from the higher Actinopterygii. Schaeffer suggests Smithts interoperculum may be modified scales in the opercular membrane. Millot and Anthony do not report an interopercular in their study of well preserved specimens and i t seems most unlikely i t occurs in Latimeria. Schaeffer (1952) interprets the suboperculum as a preopercular bone in the Coelacanthoidei. This suggestion has the merit of explaining the rigidity of the "suboperculum" and its unusual anterior position in this group. In the other crossopterygian orders the suboperculum is directly ventral. Hyoid arch; Consists of a short cylindrical hypohyal, a ceratohyal, epihyal and hyomandibular. References: Millot and Anthony (1959), Smith, J . L . B. (1940). 53 SUBCLASS DIPNEUSTI Branchiostegals three or less; median gular present or absent; 1 -2 pairs gulars present or absent (combined length of lateral gulars exceeds one half mandible length); operculum present and suboperculum present (probably inserting on hyomandibular); a hypohyal present or absent; a ceratohyal always present; an interhyal may be present. From late Early Devonian to present. The dipneustians are divided into two superorders, the Dipteri and the Ceratodi, 6 orders and 12 families. The Ceratodi have a complete branchiostegal series and full-sized operculars while the Dipteri have lost the gulars and branchiostegals and have reduced operculars. The Dipteri are readily distinguished from other fishes by their greatly expanded pentagonal operculum while the Ceratodi are distinguished by their very reduced operculum. The differences in the branchiostegal series alone certainly justify the supraordinal separation of the Ceratodi and Dipteri. The differences between the two make i t difficult to characterize the subclass Dipnoi as a whole. However, the differences should be not unduly weighted, since they are concerned mainly with a loss or reduction in characters. The suboperculum rests on the ceratohyal unlike other fishes, in which i t usually rests on the hyomandibular or preoperculum. Further the suboperculum is much more slender than in other teleostomes, except in Scaumenacia. It might therefore be questioned whether or not the dipneustian subopercle i3 not a branchiostegal. However the condition of the subopercle in Scaumenacia. where i t is of normal 3ize and shape, would seem to oppose this thesis. Further, the unusual insertion of 54 the suboperculum may be accounted for by supposing that the great expansion of the operculum in the Dipteri has necessitated a lower insertion of the suboperculum and that i t has thus been displaced down from the hyomandibular to the ceratohyal. This position has been conservatively retained in the Ceratodi even though the size of the operculum has been reduced. The possession of branchiostegals, operculum, suboperculum, median and lateral gulars by primitive dipnoans shows clearly that they are related to the other subclasses of teleostomes. But when they f irst appear in the fossil record they are already quite specialized, most branchiostegals are already lost and the gulars are more modified than primitive representatives of the other subclasses, and the operculum is considerably enlarged. Although these features distinguish them, they certainly do not warrant class distinction and the many similarities of the primitive members of the subclasses confirm their placement in a single class. The arrangement of branchiostegals and gulars in primitive dipnoans enable them to be derived from primitive Crossopterygii, but not vice versa. This opinion is in agreement with that of Westoll (1949) who believed that the Dipnoi either stemmed from the Rhipidistia, or that both groups have a common ancestor. To points listed by Westoll for such a common ancestor may be added: a median gular, a pair of lateral gulars with a pit line and broad short branchiostegals. 55 / 3UPER0RDER DIPTERI Gular plates present (Uronemiformes ?). Branchiostegals present in at least Dipteriformes and Phaneropleuriformes. Operculum large, roughly pentagonal. Middle Devonian to present. Ceratohyal present. Four orders. ^ ORDER DIPTERIFORMES (RHYWCHODIPTERIFOWIES) Branchiostegals three, a small median gular and two pairs of gular plates, the posterior bearing pit lines; a large pentagonal operculum and a small elongate suboperculum present. From the late Early Devonian to Upper Devonian. Three families. "f" Dipnorhynchidae Branchiostegals: Two branchiostegals in Dipnorhvnchu3. An elongate branchiostegal l ies below the suboperculum, which i t much resembles (in fact Westoll calls i t suboperculum 2). Laterally between the anterior and posterior gulars l ies a second triangular branchiostegal (called by Westoll a lateral gular). Gulars: Consists of a small anterior median gular, a pair of oval anterior gulars which slightly overlap a larger posterior pair of gulars, and between the two pairs of gulars l ies a second median gular. Operculars: A large subrectangular operculum lies above a narrow suboperculum, the latter much like the branchiostegal which lies below i t . Relationships: The gulars and branchiostegals are very similar to those in Dipteridae although there is one less branchiostegal. The possession of a pineal foramen, unique amongst the Di p n e u s t i , shows t h i s family i s w e l l separated. I t may, according to Berg (1947), dooerve a separate order. Westoll considers t h i s f a m i l y very p r i m i t i v e and shows indications of relationship to the R h i p i d i s t i a . I t may be noted t h a t the primitive crossopterygians also possessed a p i n e a l foramen. References: Hills (1933), Westoll (1949). ^ Rhynchodipteridae Branchiostegals: Not preserved. Gulars: Not preserved. Operculars: Operculum large, roughly quadrangular. Suboperculum not preserved. Taxonomy: Berg (1947) provisionally placed this family in its own orderj Arambourg and Guibe (1958) place i t in the Dipteriformes. References: Save-Soderbergh (1937). ^ Dipteridae P l . II Branchiostegals: In Dipterus 3 branchiostegals. Between the posterior gulars and the suboperculum lies a single spathiform elongate branchiostegal. It resembles the suboperculum except that i t is somewhat expanded distally, while the suboperculum tapers. Just anterior to this branchiostegal, between the gular and the angle of the jaw, are two very short accessory branchiostegals (about a quarter the length of the other). 57 The position of these is such that they do not f i t into a series with the branchiostegal; the second of these is below the f irst , not anterior to the f irst next to the mandible. A pit lino runs across these two accessory branchiostegals and onto the posterior gular plate. This reminds one strongly of the situation in the Crossopterygii where the second and third branchiostegals and the gular plate bear pit lines. But in the Crossopterygii the third branchiostegal is anterior, not ventral and the pit lines on the three bones are separate, not in a straight line with one another. However, the relationships seem sufficiently close to suggest that the accessory branchiostegals are homologous with the second and third branchiostegal in primitive Crossopterygii and therefore that they are really branchiostegals and not displaced mandibular elements. And finally i t appears that the posterior pair of gulars in Dipterus are derived from the lateral gulars of the Crossopterygii. The anterior pair of gulars in Dipterus may have arisen by division of the lateral crossopterygian gulars into an anterior and posterior pair. Enlargement of the anterior pair of gulars would result in the posterior displacement of the posterior pair of gulars and the accessory branchiostegals and explain why they are out of line xclth the positions in the Crossopterygii. Gulars: A small median anterior gular l ies in the fork of the mandibular rami. It is not known to bear a V-shaped pit line as in some Crossopterygii and Actinopterygii} nevertheless i t seems logical to consider i t homologous. The median gular is followed by a medium-sized pair of rhombic gulars. The anterior pair of gulars lacks a pit line; 58 the posterior pair of gulars are somewhat larger. Their posterior edges curve anteriorly much a3 those in Crossopterygii, but they overlap more. The left gular overlaps the right in both the anterior and posterior pairs. The posterior gular bears a short straight pit line anteriorly; apparently a continuation of the one running across the accessory branchiostegals. A posterior median gular l ies between the two paired gulars. References: Graham-Smith and Westoll (1937); Watson and G i l l (1923), Westoll (1949). T ORDER PHAWEROPLEURIFORMES Branchiostegals two; a small median gular and two large pairs of gular3 present; a large operculum and medium-sized suboperculum present; ceratohyal known. Upper Devonian. Three families. ^ Phaneropleuridae Branchiostegals: No data available. Gulars: A pair of long gulars, pointed anteriorly. Operculars: Operculum large and circular; suboperculum small and elongate. References: Woodward (1891), Whiteaves (1889). 59 ^ Scaumenacidae PI. II Branchiostegals: In Scaumenacia two branchiostegals. One short wide spathiform branchiostegal which widens distally, and another anteriorly between the two paired gulars ("accessory branchiostegal"). These correspond to the two in Dipterus. Gulars; A small median gular l ies in the fork of the mandible. Behind this is a relatively small anterior and a larger pair of posterior gular3. The gulars a l l correspond with those in Dipterus, although differing somewhat in form, being less elongate. Operculars; There is a large pentagonal operculum with a curved evacuation on its dorsal edge. The small wide suboperculum curves up from its lower edge to meet the operculum posteriorly. The suboperculum of Scaumenacia is nearly g the size of the operculum, the largest sub-operculum in the superorder. Relationships; The close correspondence of a l l the branchiostegal series to Dipterus leaves l i t t l e doubt that i t is related to the Dipteridae. It may be distinguished by its larger suboperculum and the presence of only one accessory branchiostegal. Reference; Stensio (1947). ^ Fleurantiidae Branchiostegals; Not preserved. 60 Gular3: In f ig . 3 of Fleurantia in Graham-Smith and Westoll, are bones which apparently represent a large oval posterior paired gular and a branchiostegal (or suboperculum ?). A paired gular in another specimen bears an arc-shaped pit l ine . Operculars: Operculum large and pentagonal, Hyoid arch: Ceratohyal short, stout, hour-glass shaped and much expanded at the posterior end, only slightly anteriorly. Relationships: Poorneas of preservation of the branchiostegals enables l i t t l e to be deduced from them. Graham-Smith and Westoll believe Fleurantia to be a secondary development from the normal dipnoan ancestor, such as Dipjterus, Scaumenacia being more advanced in fin structure. References: Graham-Smith and Westoll (1937). ^ ORDER URONEMIFORMES Branchiostegals and gulars not known; a large oval operculum and a small elongate suboperculum present; ceratohyal known. Lower Carboniferous to Upper Permian. Two families. Taxonomy: Romer does not distinguish the Uronemidae from the Dipteridae. Arambourg and Guibe synonymize the Conchopomidae with the Uronemidae and the Uronemiformes with the Ctenodontiformes. Berg»s classification is provisionally retained here. 61 ^ Uronemidae Branchiostegals: Not known. Gulars; Not known. Operculars; A large oval operculum is found in Uronemus splendens. It has a small dorsal projection. A small bone figured below may represent a suboperculum. Hyoid arch: Bones resembling the centre portion of ceratohyals have been illustrated. References: Watson and G i l l (1923), Woodward (1891). / ' Conchopomidae Branchiostegals: Not known. Gulars: Not known. Operculars: In Conchopoma gadiformis a large oval vertical operculum is known. It bears a small dorsal projection. A small angulated wing-like bone may represent a suboperculum. Hyoid arch: The ceratohyal is of even breadth anteriorly, but is much expanded posteriorly. References: Watson and G i l l (1923). / ORDER CTEMODONTIFORMES Branchiostegals not known; one pair of gular plates known; large scallop-shaped operculum; small elongate suboperculum. Lower Carboniferous 62 to Lower Triassic. One family. / Ctenodontidae PI. II Branchiostegals: Mot known. Gulars; A pair of crescentic gular plates situated anteriorly in the fork between mandibular ramii is described for Sagenodus. Mesial flanges projecting from these look as i f designed to support a pair of gulars posteriorly. Operculars; Operculum present in Ctenodus and Sagenodus. It is large, roughly circular, scallop-shaped, and bears, where i t inserts, a curved depression dorsally. On its antero-ventral border l ies a small elongate suboperculum; a ridge runs along its length externally. Relationships: Except for a dorsal evacuation the operculum of Sagenodus much resembles that of Dipterus. The anterior pair of gular plates is , however, of quite different conformation from that of Dipterus, being more cresentic. References; Woodward (1891), Watson and G i l l (1923). SUPERORDER CERATODI Gular plates and branchiostegals absent. Operculum reduced to an elongate ray which rests on the supratemporal-intertemporal (squamosal). The slender suboperculum rests on the ceratohyal. Ceratohyal and sometimes hypohyal present. Interhyal absent except in larvae. Upper Carboniferous to present. Two orders. 63 Taxonomy: Arambourg and Guibe (1958) synonymize, without discussion, the Lepidosirenformes with the Ceratodiformes. These two orders are retained here; hyoid arch differences support their separation. ORDER CERATODIFORMES A reduced operculum and suboperculum present; hypohyal, ceratohyal (expanded greatly at the upper end) and interhyal present in larvae. Upper Carboniferous to present. One family which includes a living representative Neoceratodus forsteri in Australia. Ceratodidae P l . II Operculars: In Neoceratodus operculum reduced in size and 3 h a p e d like an arrow head pointing posteriorly. The elongate oval suboperculum inserts on the expanded distal end of the ceratohyal. Under the distal end of the suboperculum is an oval cartilaginous plate. G i l l opening restricted to side of head. Hyoid arch: Composed of a small rectangular cartilaginous hypohyal and a large ossified ceratohyal. The ceratohyal is shaped as in Conchopomidae, that is of even breadth anteriorly and expanding greatly at its posterior end. Figures in Gregory, Goodrich and Holmgren and Stensio show no other hyoid bones (except a small cartilaginous hyomandibular). De Beer shows three bones, the "stylohyal", the ceratohyal and the hypohyal in a larval specimen. In the adult specimen here examined only a ceratohyal was present; the cartilaginous hypohyal, i f present, was shrivelled and there was no sign of an interhyal. 64 References: Ridewood (1894)# Gregory (1951), Holmgren and Stensio (1936), Goodrich (1958), de Beer (1937). Material examined: Neoceratodus forsteri, skeletal specimen, uncatalogued NIC, Queensland, Australia. ORDER LEPIDOSIRENIFORMES Operculum and chevron-shaped suboperculum reduced to a slender ray; only the ceratohyal (slightly expanded at the upper end) present in hyoid arch. Two families, Lepidosirenidae with a single l iving species in Brazil and Protopteridae with four living species in Africa; both in freshwater. Upper Carboniferous to present. Taxonomy: Arambourg and Guibe (1958) synonymize Protopteridae with Lepidosirenidae and Lepidosireniformes with Ceratodiformes. These groups are provisionally retained following Berg's classification. Lepidosirenidae PI. II Operculars: Operculum reduced to an elongate flat ray bearing a small cartilage under its distal tip in Lepidosiren. It inserts on the supra-temporal- intertemporal (squamosal), the hyomandibular being absent. The suboperculum is shorter but wider than the operculum. The suboperculum is a chevron-shaped bone which inserts on the upper ceratohyal. It is underlain by a broader cartilaginous base. This base is called an interoperculum by Bridge, but i t can hardly be that bone which unknown in the dipneustians (found only in the more advanced Actinopterygii). Nor are the cartilaginous distal tips of the operculum and suboperculum 65 likely to be a remnants of a hyoid ray, as he suggests, but are rather unossified portions of the operculum and suboperculum. G i l l opening restricted to side of head. Hyoid arch: Only a ceratohyal is present (even embryologically), the hypohyal, interhyal and hyomandibular being absent. The ceratohyal is large, bowed slightly downwards and expanded slightly at each end. It is attached to the parasphenoid by a ligament (the hyomandibular wanting). There is an odd patch of cartilage on the outer surface of the distal portion of the ceratohyal. Relationships: The reduced operculum and ceratohyal with posterior end unexpanded of the Lepidosireniformes are doubtless derived from the less reduced operculum and ceratohyal with expanded posterior end of the Ceratodiformes. That is the Ceratodiformes are more primitive than the Lepido sireniforme s. References: Gregory (1951), Ridewood (1894), Bridge (1898), Holmgren and Stensio (1936), de Beer (1937). Material examined: None. Protopteridae Operculars: The operculum tapers posteriorly to a point in Protopterus; i t appears to insert on the supratemporal-intertemporal (squamosal). The operculum bears an inner cartilage as in Neoceratodus. The sub-operculum is essentially similar to that in Lepidosiren. being an elongate chevron-shaped bone over a cartilaginous base. It inserts on 66 the posterior* end of the ceratohyal. G i l l opening restricted to side of head. Hyoid arch: Only a ceratohyal is present, the basihyal, hypohyal, interhyal and hyomandibular are lacking. The ceratohyal is large and slightly expanded at each end. There is a patch of cartilage on the outer surface of the anterior end of the ceratohyal. This is called a vestigial hyoidean ray by Bridge, but its position does not confirm this suggestion. Ligaments connect the ceratohyal to the skull and to the lower jaw, as in Lepidosiren. Relationship: The correspondence of the operculars and ceratohyal in Lepidosirenidae and Protopteridae suggest they are closely related. References: Ridewood (1894). Bridge (1898), Dean (1895). Material examined: None. 67 SUBCLASS BRACHIOPTERYGII ORDER P0LYPTERIF0RME3 Branchiostegals absent; median gular absent; a very large pair of lateral gulars (longer than 2/3 of mandible length); large operculum; small suboperculum present - or absent; interoperculum absent. Lower Tertiary (Eocene) to present. A single order and family. It is the only actinopterygian derived group with lateral gulars and lacking branchiostegals. Polypteridae PI. IV Branchio3tegal3: Absent. Gulars: Median gular absent. A pair of large lateral gulars which extend from the symphysis to slightly past the posterior end of the mandibles, extending to the midventral line to a point just before the posterior end of mandibles. A short transverse pit line in the form of an arc is found near the middle of each. According to Daget (1958) the gular pit line is innervated by a nerve extending posteriorly from the middle of the ramus mandibularis facialis . In Calamoichthys the gulars are slightly shorter, and the tip of the right overlaps the tip of the left . The large gular plates are reminiscent of those in Crossopterygii. Operculars: Operculum oval and larger than the triangular suboperculum, in Polypterus. Subopercle absent in Calamoichthys. G i l l membranes separate, with right over left . 6 8 Hyoid arch: A partially ossified hypohyal-, a ceratohyal and interhyal are present. The ends of the ceratohyal and interhyal are not ossified. Taxonomy: The family Polypteridae consists of two living African freshwater genera Polypterus and Calamoichthys (- lirpetoichthys). Relationships: The absence of an interoperculum relates i t to the lower Actinopterygii. The enlarged lateral gulars recall those of Pyritocephalus and Haplolepis (Haplolepiformes). These forms also have pit lines on the lateral gulars, the branchiostegals reduced and the operculum larger than the suboperculum. The similarity of the lateral gulars to the Crossopterygii and Dipnoi is doubtless a parallelism. The lateral gulars expanded to cover the region exposed by the loss of the branchiostegals. Evidence on its relations also comes from its sensory lines (Stensio, 1 9 4 7 ) . The sensory line of the cheek is of the actinopterygian type. In i ts principal features i t is much as in advanced lox*er ganoids, but in some respects i t has reached the stage of holostean ganoids. The postmaxillary line is similar in i ts connection with the dermal bones to lower ganoids and parasemionotids. They are probably derived from some sub-holostean ancestor (Stensio, 1 9 4 7 ) * References: Devilliers ( 1 9 5 8 ) , Daget ( 1 9 5 8 ) , A l l i s ( 1 9 2 2 ) , Berg ( 1 9 4 7 ) . Material examined: Calamoichthys sp., alcoholic specimen, ROM 1 8 8 7 7 , Nigeria. 69 SUBCLASS ACTINOPTERYGII Branchiostegals 0 - 5 0 ; median gular present or absent, primitively with a V-shaped pit line; lateral gulars, i f present, with length less than one half of mandible length; operculum and suboperculum usually present; in higher forms interoperculum usually present; interhyal, epihyal (in higher forms), ceratohyal and hypohyal(s) usually present. Lower Devonian to present. Five supraordinal groups with 56 orders 34 of which are l iving. It has been made apparent by various authors, Stensio, Berg, Schaeffer, Gardiner, etc., that the Chondrostei, Holostei and Teleostei (as constituted) are not natural groups, that the Holostei are polyphyletic and that some of the characters which have been used to distinguish the Holostei occur in certain chondrosteans. This has been interpreted as meaning that the Chondrostei, Holostei and Teleostei are not valid groups. However, other interpretations are possible. Firstly, that some orders have been placed in the wrong group. Secondly, that some of the characters previously used to define the groups have not been diagnostic, although they may tend to be more frequent in one group than the other. According to this interpretation the groups, i f reconstituted and redefined, would be natural. The interoperculum is an important character in diagnosing the Holostei. It is absent in the Chondrostei, present in the Holostei and present in the Teleostei (except in a few groups of the latter where i t is secondarily lost). One group presently placed in the Holostei, the Lepisosteiformes lacks and interoperculum, but has a maxillary free from 70 the cheek. However, the jaws have moved far forward, decreasing the need for an interoperculum and the preoperculum has expanded ventrally supplanting the position normally occupied by one. It may safely be suggested therefore that the interoperculum has been secondarily lost. The Platysiagiformes have been suggested to bear an incipient interoperculum. This has been demonstrated not to be so. Further, i t bears a maxillary fixed to the cheek and clearly belongs in the Chondrostei. Brough (1939) considered that the maxilla of the subholostean Luganoia was free, but his figure 15 shows the end of the maxilla snugly fitted against the large plate-like preoperculumj in other figures i t appears displaced from this notch. It is concluded the maxilla was not free. This and the lack of an interoperculum places i t in the Chondrostei. The Ospiiformes have been considered chondrosteans but their maxillary is free and i t is now clear that they possess an interoperculum (Stensio, 1932; Lehman, 1952). There now appears to be no good reason not to place the Ospiiformes in the Holostei. The Pycnodontiformes lack an interoperculum and yet have been placed in the Holostei. But recent authors (Rayner, 1941 and Gardiner, I960) and the present author are in agreement as to their placement in the chondrosteans near the platysomoids or boba3atraniiforms. When the above changes have been made the Chondrostei and the Holostei comprise homogeneous groups. A l l of the Holostei have a free maxillary and an interoperculum (except that the interoperculum is lost in the Lepisosteiformes) and-all of the Chondrostei have a fixed maxillary (except Dorypterus where the maxillary has shortened and become secondarily free) and lack an interoperculum. That there are 71 other characters which would support this grouping is suggested by Schaeffer's fine association analysis of differential characters. It is also certain that some characters, previously employed to characterize the groups (such as the relation of rays to their supports) do not definitively separate the two groups (however i t might s t i l l be possible to state that in chondrosteans fin rays exceed or equal pterygiophores and in holostei equal pterygiophores). The Teleosti may be separated from the Chondrostei and Holostei by the lack of lepidosteid tubules either in their scales or skeleton, and the primitive presence of intermuscular bones and two hypohyals. For the above reasons the modified supraordinal groups Chondrostei (Group I), Holostei (Group l l ) and Teleostei (Group III) are therefore reinstated. The Teleostei are further subdivided into the Malacopterygii (Group IIIA) and the Acanthopterygii (Group IIIB). The group Mesichthyes is discarded, a 3 wi l l be discussed later. GROUP I. CHONDROSTEI Without interoperculum. Lateral gulars present or absent. A single hypohyal. Lower Devonian to Present. /. ORDER PALAEONISCIPORMES (AEDUELLIFORMES, GYMONISCIFORMES) Branchiostegals (1)4-23, spathiform; median and a pair of lateral gulars; operculum and suboperculum present, interoperculum absent; interhyal, epihyal, ceratohyal and a hypohyal known. Lower Devonian to Lower Cretaceous. Several families. Following Lehman (1958), Bergfs Gymnonisciform.es are included in the Palaeonisciformes. The Palaeonisciformes are a diverse group which 72 will probably be broken up into other orders when a detailed taxonomic revision is made. However, the practice of removing poorly known families from Palaeonisciformes and raising them to ordinal status does not seem advisable. SUBORDER PALAEONISCOIDEI / Cheirolepidae PI. I l l Branchiostegals: 11-13 branchiostegals are found in Cheirolepis; these are short, wide and spathiform. Gulars: A median gular has not previously been reported in this group. In Traquair's (I675) figure of the ventral view of the jaws a diamond-shaped, bilaterally symmetrical bone is seen overlying the lateral gular of the right side. It seems likely that this element is a displaced median gular. As median gulars are found in other Palaeonisciformes, dipnoans and coelacanth3 i t would not be surprising to find one in Cheirolepis. A pair of wide lateral triangular gulars each with a pit line are found next to the last branchiostegal. Operculars: Operculum inclined forwards and much longer than the suboperculum. Relationship: The branchiostegals and opercular bones of Cheirolepis are l i t t l e different from those of the palaeoniscid, Glaucolepis*, most other characters are in accord with this. The small scales, as pointed out by Aldinger (1937) are however, very similar to those of acanthodians. Cheirolepis thus forms a connecting link between the 73 Teleostomi and Acanthodii. References: Traquair (1875), Watson (1925), Woodward (1898), Lehman (1947). ^ Palaeoniscidae P l . I l l Branchiostegals: (l) 4 - 2 3 usually 9-15 branchiostegals, short, wide and spathiform. In Glaucolepis at least, the last 5 branchiostegals insert on the ceratohyal, none on the hypohyal; presumably rays also seat on the epihyal (see NielsenTs excellent photo p l . 1 1 and 1 6 , 1942). Palaeoniscus 8-9, Oxygnathus 12, Gonatodus 10, Boreosomus 7-8, Hyllingea 2 3 , Ple/^molepis ca. 16, Watsonichthys 15, Acrolepis 1 4 , Diaphorognathus 7-9, Pygopteru3 20, Pteronisculus 12-22, Stegotrachelus 6, Glaucolepis 13-15, Cornuboniscu3 16, Rhadinichthys 10-11, Cycloptychius 10, Canobius 4 - 6 , Me30poma 5-7, Nematoptychius 20, Aeduella 1 . Gulars: A median gular is known from most forms which are sufficiently preserved to show it3 presence. In Hyllingea only scales are found in the gular region. In some forms a second gular is found, a small anterior gular near the symphysis overlying a larger median gular which projects posteriorly to the branchiostegals; in these the lateral gulars are absent (Bbreosomus, Diaphrognathus). In most other adequately represented forms a pair of small lateral gulars is present (Pteronisculus, Palaeoniscus, Gonatodus etc.). In Aeduella is a large diamond-shaped gular with V-shaped pit l inei The adjacent "branchiostegal" is identifiable a 3 a lateral gular by the arc-shaped pit l ine. These occupy 74 less than a quarter or third of the length of the mandibles. Operculars; Operculum and suboperculum present, the operculum generally larger and inclined forwards, except in the short-jawed forms. Hyoid arch: Watson (1925) did not find an ossified epihyal in any palaeoniscid. He described a short hypohyal and a long ceratohyal in Elonichthys. A hypohyal, long ceratohyal and short element (called 2nd ceratohyal (=epihyal, separate ceratohyal ?) by Nielsen, (1942) are evident in Glaucolepis. The ceratohyal and this element bear a lateral groove for the afferent hyoid arteries. Acrorhabdus is known to have a long hourglass-shaped ceratohyal and a triangular epihyal (?) (shaped similarly to the epihyal in higher forms such as Salvelinus); another bone, possibly the interhyal is present. Taxonomy: Palaeoniscidae is here used in the sense of Berg (1947) and includes the families Thrissonotidae (Oxygnathidae), Rhadinichthyidae, Canobiidae, Elonichthyidae, Pygopteridae, Acrolepidae, Amblypteridae, Scanilepidae, Dicellopygidae, Boreolepidae, Cocconiscidae, Cornuboniscidae, Tegeolepidae, Styracopteridae, Aeduellidae (the latter raised to ordinal status in a classification outline in Traite de Zoologie, vol. XIII, fasc. 3, 1958). Gymnoniscidae (Gymnonisciformes of Berg) is included in Palaeoniscidae following Lehman (1958); Westoll (1944) believed i t to be only the young of Sceletophorus, a palaeoniscid. Berg's lumping of so many diverse forms into the family Palaeoniscidae does not seem reasonable. Perhaps Rorner's recognition of many families would be more acceptable, although he fai ls to segregate 75 any of the lower Chondrostei into orders. However, the author did not feel in a position to accept a l l of these families, many monotypic, without analysis, Nielsen (1949) felt that Aldinger had gone somewhat too far in subdivision of the old palaeonisciid group. Berg's classification was therefore provisionally followed, rather than inadvisedly raising groups to family status. In looking at the variation of the branchiostegal series in a few of the many known paleoniscids, family recognition of certain forms seem to be suggested. Proper delimitation into families wil l of course require examination of a l l forms and looking at more than one set of characters. The two median gulars of Boreosomus and Diaphrognathus set them well aside from other palaeoniscids. Several forms have sufficiently numerous branchiostegals to separate them from other palaeoniscids: Agecocephalus, Hyllingea, Nematophychius and Pygopterus have 20 or more branchiostegals. On the other hand the few highly curved branchiostegals in Canobius and Mesopoma would seem partly to justify the family Canobiidae, Aeduella is distinct in having only a single branchiostegal. Although most of the other genera examined are united in having a median gular, a pair of lateral gulars, and 9 to 15 branchiostegals, family delimitation on other bases is of course possible. References: Berg (1947, 1955), Nielsen (1942), Moy-Thomas and Dyne (1937), Aldinger (1937), Woodward (1091), Woodward and White (1926), Lehman (1958), Brough (1933), White (1933), Gregory (1951), White (1939). 76 yf ' Coccolepidae Branchiostegals: About 14 in Coccolepis macropterus. The branchiostegals are short, spathiform and slightly curved, s:ijrdlar to those of palaeoniscids. Reference: Berg (1947). ^ Birgeriidae Branchiostegals: About 14 in Birgeria. These are of medium length and spathiform. Gulars: A small oval median gular and a pair of small lateral gulars which are almost identical to the adjacent branchiostegals were found. Pit lines absent. Operculars: The opercular bone3 are separated. The operculum is horizontal and oval; the suboperculum consists of 4 to 6 vertical triangular rays which are fused ventrally, a unique feature in the Palaeonisciformes. Hyoid arch: A long ceratohyal and a short hypohyal are known. Taxonomy: Birgeria may be a synonym of Xenestes; this, according to present rules, need not result in a change of the family name. Relationships: The unique, fan-like subopercle distinguishes this family from other palaeonisciforms. The branchiostegals and gulars are consistent with a palaeoniscid derivation. Mielsen (1949) believed the body skeleton and dermal bones of the head point to a fair ly close relationship to the Palaeoniscidae. V/atoon (1925), Stensio (1932) and Aldinger (1937) believed the chondrocranium, spiracle, position of nostril and reduced endoskeleton indicate an approach to the sturgeons. References: Nielsen (1942)* ^ Holuridae Branchiostegals: 12 in Holurus parki, 8 in Holuropsis yavorskii, these short, spathiform with lower edge curved. Operculars: Operculum and suboperculum present, rectangular. References: Moy-Thomas (1937), Berg (1955). ^ Urosthenidae Branchiostegals: Head not known. Lehman (1958) would reduce this family to a subfamily of Palaeoniscidae. References: Berg (1947, 1955). ^ Turseoidae Branchiostegals: 6-7 in Gwynoddichtis. Wide, slightly curved spathiform rays. Gulars: The small median gular is considerably behind the symphysis and sits on top of the lateral gulars. The lateral gulars extend to the symphysis. Operculars: Operculum a slender rectangle longer than the trapezoidal suboperculum. 78 Relationship; The Turseoidae differ from the palaeoniscids in the arrangement of the gular plates and in the reduced number of branchiostegals, but not so highly as to preclude immediate derivation from them. References: Bock (1959). SUBORDER PLATISOMOIDEI Although platysomoids tend to have fewer branchiostegals and more vertical operculars than the palaeoniscoids, both of these characters overlap. The two characters seem to be associated with a shortening of the jaws, requiring fewer branchiostegals to cover the throat and with erecting the suspension, making the operculars more vertical. They are doubtless palaeoniscoid derivatives, / Platysomidae Branchiostegals: 6-7 in Platysomus, about 4 in Paramesolepis tuberculata. Wide, spathiform, with ventral edge curved upwards. Operculars: Operculum and suboperculum are vertical and about equal in size. References: Lehman (1958), Moy-Thomas and Dyne (1937), Woodward (1898). ^ Amphicentridae Branchiostegals: About 5 in Cheirodopsis, 6 in Eurynotus geikei, about 7-9 in Amphicentrum, elongate, spathiform, widening distally, ventral edge straight. 79 Operculars: Operculum and suboperculum high, and oval or rectangular, about equal in 3 i z e and vertically oriented. References: Lehman (195G), Woodward (l«9l), Homer (1955), Dyne (1939). f ORDER TARRASIIF0RME3 Branchiostegals 15, spathiform; gulars not known; operculum and suboperculum present, interoperculum absent. Lower Carboniferous. A single family. ^ Tarrasiidae PI. I l l Branchiostegals: 15 in Tarrasius problematicus. Spathiform, lower edge curving upward, upper ones elongate, lower short. Gulars: Text and figures make no mention of gulars. The anterior branchiostegal is figured as broader and may possibly represent a lateral gular. Operculars: Operculum and interoperculum subrectangular and horizontally oriented. Relationships: Tarra3iu3 differs from most palaeonisciforms in the opercle and subopercle being long horizontally and the operculum not being greatly inclined forwards. Unlike other short-jawed forms the number of branchiostegals has not been reduced but they have been bunched closely together. The shortening of the jaws may be associated with the durophagus habit indicated by the teeth. The number and form of the branchiostegals are not out of line with a palaeonisciform 80 derivation. According to Moy-Thomas, "Tarrasins problematicus is certainly a Palaeoniscid, as shown by the characteristic arrangement of the bones of the skull, axial skeleton, supporting the skeleton of the median fins and microscopic structure of the scales. It is however, specialized in having continuous dorsal and ventral fins, and a diphycercal t a i l , loss of scales on the main body and shagreen-like scales on the caudal region, the peculiar shape of the pectoral, and loss of the pelvic fins". One might add to these specializations the elongate, blenny-like form. References: Traquair (1881), Moy-Thomas (1934). ^ ORDER PHANER0RHYNCHIF0R11ES A long series of branchiostegals; gulars not known; operculum and suboperculum present; interoperculum absent. Middle Carboniferous. A single family. ^ Phanerorhynchidae Branchiostegals: Phanerorhynchus is described as having a long series of branchiostegals. The two branchiostegals depicted appear narrow, pointed and quite small. Gulars: It wa3 impossible to 3ee whether lateral or median gulars were present in the specimen. Operculars: A long and slightly oblique operculum with sparse longitudinal ridges and a small subopercular present. Taxonomy; Romer (1955) places Phanerorhynchus in the Haplolepidae, apparently on the basis of the scales. But Phanerorhynchus is quite different from Haplolepidae in the possession of a long sturgeon-like snout. Westoll (1944) in a revision of the Haplolepidae states that Haplolepis is sharply distinct from Phanerorhynchus and that the possession of deep flank scales by these two groups i 3 most feasibly regarded as parallelism. Phanerorhynchus is thus here placed in its own order, as in Berg. Relationships; The branchiostegal series are too poorly known to shed light on relationships. G i l l and Watson believed the skull and shoulder girdle showed conclusive evidence of palaeoniscid affinities, but the build, character of the caudal peduncle, short-based pelvics, few large dermal rays in the fins and long undivided rostral bone were distinctive. Stensio (1932) placed i t in the vicinity of Saurichthyidae. But as Berg (1947) noted the Phanerorhynchidae are so specialized as regards to fin structure that they cannot be regarded as ancestors either of Saurichthyidae or Acipenseridae. It thus seems preferable to regard them as an independently evolved sturgeon-like form derived from the P alaeonisc iforme s. References; G i l l and Watson (1923), Stensio (1932). / ORDER HAPLOLEPTFORMES Branchiostegals 1-3 spathiform, the one next to the lateral gular expanded; large median and la'teral gulars; V-shaped pit line on median gular; opercle and subopercle present; interoperculum absent. Upper Carboniferous. A single family. 82 ^ Haplolepidae PI. I l l Branchiostegals: 1-3 in Haplolepis; 1 in Pyritocephalus. The branchiostegal behind the lateral gular (present in a l l forms) is large, rectangular or triangular and may be as large as one third of the size of the lateral gulars. The second and third branchiostegals, when present, are very broad with rectilinear edges. The exposed portion of the branchiostegals is very short. In Pyritoc ephalus there is a gap between the single branchiostegal and the suboperculum. Gulars: A large median gular 1/3 - h the length of the mandible is present; i t bears a V-shaped pit line with the apex facing posteriorly. A large pair of lateral gulars occupy about 1/3 the length of the jaws. A short transverse pit line is found in the middle of each lateral gular. The enlarged anterior branchiostegal, referred to in the literature as a posterior paired gular, is here considered a branchiostegal. Operculars: Either the operculum or the suboperculum may be larger. These are vertically oriented in this small-jawed form. Taxonomy: After Westoll's thorough revision of the family, Lehman raised i t to ordinal status. Relationships: The short jaws, vertically oriented operculars and few branchiostegals in the Haplolepidae remind one of Catopterus and Mesopoma, although the large median and lateral gulars of Haplolepidae do not appear to be present in these forms. However, the similarities in operculars and branchiostegal rays may not be an indication of close 03 common ancestry, but only parallel evolution following a reduction in jaw length. The large size of the gulars and the few branchiostegals distinguish the Haplolepidae from the palaeoniscids, perhaps at the ordinal level, although i t is , no doubt, derived from them. References: Westoll ( 1 9 4 4 ) , Lehman ( 1 9 5 8 ) . / ORDER REDFIELDIIFORMES One large spathiform branchiostegal (sometimes split) below the suboperculum, perhaps absent in some; gulars absent; opercle and sub-opercle present; interoperculum absent. Lower to Upper Triassic. A single family. Following Lehman ( 1 9 5 8 ) Brookvaliidae is synonymized with Redfieldiidae. ^ Redfieldiidae P l . V Branchiostegals: 1 large trapezoidal branchiostegal (called infraoperculum by some authors) below the subopercle in Brookvalia; a triangular one in Geitichthys, Atopocephala, Helichthys and Phylctaenichthys. In Daedalichthys is a curious oval plate (overlying the suboperculum) which is divided into two by a horizontal joint, the lower part larger. Doubtless the plate represents a single split branchiostegal. Redfieldius on the other hand, according to the figure in Berg ( 1 9 5 5 ) , has a longer suboperculum and lacks a branchiostegal. The term infraoperculum is equivalent to the branchioperculura of Hubbs; both refer to an enlarged uppermost branchiostegal. Gulars: A small narrow median bone may represent a gular in Brookvalia. 84 Gular plates do not appear to be present in other genera. Opercular bones: Opercle and subopercle approximately equal, sub-quadrangular, almost vertical. Taxonomy: Catopteridae was f irst applied to the family in 1890. In 1899 Hay discovered Catopterus Redfield 1837 was a junior homonym of Catopteru3 Agassiz 1833. Hay proposed a replacement name, Redfieldius for Catopteru3 Redfield. Then he proposed a new family name Dictopygidae, based on Dictopyge (a genus in the same family) to replace Catopteridae. But, according to Article 39a of the International Code of Zool. Nomenclature, the f amily name must be changed to one based on the valid name of the original nominal type-genus, in order to preserve the taxonomic concept. That is , the new family name must be based on the substitute name Redfieldiu3 (there being no junior synonyms available), rather than on another genus in the same family, Dictopyge. Berg was therefore correct in applying Redfieldiidae and Redfieldiiformes to this group and the name Dictopygidae is therefore a junior subjective synonym. Catopteridae has been used by most authors for this family: Stensio (1921, 1932), Wade (1935), Brough (1934), and Lehman (1958), although Romer (1955) used Dictopygidae. Most of these authors were apparently unaware of the homonymy of Catopterus. The last section of Article 39 ("The provisions of this section are not to be applied so as to upset a widely accepted family-group name that was established before 1961 under a different procedure") does not apply to Catopteridae although i t might have applied to Dictopygidae had i t been widely accepted. So Redfieldiidae and Redfieldiiformes are the proper familial and ordinal names for this group. 85 Relationships: Lehman (1958) considered the Perleidae and Catopteridae quite close, differing principally in dentition and fin anatomy. He therefore synonymized the two orders employed for them by Berg under Perleidiformes. The present study shows the branchiostegals of Bergs two order3 differ markedly. In Redfieldiiformes there io one broad branchiostegal. In Perleidiformes on the other hand there are 7-12 branchiostegals. Gular3 are unknown in Redfieldiiformes, median and lateral gulars are found in Perliediformes. Brough (1931) presents a table giving 9 differences but Schaeffer (1955) indicates that only two of these, excess of rays over radials and the number of branchiostegals are consistent and taxonomically important. The dilated branchiostegals of Redfieldiiformes are suggestive of those in Haplolepidae, which also has very few branchiostegals (one to three). Haplolepidae differ in the possession of large well developed lateral and median gulars. References: Berg (1955), Wade (1935), Brough (1931* 1934). / ORDER PERLEIDIFORMES Branchiostegals 7-12, spathiform; a median gular with a V-shaped pit line and a pair of lateral gulars; operculum and suboperculum present; interoperculum absent; ceratohyal known. Lower Triassic to Upper Triassic. A single family. Teleopterina (=Pyritocephalus) is included in the Haplolepidae following Westoll (1944). Cleithrolepidae is included in the Perleidae following Lehman (1958). 86 J 1 P e r l e i d a e P l . I l l Branchiostegals: 7 broad spathiform i n P e r l e i d u s (not i n c l u d i n g the lateral gular); 10 narrow spathiform i n Meridensia; 12 narrow spathiform in Cleithrolepis and Colobodus; at l e a s t 2 or 3 i n the fragmentary specimen of Manlietta. On the left, but not the right side, of Lehman's figure of Perleidus madagascariensis is a small branchiostegal half the length of the others; presumably this is teratological. Gulars: An oval or diamond-shaped median gular of moderate size is present. In Perleidus i t bears a V-shaped pit line, v/ith the apex facing posteriorly. The median gular of Manlietta is very large, about as long as the mandible. The lateral gulars are identical in form to the adjacent branchiostegals, which are short and wide in Perleidus; but the lateral gulars bear pit lines. Operculars: Suboperculum larger than the operculum. These are not deep. Hyoid arch: In Perleidus the ceratohyal is short aid expanded at both ends in the usual hourglass-shape. Relationships: Differences between this order and Redfieldiiformes are discussed under the latter order. Wade (1935) indicates that Cleithrolepis is probably derived from the platysomids. However, the more numerous branchiostegals of Cleithrolepis cast doubt on this view. The V-shaped pit lines on the median gulars of Haplolepiformes aid Perleidiformes would seem to indicate a common ancestory, probably from the Palaeonisciformes. Because of the more numerous branchiostegals of 87 Fer-leidiformes i t is unlikely that they were directly derived from the Haplolepiformes. The sensory canal system is very close to that of Glaucolepis (Palaeoniscoidei) (Lehman 1952). References: Lehman (1952), Brough (1939), Wade (1935), Stensio (1921), Schaeffer (1955). ^ ORDER DORYPTERIFORMES Branchiostegals absent; gulars absent; operculum, suboperculum and interoperculum absent. Upper Permian. A single family. ^ Dorypteridae PI. V Branchiostegals: Absent. Gulars: Absent. Operculars: Absent, although a small bone has been tentatively identified as an operculum in one specimen. Hyoid arch: A long rectangular bone has been termed the ceratohyal but Westoll suggests that i t is an ectopterygoid. Relationships: The reduction in the branchiostegal series suggests derivation from the Platysomoidei rather than the Palaeoniscoidei. Westoll believed only the Platysomidae could have given rise to them. Lehman (1958) places the order in the family Platysomidae. While i t is possible that Dorypterus should not be ordinally separated from the platystomids i t requires at least familial separation on the basis of: m absence of branchiostegals, absence of scales on a l l but the abdomen, the skull being roofed by a single bone aid the well developed ventrals. Further, Westoll states that Gi l l ' s recognition of i t 3 remarkable specialization can only be more strongly endorsed. In the relation of the number of fin rays to pterygiophores Dorypterus is intermediate between platysomids and Bobasatrania according to Stensio ( 1 9 3 2 ) . The order Dorypteriformes is thus provisionally retained. References: Berg ( 1 9 4 7 ) , Stensio ( 1 9 3 2 ) , G i l l ( 1 9 2 5 ) , Westoll ( 1 9 4 1 ) . ^ ORDER BOBASATRANIIFORMES Branchiostegals, i f present, 1-2 or to 4; gulars unknown; operculum present; suboperculum and interoperculum absent. Lower Triassic. A single family. ^ Bobasatraniidae P l . V Branchiostegals: One or two, i f present; in some, oval or quadrate in shape, in others there appears to be traces of at least 4 elongate spathiform branchiostegals. Gulars: Unknown. Operculars: A small narrow operculum present. Suboperculum absent; appears to have coalesced with the lower preopercular plate. Hyoid arch: A small hourglass-shaped ceratohyal. Relationships: In the reduction of the branchiostegal series Bobasatraniiformes would appear to be related to but not immediately 89 derived, from the Dorypteriformes. They probably share a derivation from the platysomoids (Stensio, Westoll). Referencesi Berg (1947), Stensio (1932), Lehman (1958), Lambe (1914), Westoll (1941). f ORDER PYCN0D0MTIF0RME3 Branchiostegals at least 2-5, acinaciform or spathiform; gulars absent; elongate vertical operculum; suboperculum replaced by preoperculum; interoperculum absent. Upper Triassic to Eocene. Three families. Relationships; The phylogenetic position of this order has been something of an enigma. Berg and Romer place i t among the holosteans, Arambroug and Bertin in their enlarged order Amiiformes. Stensio (1947), until the question of holostean classification has been solved, provisionally retains them in the Holostei. Gardiner (i960) derives them, with the Bobasatraniiformes, from an offshoot of the Dorypteriformes. Rayner (1941) places them in the subholosteans and regards them as derived from the platysomids. Because of the absence of an interoperculum they are here placed in Group I. Their skull and body structure shows similarities to the Dorypteriformes and Bobasatraniiformes. With them the Pycnodontiformes share a large ventral preoperculum which replaces the suboperculum; a dorsal preoperculum; interoperculum absent; operculum elongate and vertically suspended; branchiostegals reduced; about 32-35 neural spines; vertebral centra absent; a postabdominal bone; dorsal and anals long; median fin rays tend to equal radials; body deep. In Gyrodus and 90 Bobnsatrania a lateral line branch runs from the skull to the dorsal origin. However the Pycnodontiformes differ in possession of teeth and branchiostegals and in their abdominal pelvic fins whereas in the Dorypteriformes they are thoracic. In some Pycnondontiformes the supra-and infra.-orbital canals join (Microdon, Eomesodon), unlike Bobasatrania. It therefore seems necessary to regard the Pycnodontiformes as an offshoot of the line which gave rise to the Dorypteriformes and Bobasatraniiformes. This opinion is in agreement with that of Rayner and close to that of Gardiner (i960). Their slender curved branchiostegals, graduating to a point, distinguish them from other orders in Group I. / Gyrodontidae P l . V Branchiostegals: Unknown in Macromesodon (= Mesodon). One or two long curved sabre-like branchiostegals under the preopercle in Mesturus. Two elongate narrow rectilinear branchiostegals in Gyrodus. Gulars: Unknown in Macromesodon. Absent in Mesturus where the inter-mandibular region and throat is covered by polygonal plates. Operculars: An elongate vertically suspended operculum in Mesturus and Macromesodon; suboperculum replaced by large backwardly extending pre-operculum. Stensio (1947) believed the suboperculum and preoperculum were fused and called the bone the preoperculo-suboperculum. Hyoid arch: Poorly known but the ceratohyal is deep and a small, hypohyal is known in Mesturus. 91 References: Woodward (1895), Berg (1947), Arambourg and Bortin (1958), Ararabourg (1954), Eastman (1914)• / Coccodontidae Branchiostegals: Data on branchiostegals of Coccodus has not been found in Woodward (1895) or in any other available literature. Z i t t e l (1887) reports the operculum i s large in Zenophilus; possibly his operculum represents the preoperculo-suboperculum. "f" Pycnodontidae Branchiostegals: 4 long narrow curved sabre-like branchiostegals are apparent in one specimen of Pycnodus (plate 72, Agassiz, 1833-1843). Remains of 5 spathiform ones known in Palaeobalistum. Gulars: Unknown, presumably absent. Operculars: Operculum apparently small with ventro-posterior projection. Preoperculo-suboperculum large as in Gyrodontidae. Relationships: The available material does not indicate great differences between the three families of the order. Indeed Romer includes them in one family. References: Woodward (1895), Agassiz (1833-43). / ORDER PTYCHOLEPIFORI'IES Several spathiform branchiostegals; a median gular present; operculum and suboperculum present; interoperculum absent; ceratohyal known. Lower Triassic to Lower Jurrasic. A single family. 92 / Ptycholepidae P l . V Branchiostegals: Number not exactly indicated. In Ptycholepis f irst branchiostegal ornamented and wedge-shaped. Ventral to i t there are a series of more normal branchiostegal rays, which are moderately large and elongate, Gulars: A large median gular present. Operculars: Operculum larger than the suboperculum. Hyoid arch: Ceratohyal is a rather long flat bone somewhat constricted in the middle. Taxonomyi Brough demonstrated that Ptycholepis did not belong in the Eugnathidae and placed i t in its own family, Ptycholepidae. Lehman (1958) raised the family to ordinal status. Relationships: Brough places Ptycholepis in the Subholostei. Lehman indicates i t is probable that Ptycholepis is a descendent of the Triassic palaeoniscid genus Boreosomus or of a genus close to Boreosomus. The paucity of data on the branchiostegal series prevents further suggestions being made except to 3ay that the absence of an interoperculum precludes placement among the holostean fishes. References: Brough (1939), Lehman (1958), Woodward (1895). / ORDER, PH0LID0PLEURIF0RJIE3 Branchiostegals 6-14, spathiform; a small median gular and a pair of large lateral gulars are known; operculum and suboperculum present; 93 interoperculum absent; hypohyal and ceratohyal known. Lower to Upper Triassic. A single family. ' Pholidopleuridae Branchiostegals: 6-14. Australosomus with 6-8 short wide branchiostegals which become shorter and broader anteriorly; Macroaethes with 14 (19 in figure) elongate branchiostegals with curved edges and rounded tips; Arctosomus with about 12; Pholidopleurus with at least 4 (series incomplete) branchiostegals which are elongate, fairly broad and resemble those in Amia. In Australosomus the lower 3 branchiostegals and lateral gular appear to insert on the ceratohyal. Gulars: A small median gular in Australosomus situated anterior to the lateral gulars. A large triangular lateral gular triangular in outline i 1 1 Australosomus. A large triangular lateral gular in Macroaethes. A lateral gular is not shown in the figure of Arctosomus (Berg in Nielsen, 1949). No pit lines are found on the median or lateral gulars. Operculars: Operculum somewhat higher than the suboperculum in Australosomus but equal in Macroaethes and Pholidopleurus. Hyoid arch: In Australosomus a small curved hypohyal is placed at the upper anterior corner of the long compressed ceratohyal which is grooved for the afferent hyoid artery. Taxonomy: The few short broad branchiostegals of Australosomus appear to differ sufficiently from those of Macroaethes and Arctosomus to suggest placement in a different family or subfamily. 94 Relationships; Lehman (1952) criticizes the view that Australosomus i 3 closely related to the Saurichthyidae. Nielsen and Lehman agree that the common characters of the two groups are those shared by most primitive palaeoniscid-type fishes. The few branchiostegals and absence of a suboperculum in Saurichthys indicate i t does not have close affinities \d.th Pholidopleurus. The branchiostegal series of the Pholidopleuriformes would seem to indicate an origin either from the Perleidiformes or from the Palaeonisciformes. Stensio (1932) considers i t an offshoot from primitive palaeoniscids specialized in the direction of higher ganoids and teleosts. References: Nielsen (1949), Lehman (1952), Wade (1935). / ORDER CEPHALOXENIFORMES Poorly known, see under description of single Upper Triassic family. / Cepaloxenidae PI. V Branchiostegals: Unknown, the lower portion of the head not being represented. Gulars: Unknown. Operculars; Operculum larger than suboperculum. Presumably no inter-operculum. Taxonomy,: Lehman (1958) raises BroughTs family to ordinal status. 95 Relationships: Brough states the fins are of the holostean type, the check bones are essentially palaeoniscid. He places them in the Subholostei between Platysiagidae and Peltopleuridae. The deep flank scales resemble those of the Luganoiidae and Peltopleuridae. References: Brough (1939). t ORDER AETHODONTIFORMES Poorly known, see under description of single Upper Triassic family. / Aethodontidae Branchiostegals: Not preserved. Gulars: An oval median gular plate is known. Lateral gulars unknown. Operculars: Operculum and suboperculum about equal in size. Inter-operculum presumably absent. Taxonomy: Brough*s family was raised to ordinal status by Lehman (1958), although only fragmentary remains of the fish are known. Relationships: Tne hemiheterocercal t a i l and fins with reduced rays are subholostean in character while the cheek bones and sensory canals are palaeoniscid (supraorbital line ends on the frontal). The crushing dentition is unique among the subholostean3. References: Brough (1939). 9 6 / ORDER LUGANOIIFORMES Poorly known, see description of the single Upper Triassic family. ^ Luganoiidae PI. V Branchiostegals: Unknown. Gulars: Unknown. Operculars: The operculum and suboperculum are about equal and triangular in Luganoia, while the suboperculum is smaller and rectangular in Besania. Interoperculum absent. Taxonomy,: Lehman (1958) raised Brough's family to ordinal status. Relationships: Brough states that a l l the characters are holostean except a) the absence of an interoperculum and the g i l l cover of almost equal operculum and suboperculum, b) the plate-like preoperculum. They are amongst the most developed of the subholosteans. Subholostean characters are: the rounded, hemiheterocercal t a i l and the well spaced rays of dorsal and anal which equal in number the endoskeletal supports. The deep body scales resemble those of the Peltopleuridae and Cephaloxenidae. / ORDER PELTOPLEURIFORMES Branchiostegals 6-7 elongate spathiform; gulars unknown; operculum and suboperculum present; interoperculum absent. Upper Triassic. A single family. 9 7 / Peltopleuridae Pl . I l l Branchiostegals: In Peltopleurus about 6-7 elongate 3pathii.'orm branchiostegals, the uppermost of which is not enlarged. Gulars: Unknown. Operculars: A large triangular operculum aid a smaller rectangular suboperculum. Taxonomy: Lehman (1958) raised Brough'3 family to ordinal status. Relationships: From the branchiostegals i t can be said that i t is unlikely that the Peltopleuriformes arose from the Ptycholepiformes, Redfieldiiformes, Bobasatraniiformes, Haplolepiformes or Dorypteriformes; origin is possible from the Palaeonisciformes, Perleidiformes, or Pholidopleuriformes. According to Lehman (1958) the scales of Peltopleurus recall a l i t t l e those of Australosomus; however to the present author the scales seem more similar to those of Cephaloxenus or even Luganoia. Lehman then states that the bones of the cranial roof, with the square parietals and the externally homocercal ta i l recall those of the Perleidae. Altogether, a perleidiform ancestory of the Peltopleuriformes on a common line with the Cephaloxeniformes seems not improbable. References: Brough (1939), Lehman (1958). ye ^ ORDER PLATYSIAGIFORMES Branchiostegals about 13, spathiform; gulars unknown; operculum and suboperculum present; interoperculum absent. Upper Triassic and Lower Jurassic. A single family. ^ Platysiagidae P I . rv Branchiostegals: Number about 13 in Platysiagum minus (text, not figure of Brough). These are broad, short and spathiform. The f irst branchiostegal is subrectangular. There is a gap, probably a result of preservation, between this branchiostegal and the others. Gulars: Unknown. Operculars: The large operculum sends a wedge-shaped projection into the about equal-sized suboperculum. A small, concavity on the anterior half of the lower edge of the suboperculum provides for the presences of a small branchiostegal (called by Brough a modified branchiostegal or incipient interoperculum). It has been shown that this is an ordinary branchiostegal• Hyoid arch: Unknown. Taxonomy: Brough erected this family and Lehman raised i t to ordinal status. Relationships: The presence of about 13 branchiostegals suggests origin from the Pholidopleuriformes or Palaeonisciformes. References: Brough (1939), Lehman (1958). 99 t ORDER CHONDROSTEIFORMES Branchiostegals about 9-12, spathiform; gulars absent; operculum and suboperculum present; interoperculum absent. Lower Jurassic. Two families. ^ Chondrosteidae P l . IV Branchiostegals; 9-10 in Chondrosteus acipenseroides, about 12 in C. lindenburgi; the uppermost is slightly enlarged and the upper 6 and suboperculum are serrate basally on the ventral edge. Gulars: Gulars absent. A pair of small curved plates above the anterior end of the ceratohyal, are believed to be lateral gulars by Watson (1925), but appear to the author more likely to be hypohyals. Operculars: Operculum considerably smaller than the subopercular and shaped like an inverted comma. Suboperculum large and subquadrate and serrate on the anterior ventral corner. Hyoid arch: The "anterior branchiostegal" of Watson appears to be a hypohyal. An ordinary medium sized hourglass ceratohyal is present. The remains of an interhyal is suggested in some figures (Woodward, Watson). It was perhaps unossified. Relationships: Discussions under the Acipenseriformes show that Chondrosteidae cannot be closely related or immediately ancestral to the Acipenseriformes. From the point of view of the number and form of the branchiostegals the Chondrosteidae could have developed from the Palaeonisciformes, the 100 Perleidiformes, the Ptycholepiformes, or the Platysiagiformes. But the dorsal and anal fin rays equal the basals in the Perleidiformes, Platysiagiformes and Ptycholepiformes and the caudal is symmetrical or nearly symmetrical in Pholidopleuriformes. The latter groups are thus excluded from the ancestry of Chondrosteidae, leaving the Palaeonisci-formes. Because of the number of branchiostegals, origin is most l ikely from the suborder Palaeoniscoidei. This conclusion is in agreement with Watson (1925) who states . . . "it shows in its skull structure clear evidence of Palaeoniscid origin." Lehman (1958) isolates the Chondrosteidae in their own order. This separation may be justified by the lack of a preoperculum in the Chondrosteidae. References: Woodward (1895), Hennig (1925), Watson (1925). ^ Errolichthyidae Branchiostegals: Four elongate spathiform branchiostegals are known. There may have been more. The distal half of the branchiostegals is divided by three or four grooves - a character unique in teleostomes. Gulars: Not known* Operculars: A large operculum and suboperculum are known. These bear grooves on their posterior half similar to those on the branchiostegals. Interoperculum absent. Hyoid arch: Unknown 101 Relationships: The exact phylogenetic position of this family is uncertain. The absence of an interoperculum indicates its placement in the chondrostean fishes (Group I). The discussion under Acipenserif orme3 demonstrates that i t i3 not closely related or ancestral to that group. Lehman (1952) is followed in placing the Errolichthyidae in the Chondrosteiformes. ^ ORDER SAURICHTHYIFORMES Branchiostegals about 1-3; gulars absent; large semicircular operculum; suboperculum and interoperculum absent. Lower Triassic to Lower Jurassic. A single family. / Saurichthyidae P l . IV Branchiostegals; One long slender lanceolate branchiostegal in Saurichthys ornatus. About three long wide branchiostegals questionably figured for JS. madagascarensis; in this species they are indicated as being at an angle of about 45° to the mandible, unlike S. ornatus where the single branchiostegal is parallel to the mandible. Gular; Absent. Operculars: Operculum large and semicircular in Saurichthys, smaller in the Lower Jurassic species. Suboperculum absent (or may be incorporated into the operculum). Hyoid arch: A short curved hypohyal and a very long slender ceratohyal with the middle slightly constricted, are known to be present. 102 Relationships: The branchiostegal form in S. madagascarensis is uncertain. That of S. ornatus is unlike that of any other chondrostean in its extremely elongate form. From the number of branchiostegals i t is unlikely that the Saurichthyiforme3 are related to the Dorypteriformes, Bobasatraniiformes, Redfieldiiformes or Haplolepiformes. Stensio (1925) indicated that many osteological characters and sensory canal characteristics indicate a close relationship to the Acipenseriformes and concludes they are intermediate between the Palaeonisciformes and the Acipenseriformes. However, the Saurichthyiformes, lacking a suboperculum, cannot have given rise to Acipenseriformes which possesses one. Further the supraorbital canal of Saurichthyiformes ends on the frontal while in Acipenser i t joins the infraorbital canal. Thus the Saurichthyiformes cannot be immediately ancestral to the Acipenseriformes and the long rostra may be a parallelism. References: Stensio (1925), (1932), Lehman (1952), (1958), Berg (1947), Griffith (1962). ORDER ACIPENSERIFORMES One branchiostegal; operculum deeply engraved or incised; sub-operculum present or absent; interoperculum absent; gulars absent; interhyal, ceratohyal and a hypohyal present. Upper Cretaceous to present. Two families. Several authors, Traquair (1887), Watson (1925), Aldinger (1937), Stensio (1932) and Lehman (1952), (1958) have forwarded views that Chondrosteidae are intermediate between the palaeoniscoids and sturgeons or that they were ancestral to sturgeons. Wilimovsky (1956) on the other 1 0 3 hand expressed doubt that the unarmoured Chondrosteidae were ancestral to the armoured Acipenseridae. There are further difficulties to supposing that Chondrosteidae were ancestral, since they lack a sub-operculum, clavicle, preoperculum and ribs, a l l of which are found in Acipenseridae. The caudal is symmetrical in Chondrosteidae, heterocercal in Acipenseriformes. The supraorbital canal according to Traquair's figures of Chondrosteidae ends on the parietal while in the Acipenserif ormes i t joins the infraorbital canal (liacAlpin, 1947). Thus i t seems doubtful that the Chondrosteidae were ancestral to the Acipenseriformes (although i t is possible, though not l ikely, that they had a common ancestor not in the distant past). Lehman (1952) described a new fossil fish, Errolichthys which he placed in a separate family in association with Chondrosteidae and Acipenseridae, while (1958) he stated that i t was a representative of an order having characters in common with the Chondrosteidae and Palaeoniscidae. However, like Chondrosteidae the supraorbital canal fai ls to join the infraorbital canal and so cannot be close to the Ac ipen seriforme s. The only chondrosteans in which the infraorbital and supraorbital canals unite are the Redfieldiiformes. The large eyes, short snout, terminal mouth, developed teeth of the Redfieldiiformes differ from Acipenseriformes. The Phanerorhynchiformes considerably resemble the Acipenseriformes in their long snout, mouth and rows of scutes but are too specialized in their fins to have been ancestral. However, i t i s possible that a less specialized ancestor of either of these two orders gave rise to the Acipenseriformes. 104 Finally the presence of an operculum, suboperculum, interhyal and hypohyal argue strongly against Severtzoff*s association of the Acipenseriformes with the Selachii. These bones are not known in the Selachii (although Pleurocanthus has a hypohyal), but are found in the Teleostomi. As ably pointed out in Stensio's (1932) critique of Severtzoff, the "selachian" characters of Acipenseriformes such as the heterocercal t a i l and spiracle are primitive characters also shared by generalized Actinopterygii. Acipenseridae P l . IV Branchiostegals; Most previous authors have stated that branchiostegals are absent. However, examination of a specimen of Acipenser fulvescens shows a chevron-shaped bene exposed below the operculum. It is stubby but definitely a branchiostegal. Dissection reveals a plate of slightly larger extent which sends a slight dorsal flange up under the suboperculum, which is completely skin-covered in this species. Holmgren and Stensio (1936) describe and figure the branchiostegal and suboperculum as being at least partly exposed in Acipenser sturio. The chevron-shaped branchiostegal of Acipenser is quite reminiscent of the forked branchiostegal found in Polyodontidae. The branchiostegal appears to seat on the interhyal. Gulars: Absent. Opercularst A large grooved circular operculum with an anterior projection, partly skin covered, is found. Below this is a small, narrow, cartilage, the suboperculum. Its dorsal end underlies the 105 operculum. There has been debate as to whether the large upper bone was an operculum or suboperculum. But the discovery by Holmgren and Stensio (1936). of a lower g i l l cover bone, the suboperculum, clearly identifies the large upper element i n Acipenser as an operculum. G i l l membranea joined to isthmus. Hyoid arch: The hyoid arch is unossified. A short hypohyal, a short ceratohyal, round in cross-section, and cylindrical interhyal are found. Relationships: The extra-ordinal relationships have already been discussed. Although the families Acipenseridae and Polyodontidae are quite ancient, both reaching back to the Upper Cretaceous, the branchiostegal apparatus shows considerable similarity. The deeply grooved operculum of Acipenser is reminiscent of the incised operculum of Polyodon which has degenerated further. The single chevron-Bhaped branchiostegal of Acipenser is not unlike the lone forked branchiostegal of Polyodon. The hyoid arches of both contain the same elements and are more or less unossified. Acipenser does differ in the possession of small suboperculum. The ceratohyal of Polydon is longer, a point doubtless related to its longer mandible. Despite the similarities there are many profound differences—scales, rostrum eto. which are found in the oldest fossil forms known and subordinal status may be warranted. Material examined: Acipenser fulvescens, NMC alizarin specimen, Ontario. 106 Polyodontidae PI. IV Branchiostegals; In Polyodon a single stout branchiostegal which divides distally, after a slight upward bend, into four prongs. MacAlpin (1947) reports three or four branchiostegals in the fossil Palaeosephurus. From his photographic plate of the specimen and his drawing, i t would appear that he may have mistaken the distal divisions of the branchiostegal as separate branchiostegals. It i s therefore suggested that Palaeosephurus possessed only a single branchiostegal. Gulars; Absent. Operculars: A much incised large cartilaginous operculum is known in Polyodon and Palaeosephurus. A suboperculum is absent but by comparison with Acipenser the g i l l cover bone can safely be identified as an operculum. Hyoid arch: A short hypohyal, a long ceratohyal and a large quadrate interhyal comprise the hyoid arch. The centre portion of the ceratohyoid is ossified. Relationships: Discussion of relationships is found under Acipenseridae and the order. References: MacAlpin (1947), Berg (1947), Gregory (1933), Lehman (1952), Holmgren and Stensio (1932), Holly (1936). Material examined: Polyodon spathula, UMMZ alcoholic specimen, U. S. A. 106 a GROUP II. HOLOSTEI With interoperculum (or secondarily lost). Lateral gulars absent. One or two hypohyals (or secondarily lost). Lower Triassic to present. ^ ORDER OSPIIFORMES Branchiostegals 9-12, spathiform; one or two median gulars present, primitively with V-3haped pit line; operculum and suboperculum of about equal size and interoperculum present. Lower to Upper Triassic. A single family. ^ Ospiidae (Parasemionotidae, Tungusichthyidae) P l . IV Branchiostegals: Number 9-12. Watsonulus 11 with lower short, upper 3 elongate; Parasemionotus 12 short and broad; Stensioenotus ca. 11; Ospia ca. 11; Jacobulus 11; Tungusichthys 9; Promecosomina 10. Branchiostegals with straight or curved edges and spathiform. About 8 of the 10 branchiostegals insert on the ceratohyal, the other 2 presumably on the epihyal. Gulars: Watsonulus has an 8-shaped median gular. In Parasemionotus the gular appears to have separated into two portions, a small triangular anterior one and an elongate posterior one slightly notched for reception of the anterior element. The posterior element bears a V-shaped pit line anteriorly; there are irregular traces of this pit line on the gular of Watsonulus. The gular of Jacobulus also bears a V-shaped pit l ine, as does the large gular (4/5 of the mandible length) of Promescosomina. 107 Operculars: Operculum and suboperculum about equal-sized. The inter-operculum appears for the f irst time in the Ospiidae. The interoperculum is well illustrated in Lehman's fine (1952) photographs of specimens (pis. 39, 40)• Here the shape of the interoperculum is identical to that of the adjacent branchiostegal, both having a downcurving expanded anterior end. The interoperculum differs only in that i t is shorter, in that i ts posterior end is in contact with the front half of the suboperculum, and in that i t separates the suboperculum from the jaws. The surface pattern of the interoperculum lacks the concentric pattern of the suboperculum. The form, position and decoration a l l militate in favor of development of the interoperculum from the uppermost branchiostegal, rather then from fragmentation of an arm of the suboperculum. In Parasemionotus the interoperculum has assumed a more advanced form, being triangular and being more broadly in contact with the suboperculum, thus resembling the interoperculum of the Amiiformes. But its position and form s t i l l suggest a branchiostegal origin (pi. 42, Lehman, 1952). Hyoid arch: In Broughia the elongate ceratohyal expands posteriorly without the usual constriction in the middle; i t is connected via the "stylohyal" ( a interhyal) directly to the hyomandibular. In Parasemionotous the ceratohyal is much shortened. In Promecosomina i t is heavy, non-constricted and trapezoidal; i t is in contact with a triangular epihyal. Taxonomy: Lehman (1952) unites the Ospiidae and Parasemionotidae, an action which the author agrees with. Romer (1955) is followed in including the Tungusichthyidae in the Parasemionotidae. Ospiidae has priority. 108 Promecosomina was raised to family level out of the Semionotidae by Wade. Westoll (1944) doubtfully placed i t in his order Amioidae with the Eugnathidae, Amiidae and Macrosemiidae. Gardiner (i960) believes i t deserves a separate family. Lehman (1952) however, believes that i ts characters conform to those of the Parasemionotidae (=0spiidae). The author concurs with the latter view; the large gular, the 10-12 branchiostegals, and V-shaped gular pit line of Promecosomina agree closely with the Ospiidae. Although a very large gular plate is also found in the Eugnathidae, they apparently lack the V-shaped pit line on the gular. But i t is also possible that i t could be associated with the Catervariolidae of the Amiiformes which have such a V-shaped pit line or with the Semionotidae which also have a V-shaped pit l ine. It is difficult to t e l l with which family i t should be associated as the specimen of Promecosomina is incomplete, but the number of branchiostegals would suggest placement with Ospiidae (9-12) rather than Semionotidae (4-6) or Catervariolidae (6). Relationships; The V-shaped pit lino and fairly numerous branchiostegals (7-12) of the Ospiiformes recall those of the Perleidiformes. The relations of the fin rays to radials, reduction or absence of the clavicle, scales of lepisosteid type but with vestiges of cosmine layer, a single row of ossified dorsal radials, caudal abbreviate-heterocercal, hyomandibular with opercular process and foramen for truncus hyoideo-mandibularis n. facialis, sensory canals basically paleoniscid, a l l common to both groups also suggest that the-Ospiiformes are derived from the Perleidiformes. Stensio has proposed that this group evolved from 1 0 9 Perleidae or perleid-like Chondrostei. The Parasemionotiformes probably gave rise to the Amiiformes. References: Lehman (1952), Stensio (1932), Berg (1955), Wade (1935). ORDER AMIIFORMES (SEMIONOTIFORMES) Branchiostegals 5 or 6 - 3 0 ; median gular present, with or without V-shaped pit line; lateral gulars absent; operculum, suboperculum and interoperculum present. Lower Triassic to present. Ten families, one l iving. The number of branchiostegals, shape of opercular bones and large gular are characters indicating that the Amiiformes were derived from the Ospiiformes or had a common ancestor with them. In the lower Amiiformes (Furidae, Semionotidae, Catervariolidae, Signeuxellidae) the supraorbital canal ends on the parietal as in the Parasemionotiformes. Both Semionotidae and Catervariolidae possess a V-shaped pit line on the gular, as the Ospiiformes. In both groups there is a supramaxillary and the lower jaw is complex; the fins are holostean; postrostrals are absent; the nasals meet on the midline. But in the Ospiiformes a clavicle i s lacking and the preoperculum is platelike. It would seem most l ikely that the Ospiiformes and Amiiformes descended from a common ancestor. Brough (1939) and Gardiner (I960) are also of this opinion. / Furidae (Caturidae, Eugnathidae) PI. V Branchiostegals: 6 - 2 4 . Furp (=Eugnathus) with 6-12; Euognathus 8 or 10; Hacrepistius about 10; Isopholis 12; Megalurus about 14; Lophiostomus 13; Brachichthys (=Heterolepidotus) 1 6 ; Caturus 1 6 - 3 0 . In Lophiostomus 110 about 13 with 3 epihyal and 10 ceratohyal; in Caturus 25 with about 4 epihyal and 21 ceratohyal. The upper branchiostegal (branchioperculum) o f Brachichthys and Caturus is expanded. Branchiostegals spathiform, elongate and with curved tips. Gulars: A large median gular present, apparently lacking a v-shaped pit l ine. In Furq i t almost covers the space between the mandibular rami, in Caturus i t is about 1/3 to 5/8 the length of the mandible, in Isophili3 i t is about J the length of the mandible, in Euognathides i t is diamond-shaped, 1/3 of the length of the mandible and without a pit l ine. Said to be small in Brachichthys. Large, about 4/5 of the mandible, in Lophiostomus. Lateral gulars undescribed and presumably absent. Operculars: Operculum tending to be larger than suboperculum, the border between them straight or curved, not wedge-shaped, (except in Heterolepidotus). Interoperculum present and triangular. The figure of Heterolepidotus (Stensio, 1947) appears to show two interopercula. Hyoid arch: A small hypohyal, large, deep ceratohyal and a triangular epihyal known in Furo. In Lophiostomus the ceratohyal is angled and at the anterior end is expanded. Relationships: The range of branchiostegals in this family is quite large. Brachichthys and Caturus with 16-30 have more than the other genera known which have 14 or less. These two genera provide one of the few examples in which the number of branchiostegals have apparently increased. Rayner believed that the Furidae were the most primitive of I l l the holosteans. References: Arambourg and Bertin (1958), Berg (1955), Brough (1939), Eastman (1914), Gregory (1923), Lehman (1949), Rayner (1941), Romer (1955), Schaeffer (I960), White and Moy-Thomas (1940), Woodward (1895), Woodward (1902-12), Zittel (1887). / Acentrophoridae Branchiostegals: 9-10 curved, elongate, spathiform branchiostegals in Acen^ronhorus,. The branchiostegals a l l insert on the ceratohyal. Gulars: Gulars have not been reported in this group though the bone "x" in f ig . 13 of G i l l may represent one. Operculars: Operculum and suboperculum of about equal size; operculum not sending wedge-shaped projection into suboperculum. Interoperculum small, triangular. Hyoid arch: A hypohyal and a large, elongate, hourglass-shaped ceratohyal are known. The triangular posterior tip of the ceratohyal in Gi l l ' s f ig . 7 may represent an epihyal. Relationships: G i l l states that correspondence of fins and supports, preoperculum, opercular apparatus and absence of an infra-clavicle place this genus in the Semionotidae (this was before Berg's erection of the Acentrophoridae). The branchiostegal series do not disagree with such an alignment but indicate that the Acentrophoridae are more primitive than the Semionotidae. References: Berg (1947), G i l l (1923) 112 ^ Paracentrophoridae Branchiostegals; 10 spathiform branchiostegals in Paracentrophorus. Gulars: Not reported. Operculars: Operculum and suboperculum of equal size, the former sending a wedge-shaped projection to the latter. Interoperculum small and triangular. Relationships: According to Piveteau (1941 in Gardiner, I 9 6 0 ) belongs in Semionotidae, but according to Gardiner deserves its own family. References: Gardiner ( i 9 6 0 ) . / Semionotidae Branchiostegals: In Lepidotea (Lepidotus) about 6, the uppermost very broad; in Dapedius 6 broad; in Semionotus at least 4 to 1 4 . A figure of Lepidotes montelli shows at least 6 slender branchiostegals attached to the epihyal. Gulars: Missing in Lepidotes. Dapedius with a large median gular plate half the length of the mandible and bearing a V-shaped pit l ine. Operculars: The operculum larger than the suboperculum and extending into the suboperculum in a curve or a wedge. Interoperculum a primitive triangle, (Lepidotus) or a rectangle (Dapedius). Hyoid arch: A triangular hypohyal, an hourglass-shaped ceratohyal and an epihyal in Lepidotes. In Dapedius epihyal much contracted at proximal end and ceratohyal remarkably short and deep (in association with its 113 short jaws). Taxonomy: Berg (1947) states the Semionotidae are doubtless a heterogeneous assemblage and separates them from the Acentrophoridae. Romer places Acentrophoridae in Semionotidae. Relationships: The close similarity of fins, form and skull bones leave l i t t l e doubt of the very close affinity of Acentrophorus to Lepidotes. But Berg gives presumably adequate characters to separate them as families: frontals fused or paired; caudal heterocercal or abbreviate heterocercal; pectorals high or low. To these differences may be added the connection of the supraorbital canal with the temporal canal and the form of operculum (in Acentrophorus operculum equal to suboperculum and not sending a wedge into i t ; in Lepidotus operculum larger than suboperculum and sending a wedge into i t ) . Geologically Acentrophorus is older (Upper Permian) than Semionotidae (Lower Triassic to Lower Cretaceous). Gardiner (i960) derives Semionotus and Acentrophorus separately from the palaeoniscid fishes, not from the amiiform line. This derivation seems unlikely as i t would call for a second parallel development of the interoperculum. References: Woodward (1895), Romer (1955), Gardiner (I960), Eastman (1914). Ilk ^ Catervariolidae Branchiostegals: 6 small, elongate, spathiform branchiostegals in Catervariolus hornemani. The uppermost is covered basally by the interoperculum. Gulars: A large median gular occupies at least half the length of the mandible. A V-shaped pit l ine, apex posteriormost, is found on the anterior end. A pair of wide lateral gulars, with width occupying about 1/3-1' of the mandible, may be present. Saint-Seine states they are probably present and dots in their outline in his figures. If so„, they are the only family of Amiiformes or Group II known to possess lateral gulars. Operculars: Interoperculum denticulate. Hyoid arch: A ceratohyal and a large short wide ephihyal are known. Taxonomy: This family was recently created by Saint-Seine. Relationships: Saint-Seine placed this family in association with the Amiidae. The V-shaped pit l ine, number of branchiostegals, supraorbital pit line ending on the parietal a l l indicate relationship with the Semionotidae, rather than the Amiidae. References: Saint-Seine (1955), Arambourg and Bertin (1958). / ' Signeuxellidae Branchiostegals: A single elongate curved spathiform branchiostegal is seen in an incompletely preserved specimen; probably there were several. 115 Gulars: Unknown. Operculars: A large operculum sends a wedge into the smaller sub-operculum. Below is a fairly large triangular primitive interoperculum. Hyoid arch: Unknown. Taxonomy: Saint-Seine erected this family in 1 9 5 5 . Relationships: Other than placing the Signeuxellidae in the holostean stage Saint-Seine was uncertain of the affinities. Dorsal views of the opercular bones in Signeuxella look very similar to those in Catervariolus. The supraorbital line ceases on the parietal in both genera. These and other characters suggest they may be related. References: Saint-Seine ( 1 9 5 5 ) , Arambourg and Bertin ( 1 9 5 8 ) . J Macrosemiidae Branchiostegals: 5 - 6 , perhaps to 9 in Macrosemius rostratus; these shaped like a scimitar and inserting on the ceratohyal. Four spathiform branchiostegals with rounded tips known in M. helenae, the upper inserting on the epihyal, the others oh the ceratohyal. About 1 0 in Ophiopsis, at least 7 in Propterus. Gulars: A gular plate has been observed only in Ophiopsis and Eusemlus. Operculars: Operculum larger than the suboperculum and sending a wedge down into i t . Interoperculum fairly large, and triangular. Hyoid arch: Epihyal, ceratohyal and hypohyal present. 116 Relationships: The Macrosemiidae were doubtless derived from a semionotid such as Lepidotus, as was discussed under that family, although Rayner and Bertin and Arambourg suggest derivation from Eugnathidae (=Furidae). Although the infraorbital and supraorbital canals join in Sinamiidae, Lombardinidae and Amiidae, as well as in Macrosemiidae, Macrosemiidae does not seem closely related to any of these families. Macrosemiidae retains the branch of the supraorbital canal which terminates on the parietal, unlike the other families, a character which would seem to be primitive. The scales are rhomboid and thus more primitive than those in Amia. But on the other hand i t has fewer branchiostegal rays than Amia, References: Woodward (1895), Arambourg and Bertin (1958), Zittel (1887), Rayner (1941), Eastman (1914). ^ Sinamiidae Branchiostegals: Branchiostegals questionably suggested by dotted lines to number about 14 in restoration figure of Berg (after Stensio). Gulars: Unknown. Operculars: Operculum larger than suboperculum and sending a slight wedge into i t . Interoperculum a long triangle. The opercular bones much resemble those in Amia. Hyoid arch: Unknown. Taxonomy: Although Romer (1955) and Arambourg and Bertin (1958) place Sinamia in Amiidae, Sinamia is here retained in its ow\ family for the 117 reasons given by Berg. Relationships: The close similarities of Sinamia and Amia leave l i t t l e doubt that they are related. However, the fused condition of the parietals in Sinamia indicate that Sinamia was not directly ancestral to Amia. Their similarities must thus arise from having a close common ancestor. References: Berg (1947). ^ Lombardinidae Branchiostegals: Unknown. Gulars: Unknown. Operculars: Large vertical rectangular operculum, small, elongate, horizontal suboperculum, small primitive interoperculum. Ryold arch: Unknown. Taxonomy: This family was recently erected by Saint-Seine (1955). Relationships: Although the form of the opercular bones does not differ greatly from Catervariolidae and Signeuxellidae, the figure seems to indicate juncture of the infraorbital and supraorbital sensory canals, as in the more advanced Amiiformes. The t a i l is almost heterocercal, the scales rhomboid. Lack of data on the branchiostegals enables l i t t l e further statement to be made. References: Saint-Seine (1955), Arambourg and Bertin (1958). 118 Amiidae PI. V Branchiostegals; In Amia 10-13; in /Megalurus 11-12. Form elongate and spathiform with curved tips. The branchiostegals a l l insert on" the ceratohyal. The uppermost branchiostegal is dilated and has been termed the branchiosperculum by Hubbs. Gular; A large median gular is present in Amia. Its length is about 2/3 that of the mandibles. It is bereft of a pit l ine. Operculars: Operculum larger than the suboperculum and sending a wedge down into i t . Interoperculum small, elongate, and triangular. Hyoid arch; A triangular hypohyal, angulate large ceratohyal, triangular epihyal with a prong inserting on outer ceratohyal and small cylindrical unossified interhyal. Relationships: As stated under Sinamiidae, the Amiidae probably share a close common ancestor with that family. The ancestral form was probably near to the Furidae, judging by the branchiostegals and median gular plate. References: Berg (1947)* Jordan and Everman (1896), Hubbs (1920), Agassiz (1833-43). Material examined: Amia calva, 11 alizarin specimens, MMC58-192-S, Pt. Pelee, Ontario; 12 alizarin specimens, NMC58-209, Georgian Bay, Ontario; alcoholic specimen, BC59-426, I l l inois . 119 ORDER LEPISOSTEIPORMES (LEPIDOSTEIFORMES) Branchiostegals 3, spathiform; gular3 absent; operculum and sub-operculum present; interoperculum absent* Upper Cretaceous to present. A single family. Lepisosteidae Pl. V Branchiostegals; 3 small spathiform branchiostegals. The upper two are attached to the epihyal, the other to the ceratohyal. The uppermost is broad, the lower two attenuate. Gulars: Absent. Taxonomy: The generic name was originally spelled Lepisosteus by Lacepede. This spelling has been incorrectly emended to Lepidosteus by some authors. Arambourg and Bertin (1958) include the Lepisosteidae in the same order as Amia. The many skeletal oddities of Lepisosteidae would, in the authors opinion, justify i ts ordinal separation. Relationships: The modifications of the skull of Lepisosteiformes make its placement difficult. Most authors include i t in Holostei. The non-platelike form of the preoperculum, rudimentary clavicle, absence of cosmine layer in the scales, maxillary not immovably connected with the preoperculum and dorsal and anal pterygiophores each supporting a single ray, and t a i l not heterocercal, are characters indicating holostean placement. However, the fin rays are completely segmented and some of the above holostean characters jnight be modifications resulting from the peculiar jaws. S o the question of placement is not entirely settled. 120 Rayner (1943), from neurocranial study, believes the Lepisosteidae are derivable from the Semionotidae. Gardiner (i960) also considers such an origin l ikely. References: Berg (1947), Arambourg and Bertin (1958), Boulenger (1904), Regan (1923), de Beer (1937). Material examined: Lepisosteus osseus, alizarin specimen, NMC6O-478-A, Ontario; skeletal specimen, USNM 110191, U. S. A. Operculars: An operculum and suboperculum of approximately equal size are present, the f irst sends a wedge into the latter. The absence of an interoperculum in this presumably holostean fish has caused a number of authors to identify other elements as the inter-operculum. The preoperculum has been identified as an interoperculum by Regan (1923). The passage of the preoperculo-mandibular canal through this bone clearly identifies i t as the preoperculum however, Holmgren and Stensio (1936) cal l the small bone above the anterior end of the preoperculum, the interoperculum. However, that this bone develops in this position, l ies above the preoperculo-mandibular canal and is far from the epihyal and suboperculum which true interopercula contact. It is therefore probably the quadrato-jugal as suggested by Hammarberg (in Arambourg and Bertin, 1958). The upper branchiostegal has also been identified as the interoperculum (de Beer, 1937). But this element does not l i e in the same horizontal fold as the suboperculum, does not l i e over the outer face of the epihyal, and does not insert on the suboperculum or jaw, a l l relationships which make i t difficult to identify as an 121 interoperculum. The interoperculum must therefore be considered to have been lost (or was never present in which case Lepidosteiformes should be placed in the Chondrostei). The very anterior jaws and the downward expansion of the preoperculum into the normal position of the interoperculum are both changes which could have lead to the loss of the interoperculum. Hyoid arch: Consists of a hypohyal, a short ceratohyal round in cross section, and a short epihyal with a small backward projection on which sits the upper branchiostegal. A small cartilage on top of the epihyal may represent the interhyal. / ORDER ASP IDORHYNCHIFORMES Branchiostegals 12-13 spathiform; gulars absent; large operculum and small suboperculum present; advanced interoperculum present. Middle Jurassic to Upper Cretaceous. A single family. / ORDER ASPIDORHYNCHIFORMES Branchiostegals 12-13 spathiform; gulars absent; large operculum and small suboperculum present; advanced interoperculum present. Middle Jurassic to Upper Cretaceous. A single family. / Aspidorhynchidae P l . V Branchiostegals: 12-13 elongate spathiform branchiostegals in Aspidorhychus acutirostris; several large branchiostegals in Belonostomus. Gulars: Apparently absent 122 Operculars: In Aspldorhynchus operculum large, suboperculum small; together forming a hemicircle. Interoperculum completely under the preoperculum; without dorsal projection posteriorly; subrectangular. Hyoid arch: Ceratohyal very large, epihyal small. Relationships: The branchiostegals and opercular bones are most similar to those of the Amiiformes, particularly those of the Amiidae. There are two large postorbital bones, fulcra are lacking, and the lower jaw complex is as in Amia. Berg states the sensory canal system on the head is as in adult .Amia. Differences in the scales, dorsal fins, rostrums indicate that Amia cannot be directly ancestral to Aspidorhynchus which therefore must have descended from a common ancestor. In this regard Berg points out the parietals are fused in Aspidorhynchus, as in Sinamia and i t may be noted that the opercular bones of Aspidorhynchus are more similar to those in Sinamia. Further, the scales of Sinamia are rhombic and supraorbitals are present as in Aspiorhynchus but unlike Amia. It therefore seems that Aspidorhynchus is more closely related to the sinamiid line than to the amiid l ine. The advanced form of the inter-operculum is probably associated with the L-shape of the preoperculum. Gardiner (I960), however, suggests that the Aspidorhynchiformes maybe derived from the Pholidophoriformes; the branchiostegal series do not argue against this. References: Assman (1906; in Berg, 1947); Woodward (1895); Agassiz (1833-43). 123 / ORDER PACHYCORI^ IFORMES Branchiostegals (6+-) 30-50* spathiform; median gular present; lateral gulars appear absent; equal sized operculum and suboperculum; triangular fairly advanced interoperculum present. Upper Triassic to Upper Cretaceous. A single family. Taxonomy: Following Romer (1955) and Arambourg and Bertin (1958) Proto-sphyraenidae is synonymized with Pachycormidae. Woodward (1895) said that Protosphyraenus differed l i t t l e from Hypsocormus (Pachycormidae). However, contrary to Arambourg and Bertin, the Pachycormidae is not placed in the same order as the Amiidae. In addition to the characters by which Berg distinguished the Pachycormiformes, there are the greatly enlarged lower postorbital and the numerous branchiostegals. / Pachycormidae Branchiostegals: In Pachycormus 4°-50 with about 22 on the epihyal and 20 on the ceratohyal. Figures of Hvpsocormus show a complement, presumably incomplete, of about 6. Zittel (1887) gives the number for the family as 30-45. Branchiostegals elongate, rectilinear and spathiform. Gulars: A large median gular about ^ the length of the mandible is found in Pachycormus, a smaller oval one in Hypsocormus. A gular is present in Protosphyraena. Lateral gulars have not been reported. Operculars: Operculum very large and triangular, suspended vertically by the apex, larger in Hypsocormus than in Pachycormu3. Suboperculum equally large. Interoperculum small triangular and fairly advanced in Pachycormus. An interoperculum is not mentioned in Protosphyraena or 124 Hypsocormus nor is i t present in figures of the latter. Presumably specimen(s) of Hypsocormus are incomplete, as the interoperculum and some branchiostegals appear to be lacking. Hyoid arch: Epihyal, a broad more or less straight ceratohyal and a hypohyal are known in Pachycormus. A longitudinal groove is found on the meeting ends of the ceratohyal and epihyal. Relationships: It seems likely that the Pachycormiformes arose from the Eugnathidae which also had numerous branchiostegals (up to 24). The large postorbitals of the Eugnathidae also suggest this. Since more branchiostegals are found in Pachycormiformes this provides a second example of the rare phenomena of the branchiostegal number increasing along a phylogenetic sequence; although i t might be taken as one example, the branchiostegals increasing in the eugnathid-pachycormid l ine. In any case where parts are numerous and unspecialized the ability to increase is often retained. References: Woodward (1895, 1898), Zittel (1887), Lehman (1949). / ORDER PHOLIDOPHORIFORMES Branchiostegals 1 (?) 6-18*; narrow gular plate; operculum larger than suboperculum; interoperculum small; opercular bone borders entire; Middle Triassic to Upper Cretaceous. SLx families. 125 / Pholidophoridae PI. V Branchiostegals: 7-10 in Pholidophorus, curved and spathiform. The upper branchiostegal is axpanded distally in some species. Woodward (1895) reports at least 17 pairs of branchiostegals in Pholidophorus (?) dubius. Gulars: A narrow median gular with a median ridge in Pholidophorus bechi. Gulars unknown in other forms. Operculars: Large triangular operculum with apex ventral; triangular suboperculum with apex dorsal, slightly smaller than operculum. Inter-operculum primitive and triangular. Relationships: The Ospiiformes show some resemblances but differ in the form of the preoperculum and interoperculum. Gardiner (i960) believes the Ospiiformes gave rise to the pholidophorids. A closer resemblance is seen in the Amiiformes such as Semionotidae and Eugnathidae. Although the Macrosemiidae also resemble tho Pholidophoridae they are precluded from ancestry by the fusion of the supra- and infra-orbital canals which end on the parietal without fuoing with the infraorbital canal in the family Pholidophoridae, Pleuropholidae and Majokiidae. The close resemblance of the opercular apparatus, cephalic sensory canal3, postorbitals, irregular median line of the paired roofing bones of the skull suggest a eugnathid origin as most l ikely. References: Woodward (1895). 126 / Pleuropholidae Branchiostegals: At least 2 branchiostegals in Parapleuropholis. These are spathiform, straight, elongate and broaden distally. Gulars: Unknown and presumably absent in Pleuropholis. Operculars: Operculum very large and vertically rectangular; suboperculum small. Interoperculum long and advanced, sending a dorsal arm up behind •the preoperculum and about on level with the suboperculum. Relationships: The branchiostegals and operculars are slightly different in form from those in Pholidophoridae. According to Arambourg and Bertin Pleuropholidae are very close to Pholidophoridae. The number of branch-iostegals is considerably lower in Pleuropholidae, but i t is not certain that the series is incomplete. References: Arambourg and Bertin (1958). ^ Liguellidae Branchiostegals: In Liguella there is at least one elongate, curved, spathiform branchiostegal which expands distally. Gulars: Unknown and presumably absent. Operculars: Large rectangular operculum; small suboperculum. Inter-operculum not preserved, but presumably present. Taxonomy: Family erected by Saint-Seine. Relationships: From the fragmentary remains preserved l i t t l e can be 127 said. The branchiostegals are curved, rather than straight as in Pleuropholidae, but in both they expand distally. References: Saint-Seine (1955), Arambourg and Bertin (1958). f Majokiidae Branchiostegals: Mot preserved. Gulars: Not known. Operculars: Operculum large, quadrate; suboperculsr small, triangular and denticulate. Interoperculum ending posteriorly in a spine. Taxonomy: Family erected by Saint-Seine. Relationships: Fragmentary remains enable l i t t l e to be said about this group. References: Saint-Seine (1955), Arambourg and Bertin (1958). Oligopleuridae Branchiostegals: In Oligopleurus at least 6 elongate, curved, spathiform branchiostegals distally expanded, in Oeonoscopus about 13. Gulars: Not observed. Operculars: Opercular large, quadrate; suboperculum small. Lower limb of preoperculum large and obscuring interoperculum, i f present. Hyoid arch: Ceratohyal and epihyal present. 128 Relationships: The branchiostegals would enable this group to be derived from the Pholidophoridae. It may be noted that the Archaeomaenidae and Oligopleuridae have cycloid scales while the Pleuropholidae, Pholidophoridae, Liguellidae, and Majolciidae have rhombic scales. Berg places thi3 family in Pholidophoriformes. Miss Rayner believes the family slightly off the main Pholidophorus-Leptolepis line of evolution. Bertin and Arambourg place the family in the Leptolepiformes because of the fusion of the infra and supraorbital canals, probable loss of rostrals, and reduction of parietals and premaxilleries. But these two sensory canals do not join in Leptolepidae. References; Woodward (1895), Rayner (1948), Bertin and Arambourg (1958), Berg (1947), Woodward (1890). GROUP III. TELEOSTEI With interoperculum (or secondarily lost). Lateral gulars absent. Two hypohyals (or secondarily one or both lost). MALACOPTERYGII Branchiostegals 0-36, one or more often spathiform. Median gular only in primitive clupeiforms. Epihyal and ceratohyal separate (except Siluroidei and Beloniformes). Never spines on opercular bones. Seldom with acanthopterygian pattern of 4 external and 0-4 (5-7) ventral or internal. The division of teleost fishes into malacopterygian and acanthopterygian on the basis of branchiostegals ha.s recently received support by a study on the retractores arcuiim branchialium by Holstvoogd (1963). Holstvoogd 129 (1963 and in l i t t . ) reports the malacopterygian groups studied, Clupeiformes, Mormyriformes, Cypriniformes, Notacanthiformes, Anguilliformes (and Polypteriformes) lack the muscle retractores arcus branchialium. On the other hand, the acanthopteryian groups studied possess this muscle: Gadiformes, Gasterosteiformes, Syngnathiformes, Ophidiiformes, Percopsiformes, Cyprinodontiformes, Mugiliformes, Perciformes, Pleuronectiformes, Gobiesociformes, Tetraodontiformes, Mastastembeliformes and Batrachiiformes. The Myctophiformes also possess this muscle; as is discussed later this order may be close to the lineage that gave rise to the acanthopterygians. The Beloniformes also possess this muscle; this is puzzling since they are usually considered malacopterygians. Except for the latter, conclusions from the muscle study and the study of branchiostegals are in complete agreement. ORDER CLUPEIFORMES Branchiostegals 2-36, with 0-1 interhyal, jg-12 epihyal, 0-23 ceratohyal and 0-5 hypohyal, with 2-12 external and 0-11 ventral (or rarely internal). Median gular present only in Albuloidei and more primitive suborders. Operculars complete, without spines and usually entire (rarely crenualte). Interhyal, epihyal, ceratohyal and two hypohyals present (except interhyal and one hypohyal absent in Phractolaemidae). Lower Cretaceous to present. Fifteen suborders and 54 families, 15 of which are known only from fossils. Gosline (i960) divided the Clupeiformes into two divisions, Clupei and Osteoglossi. The Osteoglossi, with the exception of Iliodontidae, are here removed to the Mormyriformes, as discussed under that order. 130 The Hiodontidae are placed in a new suborder within the Clupeiformes. Although the Clupeiformes do show some relationship to the Pachycormiforme3, there seems l i t t l e doubt t h a t their closest relation-ships are to the Pholidophoriformes. Besides the similar branchiostegals, several other characters are shared: two supramaxillaries, lower jaw without prearticular and coronoids, and without independent supraangular. Rayner (1948) considered the two groups very close. Gardiner (i960) was of the belief that the teleosts arose from the Pholidophoriformes. Some authors have considered that the Elopoidei may belong in the Holostei (Saint-Seine, 1956, Nybelin, 1957), on the basis of such characters as the ethmoid commisure connecting the infraorbital canals and the possession of a gular. However, these characters may be regarded as primitive holdovers. The development of a second hypohyal and of intermuscular bones clearly distinguishes the members of the Clupeiformes from their predecessors. The only alternate solution would be placement of the Lycopteroidei, Leptolepoidei and Elopoidei together in their own order. But this has l i t t l e merit as the Albuloidei s t i l l retain a vestige of a median gular. So rather than establish a poorly defined order on the basis of this receding character i t appears preferable to establish an order on the sharply defined grounds of the two new characters. / SUBORDER LYCOPTEROIDEI / Lycopteridae Branchiostegals: At least 6 or & to 10 spathiform. In Lycoptera  middendorffi 10 with 3 epihyal and 7 ceratohyal. 131 Gular: A median gular present in Lycoptera, less than one half length of mandible. Operculars: Operculum large and rectangular above small suboperculum and with interoperculum. References: Woodward (1895), Berg (1948a). / SUBORDER LEPTOLEPOIDEI / Leptolepidae Branchiostegals: In Leptolepis about 21 with 9 broad and imbricating on the epihyal and about 12 spaced and delicate ones on the ceratohyal. Gular: Leptolepis with a median gular one half the length of the mandible. Operculars: Complete and entire. Hyoid arch: Ceratohyal with ordinary hourglass-shape but noteworthy for the extension of a supplementary delicate rod of bone between its extremities on the upper side and for its large central perforation; interhyal, epihyal and one large hypohyal present. References: Woodward (1895), Berg (1947), (1948), Rayner (1937), Nybelin (1963). SUBORDER ELOPOIDEI Branchiostegals at least 16-36 with 5-13 epihyal and 10-23 ceratohyal, about 12 being slightly external and 20-23 being ventral, the upper ones spathiform the lower ones becoming 3 l e n d e r . A well developed but narrow 132 median gular 1/3 to ^ the length of the mandible. Two hypohyals (except Leptolepidae has but one). Operculars complete and entire. G i l l membrane separate. Five families, two l iving. Jordan, Evermann and Clark (1930) include the elopid families and albulids in separate suborders. Berg (1947) places both in the Suborder Clupeoidei. Gosline (i960) places both in the Suborder Elopoidei (separate from the Clupeoidei) but recognizing two superfamilies, Elopoidae and Albuloidae. It is the author's opinion that Jordan, Evermann and Clark were correct in awarding subordinal status to the two groups. The two groups differ in numerous profound characters. The Albuloidei (Albulidae and Pterothrissidae) differ trenchantly from the Elopoidei (Alopidae and Megalopidae) in the following characters: 1-3 branchiostegals instead of 5-13 on the epihyal; gular plate rudimentary or absent versus well developed; one versus two supramaxillaries; pelvic rays 10-14 instead of 15-17; maxillaries toothless instead of toothed; subterminal instead of terminal or superior mouth; g i l l rakers tubercle-l ike instead of long and slender; only tvro pairs of uroneurals instead of 4; terminal vertebra with no neural arch and with a greatly expanded median crest behind i t instead of with a neural arch and a small median crest; 7 hypurals instead of 8-9; one posterminal centrum instead of two (characters from Berg, 1947, Hollister, 1939, Gosline, 1961, Gregory, 1933). These differences show the albuloids are considerably advanced with respect to the elopoids. Ridewood (1904) came to a similar opinion after study of their skulls, "There is no doubt that the ELopidae are the most archaic of existing teleosteans and that the Albulidae are in few respects more highly specialized; but the study of the skull does not 133 show any direct affinity between the two families". Nybelin (i960) although he discovered the rudimeritarj'- gular plate in Albula felt that "The detection of this plate naturally does not mean that Albula vulpes would have a closer relationship to the Elopidae and Megalopidae than what has so far been assumed, for i t is clear from other facts Elops,  Megalops and Albula a l l represent evolutionary lineages of their own". The two groups are therefore treated as separate suborders. Elopidae {f Raphiosauridae) Branchiostegals: Vary from 16 to 36, in ^Laminospondylus transversus 16 or more; in ^Rhacolepis about 20; in ^Thrissopater about 30; in /Esocelops at least 15; in /spaniodon at least 15; in ^Osmeroides 20 with 5 broad ones on the epihyal and 15 on the ceratohyal; in /Pachyrhizodus 9-10 on the epihyal; in Elops 24-36 with 12 epihyal and 20-23 ceratohyal, the upper 12 being slightly on the lateral face, the remainder on the ventral face of the arch; the upper ones being broad and expanded, the lower narrow and elongate. Gular: In Elops elongate narrow median gular occupying about one third of the mandibles; no pit line apparent on i t . In Esocelops gular at least of mandible. In Osmeroides long and narrow, its length slightly more than <?- mandible. Operculars: Complete, operculum large, interoperculum triangular and modern. Hyoid arch: Consists of 2 hypohyals, ceratohyal, epihyal and interhyal in Elops. 134 References: Springer (1957), Woodward (1901), (1902-1912), Ridewood (1904), Fowler (1936), Weber and de Beaufort (1913), Regan (1909). Material examined; 5 Elops saurus, BC55-321, Louisiana, alcoholic; USNM26218, Key West, skeleton. Megalopidae Branchiostegals: Megalops 23-27 with 13 epihyal and 10-12 ceratohyal. The range of counts in the two species atlanticus and cyprinoides are identical. Gular; A narrow elongate diamond-shaped gular without trace of pit l ine. Its length slightly exceeds half the length of the mandible. Attached by a ligament to the back of the mandibular symphysis. Operculars; Complete and entire; a narrow suboperculum forming half the posterior border of the g i l l cover behind the operculum; suboperculum narrow. Hyoid arch; An interhyal, epihyal, ceratohyal and two hypohyals present. Relationships: The number and arrangement of the branchiostegals on the hyoid arch is similar to that in Elops. The gular is somewhat longer in Megalops than Elops but other fossi l elopids have an equally long i f not longer gular. The branchiostegal series therefore do not offer support to retention of the Elopidae and Megalopidae in separate families. Some authors do not separate them. References: Fowler (1936), Day (1&75), Weber and de Beaufort (1913), 135 Ridewood (1904). Material examined; 1 Megalops atlanticus, U3NM 179715, British Guiana, skeleton. / Ganolytidae Branchiostegals: No data available. SUBORDER ALBULOIDEI Branchiostegals (4) 6-16 with 2-3 epihyal and 10-12 ceratohyal, 11 external and 1 ventral. A rudimentary gular present or absent. Operculars complete and entire. G i l l membranes separate. Interhyal, epihyal, ceratohyal and two hypohyals present. Two families. Albulidae P l . VI Branchiostegals. In ^Chanoides probably about 4; in /Ancylostylos probably 6; Albula 10-16. Albula vulpes observed (10) 11-13 left and 10-11 right with 2 epihyal and 10 ceratohyal, 11 external and 1 ventral. While (10) 11-13 branchiostegals were found in 14 specimens from the Pacific coast of America (Peru to Mexico), Day (I875), Misra (1953) and Weber and de Beaufort (1913) report 14-16 from India and the Indo-Australian Archipelago. This might suggest two species are involved. In Albula vulpes the upper branchiostegals are broadly spathiform becoming narrower ventrally; they are a l l decurved and have a median external ridge basally. The lowest ones may not contact the ceratohyal but may be free in the membrane. 136 Gular; Nybelin (i960) discovered the presence of a very slender median gular about ^ the length of the mandible in Albula vulpes. It is rudimentary compared to that of the elopoids. In 3 and 6 inch specimens only a threadlike trace was observed. The gular is contained in a gular fold which, curving anteriorly, joins the two mandibles. Its posterior border is crenulate. Operculars; Opercular bones complete and entire, suboperculum V-shaped. Hyoid arch; Consists of 2 h y p o h y a l 3 , ceratohyal, epihyal and interhyal. References: Misra (1953); Weber and de Beaufort (1913); Ridewood (1904); Woodward (1901). Material examined: 10 specimens of Albula vulpes. BC56-160, from Peru one of which was cleared and stained for detailed examination; 3 alcoholic specimens, BC59-687, from Panama, Panama; one alcoholic, BC56-162, from Talara, Peru; one alcoholic, BC60-14, Acapulco, Mexico. Pterothrissidae Branchiostegals: In ^Istieus about 10; in Pterothrissa gissu 6. Gular: Pterothrissa lacks a median gular plate. Operculars: Complete and entire. Relationships: Differences in the possession of gular plate, dorsal f in, dentition of the maxillary, etc. would certainly warrant their familial separation, although some authors have considered Pterothrissidae a subfamily of Albulidae. 137 References: Gunther (1887), Woodward (1901), Tomiyama and Abe (1958), Material examined: None. SUBORDER CLUPEIODEI Branchiostegals 5-20, with 1-5 epihyal and 0 - U ceratohyal, 2-10 external and 0-11 ventral. Gular absent. Operculars complete and entire. G i l l membranes separate. Interhyal, epihyal, ceratohyal and two hypohyals in a l l families examined. Fourteen families, four of which are known only from fossils. The branchiostegal rays of Alepocephaloidea tend to be long and slender while at least the upper branchiostegals of Clupeoidea are broad and one or more bears a clupeoid projection. The branchiostegals of Rosauridae are suggestive of the Alepocephaloidea. Superfamily Alepocephaloidea Alepocephalidae Branchiostegals: Vary from 5-9, but 13 in Leptochilichthys, a l l slender and lath-or rod-shaped. In Alepocephalus 6, Anamalopterichthys 7, Asquamiceps 5, Aleposomus 6-7, Bathytroctes 7-8, Bellocia 6, Conocara 6, Ericara 6, Grimatotroctes 7, Leptoderma 7, Leptochilichthys 13, Macromastax 9, Micrognathus 7, Mitchillina 6, Narcetes 8, Nemabathytroctes 7, Photostylus 6, Rouleina 6, Talismania 6-8, Tauredophidium 8, and Xenodermichthys 6-7. Talismania bifurcata 7 with 4 epihyal and 3 ceratohyal, 4 external and 3 ventral, slender, elongate and distally slightly laminar. Operculars: In Alepocephalus operculum crenulate, suboperculum and interoperculum entire. In the family the g i l l membranes separate, 138 sometimes united and free from the isthmus. Hyoid arch; Consists of interhyal, epihyal, ceratohyal and two hypohyals in Talismania. Relationships; Leptochilichthys differs quite strongly from other alepocephalids in having 13 branchiostegals instead of 5-9. References; Weber and de Beaufort (1913), Grey (1958), Fowler (1936), Parr (1937, 1951, 1952), Maul (1948), Beebe (1933), Garroan (1899), Tucker (1954), Gunther (1887), Misra (1953). Material examined; Bathytroctes rostratus, alcoholic specimen, USNM 137754-9, Philippines; Talismania bifurcata, alcoholic specimen, BC62-159, California. Searsiidae Branchiostegals: Vary from 4 to 8. In Persparsia 7, Pellisolus 7, Holtbyrnia 7-8, Searsia 7-8, Normichthys 8, Mirorictus 4, Platytroctes 5-6, Sagamichthys 6, Maulisia 7-8, Mentodus 7-8, Barb an tus 7-8. Relationships; Parr (1951) separated this family from Alepocephalidae. It is very closely related to the Alepocephalidae, being distinguished only by the presence of the shoulder organ. The similarity of the branchiostegal counts would confirm the closeness of their relationship. References; Parr (1951, I960), Maul (1954, 1957), Tucker (1954). 139 Macristiidae Branchiostegals: Long, slender, curved vdth about 8-10 in Macristium. Operculars: Complete and entire* G i l l membranes free. Relationships: Regan first placed this fish in the family Scopelidae; later he modified his view making i t a distinct family close to the Alepocephalidae. Berg followed the latter view but stated its position was uncertain. Marshall (1961) on the capture of a second young specimen suggested that Macristium was the survivor of the fossil ctenothrissid fishes. This author is inclined to disagree with the latter opinion. While the number of branchiostegals in the Ctenothrissidae and Macristiidae are about the same, the form of the branchiostegals is not. While the upper two branchiostegals in Ctenothrissidae are broad and spathiform, in Macristiidae a l l of the branchiostegals on the contrary are narrow. Scales are absent in Macristiidae, present in Ctenothrissidae. As Marshall points out the Macristiidae lack supramaxillaries while there are two well developed ones in the Ctenothrissidae. The number of vertebrae differ by about 20. The similarities in fin pattern and mouth angle may be a result of similarity in habits; they are not strong subordinal characteristics. On the other hand the Macristiidae are similar to the Alepocephalidae in many of the listed characters. Structure of the upper jaw and teeth, the slender branchiostegals, and the absence of scales agree with the Alepocephalidae. In some Alepocephalidae the ventrals are almost thoracic and the dorsal is central in position, thus approaching the Macristiidae. 140 Material examined: None. Superfamily Clupeoidea Dussumieriidae P l . VI Branchiostegals: Vary from 6 to 20. Spratelloides. Jenkinsia and Gilchristella with 6. Gilchristella with 2 epihyal and 4 ceratohyal, 4 external and 2 ventral, branchiostegals spathiform with clupeoid projections. Dussumieria with 12-20 with 3 i on the epihyal and 8-8^ on the ceratohyal, a l l on the external surface. Etrumeus with 13-15 bearing 4 on the epihyal and 9-10 on the ceratohyal, a l l on the external surface. ^Histiothrissa with 15. In Etrumeus most of the upper branchiostegals are broad and spathiform and bear an anterior projection at the base. Operculars: Opercular bones complete and entire. G i l l membranes separate. Hyoid arch: Consists of interhyal, epihyal, ceratohyal and two hypohyals in Etrumeus. Dusgumiera and Gilchristella. In Dussumiera the epihyal possess a foramen and i t and the ceratohyal emit small prongs toward one another on their internal face. Relationships: The more numerous branchiostegals of Dussumieriidae would seem to indicate that i t arose off the primitive clupeid stock before the Clupeidae. Chapman (1948a) agrees with this and adds several other characters to confirm this view. 141 Spratelloides, Jenkinsia and Gilchristella differ from other dussumeriids by having only 6 branchiostegals instead of at least 14 or more. The number of branchiostegals would appear to associate these fish with the Clupeidae rather than the Dussumieriidae. Spratelloides further agrees with the Clupeidae and differs from the Dussumieridae in having two instead of only one supramaxillary. Histiothrissa has too many branchiostegals for the Clupeidae and too many supramaxillaries for the Dussumieriidae; Arambourg (1954) has erected a special subfamily for i t in the Clupeidae. References; Chapman (1948a), Ridewood (1904a), Misra (1953), Munro (1955), Schultz et al (1953), Weber and de Beaufort (1913), Whitehead (1962). Material examined; Alizarin specimens of Gilchristella aestuarius, from South Africa, NMC62-141; Dussumieria acuta, alizarin specimen, NMC63-71-S, Singapore; Etrumeus teres, alizarin specimen, NMC 63-70-S, Nagasaki, Japan. Engraulididae Branchiostegals: Vary from 7-19, with 1-3 epihyal, 8£ - l l ceratohyal and 4-10 external and 0-11 ventral. Anchoviella 11-13, Setipinna 11-19, Thrissocles 10-14, Scutengraulis 12-14, Anchoa IO-15, Coila 7-13, Stolephorus 11-13, Engraulis 9-14, Lycothrissa 7-12. In Anchoa. compressa 10 with 1 epihyal and 9 ceratohyal, 10 external and one ventral, the uppermost broad and spathiform, the lovrer becoming narrower, a l l with a clupeoid projection. Engraulis encrasicholus 10 with 1 epihyal and 9 ceratohyal. Coilia nasus 11 with 2g epihyal and 8^ ceratohyal. Anchoa 142 hepsetus 15 v/ith 3 epihyal, 11 ceratohyal and 1 hypohyal, 4 external and 11 ventral, the upper ones broad and spathiform, a l l with a clupeoid projection on the base. Operculars: Opercular bones complete and entire. G i l l membranes separate or joined and free from the isthmus. Hyoid arch: Consists of interhyal, epihyal, ceratohyal and two hypohyals in Anchoa. Engraulis, Coilia. Relationships: The form of the branchiostegals indicates the Engraulididae to be related to the Dussumieriidae. References: Chapman (1944), Weber and de Beaufort (1913), Chevey (1932), Ridewood (1904a), Day (1875), Smitt (1895), Misra (1953), Inger and Kong (1962). Material examined: Anchoa hepsetus. alizarin specimen, NMC62-73, North Carolina. ^ Clupavidae Branchiostegals: Fig. 1584 of Bertin and Arambourg (1958) shows Clupavus bears at least 4 spathiform branchiostegals. Operculars: Opercular bone3 complete and entire. Relationship: According to Bertin and Arambourg may be placed in the neighbourhood of the Dussumieriidae. References: Bertin and Arambourg (1958). 143 f Pseudoberycidao Branchiostegals: Information not available. Opercularst Operculum and suboperculum entire. Relationshipst Allied to the Clupeidae according to Berg. References; Woodward (1901), Berg (1947). ^ Syllaemidae Branchiostegals: In fSyllaemu3 there are 10 delicate branchiostegals rays on the ceratohyal. Operculars: Complete and entire. Relationship: According to Berg they are allied to the Clupeidae. References: Woodward (1902^12), Berg (1947). ^ Ichthyodectidae Branchiostegals: No information available. Relationships: According to Woodward (1901) the vertebral axis, fins and scales are as in Chirocentrites. Chirocentridae Branchiostegals: Vary from 6-8. Chirocentrus with 6-8, with 2-3 on the epihyal and 4-5 on the ceratohyal; the ones on the epihyal being on the external, those on the ceratohyal on the ventral face of the arch; the upper 2-3 spathiform, the lower scythe-shaped. C. hypsoseloma with 144 6 and C. dorab with 8. The uppermost with a clupeoid projection. /Platinx with 6-7, ^Chirocentrites with about 20. Operculars; Complete and entire. G i l l membranes separate. Hyoid, arch; Consists of interhyal, epihyal, ceratohyal and two hypohyals, ventral side of ceratohyal notched for reception of branchiostegals. Relationships: The Chirocentridae differ from the Alepocephalidae and Searsiidae by the breadth of the upper branchiostegals which are narrow in the other two families. The fossil Chirocentrites with 20 branchiostegals differs trenchantly from Chirocentrus with only 6-8. In fact i t might be suggested that i t does not belong in this family. Bertin and Arambourg (1958) have placed Chirocentrites in the family /ichthyodectidae. This placement cannot be evaluated from the point of view of branchiostegals since branchiostegals are unknown in the Ichthyodectidae. References: Woodward (1901), Ridewood (1904a), Day (1875), Chevey (1932). Material examined: Chirocentrus hypsoseloma, alcoholic specimen, BC58-32, from Malaya; Chirocentrus dorab, alcoholic specimen, USNM 47990, from Cochin China. Dorosomatidae Branchiostegals: Vary from 5-6. Dorosoma 5-6 with 1 epihyal and 4 ceratohyal, 3 on the external and 2 on the ventral face of the hyoid arch; the upper three branchiostegals broad and scimitar-like the lower two acinaciform. The upper 3 have clupeoid projections at the base. 145 Nematalosa 6 with 1^  epihyal and l+h ceratohyal. Operculars: Opercular bones complete and entire. G i l l membranes separate. Hyoid arch: Consists of interhyal, epihyal, ceratohyal and two hypohyals. Relationships: Relationship to the Clupeidae, Engraulididae and Dussumieridae is clearly shown by the possession of a clupeoid projection at the base of the upper branchiostegals. References: Day (1875), Ridewood (1904a), Vladykov (1945). Material examined: Dorosoma cepedianum, alizarin specimens, NMC60-521-A, from Lake St. Claire, Canada. Clupeidae P l . VII Branchiostegals: Vary from 6 to 10 with 1^  - 3 epihyal and 4-8 ceratohyal, 3 external and 4-7 ventral. In Alosa 7; Sardinpps 7 with I5 epihyal and 5z ceratohyal; Opisthonema with 6, 2 epihyal and 4 ceratohyal; Clupea harengus pallasi i with 8-9, 2 epihyal, 7-8 ceratohyal, 3 external and 6-7 ventral; Corica, Clupeoides. Amblygaster, Pellona, Opisthopterus. Racqnda 6; Harengula 5-6; /scombroclupea, ^Diplomystus with about 10; Alosa pseudoharengus 7 with 2 epihyal and 5 ceratohyal, 3 being external and 4 being ventral, the ventral ones being inserted in notches. Opisthopterus dovi 6 with 2 epihyal and 4 ceratohyal, 3 external and 3 ventral. In these genera the upper three branchiostegals are usually wide and spathiform, the lower ones being slender. An anterior projection 146 arises from at the base of at least the upper branchiostegal. /scombroclupea according to Woodward's figure does not bear these projections. Operculars: Opercular bones complete and entire. G i l l membranes separate. Hyoid arch: Consists of interhyal, epihyal, ceratohyal and two hypohyals in Clupea, and Hyperlophus. References: Ridewood (1904a), Schultz et a l (1953), Fowler (1936), Day (1875), Chapman (1944b), Misra (1953), Woodward (1901, 1902-12). Material examined: Alosa pseudoharengus, one alizarin specimen, NMC60-452-A, from Lake Ontario, Canada; Clupea harengus p a l l a s i i , 4 alizarin specimens, BC60-326, British Columbia, Canada; Opisthopterus  dovi, alcoholic specimen, BC57-83, Bahia de Petacalco, Mexico. Denticipitidae Branchiostegals: 5 i n Denticeps clupeoides with 1 epihyal and 4 ceratohyal, a l l external, the uppermost broad mesially, the central 3 slender, the lowest broad, with a clupeoid projection, and bearing denticles along i t s anterior edge (unlike any other clupeiform). ^Palaeodenticeps with 4 or 5, the upper 3 spathiform. Operculars: Complete and entire. In Denticeps operculum denticulate, suboperculum small and edentulate, interoperculum elongate and denticulate; g i l l membranes separate. In Palaeodenticeps suboperculum denticulate. 147 Hyoid arch: In Ponticeps consists of interhyal, epihyal, ceratohyal and two hypohyals. The lower hypohyal large, the upper small and located between the upper corners of the lower hypohyal and ceratohyal. Relationships: The keeled ventral scutes on the belly, the connection between the gas bladder and ear, and the clupeoid projection are a l l characters suggesting clupeoid relationship. The reduction of caudal rays to 16 from the normal clupeoid number of 17 may represent an adaptation to miniaturization. References: Clausen (1959), Greenwood ( I960). Material examined: Denticeps clupeoides, 1 alizarin and 2 alcoholic specimens, uncatalogued NMC specimens from the Upper Yewa on the boundary between Nigeria and Dahomey, received through the kindness of Dr. Clausen; alcoholic specimen, USNM 195992, from Nigeria. Bathylaconidae Branchiostegals: 8 -9 in Bathylaco, broad, curved, scythe-like with bases bearing anterior projections. Operculars: Complete and entire except for the posterior border of the operculum which is crenulate. G i l l membranes separate. Relationships: Goode and Bean (1896) f irst placed Bathylaco in the Synodontidae of the Iniomi, probably largely on the basis of the elongate jaws and oblique opercular apparatus. Parr (1948) included i t in the Isospondyli and regarded i t as a possible intermediate between the Isospondyli and the Iniomi. Bertin and Arambourg (1958) placed i t in a 148 new suborder, Bathylaconoidei, intermediate between the Stomiatoidei and the Esocoidei. The broadness of the branchiostegals differentiates the Bathlaconidae from the Myctophiformes and from any stomiatioid or esocoid. The comma-shaped preorbital photophore need not indicate relationship to the Stomiatioidei since a similar photophore has arisen independently in the Myctophiformes. The anterior projections on the bases of the branchiostegals in Parr's figure are suggestive of the clupeids and relatives. Amongst the clupeoids the engraulidids perhaps are the most similar with the narrow elongated tooth jaws, oblique suspensorium, and large anterior eyes. The Bathylaconidae are provisionally placed in the Clupeoidei. However, until specimens of Bathylaco are available for osteological study placement wil l be uncertain. References: Goode and Bean (1896), Parr (1948), Bertin and Arambourg (1958), Jordan and Evermann (1896). Material examined: None. ^ SUBORDER TSELFATOIDEI Branchiostegals 13-14, with about 4 epihyal and 9-10 ceratohyal, a l l apparently on the ventral edge of the hyoid arch. No gulars. Uppermost branchiostegals spathiform, lower ones slender. Operculum and suboperculum present and entire; interoperculum not known. Epihyal and ceratohyal known. A single fossil family. The family Tselfatidae was found by Arambourg (1954) who placed i t in the Beloniformes. Bertin and Arambourg (1958) erected a new suborder, 149 Tselfatoidei, for its reception in the Beloniformes. Several characters do not support this ordinal placement. The Tselfatidae has too many branched caudal rays (18 instead of 13), too many pelvic rays (7 instead of 6), the dorsal and anal are very long and high instead of small low and posterior, the parasphenoid is toothed, the epihyal and ceratohyal are not sutured together, and the mouth is bordered by both the premaxillary and maxillary. All of these characters preclude its placement in the Beloniformes. The entrance of the maxillary into the gape would permit placement in Clupeiformes, Notacanthiformes, Mormyriformes, Cypriniformes or Anguilliformes. The numerous caudal rays, normal anterior vertebrae, toothed parasphenoid, presence of a supraorbital, and numerous distally spathiform branchiostegals preclude placement in any of these orders except the Clupeiformes. All of the characters of the Tselfatidae are readily accommodated in the Clupeiformes except the long pedicels of the premaxillaries. The premaxillaries are reported to have long pedicels which would indicate that they are probably protractile, a character normally acanthopterygian. The majority of characters of the Tselfatidae would appear to place them among the primitive Clupeiformes such as Albuloidei, Clupeoidei and Ctenothrissidae. / Tselfatidae Branchiostegals; Tselfatia \idth 13-14 slender branchiostegals which become spathiform dorsally. About 4 epihyal and 9-10 ceratohyal, distributed along the ventral edge of the hyoid arch. Operculars: Operculum and suboperculum entire; interoperculum not known. 150 ; Hyoid arch; Only epihyal and ceratohyal known. Epihyal and ceratohyal not sutured together. References; Arambourg (1954). Bertin and Arambourg (1958). Material examined; None. ^ SUBORDER CTENOTHRISSOIDEI / Ctenothrissidae Branchiostegals: Vary from 8 to 10 with 2 or 3 epihyal and 7 ceratohyal in Ctenothrissa. The upper branchiostegals are spathiform the lower ones become narrow and acinaciform. The uppermost branchiostegal has anterior projection at the base. Operculars; Complete and entire. Relationships; This group was f irst placed in the berycoids because of its anterior pelvics. Regan and Berg have grouped i t close to the clupeoids. The jaw structure, number form and arrangement of the branchiostegals agrees with this grouping. Bertin and Arambourg (1958) have placed the Ctenothrissidae in the Bathyclupeiformes, apparently on superficial agreements in body form and fin arrangement. The author cannot agree with this arrangement. The Bathyclupeidae have fin spines, ventrals with 15 rays, only one supramaxillary, and a perciform number, form and arrangement of the branchiostegals. While on the other hand the Ctenothrissidae lack fin spines, have two well developed supra-maxillaries and have malacopterygian branchiostegals. While i t is possible that ctenothrissid-like ancestors could have given rise to the 151 bathyclupeids, the two families definitely do not belong in the same order or to the same level of evolution. The Ctenothrissidae may be imagined to have arisen off primitive clupeoid stock, developed a shortened deepened body and with this change more anterior pelvic fins and an enlarged dorsal f in . References: Woodward (1901, 1902-1912). ^ SUBORDER SAUR0D0NT0IDEI / Saurodontidae Branchiostegals: No information available. Operculars: Complete and entire. References: Newton (1878). SUBORDER HI0D0NT0IDEI, New Suborder Branchiostegals 7-10, with 2-2^ - epihyal and 6-7 ceratohyal, 4-5 on the external and 4-5 on the ventral face of the hyoid arch. No gulars. Uppermost branchiostegals spathiform. Opercular bones complete and entire. G i l l membranes separate. Two hypohyals. One living family. Regan (1929) placed the Hiodontidae in a superfamily with the Notopteridae, with which i t bears a superficial resemblance. Berg (1947) followed Regan but raised the superfamily to a suborder. Gosline (i960) placed the superfamilies Hlodontoidae, Notopteroidae and Osteoglossoidae a l l in his division Osteoglossi of the Clupeiformes. However, the Hiodontidae would seem to be well separated from these groups, as i t is distinguished from them by the following trenchant characteristics: possession of two hypohyals instead of one; upper branchiostegals spathiform instead of acinaciform or virgaform; 8 hypurals instead of 4 152 or fewer; nasals small instead of large. A l l of these characters combine to show that the Hiodontidae i s distinct from the notopterid-osteoglossid group, and i s much more primitive. Its origin appears to l i e v/ith the albuloids or clupeoid3. As in some clupeoids a duct from the gas bladder contacts the inner ear. It differs from clupeoids and albuloids however, in lacking oviducts, in having the parapophyses coosified with the centra, and 16 branched caudal rays. These characters justify i t s sub-ordinal separation.^" Hiodontidae (Hyodontidae) P l . VII Branchiostegals; Vary from 7-10 i n Hiodon (including Amphiodon). In Hiodon tergisus 8-9 with 2 epihyal and 6-7 ceratohyal, 4 on the external and 4*-5 on the ventral surface of the hyoid arch, the uppermost 2-3 branchiostegals spathiform and expand d i s t a l l y . In Hiodon alosoides 7-10, usually 9 with 2g epihyal and 6^ ceratohyal, 5 on the external and 4 on the ventral face of the hyoid arch. A pertinent paper has just been received in f i n a l typing, Greenwood (1963). He raises Gosline»s Osteoglossi to ordinal l e v e l . The author agrees with the separation of the Osteoglossi (except Hiodontidae) from the Clupeiformes. But this author feels the Osteoglossi are sufficiently close to the Mormyriformes to be included in them, thus a new order i s not necessitated. The Hiodontidae have numerous primitive characters which are lacking in the Mormyriformes (as here construed) but which may be found in the Clupeiformes that i t i s clear they should be placed with the l a t t e r : 2 hypohyals, spathiform branchiostegals, gular fold, 8 hypurals, 3-4 uroneurals, adipose eyelid, postterminal centra, etc. Similarities of the Hiodontidae to the Notopteridae may either represent parallelisms or be evidence of distant common ancestry. Operculars; fold. 153 Complete and entire. G i l l membranes separate, with gular Hyoid arch: Interhyal, epihyal, ceratohyal and two epihyals in Hiodon tergisus. References: Ridewood (1904), (1905b), Jordan and Evermann (1896), Boulenger (1904). Material examined: Hiodon tergisus, alizarin specimen, BC58-164 from Lake Winnipeg, Manitoba; specimen BC60-250 from Lake Winnipeg, Manitoba; two alizarin specimens, NMC59-334 from Lac St, Pierre, Quebec. Hiodon  alosoides, 1 alizarin and 10 alcoholic specimens from Saskatchewan River, Manitoba. SUBORDER G0N0RHYNCH0IDEI Branchiostegals 4-5, with 4 epihyal and 0-1 ceratohyal, a l l external, the upper 2 spathiform. Opercular bones complete and entire. G i l l membranes broadly joined to isthmus. Two hypurals. A single family, Gosline (i960) united Berg's suborder Chanoidei, Phractolaemoidei, and Cromerioldei with the Gonorhynchoidei. At the same time he appeared uncertain that these groups belong together, "These five groups are so widely different that any relationship between them is difficult to comprehend. Yet the following similarities may be marshalled". He then l i s t s g i l l membranes always attached to isthmus (yet in Chanos they are united and free), the mouth small and toothless or nearly so, supra-maxillaries absent, the preopercular border free only below (Chanos), i f at a l l , and several other characters. In his key he characterizes the 154 suborder as having 3-4 branchiostegals, however, Gonorhynchidae have 4-5. This author agrees with Gosline that the Phractolaemidae, Kneriidae and Croraeriidae belong in the same suborder. These families share 3 branchiostegals, about 5-9 pelvic rays, 34-45 vertebrae, absence of axillary appendages, absence of pharyngeal teeth, g i l l opening restricted, scales cycloid or absent, presence of a gas bladder, a similar caudal skeleton (Gosline) and are a l l small African freshwater fishes. The Chanidae share with these 45 vertebrae, absence of pharyngeal teeth, cycloid scales, a gas bladder, a similar caudal skeleton (Gosline), but they differ in having 4 branchiostegals, 11-12 pelvic rays, having axillary appendages, caudal fin flaps, g i l l membranes united and free from the isthmus and being Indc—Pacific in distribution. These differences are not too trenchant. Further, Audenaerde (1961) in his osteological study of Phractolaemus states that numerous osteological and anatomical characters suggest a close relationship of Phractolaemus and Chanos. Gonorhynchidae shares only these characters: pharyngeal teeth absent, g i l l membranes joined to isthmus (but opening not restricted). It differs in having 10 pelvic rays, 54-56 vertebrae, possession of axillary appendages, ciliated scales, gas bladder absent, a different caudal skeleton (Gosline) as well as in peculiar characters such as a median rostral barbel, tongue-like and fringed flap-like structures on the roof of the mouth, rounded basi-branchial teeth, posterior side of fourth g i l l and back of branchial cavity papillose, peritoneum black, thick l ips , a pseudc—occipital condyle (Gregory, 1933), and others. It dwells in the Indo-Pacific. The characters shared with Chanidae are a 155 high number of pelvic rays, axillary appendages and number of branchiostegal rays (4 in Chanidae, 4-5 in Gonorhynchidae). But while the branchiostegal ray number is similar, as noted by Gosline, they differ in arrangement and form. In Gonorhynchidae there are 4 on the epihyal, in Chanidae only 2. Those in Gonorhynchidae lack clupeoid projections, while those of Chanidae possess clupeoid projections. The g i l l membranes differ and they differ in the peculiar characters listed above. It is concluded that Phractolaemidae, Kneriidae, Chanidae and Cromeriidae belong in one suborder, the Gonorhynchidae require a separate suborder. Gosline (I960) stated that the caudal skeleton of Gonorhynchus bore considerable resemblance to that of Dus3umieria. The dussumieriids and alepocephalids are the only clupeoids with as many as 4 branchiostegals on the epihyal like Gonorhynchus. The albuloids would also form a possible ancestor in the last respect, and somewhat resemble Gonorhynchus in dentition. Gonorhynchidae Pl. VII Branchiostegals: Vary from 4-5 in Gonorhynchus. The upper two branchiostegals broaden distally into a lamina and are more or less straight. A ridge strengthens each of the branchiostegals. 4 branch-iostegals on the epihyal and 0-1 on the ceratohyal, a l l on the external face of the arch. Operculars; Complete and entire. The operculum extends in a wedge down into the suboperculum. The suboperculum extends more than half way up 156 the posterior border of the operculum. G i l l openings wide, membranes attached to isthmus. Hyoid arch: Consists of large interhyal, epihyal, ceratohyal and two hypohyals. References: Hubbs (1920), Ridewood (1905a). Material examined: Gonorhynchus g o n o r h y n c h u 3 , alcoholic specimen, USNM 59920 from New South Wales, Australia; G. gonorhynchus, alcoholic specimen, BC56-278 from New Zealand; G. gonorhynchus, alizarin specimen, NMC62-140, from South Africa. SUBORDER CHANOIDEI Branchiostegals 3-4 with 2 epihyal and 2 ceratohyal, a l l external, spathiform, with clupeoid projections at least in Chanidae. Gular absent. Opercular bones complete and entire. G i l l opening restricted or g i l l membrane united and free from isthmus (Chanidae). Two hypohyals (Chanidae) or one (Phractolaemidae). Four families. The relationships of this suborder are discussed under the Gonorhynchoidei. This suborder appears to be distinguished by the absence of an interhyal. But this is not yet verified in the Kneriidae or Cromeriidae. Chanidae PI. VI Branchiostegals: In Chanos 4 branchiostegals on each side, a l l broad and spathiform but lower three tapering, bases broad with anterior clupeoid projection, 2 on epihyal, 2 on ceratohyal, a l l on external face of hyoid arch 157 Operculars: Complete and entire. A straight border between the operculum and suboperculum. G i l l membranes united and free from isthmus. Hyoid arch: Consists of epihyal, ceratohyal and two hypohyals. Interhyal absent, epihyal connected to hyomandibular via ligament. Relationships: The anterior projection on the base of the branchiostegal is similar to that in the clupeoids. The reflexed basicranium of Chanos is much like that of the Dussumieridae. References: Hubbs (1920), Day (1875), Jordan and Evermann (1896), Weber and de Beaufort (1913), Ridewood (1904a). Material examined: Chanos chanos, alizarin specimen, BC60-25, Mexico. Kneriidae Branchiostegals: In Kneria 3 branchiostegal rays. Operculars: Complete and entire. Operculum rounded posteriorly, sub-operculum wedge-shaped, interoperculum elongate and expanded up around posterior border of preoperculum. G i l l opening restricted to small lateral s l i t . References: Giltay (1934a), Hubbs (1920), Berg (1947). Material examined: None. Phractolaemidae Branchiostegals: 3 slender in Phractolaemus. 158 Operculars; Complete and entire. The lower limb of the preoperculum is much expanded and covers the elongate interoperculum, operculum oval, suboperculum wedge-shaped. G i l l opening restricted and reduced to a small s l i t on the side of the head. Hyoid arch; Interhyal apparently absent, epihyal, ceratohyal and one hypohyal present. Relationships: Like Mormyriformes this family has only one hypohyal. However, the opercular and caudal skeleton differ strongly indicating the loss of a hypohyal is only a parallelism. References; Hubbs (1920), Audenaerde (1961), Ridewood (1905), Poll (1957), Boulenger (1904). Material examined; None. Cromeriidae Branchiostegals; 3 curved scimitar-like branchiostegals which terminate in a point, in Cromeria. Operculars: Complete and entire. Operculum an oval, suboperculum a broad wedge, interoperculum elongate broadening posteriorly. G i l l opening restricted to a small opening on the lateral surface. Relationships: The Cromeriidae, Phractolaemidae and Kneriidae are closely related. They share three branchiostegals, a laterally restricted g i l l opening, a wedge-shaped suboperculum, an elongate interoperculum which broadens posteriorly plus characters previously mentioned. Of the three 159 the Cromeriidae are the moat degenerate, having lost the scales, supra-orbitals, suprapreopercular, etc. References: Hubbs (1920), Gregory (1933). Material examined: None. SUBORDER ST0MIAT0IDEI Branchiostegals 5-24 with 0-1 interhyal, 3-8 epihyal, 4-12 ceratohyal and 0-5 hypohyal, 3-12 external and 0-7 ventral. Gular absent. Opercular bones complete and entire. G i l l membranes separate, united and free from isthmus or joined to isthmus. Ceratohyal elongate, except in Sternoptychidae. Two hypohyals. Nine families, one wholly fossil , are placed in the family. Two additional families, one fossil are provisionally included. The Rosauridae and Protostomiatidae are only provisionally placed in this suborder. Their branchiostegal series do not confirm placement here. The remaining families share characters with one another and appear to form a natural phylogenetic unit. They are characterized by possession of numerous branchiostegals; by expansion of the upper one or two branchiostegals (except in Idiacanthidae), the remainder being slender; by the branchiostegals usually extending onto the lower hypohyal; by the deep and narrow operculum; by the reduced suboperculum and inter-operculum; by the thin and poorly ossified opercular bones; and by the photophores on the branchiostegal membrane. Tne high number of branchiostegals, elongate mouth, and deep operculum might lead one to conclude the stomiatoids were derived from 160 the Engraulididae. However, the stomiatoids have a higher number of branchiostegals on tho epihyal, making such a derivation unlikely. The high number of epihyal branchiostegals \rould be commensurate with derivation from the elopoids. This contention is supported by Regan (1923b) who found striking agreement in the skulls of Photichthys and Elogs. The unusual feeding mechanism of Chauliodontidae and Malacosteidae is described and illustrated by Tchernavin (1948), (1953) and Gunther and Deckert (1959). Gonostomatidae ( incl . Maurolicidae) Branchiostegals: Vary from 7-21. In Agyripnus 8-10, Bonapartia 11-16, Cyclothone 10-14; Danaphos 9-10, Diplophos 11-14, Gonostoma 10-14, Ichthyococcus 11-12, Margrethia 13, Maurolicus 9-10, Neophus 7-8, Photichthys 20-21, Pollichthys 11-12; Sonoda 8, Triplophus 11-14, Valencienellus 9-10, Vinciguerria 10-12, Woodsia 17, Yarella 13-16. In Vinciguerria lucetia 12 on both sides, with 4 epihyal, 7 ceratohyal and 1 hypohyal, a l l external, the upper 2 broad and scimitar-like, the lower ones becoming progressively more rod-like. Photophores occur on the branchiostegal membranes of a l l genera. Operculars: Complete and entire. Operculum elongate and vertical, suboperculum and interoperculum reduced to small lamina. G i l l openings wide, g i l l membranes separate. Hyoid arch: Consists of Vinciguerria of interhyal, epihyal, long ceratohyal and two hypohyals. Arch connected to jaws by a membrane. 161 Relationships; The maurolicine and related genera, sometimes separated as a distinct family, have 7-10 branchiostegals, while the remainder have 10-21. This would support subfamilial recognition of the two groups. References; Grey (i960). Material examined: Vinciguerria lucetia, alizarin specimen, NMC61-195, from 900 miles west of Lower California. Sternoptychid ae Pl. VII Branchiostegals: Vary from 5 to 11. In Argyropelecus 9, Polyipnus 9-11, and Sternoptyx 5» Argyropelecus sp. with 10 branchiostegals, 3 epihyal and 7 ceratohyal, 3 on external and 7 on the ventral face of the arch; the uppermost is a rounded triangle, the next two lathe-like but expanding distally, the remainder rod-like. In Sternoptyx the upper branchiostegal is expanded and wing-like. The branchiostegal membranes bear photophores, 6 in Polyipnus and Argyropelecus and 3 in Sternoptyx. Operculars: Complete and entire. Bones thin, laminar and take l i t t l e alizarin stain. Opercular narrow and vertical, suboperculum and inter-operculum small. G i l l membranes free from isthmus or attached, sometimes united and free from isthmus. Hyoid arch: Interhyal, long epihyal and ceratohyal, and 2 hypohyals present. The ceratohyal is bent in the middle with the apex upwards in Arkyropelecus. Arch connected to jaws by a membrane. 162 References: Schultz (1961), Weber and de Beaufort (1913), Fowler (1936), Garman (1899), Jordan and Evermann (1896), Gregory (1933), Misra (1953)B Material examined: Argyropelecus sp., alizarin specimen, NMC61-184, from 400 miles off California. Stomiatidae Branchiostegals: Vary from 14-19. In Stomias 14-19, Stomloides 15• Operculars: Complete and entire, g i l l membranes separate. Hyoid arch: A membrane connects the lower jaw with the hyoid arch. Relationships: Parr (1927) believed that the stomiatids deserved separation from the other two groups of stomiatoids, the gonostomatid group and the melanostomiatid group. This division is not reflected in the branchiostegal series. References: Parr (1927), (1933), Weber and de Beaufort (1913), Maul (1956a), Regan and Trewavas (1930), Regan (1923). ' Material examined: None. Chauliodontidae Branchiostegals: Vary from 15 to 21 in the genus Chauliodus. In Chauliodu3 macouni 20 branchiostegals with 5 epihyal, 11 ceratohyal and 4 hypohyal, the uppermost expanded, the remainder lathe-like. Operculars: Complete and entire. Operculum narrow and vertical, sub-operculum and interoperculum small. G i l l membranes free. 163 Hyoid arch; Interhyal, long epihyal and ceratohyal, two hypohyals. The hyoid arch and tongue are free from the jaws at the sides, being attached to the symphysis only by an elastic ligament. The resulting freedom enables the jaws to be shot forward and upward, enabling prey to be more readily caught. Tchernavin (1948, 1953) reports on this interesting feeding mechanism. References: Morrow (1961), Garman (1899). Material examined: Chauliodus macouni. alizarin specimen, NMC61-192, from 200 miles off Washington. Astronesthidae Branchiostegals: In Astronesthes 14-24. Operculars: Complete and entire. References: Fowler (1936), Gibba and Aron (I960), Maul (1956a), Weber and de Beaufort (1913). Material examined: None. Malacosteidae Branchiostegals: Vary from 8-18. In Malacosteus 8 short rod-like branchiostegals; in Aristostomias 18 with one on the interhyal, 5 on the epihyal and 12 on the ceratohyal, the upper two expanded slightly and lath-like, the remainder rod-like. Photophores on the branchiostegal membrane. 164 Operculars: Complete and entire. Operculum narrow and vertical, interoperculum and suboperculum small and covered by the preoperculum which extends back because of the prolongation of the jaw. G i l l membranes separate, not joined to jaws. Hyoid arch: The hybid arch is not connected to the front or sides of the mandible except by the long slender protractor hyoidei so that the mandible is completely free, hence the name loosejaws applied to the family. Hyoid arch consists of interhyal, long epihyal and ceratohyal and 2 hypohyals in Aristostomias and Malacosteus. References: Gunther, K. and Deckert (1959), Weber and de Beaufort (1913), Fowler (1936), Regan and Trewavas (1930). Material examined: Aristostomias scintillans. alizarin specimen, NMC61-182 from 50 miles west southwest of Cape Flattery, Washington. Melanostoraiatidae PI. VI Branchiostegals: Vary from 10-22 with 0-1 interhyal, 3-8 epihyal, 4-10 ceratohyal and 0-5 ceratohyal. In Tac to stoma 13 with 5 epihyal, 6 ceratohyal and 2 hypohyal; in Bathophilus flemingi 10 with 3 epihyal, 7 ceratohyal and 0 hypohyal, 3 external and 7 ventral; Bathophilus  metallicus 12 with 3 epihyal, 9 ceratohyal and 0 hypohyal; Chirostomias  pliopterus 22 with 8 epihyal, 10 ceratohyal, and 4 hypohyal; Leptostomias bermudensis 19 with 6 epihyal, 8 ceratohyal and 5 hypohyal; Echiostoma  tanneri 15 with 6 epihyal, 4 ceratohyal and 5 hypohyal; Melanostomias  spilorhynchus 13 with 4 epihyal, 5 ceratohyal and 4 hypohyal; Photonectes 165 dinema 15 with 5 epihyal, 6jj ceratohyal and 3h hypohyal; P. margarita 13 with 4 epihyal, 7 ceratohyal and 2 hypohyal; Flagellostomias boureei 16 with 4 epihyal, 8 ceratohyal and 4 hypohyal; Grammatostomias  flagellibarba 13 with 5^ epihyal, b\ ceratohyal and 1 hypohyal; Eustomlas  fissibarbis 16 with 5 epihyal, 8 ceratohyal and 3 hypohyal* In Bathophilu3 flemingi the upper branchiostegal expands into a lamina distally, the rest are lath-like; in Tactostoma macropus the upper one is expanded into an elongate triangle, the rest are styllform* In Flagellostomias and Photonectes the upper branchiostegal was branched, probably as a result of fusion of branchiostegals. Operculars; Complete and entire. Operculum elongate and vertical, suboperculum and interoperculum small. Hyoid arch; Consists of interhyal, epihyal, ceratohyal and two hypohyals i* 1 Bathophilus* Melanostomias. Photonectes and Tactostoma. Arch attached to mandibles by membrane. References; Gunther (1887), Parr (1933), Beebe and Crane (1939), Regan and Trewavas (1930). Material examined: Bathophilus flemingi. alizarin specimen, NMC61-80, off northern California; Tactostoma macropu3. alizarin specimen, NMC61-182, 50 miles west southwest of Cape Flattery, Washington. Idiacanthidae Branchiostegals: Vary from 12 to 18 in Idiacanthus* In Idiacanthus  fasciola 17 with 5 epihyal, 7 ceratohyal and 5 hypohyal, a l l long and slender, the upper most not expanded unlike the preceding stomiatoid families 166 Operculars; Complete and entire. Operculum elongate and vertical, others small. Hyoid arch; Consists of interhyal, long epihyal and ceratohyal and two hypohyals (basihyal of Beebe). Arch attached to mandibles by membrane. Relationship; According to Beebe, closely related to the Melanostomiatidae. Differs from other stomiatoids in the lack of an expanded upper branchiostegal. References; Beebe (1934), Garman (1899), Gunther (1887), Weber and de Beaufort (1913), Regan and Trewavas (1930). Material examined: None. / Tomognathidae Branchiostegals: At least 7 in Tomognathus mordax, the dorsal one spathiform, the remainder slender, with about 4-5 on the epihyal. Operculars: Complete and entire. Operculum deep and narrow. Relationship: The deep operculum, only the uppermost of the branchiostegals spathiform, and the high number, 4-5 of the branchiostegals on the epihyal suggest that this family is correctly placed in the Stomiatoidei. References: Woodward (1902-1912), (1936). 167 ^ Protostomidae, incertae cedis  Branchiostegals: Vary from 10-24. In Protostomias about 24, short, slender branchiostegals; in Pronotacanthus 10 slender branchiostegals with 4 epihyal and 6 ceratohyal. Operculars: Complete and entire in Prostomias; the operculum sub-triangular with the suboperculum curving up around posteriorly, the interoperculum a small triangle. Operculum not deep and narrow in Pronotacanthus. Hyoid arch; At least a long epihyal and short ceratohyal in Pronotac anthus. Relationships; Arambourg (1954) removed Pronotacanthus from the Notacanthidae and placed i t in the Stomiatoidei. The operculum and jaw apparatus of Pronot ac an thus or Prostomias does not agree with that in other stomiatoids where the operculum i s deep and narrow and the jaws and gape long making the suspensorium vertical. The arrangement of the branchiostegals, terminal mouth and absence of a "spinous" dorsal in Pronotacanthus exclude i t from the Notacanthiformes. References; Arambourg (1954), Woodward (19d). Rosauridae, incertae cedis  Branchiostegals; In Rosaura rotunda 10 long slender branchiostegals which taper distally; about 4 epihyal and 6 ceratohyal;the uppermost is bent into an S-shape. Known only from a postlarval specimen of 8.4 mm. Branchiostegals extend out behind at right angle to the jaws, the rays continuing beyond the branchiostegal membrane. 168 Operculars: Entire. Operculum a vertically suspended oval, suboperculum and elongate oval. Interoperculum not yet developed or absent. Hyoid arch; At this stage consists only of an interhyal and a ceratohyal element which has not yet divided into epihyal and hypohyal. Relationships; Tucker (1954) considered the Rosauridae was most likely related to the stomiatoids, although its affinities were difficult to trace. Unlike the stomiatoids the branchiostegals in Rosauridae project backwards horizontally, while in the stomatoids they are between the jaws pointing towards the isthmus. Usually the upper branchiostegal of the stomiatoids is enlarged; in Rosauridae i t is not enlarged. In these features of the branchiostegals Rosaura more resembles the alepocephalids. While the operculum of stomiatoids is elongate, forming most of the g i l l cover, i t is quite reduced in Rosauridae and is no larger than the suboperculum. Because of the juvenile stage of the specimen i t is difficult to come to conclusions on the relationships of Rosauridae until more material is obtained. It is hence provisionally left, incertae cedis, in the Stomiatoidei. References; Tucker (1954)* Material examined: None. 169 / SUBORDER ENCHODONTOIDEI / Enchodontidae Branchiostegals; Vary from about 9-16. In Halec 9-15; Enchodus 12-16; Eurypholis about 15 slender attenuate branchiostegals, the upper apparently not spathiform; Pantopholis 14 narrow branchiostegals. Gular absent. Operculars; Complete and entire in Halec and Rharbichthys, preoperculum obscures interoperculum in Eurypholis. In Eurypholis the suboperculum forms half, in Rharbichthys more than half of the posterior border of the g i l l cover. Relationships? Gregory (1933) placed this family in the Iniomi. Berg (1947) includes i t in the Clupeiformes. The slenderness of the upper branchiostegals and the great portion of the posterior border of the g i l l cover taken up by the suboperculum are tendencies towards the iniomous condition. But the failure of the branchiostegals to curve behind the g i l l cover and the inclusion of the toothed maxillaries in the gape place the Enchodontidae in the Clupeiformes. References: Woodward (1901), (1902-12), Davis (1887). SUBORDER SALM0N0IDEI Branchiostegals 2-19 with J-4 on the epihyal, 0-7 on the ceratohyal, 2-6 on the external and 0-6 on the ventral surface of the hyoid arch. At least the upper branchiostegals spathiform. Gular absent. Opercular bones complete and entire (except operculum crenulate in Bathylagus, Aplochiton. and some Argentinoidea. G i l l membranes separate or sometimes 170 united and free from the isthmus. Two hypohyals present. Ten living families plus one known only from fossils. The Salmonoidei break down into three natural groups, the anadromous or freshwater Salmonoidea with 7-19 branchiostegals, the litophilus, anadromous or freshwater Osmeroidea with 3-10 branchiostegals and the offshore marine Argentinoidea with 2-6 branchiostegals. The Salmonoidea further have two upturned postterminal vertebral centra while the other two superfamilies have not more than one (Gosline, I960). The branch-iostegals and caudal structure both indicate that the Salmonoidea is the most primitive of the three groups. Further indications are that the Salmonoidea, but not the others, possess an opisthotic and a supramaxilla. The families here included in the Argentinoidea have been treated as a separate suborder Opisthoproctoidei Chapman (1942* 1948)* At the other extreme Hubbs (1953) has suggested synonymizing the Macropinnidae, Dolichopterygidae, Bathylagidae, Microstomidae, Xenophthalmichthyidae, Opisthoproctidae and presumably Winteriidae with the Argentinidae. An intermediate path has been followed here, recognizing the affinities of the groups under a superfamily and yet retaining the distinctions of mo3t of the groups by familial status. Superfamily Salmonoidea Following the study of Norden (196l), Thymallidae and Coregonidae are synonymized vdth Salmonidae. Salmonidae Branchiostegals: Vary from 7-19 with 4 epihyal and 7 ceratohyal, 5 171 external and 6 ventral. In Oncorhynchus 10-20, Salmo 9-12, Salvelinus 8-14* Brachymystax 10-13, Hucho 10-14, Stenodus 8-12, Coregonus 7-10, Prosopium 7-9, Thymallus 8-12. In Salmo clarkii 11 with 4 epihyal and 7 ceratohyal, 5 external and 6 ventral, upper 4 broad and laminar, next 5 broad in the middle but narrowing at the ends, last 2 slender. In Thymallus arcticus 9-10 with 3 epihyal and 6-7 ceratohyal. In a single sample of 50 Oncorhynchus nerka, seven were found with 11 branchiostegals, nineteen with 12 and twenty four with 13. Operculars: Complete and entire. Hyoid arch: Consists of interhyal, epihyal, ceratohyal and two hypohyals in Salmo. Salvelinus and Thymallus. In Salmo and Thymallus epihyal about 2/3 the length of ceratohyal. Relationships: It may be stated that the number of branchiostegals does not give support to the recognition of the families Coregonidae and Thymallidae. The subgenus Cristivomer has 12-14 branchiostegals while the subgenus Salvelinus h a 3 9-12. While not giving basis for generic status of Cristivomer. because of overlap, i t does again, l ike other characters, point to the divergence of Cristivomer from the other species of the genus. Norden (1961) considers that the Salmonidae are the most primitive subfamily. The high number of branchiostegals are in agreement with this opinion. The branchiostegal number would favour Oncorhynchus as being a primitive genus in the subfamily but Norden (1961) considers i t the most advanced. Branchiostegals indicate Stenodus to be the most primitive of the w h i t e f i s h e 3 , an indication in accord with its well developed teeth. 172 References: Clemens and Wilby (1949), Berg (1949), Kimsey and Fisk (1960), Vladykov (1954), M i l l e r (1950), Dymond (1943), Koelz (1929), Holt (I960), Kennedy (1943), Jordan and Evermann (1896), Norden (1961), Rounsefell (1962), H i k i t a (1962). Material examined: Salmo c l a r k i i , a l i z a r i n specimen, BC54-29, B r i t i s h Columbia; Oncorhynchus nerka, 50 alcoholic specimens BC61-694, B r i t i s h Columbia; Salvelinus namaycush, ske l e t a l specimens, NM62-160-S, Ottawa f i s h market and NMC60-100 from Northwest T e r r i t o r i e s . ^ Thaumaturidae Branchiostegals t More than 5 i n Thaumaturus. Operculars: Complete and entire. Hyoid arch: At l e a s t ceratohyal and epihyal present. Relationships: Generally associated with the Salmonidae. The known number of branchiostegals i s lower than i n the Salmonidae. However, the series may be incomplete. Norden (196l) suggests from the caudal vertebra of Thaumaturus that i t i s a l l i e d to Argentinidae or Osmeridae which would agree with the known number of branchiostegals. References: Voigt (1934). Superfamily Osmeroidea Gosline (I960) includes the Osmeridae, Plecoglossidae, Salangidae, Aplochitonidae, Retropinnidae and Galaxiidae i n t h i s superfamily. He further separated them into two groups with the f i r s t three families below i n a northern hemisphere group and the l a s t three f a m i l i e s i n a 173 southern hemisphere group. Osmeridae PI. VI, VII Branchiostegals; Vary from 6-10. Hypomesus 6-8, Mallotus 8-10, Osmerus 6-8, Spirinchus 7-8, Allosmerus 6-7 and Thaleichthys 6-8. In Mallotus  villqsus 9 with 4 epihyal and 5 ceratohyal, 5 external and 4 on internal face of hyoid arch, the uppermost spathiform, the lowest virgaform. In Osmerus eperlanus mordax 7 with 4 epihyal and 3 ceratohyal, 4 on the external and 3 on the ventral face of the hyoid arch, the uppermost broad and spathiform becoming progressively narrow ventrally. In both of these species there is a slight anterior projection at the base of the branchiostegal, reminiscent of the clupeoids. The upper branchiostegal is only slightly curved in the family. Operculars: Complete and entire. G i l l membranes separate. Hyoid arch: An interhyal, epihyal, ceratohyal and two hypohyals in Thaleichthys, Mallotus and Osmerus. The epihyal large, nearly as long as the ceratohyal in these genera. References: McAllister (1963), Chapman (1941a). Material examined: Osmerus eperlanus mordax, alizarin specimen, NMC62-110, Great Lakes, Canada; Mallotus villosus, alizarin specimen, NMC60-47, Yukon, Canada; also material listed in McAllister (1963). 174 Plecoglossidae Branchiostegals; 5-6 branchiostegals in Plecoglossus with 2\ on the epihyal and 3h o n the ceratohyal. 3 on the external face and 3 on the ventral face of the hyoid arch. The upper three are broad, spathiform and crescentic, the lower three narrow. Operculars. Complete and entire. Interoperculum small, covered by the ventral arm of the preopercle. G i l l membranes separate. Hyoid arch; Consists of interhyal, epihyal, ceratohyal and two hypohyals* Epihyal about $ length of ceratohyal. Relation ship s: The number and arrangement of the branchiostegals agree with the Osmeridae although there are generally more branchiostegals in the Osmeridae. The upper branchiostegal in Osmeridae is only very slightly curved whilst in the Plecoglossidae i t is distinctly crescentic. Further, the interoperculum is normally exposed in Osmeridae whereas in the Plecoglossidae i t is hidden by the preopercle. References: Chapman (1941), Okada (i960). Material examined: None. Salangidae Branchiostegals: 4 in Salanglchthys and in family. Operculars: Complete and entire, the interoperculum well developed. G i l l membranes free from isthmus. Relationships: In the reduced number of branchiostegals, lack of scales, 175 and neotenous condition the Salangidae differ from the other northern Osmeroidea. References: Gosline (i960), Okada (i960), Hubbs (1920), Wakiya and Takahasi (1937). Retropinnidae Branchiostegalst 5-6 in Retropinna» Upper branchiostegal broad and spathiform. Operculars: Complete and entire. References: Stokell (1941)• Material examined: Retropinna osmeroides, 2 alcoholic specimens, BC56-296. Ashley River, New Zealand. Galaxiidae Pl. VI, VII Branchiostegals: Vary from 5-9. Galaxias with 5-9, Neochanna with 6-7. In Galaxias maculatus 5 with 2 epihyal and 3 ceratohyal, 3 on the external and 2 on the ventral face of the hyoid arch, the upper three spathiform. £• attenuatus 8 with 3 on the external face of the epihyal and 5 on the ventral ceratohyal, upper 3 spathiform. In Neochanna apoda. 6 with 2 epihyal and 4 ceratohyal, 3 on the external and 3 on the ventral face of the hyoid arch, the upper 3 broad laminar, spathiform, the lower 3 slender. Galaxias indicus with 9. The upper spathiform branchiostegals haspate in form. 176 Operculars: Entire and complete. G i l l membranes separate, Hyoid arch: Consists of interhyal, epihyal, ceratohyal and two hypohyals. The epihyal is about one half as long as the ceratohyal. References: Gregory (1933), Misra (1953), Stokell (1949). Material examined: Galaxias maculatus, alizarin specimen, SU 22679 from Mayne Harbor, Patagonia; Neochanna apoda. alizarin specimen, BC56-288 from Wairarapa district, New Zealand, Galaxias attenuatus, alizarin specimen, NMC62-244, Hinds River, New Zealand. Aplochitonidae (Haplochitonidae) Branchiostegals: Varies from 3-6. In Aplochiton zebra, 3 with \ epihyal and 2\ ceratohyal, 2 on external face and 1 on ventral face of the hyoid arch, a l l broad spathiform, straight; in Lovettia 6. Operculars: Complete and entire except for the border of the operculum which is crenulate in Aplochiton. Hyoid arch: Consists of interhyal, epihyal, ceratohyal and two hypohyals. Epihyal about 2/5 of length of ceratohyal. References: Chapman (1944a), Hubbs (1920), Gosline (i960). Material examined: None. 177 Superf amily Argentine-idea The classification of Cohen (1957) is followed for this group. Argentinidae (incl. Microstomidae and Xenophthalmichthyidae). Branchiostegalst Vary from 3-7* In Argentina 5-7, Glossanodon 4-5, Nansenia 3-4, Microstoma 3-4, Xenophthalmichthys 3. In Argentina  sphyraena 7 with 2 epihyal and 5 ceratohyal, 3 on the external face and 4 on the ventral edge of the hyoid arch, the upper 3 broadly spathiform, the rest slender. In Xenophthalmichthys danae, 3 spathiform branch-iostegals on the external face of the arch. Operculars; Complete in Argentina, Xenophthalmichthys and Nansenia. Opercle and subopercle crenulate in Nansenia, entire in the others. G i l l membranes separate (Argentininae) or broadly united and free from isthmus (Microstomatinae)• Hyoid arch: Consists of interhyal, epihyal, ceratohyal and two hypohyals in Argentina, and Nansenia. Relationships: According to the number of branchiostegals this family, and particularly the Argentininae, would be the most primitive of the superf amily. It is the only family of the superfamily with a post-cleithrum and has the most numerous pelvic rays. This would confirm the indication given by the branchiostegals. Cohen (1957) has already stated this position. References: Cohen (1957, 1958a, 1958b), Chapman (1942, 1942a, 1948), Bertelsen (1958), Jensen (1948), McAllister (1961), Regan (1925a), Bertin and Arambourg (1958). Material examined: Argentina silus, 4 alcoholic specimens, NMC62-79, southwest of Sable Island, Nova Scotia. Bathylagidae Branchiostegals: Constantly number 2. Leuroglossu3 2, Bathylagus 2. Two external epihyal, in B. pacificus they are broad, the breadth exceeding 1/3 the length; those in glaciali3 slender, the breadth less than 1/10 the length. Cohen (1957) indicates Beebe was in error in giving a count of 3-4 for Bathylagus and that he has never seen a single branchiostegal as reported by Hubbs (1920). But Hubbs was not in error, he stated only there was a single large branchiostegal and did not imply there was not a further slender one below. Operculars: Complete and entire in Bathylagus pacificus and B. glacialis except that there are two slots in the posterior border of the operculum. 1° Pacificus there is a groove which probably represents the point of fusion of the slots noted in glacialis. G i l l membranes united and free from the isthmus. Hyoid arch: Consists of interhyal, epihyal, ceratohyal and two hypohyals i * 1 Bathylagus. Beebe mislabels the hypohyals as basihyal. References: Beebe (1933), Chapman (1943), Cohen (1957), (1958c), McAllister (196l), Bolin (1938). Material examined: Bathylagus milleri, alizarin specimen, NMC61-183, from 100 miles off north end of Vancouver Island, British Columbia. 179 Opisthoproctidae (Dolichopterygidae, Winteriidae, Macropinnidae) Branchiostegals: Vary from 2-4. Rhynchohyalus 4, Macropinna 3, Dolichopteryx 2. Bathylychnops 2, dpisthop.roctus 2. In Macropinna 3 spathiform branchiostegals on the external surface of the ventral unossified epihyal. Operculars: Complete and entire in Macropinna; interoperculum expanded, broadly displaced anteriorly away from the suboperculum but connected to i t by a ligament; g i l l membranes united and free from isthmus. Hyoid arch: In Macropinna consists of interhyal, epihyal, ceratohyal and two hypohyals. The epihyal is ossified only on i ts dorsal edge. Relationships; The branchiostegals of Dolichopteryx definitely agree with their placement in the Cpisthoproctidae rather than in the Alepocephaloidae (in which group they had been placed). SUBORDER ESOCOIDEI Branchiostegals 4-20 with 1-8 epihyal and 3-10 ceratohyal, 3-8 on the external and 1-10 on the ventral face of the arch. Branchiostegals acinaciform to spathiform. Gulars absent. Opercular bones complete and entire. G i l l membranes separate. Hyoid arch complete with two hypohyals. Four families, one no longer l iv ing. The number of branchiostegals would indicate the Esocidae (and ^Palaeoesocidae) to be more primitive than the Umbridae and Dalliidae. But according to Gosline (I960) the caudal skeleton of Umbra seems more primitive than that of Esox. Other characters were therefore examined 180 in Chapman (1934) in an attempt to determine which family was most primitive. In checking the characters: number of pelvic rays, caudal rays and actinosts; the presence of a postcleithrum, myodome, infra-mandibular, suborbitals, nasal, septomaxillary, and basisphenoid; presence of ribs on anterior vertebrae and reduction of the preorbital; the Esocidae were found most primitive, followed by the Umbridae and lastly the Dalliidae. In this case most characters are in agreement with the number of branchiostegals (and not the caudal skeleton) in demonstrating the Esocidae to be more primitive. Pallia on the other hand has one more branchiostegal ray than does the Umbridae, in disagreement with the indications by the other characters. However, the Dallidae and Umbridae are less different from one another than from the Esocidae, In the intermediate range of branchiostegals ( 7 or over) the difference of only one branchiostegal when there is overlap is not very significant. Further, evolution may act at different rates in different organs, so that while the branchiostegals in the Dalliidae remained in a slightly more primitive condition than In the Umbridae, other characters continued advancement (advance in this case being degeneration). The number of branchiostegals in the Palaeoesocidae is about the same as in the Esocidae. Palaeoesox further agrees with the Esocidae and differs from the Dalliidae and Umbridae in that suborbitals and nasals are present, thus confirming the placement on the basis of the branchiostegals. The Esocidae and Palaeoesocidae on one hand and the Dalliidae and Umbridae on the other hand are divergent in the number of branchiostegals (10-20 as opposed to 4-8), and in the suborbitals, nasals, infra-mandibulars, etc. Therefore i t seems unwise to place the Palaeoesocidae 181 in the Umbroidea, and they are here placed provisionally in the Esocoidea. It may be noted that the subopercle in Movumbra and particularly Pallia parallels the condition in the Myctophiformes where i t sweeps up behind the operculum forming its posterior border. Similarly the presence of branchiostegals on the internal face of the ceratohyal is an iniomous condition. The numerous branchiostegals with many on the epihyal and the double hypohyals suggest that the esocoids are descended from the elopoids or primitive clupeoids, perhaps not greatly distant from the stock that gave rise to the stomiatoids. Frost (1926) notes the sagitta of Esox resembles that of Megalops, but is more highly specialized. Superfamily Esocoidea Esocidae Branchiostegals; Varies from 10 to 20 in Esox with 5-8 epihyal and 7-10 ceratohyal, 5-8 on the external face and 7-10 on the internal or ventral face of the hyoid arch. Al l of the rays are acinaciform. Crossman (I960) presents large samples showing variability within the species. Operculars: Opercular bones complete and entire, although the posterior edges of the operculum and suboperculum may be weakly crenulate. Opercular-subopercular border straight, suboperculum not extending up behind operculum. G i l l membranes separate. Hjraid arch: Consists of interhyal, epihyal, ceratohyal and two hypohyals. The epihyal is about 2/3 the length of the ceratohyal. TITO small, prongs extend towards the hypohyals from the upper anterior end of the ceratohyal. 182 References: Crossman (i960), Berg (1949). Material examined: Esox masquinongy. alizarin,specimen, NMC60-219, Ontario. / Palaeoesocidae Branchiostegals: 13 branchiostegals in Palaeoesox fritzschi with 6 on the epihyal and 7 on the ceratohyal, 6 being on the external face of the epihyal and 7 on the inner face of the ceratohyal. Branchiostegals acinaciform. Operculars: Complete and entire. Suboperculum extending diagonally up behind one third of the posterior border of the operculum. Hyoid arch: At least ceratohyal and epihyal present. Relationships: In shape, number and distribution of the branchiostegals Palaeoesocidae agree with the Esocidae and differ from the other two families. References: Voigt (1934). Material examined: None. Superfamily Umbroidea Umbridae PI. VI Branchiostegals: Vary from 4-7. In Umbra limi 4-5 with 1 epihyal and 3-4 ceratohyal, 3 external and 1-2 ventral. The upper 2 are crescentic, the lower ones are slender. In Novumbra hubbsi 6-7 with 2-3 on the epihyal and 4-5 on the ceratohyal, 4-5 external and 2 ventral. 183 Operculars: Complete and entire. In Umbra suboperculum only extending slightly up the posterior margin of the operculum, in Novumbra, extending about half way up the posterior margin. In both the operculum extends wedge-like into the hollow of the boomerang-shaped suboperculum. G i l l membranes separate. Hyoid arch: Interhyal, epihyal, ceratohyal and two hypohyals in Umbra and Novumbra. In Umbra the dorsal edge of the ceratohyal sends two prongs towards the hypohyals, the epihyal is about 2/5 the length of the ceratohyal. References: Chapman (1934), Dineen and Stokely (1954)* Berg (1949). Material examined: Umbra l imi, 5 alizarin specimens, NMC60~486-A, from Bruce Co., Ontario; 5 alizarin specimens, NMC62-135 from Ottawa, Ontario; 5 specimens, BC59-199, Silver Lake, Ontario. Superfamily Dallioidea Dalliidae Branchiostegals: Vary from 7-8 in Pallia, with 3 epihyal and 5 ceratohyal, 5 on the external and 3 on the internal face of the arch. A l l acinaciform as in Esocidae. Operculars: Complete and entire. Suboperculum extends up behind about 4/5 of the posterior border of the operculum. The operculum extends down between the two wings of the suboperculum. G i l l membranes separate. Hyoid arch: Consists of interhyal, epihyal, dumbell-shaped ceratohyal and 2 hypohyals. Dorsal edge of anterior end of the ceratohyal sends 2 prong3 towards the hypohyals. The epihyal is about 1/3 the length of 184 the ceratohyal. Referencest Berg (1949), Chapman (1934) Material examined: Pallia pectoralis, 2 alizarin specimens, NMC62~244, from tributary to Tokotna River, Alaska. ORDER MYCTOPHIFORMES (CETUNCULI, MIRIPINNATl) Branchiostegals 6-26, with 2-9 on the epihyal, 3-14 on the ceratohyal, 0-2 on the hypohyal, slender and attenuate. Gular absent. Two hypohyals, ceratohyal, epihyal and interhyal present. Ceratohyal and epihyal not sutured. Opercular bones complete and entire, never with spines. G i l l membranes separate. Upper Cretaceous to present. Seventeen living and 4 fossil families making a total of 21. These are grouped in four sub-orders. The high number of branchiostegal rays and high number of branchiostegals on the epihyal, the presence of two supramaxillaries, two postterminal centra, two hypohyals, a high number of pelvic rays, the presence of 17 branched caudal rays and 3 epurals in the primitive members of the family indicate that the Myctophiformes must have arisen from an ancestor at least as primitive as the Clupeoidei or (Ctenothrissoidei) and perhaps from the Elopoidei. If the Aulopidae, as stated by some authors, have fulcral scales above the caudal then they must be derived from a group at least as primitive as the Elopoidei. The Myctophiformes thus form a branch, about equivalent and parallel to the Clupeiformes. Certain of the order possess a branchiostegal pattern which is similar to the acanthopterygian pattern (Myctophidae, Neoscopelidae, Alepisauroidei). 185 Marshall (1955) has arranged the myctophiforra fishes in two sub-orders, the Myctophoidei and the Alepisauroidei. This arrangement agrees fairly well with what is known with the branchiostegal series. It must be admitted that data are lacking for certain of the families, most of which are deepsea forms, rare in collections. However, the following characterizations may be made at the present: Myctophoidei: subopercle and branchiostegals curl up around behind operculum exluding much of its posterior border (except Myctophidae and Neoscopelidae); branch-iostegals either a l l on the ventral or a l l on the external face of the hyoid arch (except Myctophidae and Neoscopelidae); branchiostegals 7-26. Alepisauroidei: suboperculum and branchiostegals do not curl up around posterior border of operculum, the posterior branchiostegals on the external, the anterior branchiostegals on the internal face of the hyoid arch, branchiostegals 6-9• That a gas bladder, an orbitosphenoid, 2 instead of 1 post-terminal centra may be found in some of the Myctophoidei but not in the Alepisauroidei confirms the indication of the number of branchiostegals that the Myctophoidei are more primitive. The branchiostegals series also confirm the distinctness of the two suborders. The whalefishes, comprising the three families or subfamilies, the Cetomimidae, Rondeletiidae and Barbourisiidae have been accorded different ordinal placement. Goode and Bean (1896) placed them in the Iniomi (*» Myctophiformes), Jordan (1923) in the Cetunculi, Parr (1929, 1945), distributed them among the Xenoberyces (Rondeletiidae) and Iniomi (Cetomimidae, Barbourisiidae), Bertin and Arambourg (1958) in the Stephanoberyciformes, Berg (1947) the Cetomimidae in the Myctophiformes 186 and the Rondeletiidae in the Stephanoberyciformes. Harry (1952) suggested relationship between the whalefishes and the Saccopharyngiformes. There has also been disagreement on the status of the whalefish families, Parr (1929, 1945) and Harry (1952) granting them familial status while Myers (1946) considered Barbourisiidae a subfamily of Cetomimidae. The latest author (Harry, 1952) is followed in according the three groups familial status. Unequivocal evidence for placement of these families is slight. Cetomimidae and Barbourisiidae possess non-protractile premaxillaries, a condition most