2 is VI 25 of. br Boge foe ES Ree ae ¥ Os 0 pt 5, Caitg eee Jd y The person charging this material is re- sponsible for its return to the library from which it was withdrawn on or before the Latest Date stamped below. Theft, mutilation, and underlining of books are reasons for disciplinary action and may result in dismissal from the University. UNIVERSITY OF ILLINOIS LIBRARY AT URBANA-CHAMPAIGN BUILDING use ot L161—0O-1096 Mt Ll a ‘WAY | H ' x } ite ' a , A Ik iD a) ay AG , } ; ATs ¢ | ‘ f § Digitized by the Internet Archive In 2011 with funding from University of Illinois Uroana-Champaign http://www.archive.org/details/comparativeosteo30liem The Comparative Osteology and Phylogeny of the Anabantoidei (‘Teleostei, Pisces) KAREL F. LIEM ILLINOIS BIOLOGICAL MONOGRAPHS: Number 30 THE UNIVERSITY OF ILLINOIS PRESS URBANA, 1963 $3.50 tained four issues each and were available a subscription: span wie ang : 25 (issued in 1957), each publication is numbered consecutively. No subscriptions are available, but standing orders will be accepted for forthcoming numbers. Prices of previous issues still in print are listed below, and these may be purchased ee the | h be addressed to the Exchange Department, ‘University Library, lisbane Hlinois, Lit ii RBG BAKER, FRANK COLLINS (1922): The Mol- luscan Fauna of the Big Vermilion River, Illinois, with Special Reference to Its Mod- ification as the Result of Pollution by Sew- age and Manufacturing Wastes. 15 pls. Vol. 7, No. 2. $1.25. BALDUF, W. V. (1959) : Obligatory and Facultative Insects in Rose Hips. 12 pls. No. 26. $3.50. CREGAN, SISTER MARY BERTHA (1941): Generic Relationships of the Dolichopo- didae (Diptera) Based on a Study of the Mouth Parts. 30 pls. Vol. 18, No. 1 FISHER, HARVEY I., and GOODMAN, DONALD c. (1955): The Myology of the Whooping Crane, Grus americana. 40 figs. Vol. 24, No. 2. $2. 50. GAMBILL, WILLIAM G., (1953): The Leguminosae of Illinois. "Vol 22, No. 4. $3.00. GOODNIGHT, CLARENCE JAMEs (1940): The Branchiobdellidae (Oligochaeta) of North American Crayfishes. 3 pls. Vol. 17, No. 8, $1.00. GUTBERLET, JOHN EARL (1915): On the Osteology of Some of the Loricati. 5 pls. Vol. 2, No. 2. $.50. HEISS, ELIZABETH M. (1938): A Classifica- tion of the Larvae and Puparia of the Syrphidae of Illinois, Exclusive of Aquatic Forms. 17 pls. Vol. 16, No. 4. $1.50. HIGLEY, RUTH (1918): ' Morphology and Biology of Some Turbellaria from the Mississippi Basin. 8 pls. Vol. 4, No. 8. $1.25. 1. $1.00. $4.00. Bon ‘AN os fs HOFFMEISTER, DONALD F. (1951): A at nomic and Evolutionary Study of orn Pifion Mouse, Peromyscus truei. 5 pls. ge uy ah Vol. 21, No. 4. gt es Mei. 50. pe _——— and GOODPASTER, ‘WOODROW. we i (1954): The Mammals of the Huachuca — Mountains, Southeastern Arizona. 27, ae 4 Vol. 24, No. 1. $3.00. a RUN Papillose Allocreadiidae—A Study of Their Morphology, Life Histories, and reac HOPKINS, SEWELL HEPBURN (1934): The | ‘ ships. 4 pls. 6 figs. Vol. 18, No. 2. $1, 00. r HUMES, ARTHUR GROVER ( 1942): ‘The Morphology, Taxonomy, and Bionomics of _the Nemertean Genus Ce 4 pls. 1 map. Vol. 18, No. 4. $1.50. if KENDEIGH, S. CHARLES (1941): Territorial i and Mating Behavior of the House Wren. iN 32 figs. Vol. 18, No. 3. $1.50. i ae (1952): Parental Care aul Its Evo. | lution in Birds. 85 figs. Vol. an Nos, 13. : KRAMER, soL (1950): The Morphiloge a ¥ and Phylogeny of Auchenorhynchous _ Homoptera (Insecta). 6 charts. 16 a Vol, by 20, No. 4. $2.00. 5; BR KUDO, RICHARD ROKSABRO (1944): “Mone phology and Development of Nosema notabilis Kudo; Parasitic in Sphaerospora — polymorpha Davis, a Parasite of Opsanus tau and O. beta. 12 pls. 7 figs. Nate My . No. 1. $1.25. ir MEGLITSCH, PAUL A. (1940): Cytological cy, Observations on Endamoeba blattae. 8 Spl. a4 Vol. 17, No.4. $1.00. aS THE COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI (TELEOSTEI, PIscEs ) The Comparative Osteology and Phylogeny of the Anabantoidei (Teleostei, Pisces) KAREL F. LIEM ILLINOIS BIOLOGICAL MONOGRAPHS: Number 30 2 Se Ee THE UNIVERSITY OF ILLINOIS PRESS URBANA, 1963 Board of Editors: FRANCIS J. KRUIDENIER, THEODORE DELEVORYAS, R. D. DE MOSS, J. G. STERNBURG, AND AUBREY B. TAYLOR. THIS MONOGRAPH IS A CONTRIBUTION FROM THE DEPARTMENT OF ZOOLOGY, UNIVERSITY OF ILLINOIS. DISTRIBUTED DECEMBER 1, 1962. © 1963 BY THE BOARD OF TRUSTEES OF THE UNIVERSITY OF ILLINOIS. MANU- FACTURED IN THE UNITED STATES OF AMERICA. LIBRARY OF CONGRESS CATA- LOG CARD NO. 63-7249. ACKNOWLEDGMENTS I am especially grateful to Professor George S. Myers for suggesting the topic and criticizing the manuscript, to Professor Hobart M. Smith for guidance throughout execution of the study, to Dr. Walter J. Bock for suggestions and encouragement, to Miss Alice Boatright for her help in arranging the plates, and to Miss Nancy L. Roth, who has patiently made and edited the legends for the illustrations, and has given continuous encouragement during the final stages of preparation of this manuscript. I have benefited immeasurably from numerous other authorities and friends with whom certain aspects of this study have been discussed orally or by correspondence, or from whom I have received information, radiographs, reprints, or assistance in connection with publications in- accessible to me. They include Dr. F. B. Adamstone, Dr. Robert S. Bader, Dr. Ronald A. Brandon, Mr. John C. Cralley, Dr. Warren C. Freihofer, Dr. William A. Gosline, Dr. S. Charles Kendeigh, Dr. C. Kosswig, Dr. Francis J. Kruidenier, Dr. Edward M. Nelson, Dr. C. Ladd Prosser, Dr. H. Steinitz, Dr. Stanley Weitzman, Mr. Kenneth L. Williams, and Mr. Loren P. Woods. Through the courtesy of numerous institutions and curators, 123 speci- mens have been available for osteological study. Without this aid the present study would have been impossible. My gratitude is acknowl- edged to Dr. M. Boeseman and Dr. L. D. Brongersma, Ryksmuseum van Natuurlyke Historie, Leiden, Holland; Dr. V. FitzSimons, Trans- vaal Museum, S. Africa; Dr. J. Hewitt, Albany Museum, S. Africa; Dr. A. C. Hoffman, The National Museum, S. Africa; Dr. R. A. Jubb, Rhodes University, S. Africa; Dr. George S. Myers, Stanford University; Dr. J. H. Skinner, University of Stellenbosch, S. Africa; Dr. J. L. B. Smith, Rhodes University, S. Africa; the late Margaret H. Storey, Stanford University; and Mr. A. M. R. Wegner, Museum Zoologicum Bogoriense, Indonesia. CONTENTS DESCRTERIVE OSTEOLOGY 6% 45s 45snadsbbs ei so wd we we arn ye ole Seer ie SMe BeCLORA GOUT e es iatg sh 5.4. ie 2H Sle at faa he ites eh oa AMIDE AAs CATICUIC) Saye So 9 ca eute 1e, & sy ee er igs, MRE CLOLAI GING es way..4 Sere os alert ed oe dela A See APM eI: CIC CT oa boo MNS haves So -ae dec WR oe ete bana Sates Meee SBE ONE NUDAE var sects c)-n Ans host oan ise echcacree eee ates MPa NG MIO CEATINUING etme, otis a6 rai. a, dole ice & a acelin HOM sce ances oe oe oe DR DTANCMIOCIAMNUMI. «sso oid «cals on systoko dios eeu -oraepeee SRE CLORAINOGIEG eC a. as, sca sit ideo endveseh oo) aagieey perenne ae APM e wale Cans Car schslagh ie ree scue tad nhores w Bho wale S.pogedens Oe Sector lesindle st h.l AeA a.) sm ob Ava lenn ow Mores eee AMEE LVA@ irl eV say. tea SoN Se ata, OMe a2 inchs widen eee eee VERTEBRAL COLUMN AND UNPAIRED APPENDAGES............. > OMOWHUED poo coco n boom Oo Kb OC OOK GOD OOaaDOHODD ADDO KGHDOODDD SCANWIS obo compo o ob ood T OOD DOO GOD OOa DOO HDOOSD ODO BOD DODO UDR IS CMISSTON 2 ere eens he os A reek Seis Dis whicesuants cea ots A. INTERRELATIONSHIPS AND DISTINCTIONS OF Mamiily ab elorturdaetm: Waris es ces feja.n is oaks teae one: occhmee e Family /‘Osphronemidae Bleeker. 22. i. h52 b35 238m oe Family Helostomatidae Jordan = 2c. hsisclea ties wari ore oe Vii NW Re — WwNwwnvw bd bv OU wo Ww bo Pee Oo wo nN 36 INTERRELATIONSHIPS AMONG ANABANTID GENERA............ 42 B. C. INTERRELATIONSHIPS AMONG BELONTIID GENERA............. 44 DD: PHYLOGENETIC TRENDS IN THE ANABANTOIDEI..............- 49 E. THE PHYLOGENY OF THE ANABANTOIDEI..............0..005 61 THE Status oF Luciocephalus pulcher (Gray)...............-- 69 RELATIONSHIPS OF THE ANABANTOIDEI AND THE OPHICEPHALIFORMES «6 «2% 2% Ghee sees ss sos one al: CONCEUSIONS ics os cot rari So acors eee elles onal Pe 73 LLTERATURE GIleD., cob eo Science Oe oe eee 75 PA TES cytotec kee kei Re eae Aerie ASR Se 79 TINIDEXC Sap oreo ete dos oa pee rca Rene ahh a Ae ee 141 Vili INTRODUCTION The anabantoid fishes have been grouped with the ophicephalid fishes in a separate order Labyrinthici by many early workers: Day (1876), Regan (1909), Weber and de Beaufort (1922), Jordan (1923), De- raniyagala (1929), Gregory (1933), and Smith (1945). Boulenger (1904) did not recognize the order Labyrinthici, but placed the family Anabantidae in the order Percesoces whereas the families Osphro- nemidae and Luciocephalidae were included in the order Perciformes. Berg (1940 and 1958) made radical changes in the systematics of the anabantoid fishes, placing the ophicephalids in a separate order Ophi- cephaliformes whereas the anabantoids and luciocephalids were classi- fied as two separate suborders of the order Perciformes. Recent authors, for example Forselius (1957), Bertin and Arambourg (1958), Darling- ton (1957), and Herald (1961) tend to follow Berg's classification, although most of our knowledge of the taxonomy of the anabantoids is based on the monographs of Weber and de Beaufort (1922), Bou- lenger (1916), and Barnard (1943). Further knowledge has gradually accumulated through faunal studies, nomenclatural revisions, or reports on isolated collections, for example Day (1876), Regan (1909), Myers (1923 and 1926), Herre (1924), Deraniyagala (1929), Prashad and Mukerji (1929), Shih (1936), Krull (1939), Smith (1945), Nichols (1943), and Tweedie (1952). The only osteological studies available are incomplete and scattered. Gill (1893) figured and discussed the hyobranchial apparatus of Os- phronemus goramy, and Gregory (1933), with some inaccuracies, the skulls of Anabas and Osphronemus. Berg (1958) presents a lateral view of the skull of Anabas, and Schneider (1941) a lateral view of the neurocranium of Macropodus in her discussion on hearing in ana- bantids. Zograff (1888), Henninger (1908), and Bader (1937) studied the comparative morphology of the “labyrinth organ” in great detail. The functional anatomy of the pelvic girdle of Trichogaster is discussed by Steinbach (1950). Marathe (1959) has described the cranial os- teology of Trichogaster trichopterus. Starks (1926) studied the ethmoid region in Anabas and Ophicephalus. The osteology of the Ophicephali- formes has been studied in great detail by Day (1914), Bhimachar (1932), and Srinivasachar (1955). The objectives of this study may then be stated as follows: 1. To present an illustrated description of the osteology of the repre- I 2; COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI sentative families, subfamilies, and genera of the anabantoids. The illus- trations will supply future students of both recent and fossil acanthop- teran fishes with a basis for comparison. 2. To reclassify the supergeneric groups of anabantoid fishes into well-defined families and subfamilies on osteological grounds. 3. To postulate the phylogeny of the anabantoids in the light of osteo- logical evidence, fossil records, and zoogeography. 4. To analyze evolutionary trends in several morphological units of the anabantoids and their adaptive significance. 5. To redefine the suborder Anabantoidei and to reallocate Lucio- cephalus. 6. To compare the Anabantoidei, Ophicephaliformes, and Lucio- cephaloidei. In the absence of large series of specimens, a revision of taxonomy at the specific level has not been attempted, but a basis for anabantoid phylogeny and classification has been proposed for future workers in teleost morphology, physiology, and ethology. The phylogenetic position of the Anabantoidei and Luciocephaloidei in relation to the other members of the Perciformes remains problem- atical. MATERIAL AND METHODS Detailed studies of the osteology were confined to a total of 123 specimens belonging to 26 representative species of 14 genera (Table 1). Malpulutta, Parasphaerichthys, and Parosphromenus have been de- scribed from a very limited number of specimens and are the only genera not available for osteological studies. The absence of these three genera does not seriously limit phylogenetic considerations since they are closely related and seemingly very similar to other genera ade- quately represented. All specimens have been cleared and the bones and calcified cartilage stained with sodium alizarin sulfonate (Davis and Gore, 1936). Radiographs of several specimens have also been made. The specimens were dissected and disarticulated under a stereo- scopic microscope. The drawings were made freehand at successive stages of dissection, taking great care to show correct proportions. The nomenclature for the skull bones is that of Harrington (1955) and Devillers (1958). To facilitate the descriptions and illustrations the skull has been subdivided into two highly arbitrary divisions: the neuro- cranium and branchiocranium, as defined by Gregory in 1933. The nomenclature of Merriman (1940) has been followed for the postcranial axial skeleton and pectoral girdle. The homology and terminology for the different processes of the pelvic girdle are based on Shelden (1937) and Sewertzoff (1934). Of these, the later proved the most useful. INTRODUCTION TABLE 1. L IST OF SPECIMENS STUDIED. Museum * Number of Taxa number Locality specimens Anabantidae Anabas testudineus (Bloch) UIMNH u 8 ae MZB 311 Celebes Be Hs MZB 318 Java 1 Ctenopoma muriet (Blgr.) SNHM 51345 Uganda 2 Ctenopoma multispinis (Peters) RUDI Victoria Falls 3 Sandelia bainsii (Castelnau) RUDI S. Africa 3 a TVM 8742 S. Africa 2 Sandelia capensis (Cuv. & Val.) USZI S. Africa d oe RUDI S. Africa 6 ne TVM 19821 S. Africa 2 Osphronemidae Osphronemus goramy Lacépéde MZB 433 Java 1 Helostomatidae Helostoma temmincki Cuv. MZB 510 Java 2 es AR — 4 Belontiidae Belontia hasselti (Cuv. & Val.) RNHL 10490 _- 1 Belontia signata (Giinther) SNHM 30172 Ceylon 1 Betta anabatoides Bleeker RNHL 7902 — 1 Betta picta (Cuv. & Val.) MZB 293 Situ Gunung 1 Betta splendens Regan AR — 6 Betta taeniata Regan MZB 296 Riouw 1 Betta unimaculata (Popta) MZB 294 Borneo 1 Colisa fasciata (Bloch & Schneider) AR — 6 Colisa labiosa (Day) AR — 6 Colisa lalia (Hamilton Buchanan) AR — 6 Macropodus opercularis (L.) AR = 6 Macropodus chinensis Bloch RNHL 17408 — 1 Macropodus cupanus (Cuv. & Val.) AR — 6 Sphaerichthys osphromenoides Can. RNHL 20897 — 2 rs SNHM 39450 Malaya 2 Trichopsis pumilus (Arnold & Ahl) RNHL 20980 — 1 Trichopsis vittatus (Cuv. & Val.) MZB 292 Java 5 fe MZB 297 Banda 2 RNHL 16883 — 1 Trichogaster leeri (Bleeker) AR — 6 Trichogaster trichopterus (Pallas) AR — 6 * MZB 321 Pager Alam 3 i MZB 317 Pager Alam 7 Luciocephaliformes Luciocephalus pulcher (Gray) RNHL 6371 — 1 2 SNHM 35689 Malaya 2 Ophicephaliformes Ophicephalus striatus Bloch RNHL 11875 — 2 4 AR: aquarium raised; MZB: Museum Zoologicum Bogoriense, Java; RNHL: Ryksmuseum van natuurlyke historie, Leiden; SNHM: Stanford Natural History Museum; RUDI: Rhodes University Department of Ichthyology; TVM: Transvaal Museum, S. Africa; UIMNH: Uni- versity of Illinois Museum of Natural History; USZI: University of Stellenbosch Zoological Institute. 4 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI Since otoliths have distinct patterns in suborders, families, or even genera, they were of some value in the study of interrelationships of the Anabantoidei. The otoliths of nearly all genera have been removed and studied under a binocular microscope after softening the bone with potassium hydroxide. The nomenclature of Adams (1940) has been followed closely. DESCRIPTIVE OSTEOLOGY A. ANABANTIDAE 1. NEUROCRANIUM Nasal region (figs. 1-3, 14-16, 40-42). The nasals in Anabas and Cten- opoma are large bones, suturally united to each other and overlying the ethmoid. In Sandelia the nasals are narrow and elongate, separated from each other by a fontanelle. The prefrontal and lateral ethmoid are fused as a complex bone sep- arating the nasal and orbital regions. Right and left bones are widely separated by the ethmoid. Ventrally they rest on the parasphenoid. The olfactory foramen is located near the median border of the prefrontal- lateral ethmoid complex. There are no palatine facets, although the palatine articulates against the lateral surface of the complex bone and the prevomer. The single ethmoid is a triangular cap interposed between the pre- frontal ethmoids, widely separating them. The front of the triangle slopes anteroventrad toward the prevomer. The prevomer tapers posteriorly as a point entering a shaft of the parasphenoid. Two rows of well-developed conical teeth are present on the anteroventral surface of the prevomer. Dorsomedially the bone is raised to a blunt ridge. Orbital region (figs. 1-3, 14-16, 27-30, 40-42). The paired frontals form a large portion of the roof of the cranial cavity and the central roof and margin of the orbital chamber. In Anabas they are joined by a dentate suture, whereas in Sandelia a fontanelle separates the anterior halves. Ctenopoma and Sandelia exhibit extensions of the frontals em- bracing the supraoccipital, which is not the case in Anabas. The frontals in Anabas and Ctenopoma possess on the ventral surface prominent orbital processes, articulating with the pterosphenoids with which they form an incomplete interorbital septum. The pterosphenoid forms a ventral keel on the frontal and contributes to the posterior border of the orbit. It articulates with the sphenotic, prootic, and basisphenoid. Left and right pterosphenoids do not meet. The basisphenoid is well developed and supported directly by the pterosphenoid and prootic. Its ventral process is joined to the para- sphenoid in Anabas and Sandelia, but not in Ctenopoma. The lachrymal (suborbital 1) is an elongate bone bordering the an- teroventral portion of the orbit. It articulates with the prefrontal-lateral 5 6 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI ethmoid by means of a medial spur. In Anabas the lachrymal is bordered by long radiating spines; in Ctenopoma muriei it is very weakly ser- rate, whereas in Ctenopoma multispinis and Sandelia there is no trace of any serration. There are five suborbitals (including the lachrymal) in Anabas and Ctenopoma. This number has been reduced to four in Sandelia. A very well-developed suborbital shelf is formed by promi- nent medial projections of all suborbitals in the three representative genera. In Anabas suborbitals 2-4 are very large and suturally united with the preopercular. Ctenopoma muriei shows an intermediate stage, in which the suborbitals are broad plates, reaching to nearly half the distance between orbit and preopercular. Ctenopoma multispinis and Sandelia exhibit a greater reduction, the suborbitals being narrow elongate bones. Otic region (figs. 1-3, 14-16, 27-30, 40-42, 52-54). The sphenotic is a trapezoid plate bordered by the frontal anteriorly and medially. Pos- teriorly it is joined by the supratemporal. The sphenotic does not meet the parietal and rests ventrally on the pterosphenoid and prootic. The parietals in Anabas are elongate bones, approximately of the same length as the supraoccipital, whereas in Ctenopoma and Sandelia they are half as long. The parietal in Anabas and Ctenopoma is joined me- dially with the supraoccipital, anteriorly with the frontal, laterally with the supratemporal, and posteriorly with the suprascapular. In Sandelia, however, the posterior edge of the parietal is bordered mainly by the posttemporal fossa; the extrascapular articulates only with the extreme posterior corner of the parietal. The extrascapular is best developed in Anabas and Ctenopoma, cover- ing the epiotics completely. It articulates anteriorly with the supra- temporal and parietal, and medially with the supraoccipital. The post- temporal fossa is absent in Anabas and Ctenopoma. In Sandelia the extrascapular is somewhat reduced. It overlaps the epiotic partially and does not reach the supraoccipital. A posttemporal fossa is present in Sandelia. The supratemporal is an elongate plate forming nearly half of the lateral border of the cranium. It is bounded anteriorly by the sphenotic and frontal, medially by the frontal and parietal, posteriorly by the extra- scapular, and ventrally by the exoccipital, prootic, and intercalary. The prootic is an irregularly shaped bone with an anterolaterally di- rected ridge extending from near its cranial border. The ventral half of this ridge forms the lateral wall of the myodome and also provides a slot into which the ascending process of the parasphenoid fits. Anteriorly the prootic is sutured to the pterosphenoid, with the foramen of the trigeminal nerve partly intervening, and to the dorsal portion of the basi- sphenoid. Dorsally the prootic is united to the sphenotic and supra- DESCRIPTIVE OSTEOLOGY i temporal, and posteriorly to the basioccipital, exoccipital, and in Anabas also to the ventral keel of the supraoccipital. Ventrally it is bordered by the parasphenoid. There is a very prominent bulla acustica or saccular swelling. The suture between the prootic and basioccipital is dentate in Anabas and irregular in Ctenopoma and Sandelia. The epiotic in Anabas and Ctenopoma is completely, in Sandelia only partly, overlain by the extrascapular. The bone is cone-shaped, forming the posterodorsal corner of the cranium. Dorsally it possesses an articula- tion facet for reception of the posttemporal. Medially the epiotic articu- lates with the supraoccipital, anterolaterally with the supratemporal, and posteroventrally with the intercalary and exoccipital. The epiotics do not reach the foramen magnum. The posttemporal fossa is absent in Anabas and Ctenopoma. The fossa is present in Sandelia, although rather small and shallow. The intercalary (opisthotic) is a cone-shaped bone forming in Anabas and Ctenopoma the posteroventral corner of the cranium. In Sandelia the intercalary is a small plate situated between the posterior extremity of the supratemporal and the lateral wing of the exoccipital. Dorsally the intercalary is covered by the extrascapular. The posttemporal ar- ticulates by means of its lower fork with the intercalary in all three genera. The supraoccipital is a hexagonal bone. There is no supraoccipital crest, except for a median keel on the posterior aspect of the skull. The supraoccipital does not reach the foramen magnum. In Anabas the supra- occipital possesses a very prominent ventral keel which articulates with the exoccipital and prootic, enlarging the suprabranchial chamber (fig. 14). The exoccipital is connected posterodorsally to the epiotic and inter- calary, anterodorsally to the supratemporal, anteriorly to the prootic, and ventrally to the basioccipital. A large foramen exoccipitale is lo- cated on the lateral wall of the exoccipital. In life this foramen is cov- ered by a tympanumlike membrane. It is also a passage for the glosso- pharyngeal nerve. The vagus nerve emerges from a smaller foramen just posterior and dorsal to the foramen exoccipitale. The posterior surface of each exoccipital forms a rounded condyle for articulation with the first vertebra. Dorsal wings from the exoccipitals unite above the small foramen magnum, forming an arch over the spinal cord. The suture be- tween the exoccipital and epiotic is dentate in Anabas, partially dentate in Ctenopoma, and straight in Sandelia. The posttemporal in Anabas is comparatively small and slightly forked, whereas in Ctenopoma and Sandelia the forks are well differentiated. The upper fork articulates directly with the epiotic. The lower fork is connected to the intercalary. 8 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI Basicranial region (figs. 40-42, 52-54). The basioccipital in Anabas is bordered ventrally by the parasphenoid to a greater extent than in Ctenopoma and Sandelia. The posterior end forms a condylar articula- tion with the first vertebra. The basioccipital forms nearly the whole posterior half of the otic capsule. Near its posterior end a ventrolateral pharyngeal process is differentiated at either side of the basioccipital, functioning as a support for the hyobranchial apparatus. From the posterior aspect (figs. 52-54) these pharyngeal processes of the basi- occipital appear as horizontal wings. Ventrally the basioccipital is bor- dered by the parasphenoid, anteriorly by the prootic, and dorsally by the exoccipital. The parasphenoid extends from the prevomer to near the posterior extremity of the basioccipital, underlying the greater part of the basi- occipital. Dorsally the parasphenoid sends a process to form the floor of the myodome. Near its posterior end the parasphenoid possesses very characteristic toothed transverse processes. In Anabas the conical teeth are very large, arranged in single rows of five. On the body of the parasphenoid five large teeth are present in a distinct pattern (fig. 40). In Ctenopoma and Sandelia the number of teeth has increased; in the latter genus the teeth are arranged in two rows (figs. 41, 42). 9. BRANCHIOCRANIUM Hyobranchial region. The interhyal is a small rod-shaped bone con- necting the posterodorsal margin of the epihyal to the cartilaginous block between the symplectic and the hyomandibular (figs. 76-78). The epihyal is triangular and joined loosely by a straight suture to the elon- gate ceratohyal. The dorsal and ventral hypohyals attach to the cera- tohyal with an irregular suture. The unpaired basihyal is rod-shaped, and its posterior end joins the apices of the dorsal parts of the hypohyals. The three basibranchials lie in the mid-ventral line of the pharynx. The first is situated between the hypohyals and extends anteriad ventral to the basihyal. The second basibranchial is hourglass-shaped, whereas the third is elongate. The anterior ends of the first and second hypo- branchials curve mesad and articulate with the constricted side of the second and third basibranchials. The third hypobranchial in Anabas is a large plate articulating with the third basibranchial, hypobranchial, and fourth ceratobranchial. In Ctenopoma and Sandelia the third hypo- branchial is a curved bone. The first four ceratobranchials are slender, curved, rod-shaped bones. The fifth ceratobranchials approximate each other and are covered dorsally by numerous teeth, which are arranged in ten to fourteen rows in Anabas, ten rows in Ctenopoma, and six to seven rows in Sandelia. In Anabas the first epibranchial is modified into a labyrinthine suprabranchial organ. The suprabranchial organ is less DESCRIPTIVE OSTEOLOGY 9 complex in Ctenopoma, consisting of a stem and two folded lamellae. In Anabas the stem carries at least three folded lamellae. The first epi- branchial in Sandelia is merely an enlarged but unfolded bony plate. The second and third epibranchials are rod-shaped. The third epi- branchial in Anabas possesses a dorsal process. The fourth epibranchial in Anabas is a trapezoid bone. It is a more slender and rod-shaped bone in both Ctenopoma and Sandelia. The first pharyngobranchial is want- ing. The second is a curved element bearing four teeth on the ventral surface. Pharyngobranchials 3-4 are fused indistinguishably; the ventral surface is covered with well-developed teeth arranged in four transverse rows. The urohyal is a thin elongate bone with a longitudinal ridge. Six branchiostegal rays are present in Anabas and Ctenopoma, and five in Sandelia. In Anabas the two most posterior rays articulate with the epihyal; in Ctenopoma only the sixth ray joins the epihyal. The last branchiostegal ray in Sandelia is situated between the epihyal and ceratohyal. Opercular series (figs. 27-30). The opercular has a nearly vertical anterior border which joins the posterior border of the preopercular along its entire length. The posterier border of the opercular in Anabas is deeply serrate and has two well-developed spines. In Ctenopoma the degree of serration has decreased, whereas in Sandelia there are two distinct serrate patches, each with approximately five teeth. The ventral border is at a slight angle from the horizontal. Auricular and articular processes are wanting (figs. 27-30). The subopercular has a dorsal border which corresponds to the ven- tral border of the opercular by which it is overlapped laterally. A rather short process is found at the anterodorsal corner. The posterior and ventral borders are deeply serrate in Anabas, serrate in Ctenopoma, and smooth in Sandelia. The interopercular overlaps laterally with its posterior border the anterior end of the subopercular. The posteroventral border is deeply serrate in Anabas, faintly in Ctenopoma. In Sandelia there is just a slight indication of serration on the ventral border. The preopercular is crescent-shaped. The element widens in the bend, except in Anabas. Anabas is unique in having the anterior border of the preopercular suturally united with postorbitals 2-4. Oromandibular region (figs. 64-66). The oromandibular bones are movably suspended from the neurocranium by the palatines anteriorly and by the hyomandibulars posteriorly. The palatine is a rather narrow bone bearing teeth along its antero- ventral border in Ctenopoma and Sandelia. There are no teeth on the palatine of Anabas. A slender ethmoid process originates at the antero- dorsal corner of the palatine. This ethmoid process is joined to the lat- 10 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI eral ethmoid by means of a ligament. Ventrally the palatine is joined to the ectopterygoid and entopterygoid. The ectopterygoid is a slender toothless bone tapering to a point at its anterodorsal end. It articulates with the palatine, entopterygoid, and quadrate. The entopterygoids are large elements joined medially by a mem- brane to form the roof of the buccal cavity. The entopterygoid is bounded by the palatine, ectopterygoid, quadrate, and metapterygoid. The metapterygoid is the largest of the cheek bones. Posterodorsally it is intimately joined with the symplectic process of the hyomandibular. Anteroventrally it is bounded by the entopterygoid and quadrate. Pos- teroventrally it articulates with the symplectic. The symplectic is a long curved bone the distal half of which rests on the posterior process of the quadrate. In Anabas and Ctenopoma the dorsal end of the symplectic articulates with the symplectic process of the hyomandibular, whereas in Sandelia it joins the block of cartilage which lies between the symplectic process and interhyal. The quadrate possesses a condyle on its anteroventral corner for ar- ticulation with the angular. The posterior margin is deeply incised with a slot into which the symplectic is wedged. The spinelike process caudal from the condylar region receives on its ventral surface the preoper- cular. Dorsally it articulates with the ectopterygoid, entopterygoid, and metapterygoid. The dorsal surface of the hyomandibular is faintly differentiated into an anterior and posterior head, which articulate respectively with the sphenotic and supratemporal. A short posterior process affords an ar- ticulating surface for the opercular. The premaxillary forms the anterior border of the mouth. The ascend- ing process of the premaxillaries is variable in length: one-fifth of the body of the premaxillary in Ctenopoma, one-half to one-third in San- delia, and one-third in Anabas. The processes from both sides are more or less closely conjoined in the midline and fit into a notch formed by the nasals, ethmoid, and palatines. Articular processes are wanting. The ventral surface of the premaxillary bears several rows of teeth. The toothless maxillary does not contribute to the edge of the mouth. Its narrow anterior end forms a prominent notch. The rounded inner fork slides on the prevomer. The maxillary carries the ascending process of the premaxillary with the groove between the rounded inner fork and slender (in Anabas short and rounded) outer fork. The angular is lodged in the posterior notch of the dentary. It lacks an ascending process. The dentary bears teeth on the anterior three-fourths of its dorsal surface. The ascending process of the dentary appears merely as a rounded corner. DESCRIPTIVE OSTEOLOGY 1 The retroarticular is a small irregular-shaped bone attached to the posteroventral corner of the angular, ventral to the condyle. 3. PECTORAL GIRDLE (figs. 91, 92) The supracleithrum is more or less spatulate. It is partly covered by the ventral corner of the posttemporal in Ctenopoma and Sandelia, but not in Anabas. The supracleithrum overlaps the dorsal portion of the cleithrum. The cleithrum is the largest element. Its ventral extension curves craniad and joins its partner on the opposite side by a short ligament, forming part of the floor of the hyobranchial apparatus. It possesses a very well-developed ala laminaris externa, concealing part of the post- cleithrum, scapula, and coracoid from a lateral view. The postcleithrum is divided into a dorsal and ventral portion. The dorsal one is elliptical, whereas the ventral element is elongate and tapers into a point ventrally. The scapula fits into a groove on the inner ridge of the cleithrum. There is a large foramen. Three and a half radials articulate with its posterior margin. The coracoid consists of a more or less rectangular dorsal portion articulating with the inner surface of the cleithrum and an elongate ventral portion which terminates in a short spine and extends anteriad to meet the ventral end of the cleithrum. 4, PELVIC GIRDLE (figs. 91, 92) The two basipterygia are firmly united across the midline, forming a bilaterally symmetrical unit. There is a lateral ridge, the rachis, running more or less longitudinally along each basipterygium. Medial to this ridge arises another winglike process, the processus medialis anterior, which is longer in Ctenopoma and Sandelia than in Anabas. Originating from the posteromedial corner of the basipterygium are two distinct processes: an anteriorly directed processus medialis and a posteriorly directed processus medialis posterior. A prominent processus lateralis is present at the posterolateral corner of the basipterygium in Anabas; in Ctenopoma and Sandelia it is very reduced. The conjoined anterior tips of the basipterygia are attached to the cleithral symphysis by means of a ligament. B. BELONTUDAE 1. NEUROCRANIUM Nasal region (figs. 6-11, 19-26, 45-51). The shape of the nasals varies considerably among the different members of the family. In Tricho- gaster and Colisa they are large and suturally united with each other 12, COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI ~_ in the midline, overlying the ethmoid. In Belontia, the nasals do not meet in the midline, but are widely separated by the dorsoanterior por- tion of the ethmoid. A similar condition is found in Macropodus, but the nasals here are more or less circular in shape. In Betta the pear- shaped nasals are also separated, but not by the ethmoid. In Trichopsis the nasals are elongate tubular bones, widely separated from each other and articulating only with the frontals. Similar long and narrow tubular nasals are found in Sphaerichthys. In the latter, however, the nasals ar- ticulate posteriorly with the frontals, and laterally and medially with the prefrontal-lateral ethmoid complex. Wedged between the nasals of Sphaerichthys is the prominent ethmoid. The left and right prefrontal-lateral ethmoid complexes are separated by the ethmoid. The olfactory foramen is contained in the complex. Articular facets are wanting. The ethmoid in Betta, Belontia, and Trichogaster is interposed be- tween the prefrontal-lateral ethmoid complexes. It is a subvertical plate which bends forward to the prevomer on its anteroventral portion. In Betta the ethmoid possesses a process which extends posteriorly into the orbit supporting the interorbital septum. In Trichogaster the ethmoid is covered dorsally by the nasals. In Betta and Belontia the dorsal portion of the ethmoid is exposed between the notches in the anteromedial cor- ner of the frontals. A concavity in the superior part of the ethmoid for the reception of the ascending processes of the premaxillaries is present in Betta, Belontia, and Trichogaster. In Sphaerichthys, Trichopsis, and Macropodus the ethmoid is widened dorsally but tapers off in its antero- ventral extension to join the prevomer. The widened dorsal portion of the ethmoid, in Sphaerichthys and Trichopsis, forms a rather deep fossa in which the ascending processes of the premaxillaries fit. The ethmoid is fully exposed on the dorsal surface and extends relatively farther posteriorly in Trichopsis than in Sphaerichthys. In Macropodus the ethmoid also possesses a dorsal enlargement with a concavity in which the ascending processes of the premaxillaries are lodged. The ethmoid is covered by the joined nasals and frontals in Macropodus cupanus. Macropodus opercularis differs in this respect (fig. 10). In Macropodus cupanus the ethmoid possesses at its posteromedial corner a winglike process which extends nearly halfway into the orbit, forming an incomplete interorbital septum. This process articulates with the dorsal surface of the parasphenoid. Colisa exhibits a unique condition. The ethmoid is cup-shaped, with a deep wedge-shaped cavity which is closed posteriorly. The cup-shaped ethmoid runs back between the eyes to about the middle of the orbital cavity and receives the long ascend- ing processes of the premaxillaries. The ethmoid does not articulate with the parasphenoid in Colisa. DESCRIPTIVE OSTEOLOGY i The prevomer is a more or less flat toothless bone with a laterally expanded head and a shaft which tapers posteriorly, penetrating the parasphenoid. In Macropodus cupanus and Belontia the prevomer has a dorsomedial ridge. Orbital region (figs. 6-11, 19-26, 32-39, 45-51). The frontals form three-fourths of the roof of the skull in Trichogaster, Colisa, and Sphaerichthys, two-thirds in Macropodus chinensis and opercularis, Trichopsis, and Betta, less than two-thirds in Belontia and Macropodus cupanus. A distinct notch exposing the superior concave part of the ethmoid is present at the anteromedial corner in Betta, Trichopsis, Macropodus, Sphaerichthys, and Belontia. Ventrally an orbital process of the frontal which is continuous with the pterosphenoid is present in Betta and Macropodus cupanus. Posteriorly the frontals reach the ex- treme anteromedial tip of the supratemporal. The pterosphenoid is bounded dorsally by the sphenotic, posteriorly by the prootic, and posteroventrally by the anterolateral corner of the body of the basisphenoid. Its ventral edge forms part of the boundary of the orbital opening of the braincase, and is notched for the exit of the optic nerves. The anterior border of the pterosphenoid articulates with the ventral flangelike orbital process of the frontal in Betta and Macropodus cupanus; in Belontia the pterosphenoids of the right and left side are united by winglike processes which run transversely across the orbit, forming the orbital roof. The basisphenoid is Y-shaped with one fork curved anteroventrally toward the parasphenoid. The ventral fork is sometimes united to the parasphenoid, the condition being variable within the species. In Betta the anterior border of the body of the basisphenoid sends out a promi- nent process supporting the interorbital septum. In Trichopsis the ven- tral fork is wanting. The ventral fork divides the anterior opening of the myodome into two parts, and its anterior edge receives the postero- ventral edge of the membranous interorbital septum. The body of the basisphenoid is supported by the sphenotic and prootic. The long axis of the lachrymal is nearly vertical in Macropodus, Betta, and Belontia. In the remaining genera it is horizontal. The ventral bor- der of the lachrymal is serrate except in Betta. The lachrymal articulates with the prefrontal-lateral ethmoid complex as in the family Anaban- tidae. The remaining suborbitals are five in number. Laterally they ap- pear as a chain of narrow tubular bones traversed by the infraorbital branch of the cephalic lateral line system. The suborbital shelf is well developed in all genera (fig. 10). In Trichopsis, Macropodus, Belontia, and Betta the suborbital shelf is built by prominent medial winglike processes of suborbitals 2-5, which are rigidly united with each other. Otic region (figs. 6-11, 19-26, 45-51, 57-63). In ventral view the 14 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI sphenotic is more or less hexagonal and forms part of the postorbital process. There is a prominent flangelike ridge on the ventral surface, continuous with a similar ridge on the anterolateral wall of the prootic. On the ventrolateral surface there is often an articular facet for the anterior head of the hyomandibular. The extreme posteromedial corner of the sphenotic articulates with the parietal, except in Trichogaster and Macropodus cupanus, in which the sphenotic is widely separated from the parietal. The parietals are elongate bones with the long axis arranged more or less anterolaterally. The parietals articulate posteromedially with the supraoccipital, posteriorly with the epiotic, and laterally with the supra- temporal. Anteriorly the parietals are joined to the frontals and the pos- teromedial corner of the sphenotic except in Trichogaster and Macro- podus cupanus. In the latter two the parietals do not reach the sphenotic. A prominent longitudinal ridge runs from the anteromedial corner of the parietal to the posterolateral corner. The parietal forms the antero- medial wall of the posttemporal fossa. The supratemporal forms the posterolateral corner of the skull, con- tributing to a part of the lateral wall of the neurocranium and the greater portion of the floor of the posttemporal fossa. It joins the frontal with its anteromedial corner. Anteriorly it articulates with the sphenotic, and medially with the parietal, epiotic, and exoccipital. Posteromedially the supratemporal is united with the intercalary. At the anterior portion of the supratemporal an articular facet for the posterior head of the hyomandibular is present ventrolaterally. Ventrally the supratemporal is united with the exoccipital and prootic (figs. 45-51). The prootics are large bones articulating ventrally with the para- sphenoid, posteriorly with the exoccipital and basioccipital, anteriorly with the body of the basisphenoid and pterosphenoid, and dorsolaterally with the sphenotic and supratemporal. Near its cranial border the prootic possesses a prominent anterolaterally directed process. This process joins the ascending process of the parasphenoid except in Trichopsis, Betta, and Macropodus. In the latter three genera the process extends over the ascending process of the parasphenoid, forming a more or less trans- verse septum which separates the orbital region from the otic region. The prootic also houses the greater part of the funnel-shaped myodome. The orbital opening of the myodome is triangular and is bisected by the ventral fork of the basisphenoid. The opening is surrounded ven- trally by the parasphenoid, laterally by the ascending process of the parasphenoid, and dorsally by the body of the basisphenoid and the medial processes of the prootics. The myodome expands dorsolaterally, reaching the ventral surface of the trigemino-facialis chamber. Posteriorly the myodome narrows rapidly and ends in the anterior quarter of the DESCRIPTIVE OSTEOLOGY its prootic. Just posterior to the ascending process of the parasphenoid is a foramen for the internal carotid artery, which enters the myodome at this point. After entering the myodome the internal carotid artery turns dorsally in front of the basisphenoid, penetrating the cranial cavity. The epiotic is a somewhat conical bone which adjoins the parietal anteriorly, the supraoccipital medially, and the supratemporal laterally. Ventrally it articulates with the exoccipital. In Sphaerichthys, Tricho- gaster, and Colisa the epiotic reaches the foramen magnum and forms part of its roof (figs. 61-63). On the dorsal aspect a rather prominent epiotic process arises at the posterolateral corner of the epiotic. The epaxial trunk muscles attach to these epiotic processes by means of a tendon. The processes are wanting in Sphaerichthys. The lateral wall of the epiotic forms the medial wall of the posterior half of the post- temporal fossa. The intercalary is a more or less triangular bone, wedged between the supratemporal and exoccipital. The bone is very small in Sphae- richthys, Trichogaster, and Colisa. In Belontia the bone reaches the posterolateral corner of the neurocranium. The intercalary has an ar- ticular facet for articulation with the ventral fork of the posttemporal bone. The supraoccipital consists of two portions: a dorsosuperior and a posteroinferior portion. The superior portion is diamond-shaped in Belontia, rhomboidal in Sphaerichthys, elongate and narrow in Trichop- sis, and hexagonal in Macropodus, Betta, Trichogaster, and Colisa. A well-developed supraoccipital crest is found only in Belontia (figs. 6-13). The posteroinferior portion of the supraoccipital is narrow in Belontia. In Betta, Macropodus, and Trichopsis the posteroinferior por- tion is wide dorsally and tapers ventrally, reaching the foramen mag- num with its pointed ventral end (figs. 57-60). In Sphaerichthys, Colisa, and Trichogaster the supraoccipital narrows ventrally to a much lesser extent and forms the greater part of the roof of the foramen magnum (figs. 61-63). In Sphaerichthys, Colisa, and Trichogaster the supra- occipital does not articulate with the exoccipital. In the remaining genera the supraoccipital adjoins the exoccipital along the ventral half of its posteroinferior portion. The exoccipital is bounded posterodorsally by the epiotic, laterally and anterodorsally by the intercalary and supratemporal respectively. Anteriorly it is joined to the prootic and ventrally to the basioccipital. A large foramen exoccipitale is present in the anteroventral corner. This foramen serves partly as a passage for the glossopharyngeal nerve. The vagus nerve emerges from a smaller foramen dorsoposteriorly to the foramen exoccipitale. The posterior portion of the exoccipital forms in Belontia a rounded condyle, whereas it is elongate in all the remain- 16 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI ing genera. The dorsal wings of the exoccipitals do not unite above the foramen magnum (figs. 57-63), but are separated by the posteroinferior portion of the supraoccipital. In Betta, Trichopsis, Macropodus, and Belontia the foramen magnum is medium-sized and more or less tri- angular, whereas in Sphaerichthys, Trichogaster, and Colisa the fora- men is large and more or less heart-shaped. In the latter three genera the exoccipitals do not articulate with the supraoccipital (figs. 61-63). The posttemporal is distinctly forked. The dorsal fork articulates with an articular surface on the dorsolateral corner of the epiotic; the ventral fork is connected to an articular facet on the intercalary. Basicranial region (figs. 45-51, 57-63). The basioccipital is more or less fan-shaped. Its posterior end is modified to form a vertebralike condyle for articulation with the first vertebra. Ventrally the basioccipital possesses a very prominent bilobed pharyngeal process. In Belontia, Macropodus, and Trichopsis the pharyngeal process of the basioccipital does not articulate with a similar process of the parasphenoid (figs. 19-23). In Sphaerichthys the dorsal quarter of the pharyngeal process of the basioccipital is joined to the pharyngeal process of the para- sphenoid (fig. 24), whereas in Trichogaster and Colisa nearly its entire length is joined to the parasphenoid. In the latter two genera only the distal bilobed portion of the pharyngeal process is free (figs. 25, 26). In Sphaerichthys, Colisa, and Trichogaster the anterior portion of the basioccipital is not underlain by the paraspheniod, but articulates an- teriorly with the parasphenoid. In Belontia, Trichopsis, Macropodus, and Betta the anterior half of the basioccipital is underlain by the pos- terior portion of the parasphenoid. The parasphenoid covers the ventral surface of the neurocranium through about four-fifths of its entire length in Belontia, Betta, Trich- opsis, and Macropodus, and about two-thirds in Sphaerichthys, Colisa, and Trichogaster (figs. 45-51). The bone consists of a long narrow body and two dorsolateral ascending processes. The concave anterior edge of the process forms the lateral wall of the orbital opening of the myodome. In the angle between the prootic and the posterior edge of the ascending process lies the foramen for the internal carotid artery. At its posterior end the parasphenoid develops a prominent ventrally — directed pharyngeal process. The transverse process of the parasphenoid is absent. The pharyngeal process is, in Sphaerichthys, Colisa, and Trichogaster, a thin keellike structure extending far ventrally at an angle of nearly 90° with the horizontal axis. In Belontia, Betta, Macropodus, and Trichopsis the pharyngeal process is stout but does not extend as far ventrally, and is at an angle between 15° and 70° with the hori- zontal axis (figs. 19-26). In Belontia the process is oval from a ven- tral view (fig. 45) and bears strong teeth arranged in a circle; in Betta DESCRIPTIVE OSTEOLOGY 1 the process is elongate and has five teeth of which the posterior one is stout (fig. 46); a similar condition is found in Trichopsis but the num- ber of teeth has been reduced to four; in Macropodus the teeth are weakly developed and vary in number between two and four; in Sphae- richthys the keellike process possesses only one tooth; in Colisa the teeth are very rudimentary and vary in number between zero and two; in Trichogaster, finally, the teeth are completely lost (figs. 45-51). 2. BRANCHIOCRANIUM Hyobranchial region (figs. 81-87). The hyoid apparatus is similar to that of the Anabantidae. The structure and arrangement of the three basibranchials is also as described for the Anabantidae, except in Colisa, Trichogaster, and Sphaerichthys, in which the third basibranchial is much more elongate and rod-shaped. Three hypobranchials are present: the first articulates with the second basibranchial, and the second and third with the third basibranchial. The anterolateral corner of the third hypobranchial is produced into a process which articulates with the pre- ceding hypobranchial. The first four rodlike ceratobranchials are more slender and elongate in Colisa and Trichogaster. The anteromedial ends of the fifth ceratobranchials approximate each other. The dorsal surface is covered with well-developed teeth which are arranged in ten trans- verse rows in Belontia, five rows in Macropodus and Betta, and four rows in Trichopsis. In Sphaerichthys, Colisa, and Trichogaster the teeth are weakly developed; they are arranged in a patch of two to four rows in Sphaerichthys and two rows in Colisa and Trichogaster. The first epibranchial is modified in all genera. In Sphaerichthys it consists simply of an enlarged curved plate with an epibranchial process which articu- lates with the saccular swelling of the prootic. In all other genera the first epibranchial is labyrinthine, varying in complexity in the different genera. Belontia is unique in having a very elongate suprabranchial organ; it is as long as the entire length of the hyobranchial apparatus, lodged in a suprabranchial cavity which extends posteriorly as far as the sixth vertebra, dorsal to the pleuroperitoneal cavity. The supra- branchial cavity is supported dorsally by the first four epipleural ribs, which have shifted dorsally, articulating with the neural arches of the first four vertebra, and ventrally by the fifth and sixth epipleural ribs. A similar posterior extension of the suprabranchial cavity is lacking in all other anabantoids. The first epibranchial consists, in all genera, basically of a central stem in the form of a median plate with the epi- branchial process, which articulates with the prootic. Attached to this central stem is a varying number of folded plates: two in Belontia, one in Trichopsis and Betta, three in Macropodus, three in Trichogaster, and two in Colisa. The remaining three rod-shaped epibranchials de- 18 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI crease in length from the second to the fourth. Dorsal processes are on each of the epibranchials, in various degrees of development. The first pharyngobranchial is absent. The second pharyngobranchial is a slender and slightly curved bone. The ventral surface bears eight to fifteen teeth in Belontia, four to seven in Trichopsis, two in Betta, seven in Macro- podus, three in Sphaerichthys, six in Trichogaster, and four in Colisa. A well-developed dorsal process on the second pharyngobranchial is present in Macropodus, Betta, and Trichopsis. The other genera lack similar processes. The third and fourth pharyngobranchials are fused. The dorsal surface has an articular surface for the articulation with the pharyngeal process of the basioccipital. The ventral surface is covered with teeth arranged in a varying number of transverse rows in the dif- ferent genera: seven in Belontia, five in Betta, four in Trichopsis, Macro- podus, and Sphaerichthys, and four rows of rudimentary teeth in Colisa and Trichogaster. The urohyal consists of a simple plate which gives off two lateroventral winglike processes. A cross section of the urohyal would resemble an inverted Y. In Belontia, Macropodus, Trichopsis, and Betta the urohyal is relatively short. The urohyal is more slender and elongate in Sphaerichthys, Colisa, and Trichogaster. The anterodorsal corner of the urohyal is produced into a short process. The branchios- tegal rays are well developed. Six branchiostegal rays are found in Belontia, Betta, Trichopsis, and Macropodus. The number of branchios- tegal rays is reduced to five in Sphaerichthys, Colisa, and Trichogaster. In Belontia, the sixth branchiostegal ray articulates with the epihyal. In the remaining genera all the branchiostegal rays articulate with the ceratohyal. Opercular series (figs. 32-39). In Belontia the anterior border of the opercular is relatively vertical and articulates along its entire length with the posterior border of the preopercular. In all other belontiids the anterior border of the opercular is more or less rounded and articulates only partly with the posterior border of the preopercular. The postero- dorsal corner is produced into an auricular process of variable size and shape, as either a posterior or a dorsal extension of the posterodorsal corner. In Sphaerichthys, Colisa, and Trichogaster the dorsal border of the opercular is distinctly rounded. The opercular is not serrate. The ventral border of the swbopercular is not serrate except in Tricho- gaster trichopterus. A prominent process is found at the anterodorsal corner. Posteriorly the subopercular is produced into a long and slender process in Colisa, Sphaerichthys, Trichogaster, Betta, and Belontia. The elliptical interopercular is not serrate and is overlapped dorsally by the preopercular for about three-fourths of its surface. The ventral border of the preopercular is serrate along its entire length in Trichogaster and Colisa, along half its length in Sphaerichthys, Trich- DESCRIPTIVE OSTEOLOGY 19 opsis, Belontia, and Macropodus, and not serrate in Betta. The pre- opercular is crescent-shaped and widens in the bend. Oromandibular region (figs. 69-75). The palatine consists of two com- ponents: the slender and curved ethmoid process, directed anterodor- sally, articulates with the prefrontal-lateral ethmoid complex, and the body articulates ventrally with the quadrate and posteriorly with the entopterygoid. The body of the palatine is somewhat variable in shape; mostly it is elliptical, but in Trichopsis it is triangular and in Betta irregular. The ectopterygoid is wanting. The quadrate is a more or less quadrangular bone with its antero- ventral corner modified into a condyle. The dorsal border of the bone is convex (except in Macropodus cupanus ) and articulates with the pala- tine and entopterygoid. The posterior edge is deeply notched and ven- trally produced into a prominent posterior process. The notch lodges the ventral fourth of the symplectic and may be considered as dividing the quadrate into a main body and a posterior process. The entopterygoid is a large thin bone of variable shape. Anteriorly it is bounded by the palatine, ventrally by the quadrate and metaptery- goid, and posteriorly by the metapterygoid. The metapterygoid is also variable in shape. It articulates ventrally and posteriorly with the symplectic, and anteriorly with the quadrate and entopterygoid. The metapterygoid does not reach the hyomandib- ular except in Belontia, in which it is joined with the symplectic process of the hyomandibular. The symplectic is a rather large, elongate, curved bone. Its dorsal end is bounded by the interspace of cartilage that lies between itself, the interhyal, and symplectic process of the hyomandibular. The anterior quarter of the symplectic rests on the posterior process of the quadrate. The hyomandibular consists of a body with a symplectic process an- teroventrally and an articular process posteroventrally. The angle be- tween the two stout processes in Colisa and Trichogaster is between 90° and 110°, whereas in the remaining genera it is between 30° and 70°. The dorsal surface is more or less differentiated into an anterior and posterior head, articulating respectively with the sphenotic and supratemporal. The well-developed articular process is connected to an articular facet which is situated on the medial surface in the antero- dorsal corner of the opercular. The angular is dartlike in appearance. It has a prominent process an- teriorly. The posterior end of the ventral arm of the dentary fits into the angle between the anterior process and the body of the angular. At the dorsal corner of the angular there is a prominent ascending proc- ess which reaches its highest development in Colisa, Sphaerichthys, and 20 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI Trichogaster. The angular is traversed by the posterior part of the man- dibular lateral canal. The retroarticular is a small irregular bone that fits against the postero- ventral corner of the angular, immediately ventral to the articular facet for the quadrate. The dentary is a strong V-shaped bone. The dorsal edge of the dorsal arm of the V is lined with small conical teeth. The posterodorsal corner of the dorsal arm of the V is produced into a well-developed ascending process of the dentary. The premaxillary has a well-developed ascending process, but no ar- ticular process. The ascending processes from both sides are closely con- joined and fit into a notch formed by the ethmoid and the prefrontal- lateral ethmoid complex. The ascending process is as long as the body of the premaxillary in Belontia, one-half the length of the body in Betta, Trichopsis and Macropodus cupanus, twice the length of the body in Sphaerichthys, and more than twice the length of the body in Tricho- gaster and Colisa. The toothless maxillaries are shorter and more stout than in the Ana- bantidae. The relationships of the inner and outer forks are identical to those of the Anabantidae. 3. PECTORAL GIRDLE The supracleithrum (figs. 96-101) is a small elliptical bone which articulates with the medial surface of the posttemporal. It is large in Belontia, overlapping the lateral surface of the dorsal portion of the cleithrum and postcleithrum. The cleithrum is the largest element in the pectoral girdle. The bone consists of a posterodorsal and anteroventral portion. The anteroventral portion curves craniad and joins its partner from the opposite side. A prominent ala laminaris externa as described for the Anabantidae is present only in Belontia. The process is reduced to a flangelike out- growth in the bend of the cleithrum; it continues as a ridge on the dor- sal edge of the anteroventral portion of the cleithrum. A facet for the articulation of the basipterygium is present on the mesial surface of the cleithrum about in the middle of the anteroventral portion in Trich- opsis, Macropodus, Betta, Sphaerichthys, and Belontia. In Trichogaster and Colisa a distinct fingerlike articular process is present mesially on the cleithrum, near the proximal border of the coracoid, for articulation with the basipterygia. The postcleithrum is divided into a dorsal and ventral element. The dorsal element is spindle-shaped in Colisa, Trichogaster, Belontia, Betta, and Trichopsis. In Sphaerichthys and Macropodus the dorsal element possesses a large vane on its posterior surface. The ventral element is elongate and tapers to a point ventrally. DESCRIPTIVE OSTEOLOGY 21 The coracoid consists of a dorsal, proximal, and elongate ventral por- tion which terminates posteriorly in a spine in Sphaerichthys, Colisa, Trichogaster, and Betta, but not in the remaining genera. The ventral portion of the coracoid extends anteriad to meet the ventral end of the cleithrum. The posterodorsal corner of the coracoid articulates with the largest radial. There are four hourglass-shaped radials. In Sphaerichthys the radials are elongate rodlike structures. The scapula articulates with three and a half radials. The foramen in the scapula is large in Macropodus, Belontia, and Colisa, small in Betta, Sphaerichthys, and Trichopsis. 4. PELVIC GIRDLE Belontia exhibits the most primitive stage, resembling the condition in Sandelia and Ctenopoma. In Betta and Macropodus the basipterygia are separated. The rachii of the basipterygia are also widely separated. The rachii (figs. 96-101) are directed dorsolaterally and articulate re- spectively on the mesial surface of the right and left cleithrum. The processus medialis anterior is a well-developed ridge medial to the rachis; the processus medialis is long and spinelike, whereas the pro- cessus medialis posterior is a distinct projection at the posteromedial corner; the processus lateralis is wanting. In Trichopsis the processus medialis anterior is well developed and wing-shaped, the processus me- dialis posterior is very prominent, and the processus lateralis appears as a distinct earlike projection. The most complex pelvic girdle is found in Colisa and Trichogaster. The dorsally directed rachii are slender and articulate with a fingerlike projection on the medial surface of the cleithrum. The processus medialis anterior is better developed in Colisa than in Trichogaster. In Colisa the processus medialis anterior is di- rected anteromedially and has a vanelike appendage anteriorly. The distal end of the processus medialis is curved laterally in Colisa, but not in Trichogaster. The processus lateralis is a highly developed earlike projection; the processus medialis posterior is a pointed projection in Trichogaster, whereas in Colisa it is a broad platelike process forming the larger part of the posterior border of the pelvic girdle. Trichogaster and Colisa are unique in having a processus medialis accessorius, origi- nating from the basipterygium at the level of the origin of the processus medialis anterior. The accessory medial process is arranged nearly trans- versely and is better developed in Colisa than in Trichogaster. C. OsPHRONEMIDAE 1. NEUROCRANIUM Nasal region (figs. 4, 17, 43). The nasals are elongate, slightly curved bones, widely separated form each other. The area between the nasals 22 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI is abruptly depressed to form a deep fossa for the reception of the as- cending processes of the premaxillaries. Posteriorly the nasal articulates with the frontal and prefrontal-lateral ethmoid complex. The prefrontal-lateral ethmoid complex possesses an orbital wing with a weakly developed palatine facet. The complex bone expands ventrally, joining the prevomer. Dorsally the complex reaches the posterolateral corner of the nasal. The olfactory foramen is contained in the lateral ethmoid. The complex bones are separated by the ethmoid. The ethmoid forms the anterior two-thirds of the floor of the fossa between the nasals and is but little covered by the frontals. It has a horizontal position and does not articulate with the prevomer. Its suture with the anteromedial portion of the frontal is dentate. The toothless prevomer has dorsal wings which articulate with the prefrontal-lateral ethmoid complex, forming dorsally a distinct rostral fossa. Anteriorly the prevomer curves upward and forms a distinct head. Posteriorly the shaft of the prevomer penetrates the parasphenoid. Orbital region. The frontals articulate anteriorly with the ethmoid, nasals, and prefrontal-lateral ethmoid complex, laterally with the sphe- notics and supratemporal, and posteriorly with the parietals and supra- occipital. The anteromedial portion of the frontal is depressed abruptly to form the posterior third of the floor of the fossa. The posterior portion of the frontals also forms a distinct depression for the insertion of the epaxial muscles. The lachrymal is serrate along its ventral border. It is a rhomboidal bone with its long axis directed anterodorsally, overlapping the dorsal portion of the maxillary. The remaining four suwborbitals are narrow tubular bones. The second and third suborbitals are also serrate, a unique feature among the anabantoids. Another characteristic feature of the Osphronemidae is the absence of a suborbital shelf. The pterosphenoid is deeply serrate along its ventral border. Its re- lationship with the surrounding bones is identical to the condition found in the Belontiidae. The basisphenoid is a relatively large Y-shaped bone. The shank of the Y is vertical and is joined to the parasphenoid. Otic region (figs. 4, 17, 43, 55). The sphenotic is a relatively large bone, forming the entire postorbital process. It adjoins the frontal an- teriorly and medially and the supratemporal posteriorly. At its postero- lateral surface the sphenotic possesses a shallow depression for articu- lation with the anterior head of the hyomandibular. It also bears a ventrolateral ridge which is continuous with a similar ridge on the prootic. The supratemporal is an extremely large bone exceeding the frontals in length (fig. 4). It lies at a lower level than the sphenotic and forms DESCRIPTIVE OSTEOLOGY 23 the greater part of the floor of the posttemporal fossa. The supratemporal is bounded anteriorly by the sphenotic, medially by the frontal, parietal, and intercalary, and ventrally by the anterior portion of the exoccipital and the prootic. The intercalary is wedged between the supratemporal and exoccipital. Dorsally it appears as a relatively small bone with a distinct articular facet for the ventral fork of the posttemporal. Ventrally the full extent of the bone is visible (fig. 43). The intercalary does not reach the prootic. The large parietal appears in dorsal view as an inverted V. The lateral arm of the V articulates with the frontal and supratemporal, whereas the medial arm articulates with the supraoccipital and epiotic. The posttemporal fossa is situated in the angle of the V. A prominent ridge runs along the lateral side of the medial arm of the V. The ridge is con- tinuous with a similar structure on the epiotic and serves as an insertion for the epaxial muscles. The prootic resembles that of the Belontiidae. The large spatulate supraoccipital is of the same length as the frontal (fig. 4). It possesses a very prominent supraoccipital crest. The postero- inferior portion forms the roof of the foramen magnum, separating the exoccipitals widely from each other. The epiotic is conical. It forms, together with the parietal, the lateral wall of the posttemporal fossa. An epiotic process is wanting. The pos- terolateral corner forms a distinct articular process for articulation with the dorsal fork of the posttemporal (fig. 4). The epiotic articulates ventrally with the exoccipital, anteriorly with the parietal, and medially with the supraoccipital. The epiotics do not reach the relatively small foramen magnum. The exoccipital forms posteriorly a small triangular condyle for ar- ticulation with the first vertebra. The left and right exoccipitals are widely separated by the posteroinferior portion of the supraoccipital. The exoccipitals form the lateral walls of the foramen magnum. A large foramen exoccipitale is present (fig. 17). Basicranial region (figs. 17, 43, 55). The parasphenoid consists of an elongate narrow body with dorsolateral ascending processes. The orbital portion of the parasphenoid has a dorsal, median, serrate ridge which forms, together with the pterosphenoid, an incomplete interorbital sep- tum. Posteriorly the parasphenoid is differentiated into a stout pharyn- geal process at an angle of about 80° with the horizontal axis. The pharyngeal process bears eight stout conical teeth. The basioccipital is a rather large bone articulating ventrally with the parasphenoid, anteriorly with the prootic, and dorsally with the ex- occipital. Posteriorly it forms a vertebralike condyle for articulation 24 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI with the first vertebra. Ventrally it is produced into a short and stout bilobed pharyngeal process. The anterior border of the pharyngeal proc- ess does not articulate with the posterior border of the pharyngeal process of the parasphenoid. 2. BRANCHIOCRANIUM Hyobranchial region (fig. 79). The hyoid apparatus is identical to that of the Anabantidae. The three basibranchials are hourglass-shaped. The hypobranchials resemble those of Anabas. The first four ceratobranchials are curved elongate elements. The fifth ceratobranchial is more or less triangular and bears numerous well-developed teeth. The first epi- branchial is modified into a labyrinthine suprabranchial organ. It is at- tached to the prootic by means of an epibranchial process. The supra- branchial organ consists of a main stem to which are attached three highly folded bony lamellae. The remaining epibranchials lack articular processes. The second pharyngobranchial is slightly curved and has two spinelike projections on its dorsal surface. Ventrally it bears a single row of well-developed teeth. The third and fourth pharyngobranchials are fused indistinguishably into one rectangular bone. It possesses an ar- ticular surface on its dorsal side for articulation with the pharyngeal process of the basioccipital. The pharyngobranchials bear numerous conical teeth on their ventral surfaces. The urohyal is identical to that of the Anabantidae. The sixth branchiostegal ray articulates with the epihyal, the fifth is situated between the epihyal and ceratohyal. The remaining four articulate with the ceratohyal. Opercular series (fig. 67). The opercular is oval and possesses neither an auricular nor an articular process. Its anterior border articulates only to a very limited extent with the posterior border of the preopercular. The subopercular is serrate, with about five projections. The antero- dorsal corner is produced into a process. Posteriorly the subopercular does not terminate in a process. The spatulate interopercular is serrate along its ventral border. The preopercular is crescent-shaped and expanded at its bend, the ventral edge serrate. Oromandibular region (fig. 67). The palatine adjoins the ectoptery- goid ventrally and the entopterygoid posteriorly. It articulates medially with the orbital wing of the prefrontal-lateral ethmoid complex. Antero- dorsally the palatine is produced into an ethmoid process. The ectopterygoid is somewhat triangular but its dorsal border is con- cave. It articulates dorsally with the palatine, posteroventrally with the quadrate, and posterodorsally with the entopterygoid. The entopterygoid is the largest element in the oromandibular appa- ratus. Anteriorly it meets the palatine and ectopterygoid, and ventrally the quadrate and metapterygoid. DESCRIPTIVE OSTEOLOGY PAS The metapterygoid is somewhat rectangular and articulates posteriorly with the symplectic process of the hyomandibular. Dorsally it is bounded by the entopterygoid, ventrally by the symplectic, and an- teriorly by the quadrate. The quadrate has a convex dorsal border. Posteroventrally it is pro- duced into a long, slender, posterior process. Anteroventrally the quad- rate is modified into a condyle. The dorsal portion of the curved symplectic articulates with the symplectic process of the hyomandibular. Its anterior half rests on the posterior process of the quadrate. The premaxillary possesses an ascending process which is slightly longer than the length of the horizontal limb. The maxillary is identical to that found in the Belontiidae. The dentary is rather stout and bears three rows of teeth. Postero- dorsally it is produced into a broad, rounded, ascending process. Pos- teriorly the dentary is distinctly notched, to receive a process of the angular. The angular is more or less dart-shaped. Dorsally it bears a pointed ascending process. The retroarticular is a small irregular-shaped bone at the postero- ventral corner of the angular. 3. PECTORAL GIRDLE The pectoral girdle resembles that of Belontia. 4, PELVIC GIRDLE The rachii of the basipterygia are very long and directed dorsomedi- ally. A dorsal ridge on the basipterygium (fig. 93) represents the pro- cessus medialis anterior. A long, slender, anterior projection arises from the posteromedial corner of the basipterygium and represents the processus medialis. The processus medialis posterior is very small. A distinct processus lateralis is not differentiated. The processus medialis accessorius, characteristic of Colisa and Trichogaster, is wanting. D. HELOSTOMATIDAE 1. NEUROCRANIUM Nasal region (figs. 5, 18, 44). The nasals are widely separated by the ethmoid. The nasal is bounded medially by the ethmoid, posteriorly by the frontal, and laterally by the prefrontal-lateral ethmoid complex (fig. 5). The prefrontal-lateral ethmoid complexes are separated by a ventral keel of the ethmoid. From a dorsal view (fig. 5) the complex bone ap- pears as a long and rather narrow element directed anteromedially toward the posterolateral corner of the nasal. It possesses an orbital 296 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI wing which forms part of the anterior border of the orbit; there are no articular facets. Ventrally the complex bone exhibits a unique shape (fig. 44). Posteriorly it articulates with the pterosphenoid, medially with the ventral keel of the ethmoid, and anteriorly with the dorsal wing of the prevomer. The olfactory foramen is located near the medial border of the complex bone. The ethmoid consists of a nearly horizontal body and a platelike ven- tral keel, separating the prefrontal-lateral ethmoid complexes. Postero- ventrally the ventral keel projects into the anteroventral corner of the orbit, resting on the parasphenoid. The body of the ethmoid lies at the same level as the surrounding bones, the nasals and frontals. From a dorsal view the ethmoid exhibits a very characteristic pattern (see fig. 5). The toothless prevomer is a large bone with very prominent dorso- lateral winglike processes which articulate with the nasals and the prefrontal-lateral ethmoid complexes. Posteriorly the prevomer tapers off abruptly into a point which enters a channel at the anterior end of the parasphenoid. Orbital region (figs. 5, 18, 31, 44). The frontals are joined anteriorly by the ethmoid, nasals, and prefrontal-lateral ethmoid complex, postero- medially by the supraoccipital, posteriorly by the parietal and supra- temporal, and laterally by the sphenotic and fifth suborbital. The part of the frontal posterior to the postorbital processes lies at a lower level than the anterior part, contributing to the extensive fossa for the in- sertion of the large epaxial muscles. The lachrymal is serrate along its ventral border. Dorsally it is joined to the nasal and prefrontal-lateral ethmoid complex. The lachrymal is attached to the outer edge of the orbital wing of the prefrontal-lateral ethmoid complex. The remaining four suborbitals appear in a lateral view as a chain of narrow tubular bones. The suborbital shelf is absent. The pterosphenoids are large bones, forming an incomplete inter- orbital septum. The anteroventral border is deeply serrate. Anterodor- sally it articulates with the most posterior border of the prefrontal-lateral ethmoid complex. Posteriorly it is bounded by the prootic and postero- ventrally by the basisphenoid. The pterosphenoid does not meet the small sphenotic. The basisphenoid is a stout Y-shaped bone. The shank of the Y lies vertically in the midplane of the neurocranium and joins the para- sphenoid. The arms of the Y are much broader than the shank. Antero- dorsally the basisphenoid meets the pterosphenoid and posterodorsally the prootic. DESCRIPTIVE OSTEOLOGY DAT Otic region (figs. 5, 18, 31, 44, 56). The sphenotic is a rather small bone which does not articulate with the pterosphenoid. A shallow de- pression on the posterolateral border serves as an articular facet for the anterior head of the hyomandibular. Anteriorly the sphenotic is bounded by the fifth suborbital, posteriorly by the supratemporal, and medially by the prootic. Ventrally it possesses a prominent ridge which is con- tinuous with a similar ridge on the prootic. At the dorsal surface of the skull the sphenotic does not reach the parietal, and lies at a lower level than the fifth suborbital. The supratemporal forms the posterolateral corner of the skull and contributes to the greater part of the floor of the posttemporal fossa. In length the supratemporal equals the frontal. It is bounded postero- medially by the intercalary, anteroventrally by the prootic, anterolater- ally by the sphenotic, anteromedially by the frontal and parietal, and medially by the epiotic. A unique feature is the fact that the supra- temporal is not united to the exoccipital. The latter is widely separated from the former by the large intercalary. The intercalary is a relatively large bone wedged between the ex- occipital, supratemporal, and prootic. The intercalary also contributes to the floor of the posttemporal fossa. The ventral fork of the posttem- poral rests on the dorsal surface of the intercalary. The parietal is a crescent-shaped bone contributing to a part of the medial wall of the posttemporal fossa. The parietal does not articulate with the sphenotic. It is bounded medially by the epiotic, anteriorly by the frontal, and laterally by the supratemporal. The prootics are large and more or less hexagonal bones. Each ar- ticulates posteriorly with the exoccipital and basioccipital, ventrally with the parasphenoid, anteriorly with the basisphenoid and pterosphenoid, laterally with the frontal, sphenotic, and supratemporal, and postero- laterally with the intercalary. The articulation with the intercalary and with the frontal are certainly unique features. A prominent ridge is present on the prootic at the level of the sphenotic, joining the ascend- ing process of the parasphenoid. The prootic also houses a relatively small funnel-shaped myodome which opens posteriorly through a large foramen located at the hind edge of the ascending process of the para- sphenoid. The myodome does not reach the basioccipital. The exoccipital articulates ventrally with the basioccipital, anteriorly with the prootic, laterally with the intercalary, and dorsally with the epiotic by means of a dentate suture. Posteriorly the exoccipital forms a condyle with a characteristic shape (fig. 56). The exoccipitals are widely separated from each other by the epiotics and supraoccipital (fig. 56). They form the lateral walls of the relatively large foramen 28 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI magnum. The foramen exoccipitale, found in all other anabantoid fam- ilies, is absent. The epiotics are conical bones, articulating medially with the supra- occipital, anteriorly with the parietals, laterally with the supratemporals, and ventrally with the exoccipitals. The epiotic contributes to the medial wall of the posttemporal fossa. Posteriorly it is produced into a long, slender, slightly curved epiotic process. Just lateral to the epiotic process is an articular facet for articulation of the dorsal fork of the posttemporal. The epiotics contribute to the roof of the foramen magnum. The supraoccipital is a very large bone, extending anteriorly as far as the level of the postorbital processes. It possesses an extremely well- developed supraoccipital crest. It adjoins the frontals anteriorly and the parietal and epiotic laterally. Its posteroinferior portion reaches the foramen magnum, contributing to the latter’s roof. The supraoccipital does not articulate with the exoccipitals. The posttemporal consists of a somewhat rectangular body and two very distinct forks. The dorsal fork is attached by means of a ligament to the articular facet of the epiotic, just lateral to the epiotic process. The ventral fork rests on the dorsal surface of the intercalary. The body of the posttemporal overlaps the dorsal tip of the supracleithrum. Basicranial region (figs. 18, 44, 56). The parasphenoid covers the ventral surface of the neurocranium for about four-fifths of its entire length. The long narrow bone gives off dorsolaterally directed ascending processes which form part of the lateral wall of the orbital opening of the myodome. It possesses a very prominent median keel, extending to the dorsal keels of the basibranchials, dividing the buccopharyngeal cavity nearly into two halves. The ventral keel reaches its maximum height at the level of the anterior tip of the basisphenoid. From this point the keel diminishes gradually in height and disappears at a level just posterior to the ascending process. At its posterior end the para- sphenoid is produced into a prominent keel-shaped pharyngeal process, which occasionally bears two to four very rudimentary teeth. The pharyngeal process projects as far ventrally as the median keel at its maximum height. At the anteroventral corner on either side of the pharyngeal process arises a very well-differentiated, slender, and tooth- less transverse process (fig. 56). Laterally the transverse process extends nearly to the level of the lateral border of the saccular swelling. The basioccipital is a more or less fan-shaped bone. Posteriorly it forms a vertebralike condyle for articulation with the first vertebra. Ventrally the basioccipital sends out a long, flat, pharyngeal process, which extends as far downward as the pharyngeal process of the para- sphenoid. The distal end of the process is bilobed. The pharyngeal process of the basioccipital articulates along its anterior border with DESCRIPTIVE OSTEOLOGY 29 the posterior border of that of the parasphenoid. The basioccipital ar- ticulates anteroventrally with the parasphenoid, anteriorly with the prootic, and dorsally with the exoccipital. 2. BRANCHIOCRANIUM Hyobranchial region (fig. 80). The basihyal is a rather short bone, expanded dorsoventrally. It contributes to the vertical partitioning of the buccopharyngeal cavity into two halves. The hypohyal is somewhat rectangular and articulates with a deep facet of the first basibranchial. It is composed of a dorsal and ventral portion. The ceratohyal is rather stout and articulates posteriorly with the somewhat triangular epihyal. All the elements in the hyoid apparatus are expanded dorsoventrally. The interhyal connects the hyoid apparatus with the cartilaginous block between the symplectic process of the hyomandibular and the symplectic. The first basibranchial is hourglass-shaped. The remaining two basi- branchials, on the other hand, are rod-shaped. Each basibranchial pos- sesses a very prominent median keel. These keels, together with the keel of the basihyal, the ventral keel of the parasphenoid, and the pharyn- geal processes of the parasphenoid and basioccipital, divide the bucco- pharyngeal cavity into a right and left half. The first four ceratobranchi- als are rod-shaped, slightly curved, and elongate bones. The fifth ceratobranchial is roughly triangular. Contrary to the condition in other anabantoids, the fifth ceratobranchial is toothless. The first epibranchial is modified into a labyrinthine suprabranchial organ, which articulates with the prootic by means of a distinct epibranchial process. It is com- posed of a central stem to which are attached four folded bony lamellae. The remaining three epibranchials are simple elements without any processes. The pharyngobranchials are very small. The second pharyngo- branchial is reduced to a pear-shaped toothless ossicle. The third and fourth pharyngobranchials are fused indistinguishably. They are also toothless, but bear four rows of bristlelike structures which aid in the filtering mechanism. The urohyal possesses a ventrolaterally directed ridge on either side. The six branchiostegal rays are large and slightly curved bony plates. The first four articulate with the ceratohyal, the fifth is situated between the ceratohyal and epihyal, and the sixth ar- ticulates with the epihyal. Opercular series. The opercular is a large oval bone (fig. 31). A well- differentiated auricular process is not present. The articular process appears as a dorsal earlike projection at the anterodorsal corner of the bone. The entire anterior border of the opercular articulates with the posterior border of the preopercular. The anteroventral corner of the opercular is notched to receive a process from the subopercular. The subopercular is serrate along its ventral border. Posteriorly it is 30 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI produced into a long fingerlike projection. Anterodorsally the suboper- cular sends out a prominent process which fits into the anteroventral notch of the opercular. The interopercular is spatulate and serrate along the posterior third of its ventral border. Posteriorly it overlaps the anterior end of the subopercular. The preopercular is sickle-shaped, and possesses three to four rudi- mentary spines along its posteroventral border. The preopercular over- laps the dorsal half of the interopercular. Oromandibular region (fig. 68). The large palatine possesses an an- teriorly directed ethmoid process, which articulates with the orbital wing of the prefrontal-lateral ethmoid complex and also with the inner fork of the maxillary. The palatine articulates ventrally with the ecto- pterygoid and posteriorly with the entopterygoid. The ectopterygoid is a prominent spindle-shaped bone, partly over- lapped by the quadrate. Posterodorsally it is bounded by the entoptery- goid. The large entopterygoid adjoins the palatine and ectopterygoid an- teriorly, the quadrate and part of the metapterygoid ventrally, and the metapterygoid posteriorly. The relatively small metapterygoid does not articulate with the sym- plectic process of the hyomandibular. The symplectic is a large, elongate, curved bone. It articulates with the symplectic process of the hyomandibular, the metapterygoid, and the quadrate. The anterior two-thirds of the symplectic rests on the posterior process of the quadrate. The quadrate is convex along its dorsal border. Its posterior border bears two notches. The pointed distal end of the symplectic fits into the ventral notch. Posteroventrally the quadrate is produced into a long and slender posterior process, which supports the anterior two-thirds of the symplectic. Anteriorly the quadrate forms a condyle for articula- tion with the angular. The premaxillary is a small toothless bone with a knoblike ascend- ing process of approximately the same length as the horizontal limb of the premaxillary. The maxillary is dorsally differentiated into an inner fork which ar- ticulates with the ascending process of the premaxillary and the ethmoid process of the palatine and the prevomer. The dentary is highly modified and toothless. Posteriorly it is distinctly notched, forming a movable hingelike joint with the blunt anterior proc- ess of the angular. The angular is also very modified. Its vertical position is unique among the anabantoids. Anteriorly the angular sends out a blunt process, which DESCRIPTIVE OSTEOLOGY 31 forms a highly movable hinge joint with the dentary. The ascending process of the angular is seemingly wanting, but a very prominent dorso- lateral process is differentiated near the posteroventral corner. Ven- trally the angular adjoins the small retroarticular. The hyomandibular possesses a long symplectic and a short articular process. The processes are at a 90° angle to each other. The symplectic process is expanded along its anterodorsal border and articulates with the symplectic only. Dorsally the hyomandibular is differentiated into a smaller anterior head articulating with the sphenotic and a larger pos- terior head articulating with the supratemporal. 3. PECTORAL GIRDLE The spatulate supracleithrum is small (fig. 94). The cleithrum possesses a very well-developed ala laminaris externa. The postcleithrum consists of a dorsal and ventral element. The dorsal one is oval, whereas the ventral element is elongate and tapers to a pointed ventral end. The scapula articulates with three radials. The coracoid has a slender anterior process which articulates with the lateral surface of the ventral portion of the cleithrum. Posteroventrally the coracoid is not produced into a process. Its posterodorsal corner articulates with one radial. 4, PELVIC GIRDLE The basipterygia (fig. 95) are not fused. The rachis of the basipteryg- ium is directed anteromedially and is joined to the cleithrum of the corresponding side by a ligament. Ventromedially the rachis gives rise to a prominent winglike processus medialis anterior. Two processes originate from the posteromedial corner of the basipterygium: an an- teriorly directed slender processus medialis and a posteriorly directed earlike processus medialis posterior. The processus lateralis is wanting. E. VERTEBRAL COLUMN AND UNPAIRED APPENDAGES All the vertebrae, with the exception of the first two trunk and caudal vertebrae, have centra which are quite uniform in size and form. The first centrum possesses three points of articulation with the skull: a median basal one for the basioccipital and three dorso- lateral ones for the exoccipitals. Neural arches are present and com- plete on all vertebrae. The neural arch of the first vertebra is rather weakly attached to the centrum. The prezygapophyses are short antero- dorsal projections from the neural arch. The prezygapophyses are not developed on the first centrum and the centra supporting the caudal fin. The postzygapophyses are posterodorsal spinelike projections on the neural arch. They gradually diminish in size in the posterior half of the 32 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI caudal region. The basapophyses are lacking on the first three vertebrae in the Anabantidae, the first two in Belontia, and the first one in all other anabantoids. The basapophyses are present on the other trunk vertebrae and the first caudal vertebra. In the most anterior part of the vertebral column the basapophyses are located higher on the centrum, but pro- ceeding posteriorly there is a gradual shift to a ventrolateral position. The first hemal arch possesses a very prominent hemal spine support- ing the anal fin. The anterior surface of the first hemal spine has a groove for articulation with the pterygiophores from the anterior end of the anal fin. There is a gradual decrease in size of the hemal spines proceeding caudad. Table 2 summarizes the vertebral counts made in this study. Following Schultz (1958), the first vertebra with a fully developed hemal spine, or one that reaches past the tips of the first anal pterygiophores, was considered as the first caudal vertebra. Three vertebrae are involved in the support of the caudal fin. Their centra are somewhat shorter than those of the preceding caudal verte- brae. The vertebra anterior to the antepenultimate has a hemal spine which represents a transitional stage between typical hemal spines and hypurals. The antepenultimate vertebra possesses a typical neural spine and an elongate curved hypural. The penultimate vertebra has a neural arch with a long pointed spine. The hypural of the penultimate vertebra is broader than that of the antepenultimate. The ultimate vertebra is modified posteriorly for the support of the hypurals. The third hypural does not articulate with the ultimate vertebra in Anabas, Sandelia, Be- lontia, Trichogaster, and Colisa. The third hypural is ankylosed to the ultimate vertebra in Sphaerichthys, Macropodus, Betta, and Helostoma. Hypurals 4-8 articulate with the ultimate vertebra in all Anabantoidei. Dorsal to the ultimate vertebra are two (in Helostoma, one) rod- shaped epurals. The morphology of the pterygiophores conforms with the condition in the Perciformes (Dineen and Stokely, 1956). The num- ber of interneurals and pterygiophores is summarized in Table 3. Two kinds of ribs are present: the epipleural ribs, which are ossifica- tions in the horizontal skeletogenous septum, and the pleural ribs, which are preformed in cartilage between the celomic wall and the hypaxial muscles. All trunk vertebrae carry epipleural ribs, except in Anabas and Sandelia. In the latter two the epipleurals are absent from the first two trunk vertebrae. In Belontia the first four epipleurals originate from the neural arches and the next two from the basapophyses. The first two trunk vertebrae lack pleural ribs in Anabas, Trichogaster, Trichopsis, Macropodus, Betta, and Sandelia. In Sphaerichthys, Helostoma, and Belontia, all trunk vertebrae but the first bear pleural ribs. Pleural ribs are found on all the trunk vertebrae in Colisa. Table 4 summarizes the counts on the pleural and epipleural ribs. | — —— 6 — — On wm tH OD bac aul | | he aie eae CO mn OH OD | ey fiun1ob snuauosydsc oepruauoIyds¢) 1yoUruUay DULOISO]A HT OVPI}BUIOJSOTOP] LEDS CY) Dso1gn) DSYO,) DIDVSDf D810) SNLAICOYIV] LAISDHOYIUL J, Saprouamorydso shyjyouanydy siipjnaiado snpodo.won fy snundna snpodosan py SNpD}IU Sisdoyoru, J, snyuund sisdoyort J, suapuards nyag saprojnqnun vag yjasspy DYyWOPAT EA OURS) het usurDg vYapuby sisuadno pyapuny suudsyynu puodouajy davinwu puodouay/) snaurpnisa) spqoUuy aepryuBqeuy 1g 0€& 66 86 LE 9Z OBAQOAIA [BIO], IG 0G 61 81 21 9T IVIQGOJIOA [BPNVD VI Seen OG aBIG9}JIOA YUNA, BXB YL, ‘IGGIOLNVEYNY GHL NI AVY Lag A dO UAAWAN °% ATAVY, 34 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI TABLE 3. NUMBER OF INTERNEURALS AND PTERYGIOPHORES OF THE ANABANTOIDEI. Dorsal Anal Taxa Interneurals pterygiophores pterygiophores Anabantidae Anabas testudineus 2 26 19 Ctenopoma multispinis 3 22 17 Sandelia capensis 3 21 13 Belontiidae Macropodus cupanus 4 20-21 29-30 Macropodus opercularis 3-4 20-21 29-34 Belontia hasselti 3 30 27 Trichopsis vittatus 11 9 33 Sphaerichthys osphromenoides 4-5 17 28 Trichogaster leeri 9 14 44 Betta splendens 9 11 28 Colisa lalia it 23 35 Helostomatidae Helostoma temmincki 1-2 30-31 30-32 Osphronemidae Osphronemus goramy 4 25 32 TABLE 4. NUMBER OF PLEURAL AND EPIPLEURAL RIBS IN SOME ANABANTOIDEI. Pleural ribs Epipleural ribs Taxa Trunk Caudal Total Trunk Caudal Total Anabantidae Anabas testudineus 8 12-13 20-21 8 5 13 Ctenopoma multispinis 9 13 22 10 3 13 Sandelia capensis 8 13 21 10 4 14 Belontiidae Belontia hasselti 9 2 11 10 1 11 Betta splendens 8 15 23 10 8 18 Trichopsis vittatus 7 3-0 7-10 9 6-0 9-15 Macropodus cupanus 7 8-9 15-16 9 8-9 17-18 Macropodus opercularis 7 8-9 15-16 9 6 15 Sphaerichthys osphromenoides 8 0 8 9 4 13 Trichogaster leeri 8 16 24 10 0 10 Colisa lalia 9 12-13 21-22 10 7-9 17-19 Helostomatidae Helostoma temmincki 13 9 22 13-14 0 13-14 Osphronemidae Osphronemus goramy 10 11 21 9 7 16 F. OTOLITHS Three types of otoliths can be recognized: the utriculith, sacculith, and lagenolith, located respectively in the utriculus, sacculus, and lagena DESCRIPTIVE OSTEOLOGY oD of the inner ear. In the Anabantoidei the utriculith and lagenolith prove to be of little value in establishing relationships since they are subject to intraspecific variation. The sacculith, on the other hand, proves to be useful in determining phylogenetic relationships, and is therefore the only otolith described in this study. Anabantidae (figs. 88, 90). The outer face is concave, the inner con- vex. The horizontal sulcus opens on the anterior rim and does not reach the posterior one. The sulcus has a short wide ostium and a long narrow cauda. Distinct colliculi are present. The ventral border is distinctly scalloped. Belontiidae. The outer face is slightly concave, the inner convex. The horizontal sulcus opens on the anterior border. It does not open pos- teriorly. The ostium and colliculi are wanting. The dorsal border is dis- tinctly indented. The anterior half of the dorsal border is indented in Belontia, Sphaerichthys, Colisa, and Trichogaster. In Betta and Macro- podus only the anterior third of the dorsal rim is indented. Betta is unique in having a posterior indentation in addition to the anterior one. The posterior rim is scalloped in Betta, Sphaerichthys, Colisa, and Tricho- gaster. Osphronemidae. The sacculith of Osphronemus resembles that of Colisa in many details. Helostomatidae. The sacculith of Helostoma has a very characteristic shape. It is much deeper bodied than the sacculith of other Anaban- toidei. The inner face is convex, the outer slightly concave. The hori- zontal sulcus opens at the anterior rim and does not reach the posterior border. The entire border of the sacculith is distinctly scalloped. A deep TABLE 5. ScaLteE Types Founp 1n SomME ANABANTOIDEI2 Top of Dorsal Anal Taxa Trunk head Throat fin fin Anabantidae Anabas testudineus t t t t t Ctenopoma multispinis t y t t t Sandelia capensis y y y y y Belontiidae Belontia hasselti t y t t t Sphaerichthys osphromenoides t t t t t Trichogaster trichopterus t t t t t Colisa labiosa t t t t t Helostomatidae Helostoma temmincki t y t t t Osphronemidae Osphronemus goramy t t t t t 4t=ctenoid; y = cycloid. 36 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI notch is present at the dorsoanterior corner. The lagenolith has a specific shape (fig. 88 D). G. ScALEs In all Anabantoidei the whole trunk, head, cheeks, dorsal, anal, pec- toral, and caudal fins are covered with scales. Berg (1958) stated that the scales in the Anabantoidei are characteristically ctenoid. This gen- eralization is incorrect (Table 5). DISCUSSION A. INTERRELATIONSHIPS AND DISTINCTIONS OF ANABANTOID FAMILIES Many authors, among others Giinther (1880), Regan (1909), Weber and de Beaufort (1922), Giltay (1933), Romer (1945), Bertin and Arambourg (1958), Darlington (1957), Berg (1958), and Herald (1961), recognize one family within the Anabantoidei, the Anabantidae. Gill (1893) subdivided the Anabantoidei into the Anabantidae, Osphro- nemidae, and Luciocephalidae. Boulenger (1904) considers the Ana- bantidae as a family of the order Percesoces, and the Osphronemidae as a family of the Perciformes. He included Luciocephalus in the osphro- nemids. Jordan (1923) distinguishes six families among the Labyrinthici: Luciocephalidae, Ophicephalidae, Helostomatidae, Polyacanthidae, Os- phronemidae, and Anabantidae. All previous workers have based their classification on external fea- tures and the “labyrinth organ” without proper judgment of the taxonomic values of the characters used. Consequently the phylogeny, relationships, and distinctions of the Anabantoidei appear misinterpreted and poorly understood. Recent authors, for example Forselius (1957), Nikolski (1957), Bertin and Arambourg (1958), Darlington (1957), and Herald (1961), tend to follow Berg’s classification (1940), in which he classified the Anabantoidei as a suborder of the Perciformes, with the Anabantidae as the only family. Berg separated Luciocephalus from the Anabantoidei, for the first time giving the aberrant genus a subordinal rank, Luciocephaloidei, among the Perciformes. The status of Lucio- cephalus will be discussed in a separate section. The osteological evidence obtained in this study requires distinction of four major groups among the Anabantoidei: Anabantidae: Anabas, Ctenopoma, and Sandelia. Belontiidae: Belontia, Betta, Trichopsis, Macropodus, Sphaerichthys, Colisa, Trichogaster, Malpulutta, Parasphaerichthys, and Parosphro- menus. Helostomatidae: Helostoma. Osphronemidae: Osphronemus. FAMILY ANABANTIDAE BONAPARTE Anabantidae Bonaparte, 1839, Mém. Soc. Sci. Neuchatel, 2: 13-14. The family Anabantidae (type genus Anabas Cuvier) may be con- 1 The last three were not examined osteologically. 387 38 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI sidered modern remnants of the more generalized ancestral forms. It can be easily distinguished from the remaining Anabantoidei by the following characters: 1. The prevomer bears two rows of conical teeth (fig. 40). 2. Extrascapulars are present, overlying the epiotics and intercalaries. 3. Distinct posttemporal fossae are absent. 4. The foramen magnum is small. Its lateral walls and roof are formed by the exoccipitals (fig. 52). 5. The parasphenoid underlies the greater part of the basioccipital (fig. 40). 6. The pharyngeal processes of the basioccipital are short and stout, extending laterally from the posteroventral corner of the basioccipital. 7. The parasphenoid possesses a well-developed and toothed trans- verse process (fig. 14). 8. The entire straight anterior border of the opercular is united with the posterior border of the preopercular (fig. 27). 9. The posterior border of the opercular is serrate, and so is the pos- teroventral border of the interopercular. 10. An ectopterygoid is present (fig. 64). 11. The metapterygoid is united with the symplectic process of the hyomandibular. 12. The length of the ascending process of the premaxillary is one- fifth to one-half that of the horizontal body of the premaxillary. 13. The angular and dentary lack distinct ascending processes. 14. The basipterygia are firmly joined, forming a bilateral symmet- rical unit (fig. 91). 15. The conjoined anterior tips of the basipterygia are united to the cleithral symphysis by a ligament. 16. Total number of vertebrae is 25. 17. The sacculith is not indented anteriorly and possesses an ostium and colliculi. The Anabantidae appears to be the most primitive and may have given rise to the Helostomatidae, Osphronemidae, and Belontiidae. The pres- ence of teeth on the prevomer and the palatine is, as a rule, character- istic of the least specialized representatives of the teleosts. This fact has, as most authors (Myers, 1958, and Makushok, 1958) have pointed out, auxiliary importance for the establishment of relationships. The well-developed extrascapulars, overlying the epiotics and intercalaries, resemble those described in Amia by Shufeldt (1885). The toothed trans- verse processes of the parasphenoid, palatines, and prevomer indicate that these fishes are carnivorous, as indeed is reported by Vaas, Sachlan, and Wiraatmadja (1953) for Anabas testudineus. The mouth is rela- tively large and the ascending processes of the premaxillaries are very DISCUSSION 39 short. The protrusility of the jaws is consequently very limited. This type of jaw represents a primitive condition among the acanthopterans as described by Gregory (1933) for Lates niloticus. The oromandibular apparatus is complete and has a well-developed ectopterygoid (fig. 64). Other primitive features are the articulation of the metapterygoid with the symplectic process of the hyomandibular, the connection of the ankylosed basipterygia to the cleithral symphysis by means of a liga- ment, and the presence of an ostium and colliculi and absence of an anterior indentation on the sacculith. The primitive phylogenetic features in their totality seem to indicate that the Anabantidae resemble the an- cestral stock in many important ways. FAMILY BELONTIIDAE N. FAM. The Belontiidae (type genus Belontia Myers), the largest group of the Anabantoidei, is characterized by the following features: 1. No teeth on the prevomer (figs. 45-51). 2. Extrascapulars are wanting (figs. 6-13). 3. The posttemporal fossae are well developed. 4. The posteroinferior portion of the supraoccipital forms part of the roof of the relatively large foramen magnum (figs. 57-63). 5. A prominent, bilobed, pharyngeal process of the basioccipital is present (figs. 45-51, 57-63). 6. Transverse processes of the parasphenoid are wanting. 7. The parasphenoid has a well-developed pharyngeal process (figs. 19-26). 8. The interopercular is not serrate (figs. 32-39). 9. The ectopterygoid is wanting (figs. 69-75). 10. Distinct ascending processes are present on the angular and den- tary. ll. The length of the ascending processes of the premaxillaries is one to two and a half times that of the horizontal body of the premaxillary. 12. The sacculith is indented anteriorly. It lacks an ostium and col- liculi. The Belontiidae forms the largest group of the Anabantoidei. There is a large morphological gap between the Belontiidae and the Ana- bantidae. The belontiids form a more specialized group. The specializa- tion amounts to the total loss of the extrascapulars in the otic region, and the ectopterygoid in the oromandibular region (figs. 6-13, 69-75). Other specialized features are the reduction in size of the gape of the mouth, the increase in length of the ascending processes of the pre- maxillaries, dentary, and angular, the acquisition of a protrusile jaw mechanism, the loss of teeth on the prevomer and palatine, the develop- ment of ventrally directed pharyngeal processes of the parasphenoid 40 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI and basioccipital, and the presence of an anterior indentation and loss of ostium and colliculi on the sacculith. These morphological distinc- tions can be traced invariably in all representatives of the group. The establishment of a family for this homogenous group seems to be well supported by the new osteological data. Jordan (1923) proposed the family Polyacanthidae with Polyacanthus hasselti as the type and only species. Polyacanthus, however, type species opercularis (L.), is a syn- onym of Macropodus (see Myers, 1923). Polyacanthus hasselti there- fore needed a new name, and in 1923 Myers proposed Belontia for it. According to Article 5 of the International Code of Zoological Nomen- clature as quoted by Schenk, McMasters, Keen, and Muller (1956), the family name Polyacanthidae should be changed to Belontiidae, because the name of the type genus has been changed from Polyacanthus to Belontia. This view appears to be valid under the 1961 Code also. The Belontiidae include all genera of Jordan’s family Osphronemidae, with the exception of Osphronemus. The latter resembles the belontiids super- ficially but, as discussed in the following section, the resemblances are due to convergence rather than phylogenetic relationships. FAMILY OSPHRONEMIDAE BLEEKER Osphronemoidei Bleeker, 1859, Enumeratio Specierum Piscium Archi- pelago Indico: 18. The Osphronemidae (type genus Osphronemus Lacépéde ) resembles the belontiids in several ways, but can be distinguished by the following characters: 1. The ethmoid is horizontal and does not articulate with the pre- vomer. 2. Posteriorly the shaft of the prevomer does not taper to a point; the prevomer forms dorsally a rostral fossa (fig. 43). 3. The second and third suborbitals are serrate. 4. A distinct suborbital shelf is wanting. 5. The parietals are large and appear as inverted V’s (fig. 4). 6. The supraoccipital forms the roof of the relatively small foramen magnum, separating the exoccipitals widely (fig. 55). 7. The subopercular, interopercular, and preopercular are serrate (iganGio)e 8. The ectopterygoid is present (fig. 67). 9. The metapterygoid articulates with the symplectic process of the hyomandibular. 10. The very elongate basipterygia are directed dorsomedially and are directly connected to the cleithra (fig. 93). 11. There are 29 vertebrae. 12. All scales are ctenoid. The Osphronemidae are only distantly related to the belontiids despite DISCUSSION 4l the close superficial resemblances between the two groups. Jordan (1923) did not recognize the phylogenetic heterogeneity of his family Osphronemidae. The osphronemids seem to be an early offshoot of the anabantid stock. They are readily distinguished from the belontiids by the following characters: the presence of well-developed ectopterygoids, the serrated interoperculars, second and third suborbitals, and vestigial suborbital shelf, and failure of the ethmoid to articulate with the pre- vomer. The retention of the ectopterygoid in the oromandibular region suggests that the osphronemids form one of the three main evolutionary lines derived from the Anabantidae. The loss of the suborbital shelf is also an indication that the osphronemids represent a different evolu- tionary line from the belontiids, in which the extensive suborbital shelf is retained. The Osphronemidae sensu novo thus encompasses only one genus, Osphronemus. FAMILY HELOSTOMATIDAE JORDAN Helostomidae Jordan, 1923, Stanford Univ. Publ. Biol. Ser., 3: 132. The family Helostomatidae (type genus Helostoma Cuvier) is com- posed of only one monotypic, highly specialized genus, Helostoma, which can be characterized as follows: 1. The toothless prevomer possesses prominent dorsolateral wings which articulate with the nasals and prefrontal-lateral ethmoid com- plex (fig. 18). 2. The pterosphenoid articulates with the prefrontal-lateral ethmoid complex (fig. 44). 3. The epiotics and supraoccipital form the roof of the large foramen magnum (fig. 56). 4. The parasphenoid possesses a keel-shaped pharyngeal process and long, slender, transverse processes (figs. 18, 44, 56). 5. The fifth ceratobranchial and second, third, and fourth pharyngo- branchials are toothless (fig. 80). 6. An ectopterygoid is present (fig. 68). 7. The metapterygoid does not articulate with the symplectic process of the hyomandibular. 8. The premaxillary is toothless (fig. 68). 9. The toothless dentary forms a hingelike, highly movable joint with the angular (fig. 68). 10. The suborbital shelf is wanting. The elaborate filter apparatus of the gill arches has been described by Seitz (1937). Boker (1937), and Vaas, Sachlan, and Wiraatmadja (1953) classify Helostoma principally as a plankton feeder. The helos- tomatids resemble the belontiids and osphronemids in the loss of the extrascapulars and the development of a pharyngeal process on the 49 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI parasphenoid. The group, on the other hand, differs significantly from the other Anabantoidei in the following specialized characters: the pre- vomer articulates with the nasals, and the parasphenoid possesses both the pharyngeal and transverse processes, the toothless buccopharyngeal cavity, and the highly movable hinge joint between the dentary and angular. The presence of a well-developed ectopterygoid and both the pharyngeal and transverse processes on the parasphenoid suggests that the helostomatids originated independently from the osphronemids and belontiids. B. INTERRELATIONSHIPS AMONG ANABANTID GENERA Boulenger (1916) placed the Asiatic genus Anabas with the African Ctenopoma under the former name. De Beaufort (1951), Kosswig (1954), Steinitz (1954), and Darlington (1957) followed Boulenger’s view. But several authors do distinguish the two separate genera (Regan, 1909, Jordan, 1923, Krull, 1939, Forselius, 1957, and Bertin and Aram- bourg, 1958). The two genera are undoubtedly distinct from each other and can be separated easily by the osteological characters tabulated below: Anabas (figs. 27, 40, 64). Lachrymal possesses long radiating spines. Suborbitals 2-4 suturally united with preopercular. Posttemporal slightly forked. Transverse process of parasphenoid short with five teeth. Fifth and sixth branchiostegal rays articulate with epiphyal. Opercular deeply serrate and with two spines. Palatine toothless. Ascending process of premaxillary one-third of body of premaxillary. Basipterygium with a prominent pro- cessus lateralis. Ctenopoma (figs. 28, 29, 41, 65). Lachrymal weakly serrate; in Cteno- poma multispinis no serration. Suborbitals 2-4 do not articulate with preopercular. Posttemporal distinctly forked. Transverse processes of parasphenoid long with numerous small teeth. Only sixth branchiostegal ray articu- lates with epihyal. Opercular serrate, no spines. Teeth on palatine. Ascending process of premaxillary one-fifth of body of premaxillary. Processus lateralis on basipterygium vestigial. Myers (1926) stated that Ctenopoma multispinis may be congeneric with the Asiatic Anabas because of the strongly denticulate suboper- cular. From the above mentioned differences it is evident that multispinis is not congeneric with Anabas. The serration of the subopercular cannot be considered as a characteristic of the genus Anabas. It is generally assumed (Forselius, 1957) that Anabas is a recent repre- sentative of an ancestral form of anabantid fish and is less specialized than Ctenopoma and Sandelia. Regan (1909) considers Ctenopoma to be more closely related to Anabas than to Sandelia. DISCUSSION 43 There is sufficient osteological evidence to assume safely that Anabas and Ctenopoma resemble the ancestral anabantid stock very closely. Regan’s hypothesis that Anabas is a recent representative of an ancestral anabantid is, however, difficult to support. Anabas is specialized in sev- eral ways, and originated from the hypothetical ancestral stock during the Lower Tertiary (see subsequent discussion). Ctenopoma is more primitive than Anabas in several features: the presence of teeth on the palatines, the shorter ascending process of the premaxillary, and the large number of teeth on the parasphenoid. Anabas, on the other hand, appears to be more archaic in the following characters: the posttemporal is only slightly forked, the fifth and sixth branchiostegal rays articulate with the epihyal, and the suborbitals are suturally united to the pre- opercular. Ctenopoma and Anabas seem to be derived from a common ancestral stock, which possessed the primitive characters exhibited in both genera. The genus Sandelia can be distinguished from Ctenopoma as follows: Sandelia Nasals separated (fig. 3). Very shallow posttemporal fossa pres- ent. Extrascapulars do not articulate with supraoccipital and cover epiotics partly (fig. 3). Intercalary small, wedged between supratemporal and exoccipital. First epibranchial a rectangular plate. Five branchiostegal rays (fig. 78). Subopercular not serrate (fig. 30). Ascending process of premaxillary one-half to one-third of body of premaxillary. Scales cycloid. Ctenopoma Nasals suturally joined (fig. 2). No posttemporal fossa. Extrascapulars cover epiotics and ar- ticulate with supraoccipital (fig. Pn) Intercalary cone-shaped, forming posteroventral corner of neuro- cranium. First epibranchial labyrinthine- shaped. Six branchiostegal rays (fig. 77). Subopercular serrate (figs. 28, 29). Ascending process of premaxillary one-fifth of body of premaxillary. Scales ctenoid and cycloid. Regan’s hypothesis that Ctenopoma is more closely related to Anabas than to Sandelia is supported by the data tabulated above. The loss of one branchiostegal ray, the separation of the nasals, and the simple first epibranchial are certainly specialized characters separating Sandelia widely from Ctenopoma. Sandelia seems to have developed in isolation in the southern tip of Africa (Barnard, 1943). The simple form of the suprabranchial organ seems to indicate that the air-breathing capacity of Sandelia is less efficient than that of its tropical predecessors. Sandelia is probably not exposed to such austere habitats as Ctenopoma is in the tropical swamps of central Africa. Air-breathing is of great survival value in habitats which are subjected to periodic droughts. Sandelia has 44 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI penetrated into temperate habitats in which severe periodic droughts do not occur. In these habitats air-breathing may not be a selective advan- tage, since it has lost most of its survival value. It is, therefore, to be expected that the labyrinthine suprabranchial organ could disappear during the course of evolution in environments in which the original adaptive value of air-breathing is lost. Regression of the suprabranchial respiratory organs has also been recorded in some Clariidae (Greenwood, 1961). Greenwood considers the variation of suprabranchial organs of the clariids as an example of cyclic evolution. A similar theory may be suggested for the anabantids. Jordan (1923) has apparently misplaced Sandelia in the family Helostomatidae. The differences between Sandelia and Helostoma are plainly evident from the previous section on the interrelationships of the anabantoid families. C.. INTERRELATIONSHIPS AMONG BELONTUD GENERA The more conservative representative of the belontiids appears to be the genus Belontia. It retains several ancestral features: the articulation of the entire straight anterior border of the opercular with the posterior border of the preopercular (fig. 32), the metapterygoid joined to the symplectic process of the hyomandibular (fig. 69), the relatively short ascending process of the premaxillary, and the ankylosed basipterygia with the conjoined rachii (fig. 96). These well-marked differences be- tween Belontia and the other belontiids point to a long period of inde- pendent existence. Opposed to the above mentioned conservative fea- tures, a number of specialized modifications are exhibited in Belontia. Among these are the prominent posterior extension of the suprabranchial cavity, which is supported by the first six epipleural ribs, the large supra- occipital with a high crest, and the development of longitudinal ridges on the parietals and epiotics for additional insertion of the epaxial mus- cles (fig. 19). Belontia seems to be an isolated genus representing the end product of a sterile evolutionary line. Originating from the same ancestor of Belontia is another branch of the family, of which Betta appears to be the most primitive. Betta is easily distinguished from the other genera by the loss of serration of the lachrymal and preopercular (fig. 33), the presence of a process on the ethmoid and basisphenoid supporting the interorbital septum (fig. 20), and the anterior and posterior indentations of the dorsal rim of the sacculith. In Trichopsis, the widely separated nasals become elongate and tubular (fig. 8), the ventral fork of the basisphenoid is lost (fig. 21), and the number of teeth on the pharyngeal process is reduced from five, as in Betta, to four. A third evolutionary line which probably originated from the an- cestor of Betta has been more flexible, giving rise to several divergent DISCUSSION 45 groups. The closely related genera Macropodus, Parosphromenus, and possibly Malpulutta constitute one group. Macropodus is disturbingly diverse but I am not now convinced that its partition at the generic level would be advisable. Macropodus cupanus is widely different in osteology from Macropodus chinensis and the type species Macropodus opercularis: M. opercularis and M. chinensis M. cupanus Nasals widely separated (fig. 10). Nasals suturally united (fig. 9). Ethmoid exposed dorsally. Ethmoid covered dorsally by frontals and nasals. Ethmoid lacks processes (fig. 23). Ethmoid possesses a posterior keel- like process extending into orbit (fig. 22). These osteological distinctions, combined with the resemblance of the breeding behavior to that of Betta splendens (Seleuthner, 1950, as quoted by Gilch, 1957) and the absence of oil containers in the eggs (Oelschlager, 1925, and Seleuthner, 1950, as quoted by Gilch, 1957), collectively suggest that cupanus represents a separate genus or subgenus. Seleuthner (1950, as quoted by Gilch, 1957) suggested that cupanus represents an intermediate stage between Betta and Macro- podus. The species cupanus is definitely more primitive than the other forms examined of Macropodus, but it also differs significantly from Betta. Polyacanthus is not available since its type is Chaetodon chinensis Bloch, a synonym of Labrus opercularis L. The type of Macropodus is M. viridi-auratus, which is, however, conspecific with Labrus oper- cularis (Myers, 1923). Accordingly a new name would be required if cupanus were to be regarded as distinct generically or subgenerically from other species now assigned to Macropodus. I here refrain from erection for cupanus of (a) a new genus because chinensis is in several ways intermediate in osteology between cupanus and opercularis, and (b) a new subgenus because the differences are not in my opinion com- mensurate with the conventionally accepted requirements for a mono- typic group. Macropodus is more specialized than Betta and Trichopsis because it acquired a greater reduction in size and number of teeth on the pharyngeal process of the parasphenoid and a greater angle between the pharyngeal process of the parasphenoid and the horizontal axis of the basioccipital. The position of Malpulutta and Parosphromenus is still uncertain. The lack of material for anatomical study renders impossible determina- tion of the exact relationships. Deraniyagala (1937) stated that Mal- pulutta is related to Belontia and Macropodus, and that Sphaerichthys is the genus most closely related to Malpulutta. There is, however, no supporting evidence for this hypothesis. Belontia and Macropodus are 46 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI very distantly related to each other, as shown in the beginning of this section. Sphaerichthys differs in many ways from Macropodus. Malpulutta and Parosphromenus are tentatively placed between Macro- podus and Sphaerichthys. The rest of the belontiids belongs to an evolutionary line which, seem- ingly, has been derived from an ancestral Macropodus. This evolution- ary line is comprised of the genera Sphaerichthys, Parasphaerichthys, Colisa, and Trichogaster. Sphaerichthys probably originated from a Macropodus-like stock and became specialized as follows: the nasals became long, narrow, tubular, and widely separated (fig. 11); the eth- moid became fully exposed dorsally; the epiotic processes were lost; the keellike pharyngeal process of the parasphenoid turned to a straight angle with the horizontal axis of the basioccipital (fig. 24); the number of teeth on the pharyngeal process of the parasphenoid was reduced to one; the dorsal quarter of the pharyngeal process of the basioccipital acquired a connection with that of the parasphenoid; the radials became elongate rod-shaped structures; the number of ver- tebrae became reduced to 25; and the pleural and epipleural ribs were lost in the caudal region. In the second evolutionary line, derived from the Macropodus-like stock and leading to Colisa and Trichogaster, the nasals remain large plates overlying the ethmoid completely, the pleural ribs in the caudal region are retained, and the number of vertebrae increases. An accentuation of the evolutionary trend already present in Sphae- richthys can be traced in the following morphological characters: the pharyngeal process of the basioccipital is joined along its entire an- terior border to that of the parasphenoid, the pharyngeal process of the parasphenoid possesses zero to two vestigial teeth (figs. 25, 26), and the number of teeth on the fifth ceratobranchial is greatly reduced (figs. 86, 87). Trichogaster and Colisa can be separated as follows: Trichogaster Ethmoid subvertical shaped. and not cup- No teeth on pharyngeal process of parasphenoid. Second pharyngobranchial bears six teeth. Five pelvic fin rays and spine present. Six to ten interneurals. Total number of vertebrae 30 to 31. Epipleural ribs not present in caudal region. Colisa Ethmoid vertical and cup-shaped, ex- tending posteriorly between eyes to middle of orbit. Pharyngeal process of parasphenoid with one to two vestigial teeth. Second pharyngobranchial bears four teeth. One pelvic fin ray, spine rudimentary. One to two interneurals. Total number of vertebrae 26 to 29. Seven to nine epipleural ribs in cau- dal region. DISCUSSION 47 The ancestral Macropodus-like stock may have separated into two lines. One of these, the genus Colisa, shows a tendency toward a pos- terior shift and modification of the ethmoid, a marked reduction in number of interneurals, a retention of the epipleurals in the caudal region, and a reduction in the number of pelvic fin rays. In the other line of development, the genus Trichogaster, the ethmoid remained subvertical, the number of interneurals increased, the epipleural ribs in the caudal region were lost, and all five pelvic fin rays were retained. Many important similarities between Colisa and Trichogaster, as for example the large size of the foramen magnum, the articulation of the pharyngeal process of the basioccipital with that of the parasphenoid, and the contribution of the epiotics in the formation of the roof of the foramen magnum, strongly indicate the close relationship of the two genera. The status of Parasphaerichthys remains to be investigated. Prashad and Mukerji (1929) stated that it is closely related to Sphaerichthys. They summarize the two genera as follows: Sphaerichthys Parasphaerichthys Jaws equal. Jaws unequal. Lateral line vestigial. Lateral line absent. Dorsal fin with 8 to 12 spines and _ Dorsal fins with 4 spines and 6 rays. 7 to 10 rays. Anal fin with 8 to 10 spines and 18 Anal fin with 13 spines and 12 rays. to 22 rays. Ventrals originate slightly before pec- Ventrals originate slightly behind pec- torals. torals. Parasphaerichthys is here tentatively placed close to Sphaerichthys. Within the Belontiidae, then, three well-defined groups can be recog- nized. The establishment of three new subfamilies not only illustrates the natural relationships between the belontiids, but also represents the three major evolutionary lines of the family. The three new subfamilies are: 1. Belontiinae n. subfam., which includes just one genus, Belontia. 2. Macropodinae n. subfam., comprised of the following genera: Betta, Trichopsis, Macropodus, Parosphromenus, and Malpulutta. 3. Trichogasterinae n. subfam., which encompasses Sphaerichthys, Parasphaerichthys, Colisa, and Trichogaster. The morphological distinctions between the three subfamilies, consti- tuting their diagnoses, are tabulated below (figs. 19-26, 32-39, 45-51, 57-63, 69-75, 81-87, 96-102): 48 Belontiinae Eight to ten teeth on pharyngeal process of parasphenoid. Pharyngeal process of basioccipital does not articulate with that of parasphenoid. Entire anterior border of opercular articulates with preopercular. Metapterygoid articu- lates with symplectic process of hyoman- dibular. Ascending process of premaxillary shorter than its body. Posterior extension of suprabranchial cavity supported by first six epipleurals. Pterosphenoids united. Six branchiostegal rays, of which sixth is joined to epihyal. Suborbital shelf formed by suborbitals 2-5. Epiotics do not reach foramen magnum. Supraoccipital articu- lates with exoccipi- tals. Supraoccipital crest present. Macropodinae Two to five teeth on pharyngeal process of parasphenoid. As in Belontiinae. Anterior border of oper- cular articulates partly with preoper- cular. Metapterygoid does not articulate with hyo- mandibular. Ascending process of premaxillary one to one and a half times its body. No posterior extension of suprabranchial cav- ity. Pterosphenoids rated. Six branchiostegal rays joined to ceratohyal. sepa- As in Belontiinae. As in Belontiinae. As in Belontiinae. Supraoccipital crest ab- sent. COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI Trichogasterinae Zero to two rudimentary teeth on pharyngeal process of parasphe- noid. Pharyngeal process of basioccipital articu- lates with that of parasphenoid. As in Macropodinae. As in Macropodinae. Ascending process of premaxillary at least twice as long as its body. As in Macropodinae. As in Macropodinae. Five branchiostegal rays joined to ceratohyal. Suborbital shelf formed by suborbital 2. Epiotics reach large foramen magnum. Supraoccipital does not articulate with ex- occipitals. As in Macropodinae. The evolution of the belontiids may be briefly reviewed as follows. The Belontiinae may have separated at an early stage from the ances- tral anabantid stock, retaining several anabantid features: the articula- tion of the entire straight border of the opercular with the posterior border of the preopercular, the union of the metapterygoid to the symplectic process of the hyomandibular, and the articulation of the sixth branchiostegal ray with the epihyal. The Belontiinae acquired several specializations during their course of evolution: the suprabran- chial cavity extends far posteriorly and is supported by the first six epi- DISCUSSION 49 pleural ribs, the pterosphenoids are united, and the supraoccipital crest becomes well differentiated. The Macropodinae may have a common ancestral origin with the Belontiinae. The Macropodinae, however, be- came specialized in a different way: the suprabranchial cavity becomes reduced in size and shifts into a more ventral position, the supraoccipital crest is not differentiated, the jaws become more protrusile, the teeth on the parasphenoid become vestigial and reduced in number. The Tricho- gasterinae originated most likely from a Macropodus-like ancestor and became modified in the following characters: branchiostegal rays are reduced to five, the teeth on the pharyngeal process of the parasphenoid are vestigial or lost, the supraoccipital is separated from the exoccipitals by the epiotics, the suborbital shelf becomes limited to suborbital 2, and the epiotics reach the very large foramen magnum. D. PHYLOGENETIC TRENDS IN THE ANABANTOIDEI Gosline (1959) has stressed the importance of a proper understanding of the mode of life and functional morphology for the student of phylo- genetic histories of teleosts. A thorough knowledge of the functional systems is especially crucial in the phylogenetic studies of modern teleosts. It seems therefore appropriate to analyze the phylogenetic trends in some morphological and functional units of the Anabantoidei before entering into the final synthesis of the phylogeny. Protrusility of the jaws (figs. 64-75). The incipient stage is found in the family Anabantidae. In Ctenopoma the ascending process of the premaxillary is very short, not exceeding one-fifth of the length of the horizontal body of the premaxillary. The length of the ascending process in Anabas and Sandelia varies between one-third and one-half of that of the body of the premaxillary. The relatively short ascending processes are more or less conjoined in the midline, forming an inverted V, the top of which articulates with a notch formed by the nasals and ethmoid. The conjoined ascending processes form a “sliding stalk” (Delsman, 1925). This stalk slides forward and backward in the rostral notch. The sliding of the stalk takes place automatically when the dentary and angular are depressed, because of the presence of a ligament on each side which links the freely movable premaxillary with the lower jaw. The adductor mandibulae muscles consist of two parts: A, inserts on the medial face of the maxillary, retracting it and indirectly the man- dible and premaxillary; A, inserts on the ascending processes of the dentary and angular, and pulls on the mandible in closing the mouth. The jaw mechanism is thus of the type common to most acanthopteran fishes (Eaton, 1943). The short ascending processes of the premaxillaries in the primitive Anabantidae limit the protrusility of the jaw to a mini- 50 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI mum, because the degree of protrusility of the jaws is directly propor- tional with the length of the ascending processes of the premaxillaries (Delsman, 1925, Gregory, 1933, and Eaton, 1943). A nicely graded series toward increased protrusility of the jaws can be followed in the Belontiidae. Betta and Trichopsis represent the most primitive belontiids. The mouth is of moderate size. The ascending process of the premaxillary in Betta is slightly longer than its horizontal body. In Trichopsis the length of the ascending process is approximately one and a half times that of the body. Macropodus represents an inter- mediate stage between the moderately protrusile jaws of Betta and Trichopsis and the highly protrusile jaws of the Trichogasterinae, in which the ascending process is at least twice as long as the body of the premaxillary. Correlated with the increased protrusility is a gradual development of the ascending processes of the dentary and angular for insertion of part A, of the adductor mandibulae muscle. Increased pro- trusility of the jaws is usually accompanied by an increase in size of the mouth (Al-Hussaini, 1947, Delsman, 1925, and Eaton, 1943). In the belontiids the opposite is true. In the sequence leading from the Ana- bantidae to the Trichogasterinae, the relatively long slender anabantid jaws become greatly shortened. A similar trend has been described for many actinopteran fishes by Gregory (1933). In the Osphronemidae the mouth is relatively large and the length of the ascending process of the premaxillary is about one and a half times that of the body. This “sliding stalk” fits in the rostral notch formed by the ethmoid. The premaxillary is moderately protrusile. Helostoma (fig. 68) represents an aberrant form, which, seemingly, has split at an early evolutionary stage from the ancestral anabantid stock. The edentulous premaxillary and dentary are very reduced in size. The jaws are highly protrusile. The protrusility has been estab- lished by a highly movable hinge joint between the dentary and angular. The movement between the dentary and angular is accomplished by the differentiation of a unique dorsolateral process on the angular. The dorsolateral process of the angular serves as an insertion for part A, of the adductor mandibulae muscles. The only function of the premaxil- lary is to support the thick “lips.” Its role in controlling the degree of protrusility of the mouth has been taken over by the modified dentary and angular. Dentition (figs. 40-51, 64-75, 76-87). The most primitive condition is found in Ctenopoma and Sandelia, in which both the prevomer and palatine bear teeth. The teeth on the prevomer are retained in Anabas, but in the latter the palatine has lost any trace of dentition. The ana- bantids possess many well-developed teeth on the parasphenoid, pre- maxillary, and dentary. In the three divergent lines derived from the DISCUSSION Dill ancestral anabantids there is a definite evolutionary trend toward re- duction in size and number of teeth. In the osphronemids the prevomer and palatine are devoid of any teeth, whereas the parasphenoid bears eight stout conical teeth. The helostomatids exhibit an extreme edentu- lous stage. The premaxillary, dentary, and palatine have lost their teeth, and the parasphenoid bears occasionally one to two vestigial teeth. A similar edentulous condition has been developed independently in vari- ous unrelated teleosts, as for example in Brevoortia (Gregory, 1933), Clupea (Kirchhoff, 1958), Aeoliscus, Hippocampus, and Syngnathoides (Al-Hussaini, 1947). A gradual and progressive reduction in dentition can be traced in the evolutionary line leading from the anabantids to the Belontiinae, Macro- podinae, and Trichogasterinae. In all three subfamilies the teeth are retained on the premaxillary and dentary, whereas the teeth on the prevomer and palatine are lost. The Belontiinae represent the most “primitive” stage in having eight strong teeth on the parasphenoid. Among the Macropodinae Betta possesses five teeth on the parasphenoid, Trichopsis has four, and Macropodus has two to four. The Trichogaster- inae appear to be the most advanced group. Sphaerichthys possesses one tooth on the parasphenoid, and Colisa has zero to two vestigial teeth. The most progressive evolutionary stage is represented by Tricho- gaster, in which the dentition of the parasphenoid has been lost com- pletely. The evolutionary trend toward reduction in dentition of the prevomer, palatine, and parasphenoid seems widely spread among teleosts. It has been traced in many evolutionary sequences, leading to specialization (Gregory, 1933, Myers, 1958, and Makushok, 1958). It is also generally accepted that the predaceous type of mouth is primitive, from an evo- lutionary point of view, but this generalization is not necessarily true in all cases. In anabantids the complete armament of teeth on the margins of the mouth, on the palate, and in the pharynx is highly adapted to a predaceous behavior. Vaas, Sachlan, and Wiraatmadja (1953) reported that Anabas testudineus is typically carnivorous in its natural habitat, feeding mainly on insects and small fish. The same workers reported that Osphronemus goramy, in which the palatal teeth are lost, is predomi- nantly herbivorous. The planktonic feeding behavior of Helostoma temmincki, which possesses an edentulous mouth, palate, and pharynx, was well established by Boker (1937), Seitz (1937), and Vaas, Sachlan, and Wiraatmadja (1953). The latter three authors made a careful analysis of the natural feeding habits in nature of only three members of the belontiids. Betta unimaculata, with five rather well-developed teeth on the parasphenoid, seems to be carnivorous. Its diet consists mainly of insects obtained at the surface and in the adjacent air. The UNIVERSITY: Ur ILLINOIS LIBRARY, SY COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI TaBLE 6. NUMBER AND Form oF GILL RAKERS IN SOME REPRESENTATIVE ANABANTOIDS. Many, Extremely Few and moderately Many numerous, Taxa course course fine very fine Anabantidae Anabas Ra Ctenopoma Sandelia * Osphronemidae Osphronemus Belontiidae Belontia Betta Trichopsis Macropodus Sphaerichthys Colisa * Trichogaster * Helostomatidae Helostoma * * * CD Gs tot a try! stomach contents of Trichogaster trichopterus and Trichogaster leeri, in which the dentition of the parasphenoid is lost, are composed of plant parts, filamentous algae, and detritus. The stomach contents of several specimens of Belontia hasselti, Trichopsis vittatus, and Macropodus opercularis have been analyzed. All three species appear to be omniv- orous, feeding on plant parts, epiphytes, and insects. Insects, small fish, and shrimp were found in the stomach of one specimen of Ctenopoma and four specimens of Sandelia. Evolutionary patterns, similar to those of the buccal and palatal den- tition, are found in the hyobranchial apparatus. Table 6 summarizes the data obtained by the study of the gill rakers. A definite correlation exists between the structure of the gill rakers and pharyngeal teeth on one hand, and feeding habits on the other hand (Suyehiro, 1941, Al-Hussaini, 1947, and Girgis, 1952). In the carnivorous anabantids the gill rakers are few and coarse, and the teeth on the third, fourth, and fifth pharyngobranchials and fifth ceratobranchi- als are numerous and stout. That this anatomical arrangement is adapted to predaceous feeding habits has been confirmed by the study of stomach contents. In the herbivorous osphronemids the pharyngeal teeth are still numerous and stout, aiding efficiently in grinding plant parts; the many gill rakers are moderately coarse to protect the gill filaments. The gill rakers form a straining mechanism which bars the entry of food par- DISCUSSION 53 ticles into the gill filaments. The omnivorous Belontiinae and Macro- podinae resemble the osphronemids very closely, but the teeth on the pharyngeal elements become more villiform. In the Trichogasterinae the pharyngeal skeleton becomes much more slender, the gill rakers much finer and more numerous, and the pharyngeal teeth decrease in size and number, obviating much of the grinding capacity and adapting to fine soft food. Helostoma is highly specialized to planktonic feeding habits. Both the buccal cavity and pharynx are devoid of any teeth. The gill rakers form a very efficient straining mechanism, collecting microscopic food particles (described in great detail by Seitz, 1937). Helostoma also scrapes algae from stones, etc., by means of highly movable “lips” which possess small horny teeth. A definite correlation between diet and dentition has been found in the anabantoids. Barrington (1957), however, pointed out that many species of fishes can clearly make use of a wide variety of foods and that the ecological situation of fishes may often favor a generalized rather than a highly specialized digestive system. Observations on the feed- ing behavior of several members of the belontiids confirm Barrington’s statements, although unquestionably some species are adapted predomi- nantly to particular categories of food. The dentition, size of the mouth, and structure of the hyobranchial apparatus impose definite limitations on the size and structure of the food and a definite evolutionary trend toward the loss of teeth is present in all three families derived from the predaceous anabantids. The Belontiinae and Macropodinae are likely capable of omnivorous feeding habits. This phenomenon should not be considered as a preadaptive polyphagous stage in the sense of Cuénot (1951), but rather as a broadly adaptive character which may permit a narrowing in realization of prospective functions. The condition in the Trichogasterinae, which are derived from the Macropodinae, illus- trates beautifully the narrowing in function or specialization from omniv- orously adapted forms to purely herbivorously adapted forms. The evolution of the dentition and the hyobranchial apparatus can be briefly explained by assuming a more or less simultaneous divergence of three evolutionary lines, the osphronemids, helostomatids, and belontiids, from a similar ancestral adaptive type, represented by the carnivorous anabantids. The three independent lines enter into three diverging adap- tive zones. The helostomatids became very specialized plankton feeders, the osphronemids purely herbivorous, and the belontiids mainly omnivo- rous. Within the belontiids there is a trend toward a progressive occupa- tion of new herbivorous zones by the Trichogasterinae, which have originated from the prospectively adapted omnivorous Macropodinae. It would seem that the adaptive zones of the Trichogasterinae and osphronemids overlap. This is, however, not the case. The osphronemids 54 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI feed on much larger food than the Trichogasterinae, which are charac- terized by a very small mouth and a nearly edentulous palate and pharynx. The Trichogasterinae are adapted to fine and soft herbivorous food, approaching the planktonic type. Suprabranchial cavity (figs. 14-26). The suprabranchial cavity houses the suprabranchial organ, which is able to perform air-breathing func- tions (Zograff, 1888, Henninger, 1908, and Bader, 1937). In the in- cipient stage the suprabranchial cavity extends far dorsally and is en- closed medially by the exoccipitals, supraoccipital, and prootics, dor- sally by the frontal, supratemporal, and parietal, posteriorly by the epiotic and intercalary, and laterally to a great extent by the supratem- poral and sphenotic. This extreme dorsal position of the suprabranchial cavity is found in Anabas, and to a lesser extent in Ctenopoma and Sandelia. In the latter two genera there is a tendency toward a more ventral position and a reduction in size of the cavity, which is bounded laterally by the hyomandibular and opercular. The supratemporal and sphenotic form the roof rather than the lateral wall of the cavity. There is a definite trend toward a ventral shift of the suprabranchial cavity in all three phyletic lines derived from the ancestral anabantid stock. Among the belontiids, the Belontiinae are unique. The suprabranchial cavity has shifted slightly to a more ventral position, but also extends far posteriorly, where it is supported by the first six epipleural ribs. In the Macropodinae the gradual ventral shift of the suprabranchial cavity is acquired by the development of pharyngeal processes of the para- sphenoid and basioccipital. The gradual development of the processes, accompanied by the ventral shift of the suprabranchial cavity, is well illustrated in the sequence Betta, Trichopsis, and Macropodus. The peak of this evolutionary trend is reached in the Trichogasterinae. With the differentiation of the pharyngeal processes, the hyobranchial apparatus, which is suspended mainly from the pharyngeal processes of the basi- occipital, shifts, correspondingly, to a more ventral position. Concomi- tant with the ventral shift of the hyobranchial apparatus and supra- branchial cavity is the gradual change from a cylindrical skull to a deep- bodied compressed skull. Gregory’s (1933) statement that the opercular region becomes enlarged in Osphronemus and other anabantoids in con- nection with the development of the suprabranchial organ seems to be erroneous. In Anabas the adductor arcus palatini and second levator arcus branchi- alis muscles form the contractile floor of the suprabranchial cavity. The modified muscles represent the principle components of air ventilation. It is evident that, in Anabas, the large dorsal portion of the supra- branchial cavity, which is enclosed by a bony capsule formed by the supratemporal and sphenotic, epiotic, and intercalary, will account for DISCUSSION 55 a large volume of residual air. In the three families derived from the ancestral anabantids the noncontractile bony capsule is gradually elimi- nated. In the osphronemids, helostomatids, and belontiids, the bony pos- terior wall has disappeared, and the supratemporal and sphenotic form only the dorsal wall of the suprabranchial cavity. Bader (1937) noted that the well-differentiated musculature of the suprabranchial cavity in Macropodus, Trichopsis, Colisa, and Trichogaster extends dorsolater- ally and posteriorly. A similar condition is found in Helostoma and Os- phronemus. The contractile and muscular lateral, posterior, and ventral walls of the suprabranchial cavity in the osphronemids, helostomatids, and belontiids resulted, undoubtedly, in a more efficient ventilation than in the ancestral anabantids. The volume of residual air in the belontiids, helostomatids, and osphronemids (Bader, 1937) seems to be smaller than in the anabantids. A mechanical concomitant of these anatomical changes in the suprabranchial cavity is the reduction in size of the pharyngeal cavity, which in turn controls the size limit of the food particles. Body form. Myers (1958) listed the deepening and compression of the body as one of the evolutionary trends leading from generalized teleosts to more specialized ones. This trend can also be followed in the evolution of the anabantoid fishes. The ancestral anabantids are short-bodied. The osphronemids and helostomatids are compressed, deep-bodied fishes. The gradual deepening of the body can be traced in the sequence leading from the anabantids to the Trichogasterinae. Among the Macropodinae the most primitive short-bodied forms are represented by Betta and Trichopsis. A slight deepening and compression of the body is found in Macropodus, which forms an intermediate stage between Betta and the distinctly compressed and deep-bodied Sphae- richthys, and Colisa. The Belontiinae, although primitive in many ways, are compressed and deep-bodied. The body form seems to be correlated with the degree of development of the pharyngeal processes of the parasphenoid and_basioccipital. Parallel with the deepening of the body is the differentiation of post- temporal fossae. The fossae are wanting in the short-bodied Anabas and Ctenopoma, in which the epaxial muscles are inserted on the posterior surface of the skull, formed by the supraoccipital, epiotics, intercalaries, and exoccipitals. In Sandelia the posttemporal fossae are in a rudimen- tary stage. The epaxial muscles pass ventral to the extrascapulars, insert- ing on the shallow fossa formed by the supratemporals, epiotics, and intercalaries. In the osphronemids and helostomatids the extrascapulars are lost. The neurocranium is better adapted to resist the stresses caused by the thrusts coming from the locomotory, masticatory, and respiratory apparatus as discussed for the generalized percoid skull by Gregory 56 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI (1933). The neurocranium is also more compact than that of the ana- bantids because the epaxial muscles pass above the dermal skull roof, inserting on the frontals, supraoccipital, epiotics, and intercalaries. The posttemporal fossae are well developed and appear as distinct depres- sions on the posterolateral corners of the skull roof. The posteriorly projecting epiotic processes play a very important role in conveying wrenching and pulling forces from the pectoral girdle. Among the belontiids, the Belontiinae are unique in having very prominent longi- tudinal ridges on the parietals, epiotics, and supraoccipital. The Macro- podinae and Trichogasterinae have well-differentiated posttemporal fossae and epiotic processes, which are lacking in Sphaerichthys. In none of the belontiids do the epaxial muscles extend as far anteriorly as in the helostomatids and osphronemids. Correlated with the ap- pearance of posttemporal fossae is the increase in size of the foramen magnum. A nicely graded series is present: in the anabantids the foramen is very small and is surrounded by the exoccipitals; in Betta and Trichopsis the foramen is slightly larger and surrounded by the exoccipitals and supraoccipital; Macropodus represents an intermediate stage; the fora- men becomes very large in the Trichogasterinae, in which the epiotics form part of the roof of the foramen. Suborbital shelf. The suborbital shelf is a bony platform formed by fused medial wings of the suborbitals. The function of the shelf is still unknown (Makushok, 1958, and Smith and Bailey, 1962). In the incipient stage represented by the anabantids, suborbitals 2-5 contribute to the formation of the extensive and rigid suborbital shelf. During the evolution of the belontiids there is a gradual decrease in rigidity and extent of the suborbital shelf. The Belontiinae and Macro- podinae represent a more primitive condition, in which suborbitals 2-5 still join in the formation of the rather extensive suborbital shelf. In the Trichogasterinae the less extensive suborbital shelf is made up only by suborbital 2. Suborbitals 3 and 5 are tubular and have lost their medial winglike processes. The helostomatids resemble the osphronemids in respect to the structure of the suborbital shelf. In the evolutionary lines leading to the osphronemids and helostomatids the suborbital shelf becomes vestigial. In Osphronemus the suborbital shelf consists of an insignificant medial projection of the second suborbital. Branchiostegal rays (figs. 76-87). In the primitive genus Anabas the first four branchiostegal rays articulate with the ceratohyal and the two posterior ones with the epihyal. A gradual forward shift of the branchios- tegal rays can be traced. In Ctenopoma only the last branchiostegal ray articulates with the epihyal; the anterior five are joined to the ceratohyal. In Sandelia all five branchiostegal rays are connected with the cerato- hyal. Among the belontiids, Belontia represents an archaic stage in hav- DISCUSSION aT ing the last branchiostegal ray joined to the epihyal. In the remaining belontiids the branchiostegal rays articulate solely with the ceratohyal. In the osphronemids and helostomatids the sixth branchiostegal ray is connected to the epihyal, the fifth is situated between the epihyal and ceratohyal, and the remaining four are joined to the ceratohyal. Loss of skull elements and polyisomerism. A progressive reduction in the number of skull elements can be traced in the series Anabantidae- Macropodinae-Trichogasterinae. The evolutionary significance of this phenomenon is still unknown. In the Belontiinae and Macropodinae the extrascapulars and ectopterygoids are lost. Correlated with the loss of the extrascapulars is an increase in size of the supratemporal, epiotics, and posttemporal fossae. The place of the lost ectopterygoid is taken by the enlarged body of the palatine and entopterygoid. The most advanced stage is represented by the Trichogasterinae, in which not only the extrascapulars and ectopterygoids are lost, but also one of the branchios- tegal rays. Hubbs (1919) stated that the Acanthopterygii have, almost without deviation, typically six branchiostegal rays. The general arrange- ment described for the Acanthopterygii is also found in the anabantoids, with the exception of the Trichogasterinae and Sandelia, in which the number of branchiostegal rays has been reduced to five. In the osphronemids and helostomatids the loss of skull elements has been restricted to the extrascapulars. Gregory (1951) observed that secondary polyisomerism frequently occurs in the teleostean vertebral column. Secondary polyisomeres arise by budding or by subdivision of older polyisomeres. Secondary poly- isomerism seems to occur in the evolution of the postcranial axial skele- ton of the anabantoids. The anabantids resemble the ancestral percoids very closely in having 10 trunk and 15 caudal vertebrae. The ancestral percoids possess 10 trunk and 14 caudal vertebrae (Gregory, 1951). In all three phyletic lines derived from the anabantids there is a definite trend to increase the total number of vertebrae (Table 2). The osphro- nemids and helostomatids possess a total of 29 vertebrae. Helostoma is unique in having secondary polyisomerism in the trunk region only. The trunk vertebrae in Helostoma increased in number to 14. In all remain- ing anabantoids secondary polyisomerism is restricted to the caudal region. The belontiids exhibit a rather wide variation in number of vertebrae. Belontia, Macropodus, Colisa fasciata, and Colisa labiosa possess 27 vertebrae. In Trichopsis, Betta, and Trichogaster the total number of vertebrae has increased to, respectively, 28, 29, 31. Sphae- richthys and Colisa lalia form remarkable exceptions in this otherwise graded series. Sphaerichthys has the same number of vertebrae as the anabantids, namely 25, whereas Colisa lalia possesses 26 vertebrae. The latter two genera have probably evolved from a Macropodus-like 58 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI stock with 27 vertebrae by the loss of one to two vertebrae, returning to the condition found in the ancestral anabantids. This seems to indicate that reversed evolution or negative anisomerism as defined by Gregory (1951) has taken place. The reduction of vertebrae may be correlated with the small size of Sphaerichthys and Colisa lalia. Myers (1958) ob- served that there are morphological trends correlated with small size in teleosts, as for example reduction in number of fin rays and scales, and reduction or suppression of the lateral line. The lateral line of Sphaerichthys is vestigial. The low number of vertebrae in Sphaerichthys and Colisa is also correlated with the quick-dodging method of loco- motion. Members with a high number of vertebrae, as for example Trichogaster, Betta, and Trichopsis, move in a more sinuous or eellike way. Secondary polyisomerism can also be traced in the evolution of the anal pterygiophores (Table 3). In the ancestral anabantids the total number of anal pterygiophores varies between 13 and 19. This num- ber has increased in all three derivatives of the anabantids, as fol- lows: Belontiidae, 28-35, Helostomatidae, 30-32, and Osphronemidae, 32. In studying the evolutionary changes in the postcranial axial skele- ton of teleosts it is essential to take the environmentally induced varia- tions into consideration. Lindsey (1954) has studied the temperature- controlled meristic variation in Macropodus opercularis. His results indi- cate that at developmental temperatures of 31° C, 28° C, and 24° C the average total number of vertebrae is, respectively, 28.78, 28.00, and 28.49. Lindsey included the ultimate vertebra in his counts. In my counts I have omitted the ultimate vertebra. Taning (1950) also showed that there is a definite influence of temperature on the number of vertebrae in teleosts. The variation in vertebral counts in the anabantoids, how- ever, does not seem to be induced environmentally, because of the fol- lowing two reasons: the anabantoids occur in the tropical habitat, in which there is a rather uniform temperature, and the degree of varia- tion is much greater than the experimentally induced differences. Pelvic girdle (figs. 91-102). The position of the pelvic girdle in the anabantids is nearly horizontal. It is attached indirectly to the cleithral symphysis by means of a ligament. This arrangement is characteristic of the ancestral percoid and represents the incipient stage in the evolu- tion of the anabantoid pelvic girdle. The pelvic girdle in the Belontiinae resembles that of the anabantids rather closely, but differs in the direct attachment of the conjoined rachii of the basipterygia to the cleithral symphysis. A progressive change in the position of the basipterygia, ac- companied by the development of various processes, can be found in the Macropodinae and Trichogasterinae. The rachii of the basipterygia are projected dorsolaterally and articulate with the cleithra of the DISCUSSION 59 corresponding sides in Betta, Trichopsis, Macropodus, and Sphaerichthys. The approximate angle between the horizontal axis and the dorso- laterally directed rachii in Betta, Trichopsis, Macropodus, and Sphae- richthys is, respectively, 30°, 45°, 60°, and 75°. Concomitant with the change in form and position of the rachii is the differentiation of a prominent processus medialis anterior and a marked elongation of the first pelvic fin ray. This anatomical arrangement enables the fish to move the pelvics anteriorly at an angle of 90° with the long axis of the body. The functional significance of this mobility of the pelvics has not been experimentally studied. The bright colors on the pelvics indicate that it may play a role in the reproductive and aggressive behavior of these fishes. Trichogaster and Colisa represent the most specialized stage in this evolutionary series. The elongate rachii of the basipterygia are di- rected straight dorsally and articulate with a special fingerlike process on the cleithra. The processus medialis anterior is extremely well de- veloped in Colisa. A new process, the processus medialis accessorius, is developed in service as the origin of the superficial abductor muscles. The first fin ray is very elongate in both genera. The remaining pelvic fin rays become vestigial in Trichogaster and are completely lost in Colisa. Steinbach (1950) studied the mechanical aspects of the pelvic appendages of Trichogaster. Both Trichogaster and Colisa can move the elongated fin ray in any direction. Scharrer, Smith, and Palay (1947) described taste buds on the elongate pelvic fin ray which is innervated by the facial nerve. The new function of the pelvic fins of Trichogaster and Colisa is both gustatory and tactile. Taste buds have not been found in the other belontiids. The new adaptive structure, the modified pelvic fin ray in Trichogaster and Colisa, has developed by an intensi- fication of a former morphological trend already present in its ances- tors, as for example in Macropodus. The new adaptive structure serves a quite different function from the original locomotory and ethological purposes. The morphological trend involves a hypertrophy of the first pelvic fin ray and atrophy of the remaining four rays. This evolutionary trend furnishes an example whereby a structure adaptive in one way in an ancestral group, namely the typical locomotory function of the pelvics in the anabantids, is transformed to serve a different function in a descendent group, in this case the combined tactile and gustatory pelvics in Trichogaster and Colisa. In the helostomatids the rachii are widely separated and articulate directly with the cleithra at an angle of approximately 15°. The pelvic fin rays of the osphronemids exhibit a superficial resemblance to that of Trichogaster. The first pelvic fin ray is very elongate and the remaining rays vestigial. The internal anatomy of the girdle, however, differs considerably from that of Trichogaster. The processus medialis accessorius is lacking, the processus medialis 60 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI anterior is merely a ridge on the basipterygium, the rachii of the basip- terygia are directed dorsomedially, and the pelvic spine is well de- veloped. The evolution of the pelvic girdle seems to indicate that the belontiids and osphronemids have arisen independently from each other. General remarks. It should be emphasized that the evolutionary trends discussed do not occur singly. It is a general rule that a change in one character almost always carries others along with it, by genetic corre- lations, by ontogenetic mechanisms, and by adaptive correlations. As Simpson (1953) pointed out, the interdependency in the total adaptive situation is generally too complex to permit definite determination of whether a given trend is really primary in an absolute sense or that one of two simultaneous trends is independent and the other dependent. It is plainly evident that the trends in protrusility of the jaws and the dentition are closely related to each other and produced by func- tional and adaptive mechanisms. It is a congruence of functions corre- lated with a change in feeding habits. It is, however, quite possible that the change in position and extent of the suprabranchial cavity is corre- lated with the changes in dentition and protrusility. The change in size and position of the suprabranchial cavity is made possible by the de- velopment of the pharyngeal processes, which in turn limits the size of the food. In Sandelia and Ctenopoma the suprabranchial cavity has shifted to a slightly more ventral position without the development of pharyngeal processes. The development of the pharyngeal processes on the parasphenoid and basioccipital is accompanied by a simultaneous shift and reduction in size of the suprabranchial cavity in all three phyletic lines derived from the anabantid stock. This fact seems to indi- cate a definite correlation between the two changing features. The de- velopment of the posttemporal fossae and the deepening of the body are certainly correlated. The ventral shift of the suprabranchial cavity is a mechanical concomitant with the formation of the posttemporal fossae, which takes the space formerly occupied by the cavity. An im- mediate result of these interdependent trends is a gradual deepening of the skull, which in turn depends on the shape of the body (Gregory, 1933). The body form is probably controlled by an ontogenetic mecha- nism, namely a distinctive growth pattern, which changes because of — genetic alterations produced by selection. The simultaneous changes in dentition, protrusility of the jaws, the shift and extent of the supra- branchial cavity, the development of the posttemporal fossae, and the body form seem to provide a complex example of correlated changes in which genetic, ontogenetic, and functional correlations are simultane- ously involved. The attempt to isolate intricately interdependent features and factors is therefore highly artificial. An apparent uniformity of trends is present in the phylogeny of the DISCUSSION 61 anabantoids. Parallel changes in dentition, position and size of the suprabranchial cavity, posttemporal fossae, body form, arrangement of branchiostegal rays, and secondary polyisomerism in the vertebral column occur in all three phyletic lines derived from the ancestral ana- bantid stock. Simpson (1953) indicated that the occurrence of similar trends among members of a given group is quite common. The phe- nomenon that diverse descendent lines of common origin show parallel trends is often called “programme evolution” (George, 1948). An apparent basis for parallel changes in related forms lies in the fact that the organisms share a rather similar genotype. A group of related and genetically similar forms would be similar in their pheno- typical expression (Bock, 1960). The interdependence between the different structures should not be interpreted to mean that they do not change independently. The indepen- dent changes of structure within evolving groups are well known under the concept of mosaic evolution. The dependency and interdependency of evolving structures have an important bearing on their taxonomic usefulness. Interdependence is related to grouping of structures into functional complexes. Independence, on the other hand, is shown by the fact that functional complexes can break down and can be reorganized during evolutionary change. E. THe PHYLOGENY OF THE ANABANTOIDEI The proposed phylogeny of the anabantoids, based on the morpho- logical data and interpretations outlined in previous sections, is sum- marized in figs. 103 and 104. The anabantids represent the most primitive family of the Anaban- toidei. The presence of large extrascapulars, the toothed prevomer, pala- tine, and parasphenoid, and the articulation of the metapterygoid with the symplectic process of the hyomandibular are indicative of the initial stage in anabantoid phylogeny. Anabas testudineus is known from Plio- cene and Pleistocene deposits of Java (Sanders, 1934). A well-preserved specimen of Osphronemus goramy? has been found in Sumatran marl shales dating from the early Tertiary (Sanders, 1934). According to Ro- mer (1945), a wide variety of teleosts evolved in the Cretaceous, and all major teleost groups were established by the Eocene. The possible occurrence of the highly specialized Osphronemus goramy in the early Tertiary suggests that the four anabantoid families were already differ- entiated by that time. The ancestral anabantids seem to have originated from a percoid stock during the Upper Cretaceous or Paleocene, in the tropical portion of the Oriental region. The great abundance and di- * The possibility that this fossil does not represent Osphronemus goramy but that it has paralleled this species closely must be left open. 62 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI versity of the recent Anabantoidei in Asia supports the hypothesis that the geographical origin of this group has been most likely tropical Asia (Darlington, 1957). It is also generally accepted that the represen- tation of Anabas (erroneously taken to include Ctenopoma) in Africa is evidence of a late post-Tethys immigration of the anabantids into Africa (Kosswig, 1954, and Steinitz, 1954). But it has been shown that Anabas and Ctenopoma are widely separated genera. Anabas is not represented in Africa. Anabas is certainly not the direct ancestor of Ctenopoma, since it possesses several specialized features not found in Ctenopoma. The hypothesis that Anabas and Ctenopoma are derivatives of a common an- cestral anabantid stock is supported with strong osteological evidence. It is assumed that this ancestral form evolved during the Upper Cretaceous or Paleocene in the tropical part of the Oriental region, from where it radiated into Africa. The invasion into Africa probably occurred in the early Tertiary (Darlington, 1957). Darlington also pointed out that from the Mesozoic into the Tertiary the Tethys Sea stretched across the whole of southern Europe and Asia to the East Indies. But the Tethys was not a stable sea, and the region was constantly changing, with changing barriers and bridges, and an immense changing frontage between salt and fresh water. There is considerable evidence from mammalian fossils in the Upper Eocene and Lower Oligocene of El Faiyum (northern Egypt) that Africa and Eurasia were united in the late Eocene. Picard (1943) also pointed out that land seems to have existed in the higher parts of Transjordan and Cisjordan since the Upper Eocene. The ancestral anabantid form seems to have invaded Africa during the Upper Eocene. It differentiated into the genus Ctenopoma in Africa, whereas in Asia it gave rise to (1) Anabas, known since the Pliocene and Pleistocene, (2) the osphronemids, known from the early Tertiary (Sanders, 1934), (3) the helostomatids, and (4) the belontiids. No fossil records exist of the latter two families. Steinitz (1954) pointed out that the majority of the fishes found in both Asia and Africa supports the hypothesis of a connection between the two continents during the Miocene. Kosswig (1954) favors a Pliocene land connection. Steinitz remarked that a continuous land mass between the Asiatic and African continents became established only during the Lower Miocene. It should be emphasized, however, that the Tethys Sea was unstable, and that the bridges and barriers were constantly changing. The ancestral ana- bantid form might have been able to cross barriers which were in- accessible to other fresh-water fishes. Anabas and Ctenopoma are extraor- dinarily hardy fish, surviving in spite of very poorly oxygenated water because of their remarkable air-breathing capacity. They are able to live for prolonged periods outside the water. Anabas testudineus has more than usual salt tolerance (Weber and de Beaufort, 1922, Giltay, DISCUSSION 63 1933, Sanders, 1934, and Delsman, 1951). Numerous specimens are col- lected daily from the Bay of Djakarta (Java). Anabas behaves as a secondary division fresh-water fish and is, very likely, able to cross narrow sea barriers as reported for Aplocheilus, Rasbora, and Puntius (Myers, 1951, and Darlington, 1957). The assumption that Anabas has been carried by man across Wallace’s line to the eastern part of the Indo-Australian Archipelago (Myers, 1937) is difficult to support. Os- phronemus goramy, which is as hardy a fish as Anabas but is far su- perior as a food fish, has not been carried across Wallace’s line by man. Osphronemus would probably be preferred above Anabas for trans- portation by the natives. It seems therefore less questionable to ascribe the wide distribution of Anabas to its exceptional physiological charac- teristics. Similar physiological traits would enable the ancestral anaban- tid form to invade Africa during the Eocene whether there was a con- tinuous land bridge between Africa and Asia as proposed by Darlington (1957) or not, as hypothesized by Steinitz (1954). The presence of a broad connection, Lemuria, between Africa, Madagascar, and India in the Tertiary has been conclusively disproved (de Beaufort, 1951). Following the paleogeographic suggestions by Steinitz (1954) we may assume that the ancestral anabantid stock migrated into Africa during the Miocene. After the complete breakdown of Africa’s land connection with Asia by the Erythrean Rift Valley in the Pliocene, the ancestral stock differentiated into the genus Ctenopoma in Africa and Anabas in Asia. The presence of Osphronemus goramy in the early Tertiary does not support this hypothesis, unless we assume that the ancestral anaban- tid had already split into the three more specialized families before or during the period of radiation. Darlington (1957) favors an early Tertiary migration of the ancestral anabantid stock into Africa. The westbound migration of the ances- tral form would have taken place simultaneously with the differentiation of the osphronemids, helostomatids, and belontiids in Asia. Parallel evolutionary patterns are found in the Ostariophysi (Darlington, 1957). The evolution of the Anabantoidei corresponds with the general evolu- tionary patterns of fresh-water fishes on the Asiatic and African conti- nents. Africa has been relatively stable, favoring the survival of archaic forms, whereas Asia has somehow favored the evolution of new groups of fishes. There is decisive evidence against the direct derivation of the osphronemids, helostomatids, and belontiids from Anabas. Anabas prob- ably evolved from the ancestral stock after Africa’s land connection with Asia was broken down during the Lower Pliocene (Steinitz, 1954). Osphronemus goramy was already present during the Lower Tertiary; these facts indicate that Osphronemus evolved before Anabas, from a common anabantid stock. The osteological and zoogeographical evidence 64 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI seems to indicate that the archaic recent Anabantidae, the specialized Osphronemidae, Helostomatidae, and Belontiidae arose during the Lower Tertiary from a common ancestral anabantid stock. No attempt has been made to reconstruct the hypothetical common ancestor. The set of characters found in both Ctenopoma and Anabas may be regarded as those which the group owed to its common ancestor. The common ancestor constructed by abstraction from known forms represents an archetype as understood by the Naturphilosophen and Owen, and may differ considerably from the actual common ancestor. The genus Sandelia, isolated on the southern tip of South Africa (Barnard, 1943), has been derived from Ctenopoma. Sandelia exhibits several specializations, as for example the loss of one branchiostegal ray, the reduction in size of the suprabranchial cavity, the simple first epibranchial, and the separated nasals. But the toothed palatine and prevomer, the spineless serrate opercular, and the arrangement of the pharyngeal processes on the basioccipital indicate the close relationship with Ctenopoma. There is also strong zoogeographical evidence for the direct derivation of Sandelia from Ctenopoma. The present distribu- tional pattern of the African anabantids indicates that Ctenopoma has spread along a principal route which leads from an extensive, favorable area of the African tropics into a smaller, less favorable area of tem- perate South Africa. The expanding Ctenopoma population apparently crossed the climatic barrier with difficulty and the group across the barrier became isolated and differentiated independently from its an- cestors. As Darlington (1957) remarked, the South African fauna as a whole has been formed by southward extension of parts of the tropical African fauna. It appears that Anabas and Ctenopoma evolved by schistic (Smith, 1956 and 1960) evolution or splitting from a common ancestor. Geographic isolation seems to have brought about differences between the two separated populations. Sandelia, on the other hand, evolved by phyletic evolution from Ctenopoma. A progressive change in some features can be followed in the series Ctenopoma muriei from central tropical Africa, Ctenopoma multispinis from the Zambezi River system, and Sandelia from temperate South Africa: the lachrymal is weakly serrate in muriei, not serrate in multispinis and Sandelia; suborbitals extend to nearly half the distance between orbit and preopercular in muriei, suborbitals narrow in multispinis, suborbitals very narrow in Sandelia; a gradual but progressive reduction in size of the suprabran- chial cavity and an increase in length of the ascending process of the pre- maxillary in the sequence muriei, multispinis, and Sandelia. During the Lower Tertiary three divergent lines radiated more or less simultaneously from the ancestral anabantid stock. In the first line, the osphronemids, the transverse processes of the parasphenoid are DISCUSSION 65 lost, the ectopterygoids retained, the suborbital shelf becomes greatly reduced, the basipterygia and first pelvic fin ray greatly elongate, and the metapterygoid is divested of its articulation with the symplectic process of the hyomandibular. The second line is represented by the highly specialized helostomatids, in which transverse processes and pharyngeal processes of the parasphenoid are retained, the ectoptery- goid is not lost, the dentition is completely lost, the dentary and angular acquire a highly movable hinge joint, and the metapterygoid does not articulate with the symplectic process of the hyomandibular. The third line represents the largest group, the belontiids, in which the ectoptery- goids are lost and the transverse processes of the parasphenoid are not retained. The three groups seem to have originated by schistic evolu- tion from the same ancestral adaptive type, entering into diverging adaptive zones. Osphronemus goramy made a rather sudden appearance in the Lower Tertiary, most likely the Upper Eocene, and has not changed since. The fossil specimen resembles the recent forms in great detail (Sanders, 1934). If we assume that the fossil is actually Osphro- nemus goramy then the species did not change during a period of ap- proximately 36 million years. The rather sudden appearance of the osphronemids from the ancestral anabantids and the complete absence of intermediate stages seem to indicate that the change from the carniy- orous ancestral adaptive type to the large-sized omnivorous forms might have taken place quickly. This evolutionary change at an excep- tionally high rate might be considered as tachytelic. Tachytely often occurs when populations shift from one major adaptive zone to another (Simpson, 1953). In this particular case the shift is from purely pre- daceous feeding habits to omnivorous adaptive types. The very elongate first pelvic fin ray, proven to be gustatory and tactile, probably evolved as a concomitant of the feeding behavior. The possible tachytelic line leading toward the differentiation of the osphronemids became brady- telic. Osphronemus has not changed morphologically since the Upper Eocene. Osphronemus is rather broadly adapted: it can live on a variety of foods (Weber and de Beaufort, 1922), is able to survive in poorly oxygenated water, and has a high salt tolerance. Environmental factors may therefore fluctuate greatly without requiring any adaptive change for survival. These factors may have contributed to the arrested evolu- tion of Osphronemus since the Upper Eocene. Helostoma temmincki represents the most specialized member of the anabantoids. The morphological gap between Helostoma and the an- cestral stock is exceptionally large. The edentulous condition, the unique movable hinge joint between the dentary and angular, and the elaborate filtering apparatus in the hyobranchial region are, undoubtedly, ad- vanced features adaptive to a highly specialized feeding behavior. 66 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI Transitional forms between Helostoma and the predaceous ancestral stock are lacking. This sudden evolutionary change from the predaceous adaptive zone to the planktonic adapted forms seems to indicate that an all-or-none reaction is involved whereby no intermediate stages per- sisted. Once the direction of evolution of the feeding mechanism was established, natural selection continuously acted in that direction until completion of the trend. We are possibly confronted with a beautiful example of quantum evolution. The all-or-none element in quantum evolution arises in general from discontinuity between adaptive zones. It seems that the direction of the relatively rapid change in the evolu- tionary history of Helostoma was rigidly adaptive as a result of strong selection pressure. In this case quantum evolution has, possibly, led to a group of fish whose adaptive zone and correlated morphology are such that a family rank should be established. After penetration into the specific adaptive zone during the Lower Tertiary, the evolution of the helostomatids seems to have been arrested. Helostoma seems to be perfectly adapted to its narrow zone. It has been pointed out by Simpson (1953) for animals and Stebbins (1949) for plants that the lines with arrested evolution were specialized when their evolution was arrested. The evolutionary pattern of Helostoma seems to fit the generalization that the most slowly evolving groups seem to be very highly and spe- cifically adapted to a particular zone. The third evolutionary line originating from the ancestral anabantid stock gave rise to the Belontiidae. The evolutionary pattern of the belontiids is mainly a combination of phyletic and schistic evolution. The Belontiinae have several features in common with the ancestral anabantid stock, among which the closely conjoined basipterygia and the articulation of the metapterygoid with the symplectic process of the hyomandibular are the most important. Belontia seems to resemble an intermediate stage between the ancestral anabantid stock and the more advanced belontiids. Some specialized features not found in any other belontiid, as for example the posterior extension of the suprabranchial cavity supported by the first six epipleurals and the high supraoccipital crest, indicate strongly that Belontia forms an early, sterile, side branch of the main evolutionary line. The main line, originating from the same ancestral stock as the Belontiinae, leads to a progressively changing series, which consists of Betta, Trichopsis, and Macropodus. The latter genera represent the subfamily Macropodinae, which differs from the Belontiinae in lacking a posterior extension of the suprabranchial cavity and a distinct supraoccipital crest, and the loss of the articulation be- tween the metapterygoid and the symplectic process of the hyomandib- ular. These differences suggest that the Belontiinae and Macropodinae represent two divergent lines. In the belontiine line there is a tendency DISCUSSION 67 toward deepening of the skull by the differentiation of a high supra- occipital crest and a posterior extension of the suprabranchial cavity, whereas in the macropodine line the deepening of the skull is the result of the development of pharyngeal processes on the parasphenoid and basioccipital. We can trace the following gradual and _ progressive changes: reduction in dentition of the parasphenoid, deepening of the body, increase in protrusility of the jaws, and increase in size of the pharyngeal processes of the parasphenoid and basioccipital. These pro- gressive changes are evidently nonrandom and cumulative and appar- ently adaptive. The morphological changes are correlated with the changing feeding habits. The Macropodinae are omnivorous, and there is a trend toward specialization for eating small-sized food. It should be emphasized that the macropodine evolutionary pattern is not purely phyletic, but the longest line leading to Malpulutta takes a branch from Trichopsis, which was produced by splitting as indicated by the aberrant elongate and tubular nasals. The Trichogasterinae forms a homogenous group in which the trends present in the Macropodinae reach their peaks. The group probably originated from a Macropodus-like stock, by an intensification of the trends already present in the Macropodinae. The occupation of con- tiguous and successively higher and narrower zones by a sequence of populations splitting off from the next lower zones seems to have taken place during the evolution of the Macropodinae and Trichogasterinae. The progressive specialization culminated in the penetration of the narrow adaptive zone by the herbivorous Trichogasterinae, which feed on a diet of small plant parts, epiphytes, plankton, filamentous algae, and detritus. It seems that all zones remain in existence and are occu- pied. In the line leading from Betta to the Trichogasterinae, we can trace a sequential occupation of zones decreasing in width. Betta is omniv- orous, whereas the Trichogasterinae are herbivorous and nearly plank- tonic. The morphological differences between the Macropodinae and the Trichogasterinae are of such magnitude that the rank of subfamilies should be recognized. The position of the anabantoids among the Perciformes is still ques- tionable. The ancestors of the anabantoids are still unknown, but are perhaps to be found among the Percoidei, which occupies a central place in the order Perciformes. The anabantoids arose during a major episode of proliferation of teleostean suborders, families, and genera. During this “explosive phase” of adaptive radiation (Upper Cretaceous and Lower Tertiary) of teleosts there was, probably, a considerable phyletic divergence among the evolving lineages, penetrating into a great variety of adaptive zones by successive quantum shifts. Among the crowding lines parallelism and convergence were at work. The 68 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI phylogeny of teleosts and phylogeny in general are very poorly under- stood. In this discussion I have attempted to analyze some of the inter- woven modes and factors of the evolution of the anabantoids. But as Simpson states, relative to adaptive radiation: “The total process cannot be made simple, but it can be analyzed in part. It is not understood in all its appalling intricacy, but some understanding is in our grasp, and we may trust our own powers to obtain more.” THE STATUS OF LUCIOCEPHALUS PULCHER (GRAY) The taxonomic status of the monotypic genus Luciocephalus among the teleosts seems to be problematical. Earlier workers (Day, 1876, Regan, 1909, and Weber and de Beaufort, 1922) lumped Luciocephalus with the anabantoids. Boulenger (1904), however, established a sep- arate family Luciocephalidae. Berg (1940) made a study of the skeleton of Luciocephalus, using two radiographs from one specimen. He as- signed it to a new monotypic suborder, the Luciocephaloidei. A full description of the structure of Luciocephalus with a discussion on its taxonomic status will be presented elsewhere (Liem, unpublished ). Only a brief comparison with the Anabantoidei is given here: Luciocephalus Anabantoidei Median gular element present. Median gular element absent. No pharyngeal processes on para- Pharyngeal processes present on para- sphenoid and basioccipital. sphenoid and basioccipital. No swim bladder. Swim bladder extending into caudal region. No dorsal and anal spines. Dorsal and anal fins with spines. Suprabranchial organ rectangular, ar- Suprabranchial organ labyrinthine, ticulating directly with prootic. articulating by means of an epi- branchial process. No ethmoid process on palatine. Ethmoid process present on palatine. Anteroventral and posteroventral cor- Dorsoanterior corner of sacculith in- ners of sacculith distinctly in- dented. dented (fig. 90 D). Luciocephalus has been classified with or near the anabantoids be- cause of the presence of a suprabranchial organ. Previous workers made the assumption that the suprabranchial organ provides a good character for establishing relationships of the ophicephalids, anabantoids, and luciocephaloids. It has been pointed out (Bock, 1960) that the taxonomic value of a structure depends upon the nature of the controlling selec- tion forces. If these selection forces have a tight control over the structure and if they have arisen repeatedly then that structure has little taxonomic value. In order to evaluate the proper taxonomic relation- ships between the anabantoids and Luciocephalus we have to distin- guish between homology and convergence among the similarities present in the two groups. Air-breathing has evolved, independently, in many bony fishes, as an adaptation to environmental conditions which render aquatic respiration impossible or inefficient. A great variety of organs be- came transformed into air-breathing organs (Carter, 1957), as for ex- ample the pharyngeal epithelium and diverticula of the pharynx in Peri- ophthalmus, Electrophorus, and Monopterus; swim bladder and lungs in 69 70 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI Polypterus, Amia, the recent Dipnoi, Lepisosteus, Umbra, Erythrinus, and Gymnarchus; the esophagus in Monopterus (Liem, 1961); stomach and intestine in Misgurnus, Doras, etc.; the skin in Anguilla, Periophthalmus, and Boleophthalmus; and diverticula of the branchial chamber in Clarias, Saccobranchus, Channa, Luciocephalus, and the anabantoids. In the South African genus Sandelia the suprabranchial organ became sec- ondarily reduced, indicating that the controlling selection forces have reversed their direction during the evolution of the anabantoids. It seems that the controlling selection forces guiding the evolution of the diver- ticula of the branchial chamber have arisen repeatedly and have re- versed their direction several times. Therefore, one has to ascribe little value to the suprabranchial organ in showing relationships between Luciocephalus, anabantoids, and ophicephalids. Independent convergent adaptations to air-breathing have evidently evolved in many bony fishes of quite different ancestries. Convergence generally involves adaptations to ecologically similar conditions by various groups of oganisms of dis- tinct ancestries (Simpson, 1961). Luciocephalus resembles the anaban- toids superficially in respect to the modified first epibranchial, which is specifically adaptive to environments in which air-breathing has a great selective value. The two groups resemble each other only in one way, specifically adaptive to a shared ecology. In characteristics not related to that adaptation, for example the otoliths, the oromandibular appa- ratus, and the form and extent of the frontals, the two groups differ greatly. The resemblance between Luciocephalus and the anabantoids seems, therefore, to be a case of convergence without any bearing on propinquity of descent. It is impossible, at this stage, to determine the taxonomic status of Luciocephalus with any certainty. There is evidence that Luciocephalus is not related to the anabantoids. The combination of primitive iso- spondylous and advanced acanthopteran characters gives Luciocephalus a unique position among the teleosts. It is well known that many origi- nally soft-rayed stocks have progressed in certain features toward the spiny-rayed stage of evolution, producing parallelisms and convergence on a grand scale in many different lines. Luciocephalus might represent a relict of the original soft-rayed stock which has progressed in certain features toward the spiny-finned stage of evolution. It seems possible that we are dealing with a form with ordinal affiliation. RELATIONSHIPS OF THE ANABANTOIDEI AND THE OPHICEPHALIFORMES Many authors unite the ophicephalids with the anabantoids (Gregory, 1951). The osteology of the Ophicephaliformes (Channiformes) has been described in great detail by Day (1914) and Bhimachar (1932). The morphological differences between the anabantoids and the Ophi- cephaliformes are: Ophicephaliformes Fins without spines. Metapterygoid articulates with fron- tal and sphenotic. Pharyngeal processes of parasphenoid and basioccipital absent. First epibranchial modified, articulat- ing with supratemporal. Frontals articulate with parasphenoid. Outer wall of otic bulla formed al- most completely by prootics. Anabantoidei Fins with spines. Metapterygoid small, widely sepa- rated from sphenotic and frontal. Pharyngeal process present on para- sphenoid and basioccipital. First epibranchial modified, articulat- ing with prootic. Frontals do not articulate with para- sphenoid. Outer wall of otic bulla formed by prootics, basioccipital, and ex- occipital. These osteological differences support Berg’s (1940) proposal of a new order Ophicephaliformes. The inclusion of the ophicephalids in the anabantoids was based solely on the presence of a modified first epi- branchial. The modification is an adaptive device for air respiration, which provides a considerable selective advantage in natural selection for fishes in general, especially in habitats with periodic droughts dur- ing which high temperatures decrease the availability of oxygen in the water. The superficial resemblance between the ophicephalids and ana- bantoids is limited to one structure which is specifically adaptive to a shared ecology. A closer study of the morphology of the suprabranchial organ of the two groups reveals several significant differences. In the Ophicephaliformes the suprabranchial organ is formed by the epi- branchial and an expansion of the hyomandibular and articulates with the supratemporal. In the anabantoids the suprabranchial organ is formed by the first epibranchial only and articulates with the prootic. The first epibranchial becomes modified in different ways in the ana- bantoids, luciocephalids, and Ophicephaliformes. Functionally, the first epibranchial is not associated with the digestive system as is the case in the three remaining epibranchials. The first epibranchial seems there- fore “immediately available” for performing a respiratory function with- 71 2, COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI out affecting the digestive mechanisms. The Ophicephaliformes and the anabantoids differ in many features which are not related to the shared ecology. Berg’s (1940) observation that the Ophicephaliformes and ana- bantoids are not related and that their resemblance is due to convergence is strongly supported by the new morphological evidence. The Ophi- cephaliformes resemble the Perciformes in lacking an orbitosphenoid, the absence of the mesocoracoid in the pectoral girdle, and the arrange- ment of the premaxillary and maxillary. But they possess several primi- tive features which separate them from the Perciformes, as for ex- ample the absence of spines in the fins, the fact that the pelvic girdle is not directly attached to the cleithra, and the arrangement of the branchiostegal rays. The loss of spines in the fins and the separation of the basipterygium from the cleithrum might have occurred secondarily. It seems that specu- lation on the ordinal affiliation of the ophicephalids is premature at the present time. A knowledge of the different selection forces which were operative during the evolution of the different anatomical struc- tures of the ophicephalids is essential for determining the proper taxo- nomic value of the structures. The “labyrinth organ” should not be used as the sole character for determining relationships between the ophi- cephalids and anabantoids. There seems to be some indication that the selection forces have a tight control over the “labyrinth organ.” The modified epibranchial would therefore have little taxonomic value. The use of a large number of different taxonomic characters will reduce the chance of error considerably. The Ophicephaliformes (Channiformes ) differ significantly from the anabantoids in characters which are not related to the adaptation to air-breathing. The taxonomic distinctness of the ophicephalids from the anabantoids is evident from the articula- tion of the metapterygoid with the frontal and sphenotic, the absence of the pharyngeal processes of the parasphenoid and basioccipital, the connection between the frontals and the parasphenoid, and the con- struction of the otic bulla by the prootic. The taxonomic position of the ophicephalids as a separate order as advocated by Berg seems to be justified, although more supporting evidence is necessary. CONCLUSIONS 1. Based on the obtained osteological data the suborder Anaban- toidei is recharacterized as follows. A suprabranchial organ, sometimes labyrinthine, is formed by expansion of first epibranchial. Nasals vari- able in shape and size, united or separated. Swim bladder divided pos- teriorly, extending into caudal region. Opercular bones serrate or entire. Pharyngeal processes present on parasphenoid and basioccipital. Ventrals thoracic, I 1-5, spine sometimes vestigial. Dorsals and anals always with spines. Scales ctenoid and cycloid. Vertebrae 25-31. 2. Four families are recognized within the Anabantoidei: a) Anabantidae, including the genera Anabas, Ctenopoma, and Sandelia. Characterized by teeth on prevomer, extrascapulars, toothed transverse processes on parasphenoid, ectopterygoids, and suborbital shelf present. b) Osphronemidae, including the genus Osphronemus. Character- ized by no teeth on prevomer, no extrascapulars, toothed pharyngeal process on parasphenoid, transverse processes of parasphenoid absent, ectopterygoids present, and suborbital shelf vestigial. c) Helostomatidae, including the genus Helostoma. Characterized by premaxillaries, dentaries, palate and pharynx devoid of teeth, dentary and angular articulate by a hinge joint, pharyngeal and transverse proc- esses present on parasphenoid, ectopterygoids present, suborbital shelf absent. d) Belontiidae, n. fam., characterized by the absence of teeth on prevomer, no extrascapulars, pharyngeal process present on parasphe- noid, ectopterygoids absent. Three new subfamilies are proposed within the Belontiidae: (1) Belontiinae, n. subfam., including one genus, Belontia. Dis- tinguished by: metapterygoid articulates with symplectic process of hyomandibular, suprabranchial cavity extends far posteriorly, supported by six first epipleurals, supraoccipital crest well developed, epiotics do not reach foramen magnum, six branchiostegal rays, last of which articu- lates with epihyal, and suborbital shelf formed by suborbitals 2-5. (2) Macropodinae, n. subfam., including the genera Macropodus, Betta, Trichopsis, Parosphromenus, and Malpulutta. Distinguished by: metapterygoid does not articulate with symplectic process of hyoman- dibular, suprabranchial cavity not supported by epipleurals, no supra- occipital crest, epiotics do not reach foramen magnum, six branchiostegal rays articulating with ceratohyal, suborbital shelf formed by suborbit- als 2-5. 73 74 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI (3) Trichogasterinae, n. subfam., which includes Sphaerichthys, Parasphaerichthys, Colisa, and Trichogaster. Distinguished by: metap- terygoid does not reach symplectic process of hyomandibular, supra- branchial cavity not supported by epipleurals, no supraoccipital crest, epiotics reach foramen magnum, five branchiostegal rays articulating with ceratohyal, suborbital shelf formed by second suborbital. 3. Macropodus opercularis differs significantly from Macropodus cupanus, but neither generic nor subgeneric segregation is here recom- mended. 4. Phylogenetic trends have been traced in three evolutionary lines derived from the anabantids in respect to (a) a gradual and progressive increase in the protrusility of the premaxillary; (b) a gradual reduction in dentition; (c) an increase in number of gill rakers; (d) a ventral shift and reduction in size of the suprabranchial cavity; (e) an increase in size and change in position of the pharyngeal processes of the para- sphenoid and basioccipital; (f) a progressive change from short-bodied cylindrical to deep-bodied compressed forms; (g) a gradual develop- ment of posttemporal fossae; (h) a reduction in the extent of the sub- orbital shelf; (i) a forward shift of the branchiostegal rays; (7) a re- duction in number of bones in the skull with increasing individuality of those remaining; (k) polyisomerism in the vertebral column and anal pterygiophores; (1) a gradual change in position and form of the basip- terygia. 5. The complex interdependency in the total adaptive situation is discussed. Parallel changes occur in the three lines derived from the anabantids. 6. The phylogeny of the Anabantoidei is summarized (figs. 103, 104). 7. The anabantids resemble the ancestral stock from which the os- phronemids, helostomatids, and belontiids evolved during the Lower Tertiary. 8. The presence of Osphronemus goramy in the Upper Eocene seems to indicate that the original tachytelic line became bradytelic. 9. The Helostomatidae seems to have evolved by a rapid shift into a new zone. The discontinuity between the adaptive zones, correlated with the large morphological gap without any intermediate stages, seems to indicate that the shift took place by quantum evolution. 10. Several aspects of the paleogeography and zoogeography have been incorporated in the discussion of the phylogeny. ll. The suprabranchial organ of the anabantoids differs significantly from that of the luciocephalids and Ophicephaliformes. 12. Supporting morphological evidence is given for Berg’s proposal that the Ophicephaliformes represent a separate order. 13. The resemblance of the Anabantoidei, luciocephalids, and Ophi- cephaliformes is due to convergence rather than phylogenetic relation- ship. LITERATURE CITED Apams, L. A. 1940. Some characteristic otoliths of American Ostariophysi. J. Morph., 66: 497-527. Au-Hussaint, A. H. 1947. The feeding habits and the morphology of the alimentary tract of some teleosts. Publ. Mar. Biol. Station Ghardaqa (Red Sea), 5: 4-61. Baper, R. 1937. Bau, Entwicklung und Funktion des akzessorischen Atmungs- organs der Labyrinthfische. Zeitsch. wiss. Zool., 149: 323-401. Barnarp, K. H. 1943. Revision of the indigenous freshwater fishes of the S.W. Cape region. Ann. S. African Mus., 36: 101-262. BARRINGTON, E. J. W. 1957. Alimentary canal and digestion. In Physiology of fishes, I. Academic Press, New York, 447 p. BeaurorT, L. F. pe. 1951. Zoogeography of the land and inland waters. Sidgwick and Jackson, London, 208 p. Berc, L. S. 1940. Classification of fishes both recent and fossil. Edwards Brothers, Ann Arbor, Michigan, 517 p. . 1958. System der rezenten und fossilen Fischartigen und Fische. Veb. Deutscher Verlag der Wissenschaften, Berlin, 310 p. Bertin, L., and C. ARAMBouRG. 1958. Super-ordre des Téléostéens. In P. P. Grassé et al., Traité de zoologie, 13. Masson, Paris, 2758 p. Buimacuar, B. C. 1932. The cranial osteology of Ophicephalus striatus Bloch. J. Mysore Univ., 6: 72-86. Bock, W. J. 1960. The palatine process of the premaxilla in the Passeres. Bull. Mus. Comp. Zool. (Harvard), 122: 361-488. Boxer, H. 1937. Einfiihrung in die vergleichende biologische Anatomie der Wirbeltiere, 2. Gustav Fischer, Jena, 258 p. BouLENcER, G. A. 1904. Fishes. The Cambridge Natural History, 7: 541-727. . 1916. Catalogue of the freshwater fishes of Africa in the British Mu- seum (Natural History), 4: 1-392. Carter, G. S. 1957. Air Breathing. In Physiology of fishes, I. Academic Press, New York, 447 p. CurenoT, L. 1951. L’évolution biologique, les faits, les incertitudes. Masson, Paris, 592 p. DaruincTon, P. J. 1957. Zoogeography. John Wiley, New York, 675 p. Davis, D. D., and U. R. Gore. 1936. Clearing and staining skeletons of small vertebrates. Field Mus. Nat. Hist., Tech. Ser., No. 4: 3-15. Day, A. L. 1914. The osseous system of Ophicephalus striatus Bloch. Philip- pine J. Sci., 9D: 19-55. Day, F. 1876. The fishes of India. Taylor and Francis, London, 509 p. DeusMan, H. C. 1925. Fishes with protrusile mouths. Treubia, 6: 98-106. . 1951. Dierenleven in Indonesié. Van Hoeve, ’s-Gravenhage-Bandung, 344 p. DerantyvacGaLa, P. E. P. 1929. The Labyrinthici of Ceylon. Spolia Zeylan., 15. 79s . 1937. Malpulutta kretseri—a new genus and species of fish from Cey- lon. Spolia Zeylan., 20: 351-353. 75 76 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI DeEvILLERS, Cu. 1958. Le crane des poissons. In P. P. Grassé et al., Traité de zoologie, 13. Masson, Paris, 2758 p. DinEEn, C. F., and P. S. Sroxety. 1956. The osteology of the Sacramento Perch, Archoplites interruptus (Girard). Copeia: 217-230. Eaton, T. E. 1943. An adaptive series of protractile jaws in cichlid fishes. J. Morph., 72: 183-190. Forsetius, S. 1957. Studies of anabantid fishes. Zool. Bidrag Uppsala, 32: 93-597. Georce, T. N. 1948. Evolution in fossil communities. Proc. Roy. Phil. Soc. Glasgow, 72: 23-42. Gitcu, G. 1957. Vergleichende Untersuchungen eines hydrostatischen Ap- parates larvaler Labyrinthfische. Zool. Jahrb. Anat., 76: 1-62. Gitt, T. N. 1893. Families and subfamilies of fishes. Mem. Nat. Acad. Sci., 6: 127-138. Gittay, L. 1933. Resultats scientifiques du voyage aux Indes Orientales Néer- landaises. Poissons. Verh. Kon. Natuurh. Mus. Belgié., 5 (3): 1-129. Gircis, S. 1952. The bucco-pharyngeal feeding mechanisms in an herbivorous bottom-feeding cyprinoid, Labeo horie. J. Morph., 90: 221-315. Goopricu, E. S. 1931. Introduction to the structure and development of ver- tebrates. Macmillan, London, 837 p. Gos.inE, W. A. 1951. The osteology and classification of the Ophichthid eels of the Hawaiian Islands. Pacific Sci., 5: 298-320. . 1953. Hawaiian shallow-water fishes of the family Brotulidae, with the description of a new genus and notes on brotulid anatomy. Copeia: 215-225. . 1959. Mode of life, functional morphology, and the classification of modern teleostean fishes. Systematic Zool., 8: 160-164. GreENwoop, P. H. 1961. A revision of the genus Dinotopterus Blgr. (Pisces, Clariidae). With notes on the comparative anatomy of the suprabranchial organs in the Clariidae. Bull. Brit. Mus. (Nat. Hist. Zool.), 7: 217-241. Grecory, W. K. 1933. Fish skulls. A study of the evolution of natural mecha- nisms. Trans. Amer. Phil. Soc., 23: 75-481. . 1951. Evolution emerging. Macmillan, New York, 2 vols. Guntuer, A. C. L. G. 1880. An introduction to the study of fishes. A. and C. Black, Edinburgh, 720 p. HarrinctTon, R. W., Jr. 1955. The osteocranium of the American cyprinid fish, Notropis bifrenatus, with an annotated synonymy of teleost skull bones. Copeia: 267-290. HENNINGER, G. 1908. Die Labyrinthorgane bei Labyrinthfische. Zool. Jahrb. Anat., 25: 251-304. HerRALp, E. S. 1961. Living fishes of the world. Doubleday, Garden City, New York, 304 p. Herre, A. W. C. T. 1924. Distribution of the true freshwater fishes in the Philippines. Labyrinthici, Clariidae and Siluridae. Philippine J. Sci., 24: 683-707. Husss, C. L. 1919. A comparative study of the bones forming the opercular series of fishes. J. Morph., 33: 61-71. Jorpan, D. S. 1928. A classification of fishes. Including families and genera as far as known. Stanford Univ. Publ., Biol. Ser., 3: 79-243. KircHuHorFr, H. 1958. Funktionell-anatomische Untersuchung des Visceral- apparates von Clupea harengus L. Zool. Jahrb. Anat., 76:461-540. Kosswic, C. 1954. Contributions to the historical zoogeography of African LITERATURE CITED eh freshwater fishes. Publ. Hydrobiol. Res. Inst., Univ. Istanbul, Ser. B, 2: 82-91. Krux, H. 1939. Namenliste der bekannteren Labyrinthfische zur Ermitte- lung der giiltigen Namen. Zool. Anz., 125: 325-334. Liem, K. F. 1961. Tetrapod parallelisms and other features in the functional morphology of the blood vascular system of Fluta alba. J. Morph., 108: 131-143. . The osteology of Luciocephalus pulcher with a discussion on the gular element and taxonomic status. Unpublished manuscript. Linpsey, C. C. 1954. Temperature-controlled meristic variation in the paradise fish Macropodus opercularis (L.). Canad. J. Zool., 32: 87-98. MaxusHox, V. M. 1958. The morphology and classification of the northern blennioid fishes (Stichaeoidei, Blennioidei, Pisces). Proc. Zool. Inst. USSR, 25: 3-129 (in Russian). Maratue, V. B. 1959. The cranial osteology of Trichopodus trichopterus (Pall.). J. Univ. Bombay, n.s., 27B: 70-79. MeErRRIMAN, D. 1940. The osteology of the striped bass (Roccus saxatilis). Ann. Mag. Nat. Hist., 5: 55-64. Myers, G. S. 1923. Notes on the nomenclature of certain anabantids and a new generic name proposed. Copeia: 62-63. . 1926. Notes on anabantids III. Copeia: 97-100. . 1937. Fresh-water fishes and West-Indian zoogeography. Smithsonian Report: 339-364. . 1951. Fresh-water fishes and East-Indian zoogeography. Stanford Ichth. Bull., 4: 11-21. . 1958. Trends in the evolution of teleostean fishes. Stanford Ichth. Bull., 7: 27-30. Nicuots, J. T. 1943. The fresh-water fishes of China. Amer. Mus. Nat. Hist., 9: 1-322. Nixotskr, G. W. 1957. Spezielle Fischkunde. Veb. Deutscher Verlag der Wissenschaften, Berlin, 632 p. Picarp, L. 1943. Structure and evolution of Palestine. Bull. Geol. Dept. He- brew Univ. Jerusalem, 4 (2): 134. PrasHaD, B., and D. D. Muxkeryt. 1929. The fish of the Indawgyi Lake and the streams of the Myitkyina District. Rec. Ind. Mus., 31: 161-223. Recan, C. T. 1909. The Asiatic fishes of the family Anabantidae. Proc. Zool. Soc. London: 767-787. Romer, A. S. 1945. Vertebrate paleontology. University of Chicago Press, Chicago, 687 p. SANDERS, M. 1934. Die fossilen Fische der alttertiaéren Siisswasserablagerungen aus Mittel-Sumatra. Verh. geol.-mynb. Genoot. Ned. Kolon., 11 (1): 1-144. SCHARRER, E.., S. W. Smiru, and S. L. Patay. 1947. Chemical sense and taste in the fishes Prionotus and Trichogaster. J. Comp. Neur., 86: 183-198. SCHENK, E. T., J. H. McMastrers, A. M. KEEN, and S. W. Mutter. 1956. Procedure in taxonomy. Stanford University Press, Stanford, 119 p. SCHNEWER, H. 1941. Die Bedeutung der Atemhohle der Labyrinthfische fiir ihr H6rvermogen. Zeitsch. vergl. Physiol., 29: 172-194. ScuuLtz, L. P. 1958. Review of the parrotfishes family Scaridae. U.S. Nat. Mus. Bull., 214: 1-143. Seitz, K. 1937. Untersuchungen zur biologischen Anatomie der Ernahrung bei Helostoma temmincki. Anat. Anz., 85: 230-238. 78 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI SEWERTZOFF, A. N. 1934. Evolution der Bauchflossen der Fische. Zool. Jahrb. Anat., 58: 415-500. SHELDEN, F. F. 1937. Osteology, myology and probable evolution of the Nematognath pelvic girdle. Ann. N.Y. Acad. Sci., 37: 1-96. Suin, H. J. 1936. Notes on the labyrinth fishes of China. Bull. Fan Memorial Inst. Biol. Peiping Zool., 72: 67-97. SHUFELDT, R. W. 1885. The osteology of Amia calva. Ann. Report Commis- sioner of Fish and Fisheries for 1883: 1-94. Simpson, G. G. 1953. The major features of evolution. Columbia University Press, New York, 434 p. . 1961. Principles of animal taxonomy. Columbia University Press, New York, 247 p. SmiTH, C. Lavett, and R. M. Baitey. 1962. The subocular shelf of fishes. Je Morphs 110: eli: SmitH, Hopart M. 1956. Schistic evolution. Evolution, 10: 228-229. . 1960. Evolution of chordate structure. Holt-Dryden, New York, 529 p. SmirH, Hucu M. 1945. The freshwater fishes of Siam. Smithsonian Inst. Bull., 188: 1-622. SRINIVASACHAR, H. R. 1955. The skull of Ophicephalus. Proc. Ind. Acad. Sci., 42B: 226-237. Srarks, E. C. 1926. Bones of the ethmoid region of the fish skull. Stanford Univ. Publ., Biol. Ser., 4: 139-338. STEBBINS, G. L., Jr. 1949. Reality and efficacy of selection in plants. Proc. Amer. Phil. Soc., 93: 501-5138. STEINBACH, G. 1950. Funktionelle und vergleichende anatomische Untersuch- ungen iiber die Bauchflossen von Trichogaster trichopterus Pallas. Deutsche zool. Zeitsch., I: 24-60. Sternitz, H. 1954. The distribution and evolution of the fresh-water fishes of Palestine. Publ. Hydrobiol. Res. Inst., Univ. Istanbul, Ser. B, I: 225- IS, SuyEHIRO, Y. 1941. A study on the digestive system and feeding habits of fish. Japan. J. Zool., 10: 1-303. Taninc, A. V. 1950. Experimental study of meristic characters in fishes. Biol. Rev. Cambridge Phil. Soc., 27: 169-193. TweepiE, M. W. F. 1952. Notes on Malayan fresh-water fishes. Bull. Raffles Mus., 24: 63-95. Vaas, K. F., M. Sacuian, and G. WrraaTmapyA. 1953. On the ecology and fisheries of some inland waters along the rivers Ogan and Komering in southeast Sumatra. Contr. In]. Fish. Res. Station, Bogor, No. 3: 1-32. WepeErR, M., and L. F. pe Beaurort. 1922. The fishes of the Indo-Australian Archipelago, 4. Brill, Leiden, 410 p. ZocrAFF, N. 1888. On the construction and purpose of the so-called labyrin- thine apparatus of the labyrinthic fishes. Quart. J. Micr. Sci., 28: 501-512. PLATE 1 Fig. 1. Dorsal aspect of skull of Anabas testudineus. Right sensory canals omitted. Fig. 2. Dorsal aspect of skull of Ctenopoma muriei. Right sensory canals omitted. Fig. 3. Dorsal aspect of skull of Sandelia capensis. Right sensory canals omitted. Fig. 4. Dorsal aspect of skull of Osphronemus goramy. Abbreviations: E, ethmoid; EO, epiotic; F, frontal; IC, intercalary; L, lachrymal; LE, lateral ethmoid; MX, maxillary; N, nasal; PA, parietal; PF, prefrontal; PM, premaxillary; PP, extrascapular; PTF, posttemporal fossa; PTT, posttemporal; SO;, suborbital; SOC, supraoccipital; SPH, sphenotic; ST, supratemporal. 80 PM 0.5 CM 81 PLATE 2 Fig. 5. Dorsal aspect of skull of Helostoma temmincki. Right sensory canals omitted. Fig. 6. Dorsal aspect of skull of Belontia hasselti. Right sensory canals omitted. Fig. 7. Dorsal aspect of skull of Betta splendens. Fig. 8. Dorsal aspect of skull of Trichopsis vittatus. Right sensory canals omitted. Fig. 9. Dorsal aspect of skull of Macropodus cupanus. Right sensory canals omitted. Abbreviations: D, dentary; E, ethmoid; EO, epiotic; EOC, exoccipital; EOP, epiotic process; F, frontal; IC, intercalary; L, lachrymal; MX, maxillary; N, nasal; PA, parietal; PF, prefrontal; PM, premaxillary; PTF, posttemporal fossa; SO;, sub- orbital; SOC, supraoccipital; SPH, sphenotic; ST, supratemporal. PLATE 3 Fig. 10. Dorsal aspect of skull of Macropodus opercularis. Right sensory canals omitted. Fig. 11. Dorsal aspect of skull of Sphaerichthys osphromenoides. Left lachrymal and right sensory canals omitted. Fig. 12. Dorsal aspect of skull of Colisa fasciata. Right sensory canals omitted. Fig. 13. Dorsal aspect of skull of Trichogaster trichopterus. Right sensory canals omitted. Abbreviations: D, dentary; E, ethmoid; EO, epiotic; EOC, exoccipital; EOP, epiotic process; F, frontal; IC, intercalary; L, lachrymal; LE, lateral ethmoid; MX, maxil- lary; N, nasal; P, palatine; PA, parietal; PF, prefrontal; PM, premaxillary; PTF, posttemporal fossa; SO;, suborbital; SOC, supraoccipital; SPH, sphenotic; ST, supratemporal. 84 85 PLATE 4 Fig. 14. Lateral aspect of neurocranium of Anabas testudineus. Supratemporal- intertemporal removed. Fig. 15. Lateral aspect of neurocranium of Ctenopoma multispinis. Fig. 16. Lateral aspect of neurocranium of Sandelia capensis. Extrascapular omitted. Fig. 17. Lateral aspect of neurocranium of Osphronemus goramy. Abbreviations: APPS, ascending process of parasphenoid; BOC, basioccipital; BS, basisphenoid; E, ethmoid; EO, epiotic; EOC, exoccipital; F, frontal; FEOC, fora- men exoccipitale; IC, intercalary; ICA, internal carotid artery; LAG, lagenar swelling; LE, lateral ethmoid; MY, myodome; N, nasal; PA, parietal; PF, pre- frontal; PO, prootic; PP, extrascapular; PPBOC, pharyngeal process of basioc- cipital; PPPS, pharyngeal process of parasphenoid; PS, parasphenoid; PTF, post- temporal fossa; PTS, pterosphenoid; PV, prevomer; S, swelling of sacculus; SOs, suborbital; SOC, supraoccipital; SPH, sphenotic; ST, supratemporal; TPPS, trans- verse process of parasphenoid; U, swelling of utriculus; X, foramen of vagus nerve. 86 0.5 CM SPH PA SOC | 17 87 PLATE 5 Fig. 18. Lateral aspect of neurocranium of Helostoma temmincki. Fig. 19. Lateral aspect of neurocranium and first six vertebrae of Belontia hasselti. Fig. 20. Lateral aspect of neurocranium of Betta splendens. Abbreviations: APPS, ascending process of parasphenoid; BAP, basapophysis; BOC, basioccipital; BS, basisphenoid; C, centrum; DPT, dorsal pterygiophores; E, eth- moid; EO, epiotic; EOC, exoccipital; EOP, epiotic process; EPLR, epipleural ribs; F, frontal; FEOC, foramen exoccipitale; IC, intercalary; ICA, internal carotid artery; IN, interneural; LE, lateral ethmoid; MY, myodome; N, nasal; NS, neural spine; PA, parietal; PF, prefrontal; PLR, pleural ribs; PO, prootic; POZ, postzyg- apophysis; PPBOC, pharyngeal process of basioccipital; PPPS, pharyngeal process of parasphenoid; PRZ, prezygapophysis; PS, parasphenoid; PTF, posttemporal fossa; PTS, pterosphenoid; PV, prevomer; SO;, suborbital; SOC, supraoccipital; SPH, sphenotic; ST, supratemporal; TPPS, transverse process of parasphenoid. 88 PLATE 6 Fig. 21. Lateral aspect of neurocranium of Trichopsis vittatus. Fig. 22. Lateral aspect of neurocranium of Macropodus cupanus. Fig. 23. Lateral aspect of neurocranium of Macropodus opercularis. Abbreviations: APPS, ascending process of parasphenoid; BOC, basioccipital; BS, basisphenoid; E, ethmoid; EO, epiotic; EOC, exoccipital; EOP, epiotic process; F, frontal; FEOC, foramen exoccipitale; IC, intercalary; ICA, internal carotid artery; LAG, lagenar swelling; LE, lateral ethmoid; MY, myodome; N, nasal; PA, parietal; PF, prefrontal; PO, prootic; PPBOC, pharyngeal process of basi- occipital; PPPS, pharyngeal process of parasphenoid; PS, parasphenoid; PTF, post- temporal fossa; PTS, pterosphenoid; PV, prevomer; S, swelling of sacculus; SOs, suborbital; SOC, supraoccipital; SPH, sphenotic; ST, supratemporal; U, swelling of utriculus; X, foramen of vagus nerve. 90 MY APPS PLATE 7 Fig. 24. Lateral aspect of neurocranium of Sphaerichthys osphromenoides. Fig. 25. Lateral aspect of neurocranium of Colisa lalia. Fig. 26. Lateral aspect of neurocranium of Trichogaster trichopterus. Abbreviations: APPS, ascending process of parasphenoid; BOC, basioccipital; BS, basisphenoid; E, ethmoid; EO, epiotic; EOC, exoccipital; EOP, epiotic process; F, frontal; FEOC, foramen exoccipitale; IC, intercalary; ICA, internal carotid artery; LE, lateral ethmoid; N, nasal; PA, parietal; PF, prefrontal; PO, prootic; PPBOC, pharyngeal process of basioccipital; PPPS, pharyngeal process of para- sphenoid; PS, parasphenoid; PTF, posttemporal fossa; PTS, pterosphenoid; PV, prevomer; SO;, suborbital; SOC, supraoccipital; SPH, sphenotic; ST, supratem- poral. PLATE 8 Fig. 27. Lateral aspect of intact skull of Anabas testudineus. Suborbitals 3-5 un- labeled. Fig. 28. Lateral aspect of intact skull of Ctenopoma muriei. Suborbitals 3-5 un- labeled. Fig. 29. Lateral aspect of intact skull of Ctenopoma multispinis. Suborbitals 3-5 unlabeled. Fig. 30. Lateral aspect of intact skull of Sandelia capensis. Suborbitals 3-5 unlabeled. Abbreviations: A, angular; BS, basisphenoid; D, dentary; ECT, ectopterygoid; ENT, entopterygoid; EO, epiotic; F, frontal; HM, hyomandibular; IOP, interopercular; L, lachrymal; LE, lateral ethmoid; MPT, metapterygoid; MX, maxillary; N, nasal; OP, opercular; PA, parietal; PF, prefrontal; PM, premaxillary; POP, preopercular; PP, extrascapular; PS, parasphenoid; PTS, pterosphenoid; PTT, posttemporal; Q, quadrate; RA, retroarticular; SO:, suborbital; SOC, supraoccipital; SOP, sub- opercular; SPH, sphenotic; ST, supratemporal; SY, symplectic. 94 PLATE 9 Fig. 31. Lateral aspect of intact skull of Helostoma temmincki. Suborbitals 3-5 un- labeled. Fig. 32. Lateral aspect of intact skull of Belontia hasselti. Suborbitals 3-5 unlabeled. Fig. 33. Lateral aspect of intact skull of Betta splendens. Suborbitals 3-5 unlabeled. Abbreviations: A, angular; BS, basisphenoid; D, dentary; E, ethmoid; ECT, ecto- pterygoid; ENT, entopterygoid; EO, epiotic; EOC, exoccipital; F, frontal; HM, hyomandibular; IC, intercalary; IOP, interopercular; L, lachrymal; LE, lateral ethmoid; MPT, metapterygoid; MX, maxillary; N, nasal; OP, opercular; PA, parie- tal; PM, premaxillary; POP, preopercular; PS, parasphenoid; PTS, pterosphenoid; Q, quadrate; RA, retroarticular; SO., suborbital; SOC, supraoccipital; SOP, sub- opercular; SPH, sphenotic; ST, supratemporal; SY, symplectic. 96 Q ENT MPT SY 33 97 PLATE 10 Fig. 34. Lateral aspect of intact skull of Trichopsis vittatus. Suborbitals 3-5 un- labeled. Fig. 35. Lateral aspect of intact skull of Macropodus opercularis. Suborbitals 3-5 unlabeled. Fig. 36. Lateral aspect of intact skull of Sphaerichthys osphromenoides. Suborbitals 3-5 unlabeled. Abbreviations: A, angular; BS, basisphenoid; D, dentary; E, ethmoid; ENT, ento- pterygoid; EO, epiotic; EOC, exoccipital; F, frontal; HM, hyomandibular; IC, intercalary; IOP, interopercular; L, lachrymal; LE, lateral ethmoid; MPT, metap- terygoid; MX, maxillary; N, nasal; OP, opercular; P, palatine; PA, parietal; PM, premaxillary; POP, preopercular; PS, parasphenoid; PTS, pterosphenoid; Q, quadrate; RA, retroarticular; SC, sensory canal; SO2, suborbital; SOC, supra- occipital; SOP, subopercular; SPH, sphenotic; ST, supratemporal; SY, symplectic. 0.5 CM 0.5 CM Puate 11 Fig. 37. Lateral aspect of intact skull of Trichogaster trichopterus. Suborbitals 3-4 unlabeled. Fig. 38. Lateral aspect of intact skull of Trichogaster leeri. Suborbitals 3-4 removed. Fig. 39. Lateral aspect of intact skull of Colisa fasciata. Suborbitals 3-5 unlabeled. Abbreviations: A, angular; BS, basisphenoid; D, dentary; E, ethmoid; ENT, ento- pterygoid; EO, epiotic; EOC, exoccipital; EOP, epiotic process; F, frontal; HM, hyomandibular; IC, intercalary; IOP, interopercular; L, lachrymal; LE, lateral ethmoid; MPT, metapterygoid; MX, maxillary; N, nasal; OP, opercular; PA, parietal; PF, prefrontal; PM, premaxillary; POP, preopercular; PS, parasphenoid; PTF, posttemporal fossa; PTS, pterosphenoid; Q, quadrate; RA, retroarticular; SO:_.;, suborbital; SOC, supraoccipital; SOP, subopercular; SPH, sphenotic; ST, supratemporal; SY, symplectic. 100 PLATE 12 Fig. 40. Ventral aspect of left half of neurocranium of Anabas testudineus. Fig. 41. Ventral aspect of left half of neurocranium of Ctenopoma multispinis. Fig. 42. Ventral aspect of left half of neurocranium of Sandelia capensis. Fig. 43. Ventral aspect of left half of neurocranium of Osphronemus goramy. Fig. 44. Ventral aspect of left half of neurocranium of Helostoma temmincki. Abbreviations: APPS, ascending process of parasphenoid; BOC, basioccipital; BS, basisphenoid; E, ethmoid; EOC, exoccipital; F, frontal; IC, intercalary; ICA, in- ternal carotid artery; LAG, lagenar swelling; LE, lateral ethmoid; MY, myodome; N, nasal; OF, olfactory foramen; OPF, orbital process of frontal; PO, prootic; PPBOC, pharyngeal process of basioccipital; PPPS, pharyngeal process of para- sphenoid; PS, parasphenoid; PTS, pterosphenoid; PV, prevomer; S, swelling of sacculus; SO;, suborbital; SPH, sphenotic; ST, supratemporal; TPPS, transverse process of parasphenoid; X, foramen of vagus nerve. PLATE 13 Fig. 45. Ventral aspect of left half of neurocranium of Belontia hasselti. Fig. 46. Ventral aspect of left half of neurocranium of Betta splendens. Fig. 47. Ventral aspect of left half of neurocranium of Trichopsis vittatus. Fig. 48. Ventral aspect of left half of neurocranium of Macropodus opercularis. Abbreviations: APPS, ascending process of parasphenoid; BOC, basioccipital; BS, basisphenoid; E, ethmoid; EOC, exoccipital; F, frontal; IC, intercalary; ICA, internal carotid artery; LAG, lagenar swelling; LE, lateral ethmoid; MY, myo- dome; N, nasal; OF, olfactory foramen; OPF, orbital process of frontal; PO, prootic; PPBOC, pharyngeal process of basioccipital; PPPS, pharyngeal process of parasphenoid; PS, parasphenoid; PTS, pterosphenoid; PV, prevomer; S, swelling of sacculus; SPH, sphenotic; ST, supratemporal; U, swelling of utriculus; X, fora- men of vagus nerve. 104 EOC 105 PLATE 14 Fig. 49. Ventral aspect of left half of neurocranium of Sphaerichthys osphromen- oides. Fig. 50. Ventral aspect of left half of neurocranium of Colisa fasciata. Fig. 51. Ventral aspect of left half of neurocranium of Trichogaster trichopterus. Abbreviations: APPS, ascending process of parasphenoid; BOC, basioccipital; BS, basisphenoid; E, ethmoid; EOC, exoccipital; F, frontal; FEOC, foramen exoc- cipitale; IC, intercalary; ICA, internal carotid artery; LAG, lagenar swelling; LE, lateral ethmoid; MY, myodome; N, nasal; OF, olfactory foramen; PO, prootic; PPBOC, pharyngeal process of basioccipital; PPPS, pharyngeal process of para- sphenoid; PS, parasphenoid; PTS, pterosphenoid; PV, prevomer; S, swelling of sacculus; SPH, sphenotic; ST, supratemporal; U, swelling of utriculus; X, foramen of vagus nerve. 106 107 PLATE 15 Fig. 52. Posterior aspect of neurocranium of Anabas testudineus. Left extrascapular and posttemporal removed. Fig. 53. Posterior aspect of neurocranium of Ctenopoma multispinis. Right extra- scapular omitted. Fig. 54. Posterior aspect of neurocranium of Sandelia capensis. Left extrascapular removed. Fig. 55. Posterior aspect of neurocranium of Osphronemus goramy. Fig. 56. Posterior aspect of neurocranium of Helostoma temmincki. Fig. 57. Posterior aspect of neurocranium of Belontia hasselti. Abbreviations: BOC, basioccipital; EO, epiotic; EOC, exoccipital; EOP, epiotic process; FEOC, foramen exoccipitale; IC, intercalary; PO, prootic; PP, extra- scapular; PPBOC, pharyngeal process of basioccipital; PPPS, pharyngeal process of parasphenoid; PS, parasphenoid; PTF, posttemporal fossa; PTT, posttemporal; SOC, supraoccipital; ST, supratemporal; TPPS, transverse process of parasphenoid. 108 109 Fig. Fig. Fig. Fig. Fig. Fig. Posterior Posterior Posterior Posterior Posterior Posterior aspect aspect aspect aspect aspect aspect PLaTE 16 of neurocranium of Betta splendens. of neurocranium of Trichopsis vittatus. of neurocranium of Macropodus opercularis. of neurocranium of Sphaerichthys osphromenoides. of neurocranium of Colisa fasciata. of neurocranium of Trichogaster leeri. Abbreviations: BOC, basioccipital; EO, epiotic; EOC, exoccipital; EOP, epiotic process; FEOC, foramen exoccipitale; IC, intercalary; PA, parietal; PO, prootic; PPBOC, pharyngeal process of basioccipital; PPPS, pharyngeal process of para- sphenoid; PTF, posttemporal fossa; SOC, supraoccipital; ST, supratemporal. 110 111 PLATE 17 Fig. 64. Lateral aspect of oromandibular region of Anabas testudineus. Fig. 65. Lateral aspect of oromandibular region of Ctenopoma multispinis. Fig. 66. Lateral aspect of oromandibular region of Sandelia capensis. Fig. 67. A. Lateral aspect of oromandibular region of Osphronemus goramy. B. Opercular series of Osphronemus goramy. Abbreviations: A, angular; APA, ascending process of angular; APD, ascending process of dentary; APPM, ascending process of premaxillary; D, dentary; ECT, ectopterygoid; ENT, entopterygoid; EPP, ethmoid process of palatine; HM, hyo- mandibular; IOP, interopercular; MPT, metapterygoid; MX, maxillary; OP, oper- cular; P, palatine; PM, premaxillary; POP, preopercular; Q, quadrate; RA, retro- articular; SOP, subopercular; SY, symplectic. 113 PLATE 18 Fig. 68. Lateral aspect of oromandibular region of Helostoma temmincki. Fig. 69. Lateral aspect of oromandibular region of Belontia hasselti. Fig. 70. Lateral aspect of oromandibular region of Betta splendens. Fig. 71. Lateral aspect of oromandibular region of Trichopsis vittatus. Abbreviations: A, angular; APA, ascending process of angular; APD, ascending process of dentary; APPM, ascending process of premaxillary; D, dentary; DLPA, dorsolateral process of angular; ECT, ectopterygoid; ENT, entopterygoid; EPP, ethmoid process of palatine; HM, hyomandibular; MPT, metapterygoid; MX, maxillary; P, palatine; PM, premaxillary; Q, quadrate; RA, retroarticular; SY, symplectic. 114 APPM EPP P 0.5 CM appm EPP APD Fig. 72. Fig. 73. Fig. 74. Fig. 75. PLATE 19 Lateral aspect of oromandibular region of Macropodus cupanus. Lateral aspect of oromandibular region of Sphaerichthys osphromenoides. Lateral aspect of oromandibular region of Colisa fasciata. Lateral aspect of oromandibular region of Trichogaster leeri. Abbreviations: A, angular; APA, ascending process of angular; APD, ascending process of dentary; APPM, ascending process of premaxillary; D, dentary; ENT, entopterygoid; EPP, ethmoid process of palatine; HM, hyomandibular; MPT, metapterygoid; MX, maxillary; P, palatine; PM, premaxillary; Q, quadrate; RA, retroarticular; SY, symplectic. 116 0.5 CM PLATE 20 Fig. 76. Hyobranchium of Anabas testudineus. A. Dorsal aspect of left half of hyobranchium, first epibranchial reflected laterally. B. Lateral aspect of modified first epibranchial (suprabranchial organ). C. Lateral aspect of urohyal. D. Lateral aspect of hyoid apparatus. Fig. 77. Hyobranchium of Ctenopoma multispinis. A. Dorsal aspect of left half of hyobranchium, first epibranchial seen from the medial aspect. B. Lateral aspect of modified first epibranchial. C. Ventral aspect of pharyngobranchials. D. Lateral aspect of hyoid apparatus. E. Lateral aspect of urohyal. Fig. 78. Hyobranchium of Sandelia capensis. A. Lateral aspect of modified first epibranchial. B. Dorsal aspect of left half of hyobranchium. C. Ventral aspect of pharyngobranchials. D. Lateral aspect of hyoid apparatus. E. Lateral aspect of urohyal. Abbreviations: BB, basibranchial; BH, basihyal; BSR, branchiostegal rays; CB, ceratobranchial; CH, ceratohyal; EB, epibranchial; EBP, epibranchial process; EH, epihyal; HB, hypobranchial; HHD, dorsal hypohyal; HHV, ventral hypohyal; IH, interhyal; PB, pharyngobranchial; URH, urohyal. 118 Fig. QUmbDOePS OmmMoOW> PLATE 21 79. Hyobranchium of Osphronemus goramy. Medial aspect of modified first epibranchial (suprabranchial organ). Dorsal aspect of left hyobranchium. Lateral aspect of first epibranchial. . Ventral aspect of pharyngobranchials 2-4. Lateral aspect of urohyal. Dorsal aspect of fifth ceratobranchial. . Lateral aspect of hyoid apparatus. 80. Hyobranchium of Helostoma temmincki. Medial aspect of modified first epibranchial (suprabranchial organ). Dorsal aspect of left hyobranchium. . Lateral aspect of first epibranchial. . Ventral aspect of epibranchials 2-4 and pharyngobranchials. Dorsal aspect of fifth ceratobranchial. . Lateral aspect of hyoid apparatus. . Lateral aspect of urohyal. Abbreviations: BB, basibranchial; BH, basihyal; BSR, branchiostegal rays; CB, cerato- branchial; CH, ceratohyal; EB, epibranchial; EBP, epibranchial process; EH, epi- hyal; HB, hypobranchial; HHD, dorsal hypohyal; HHV, ventral hypohyal; IH, interhyal; MR, median ridge of basibranchials; PB, pharyngobranchial; URH, urohyal. PLATE 22 Fig. 81. Hyobranchium of Belontia hasselti. AOE E am OW > Dorsal aspect of left half of hyobranchium, first epibranchial reflected laterally. Lateral aspect of modified first epibranchial (suprabranchial organ). Ventral aspect of pharyngobranchials. . Dorsal aspect of fifth ceratobranchial. . Lateral aspect of hyoid apparatus. . Lateral aspect of urohyal. 82. Hyobranchium of Betta splendens. Dorsal aspect of left half of hyobranchium, first epibranchial reflected laterally. . Lateral aspect of modified first epibranchial (suprabranchial organ). Ventral aspect of pharyngobranchials. Dorsal aspect of fifth ceratobranchial. Lateral aspect of urohyal. . Lateral aspect of hyoid apparatus. Fig. 83. Hyobranchium of Trichopsis vittatus. A. B. C. 1D), i. Lateral aspect of modified first epibranchial. Dorsal aspect of left half of hyobranchium. Ventral aspect of pharyngobranchials. Lateral aspect of hyoid apparatus. Lateral aspect of urohyal. Abbreviations: BB, basibranchial; BH, basihyal; BSR, branchiostegal rays; CB, ceratobranchial; CH, ceratohyal; EB, epibranchial; EBP, epibranchial process; EH, epihyal; HB, hypobranchial; HHD, dorsal hypohyal; HHV, ventral hypohyal; IH, interhyal; PB, pharyngobranchial; URH, urohyal. Fig. A. B. C, D. E. Fig. Onna >e ymbOW> PLATE 23 84. Hyobranchium of Macropodus opercularis. Lateral aspect of modified first epibranchial (suprabranchial organ). Dorsal aspect of left half of hyobranchium, first epibranchial reflected laterally. Ventral aspect of pharyngobranchials. Lateral aspect of urohyal. Lateral aspect of hyoid apparatus. 85. Hyobranchium of Sphaerichthys osphromenoides. . Posterior aspect of modified first epibranchial (suprabranchial organ). . Lateral aspect of first epibranchial. Dorsal aspect of left half of hyobranchium. . Ventral aspect of epibranchials 2-4 and pharyngobranchials. Dorsal aspect of fifth ceratobranchial. Lateral aspect of hyoid apparatus. 86. Hyobranchium of Colisa lalia. Medial aspect of modified first epibranchial (suprabranchial organ). Dorsal aspect of hyobranchium. Branchial element from right side not shown. . Lateral aspect of modified first epibranchial. . Ventral aspect of pharyngobranchials. Dorsal aspect of fifth ceratobranchial. . Lateral aspect of hyoid apparatus. . Lateral aspect of urohyal. Abbreviations: BB, basibranchial; BH, basihyal; BSR, branchiostegal rays; CB, ceratobranchial; CH, ceratohyal; EB, epibranchial; EBP, epibranchial process; EH, epihyal; HB, hypobranchial; HHD; dorsal hypohyal; HHV, ventral hypohyal; IH, interhyal; PB, pharyngobranchial; URH, urohyal. 124 PLATE 24 Fig. 87. Hyobranchium of Trichogaster trichopterus. . Medial aspect of modified first epibranchial (suprabranchial organ). . Dorsal aspect of hyobranchium. Branchial elements from right side not shown. . Lateral aspect of first epibranchial. . Ventral aspect of epibranchials 2-4 and pharyngobranchials. . Dorsal aspect of fifth ceratobranchial. . Lateral aspect of hyoid apparatus. . Lateral aspect of urohyal. Abbreviations: BB, basibranchial; BH, basihyal; BSR, branchiostegal rays; CB, ceratobranchial; CH, ceratohyal; EB, epibranchial; EH, epihyal; HB, hypobran- chial; HHD, dorsal hypohyal; HHV, ventral hypohyal; IH, interhyal; PB, pharyngo- branchial; URH, urohyal. QnAOAOmS C 0.5 CM PLATE 25 Lateral aspects of left otoliths. Sequence of otoliths from left to right is: utriculith (Japillus), sacculith (sagitta), lagenolith (asteriscus ). Fig. 88. A. Anabas testudineus. Ctenopoma muriei. . Sandelia capensis. Helostoma temmincki. Belontia hasselti. Macropodus opercularis. Betta splendens. . Sphaerichthys osphromenoides. . Colisa lalia. . Trichogaster trichopterus. . Ctenopoma multispinis. . Luciocephalus pulcher. Fig. 89. Fig. 90. DUAerbaowsrvawPe 128 PLATE 26 Fig. 91. Pectoral and pelvic girdles of Anabas testudineus. A. Lateral aspect of pectoral girdle. B. Dorsal aspect of pelvic girdle. C. Ventral aspect of pelvic girdle. Fig. 92. Pectoral and pelvic girdles of Sandelia capensis. A. Lateral aspect of pectoral girdle. B. Ventral aspect of pelvic girdle. C. Dorsal aspect of pelvic girdle. Abbreviations: BSP, basipterygium; CL, cleithrum; CO, coracoid; PCL, postclei- thrum; PL, processus lateralis; PMA, processus medialis anterior; PME, processus medialis; PMP, processus medialis posterior; PTT, posttemporal; RAD, radial; RBSP, rachis of basipterygium; SCA, scapula; SCL, supracleithrum. 131 PLATE 27 Fig. 93. Ventral aspect of pelvic girdle of Osphronemus goramy. Fig. 94. Pectoral girdle of Helostoma temmincki. A. Lateral aspect. B. Medial aspect. Fig. 95. Pelvic girdle of Helostoma temmincki. A. Ventral aspect. B. Dorsal aspect. Fig. 96. Pectoral and pelvic girdles of Belontia hasselti. A. Lateral aspect of pectoral girdle. B. Ventral aspect of pelvic girdle. Fig. 97. Pectoral and pelvic girdles of Trichopsis vittatus. A. Lateral aspect of pectoral and pelvic girdles. B. Ventral aspect of pelvic girdle. Abbreviations: BSP, basipterygium; CL, cleithrum; CO, coracoid; PCL, postclei- thrum; PL, processus lateralis; PMA, processus medialis anterior; PME, processus medialis; PMP, processus medialis posterior; PTT, posttemporal; RAD, radial; RBSP, rachis of basipterygium; SCA, scapula; SCL, supracleithrum. BSP. RBSP 133 PLATE 28 Fig. 98. Pectoral and pelvic girdles of Macropodus opercularis. A. Medial aspect of pectoral and pelvic girdles. B. Ventral aspect of pelvic girdle. Fig. 99. Pectoral and pelvic girdles of Sphaerichthys osphromenoides. A. Medial aspect of pectoral and pelvic girdles. B. Ventral aspect of pelvic girdle. Fig. 100. Pectoral and pelvic girdles of Colisa lalia. A. Medial aspect. B. Lateral aspect. Abbreviations: BSP, basipterygium; CL, cleithrum; CO, coracoid; PCL, postclei- thrum; PL, processus lateralis; PMA, processus medialis anterior; PMAS, processus medialis accessorius; PME, processus medialis; PMP, processus medialis posterior: PTT, posttemporal; RAD, radial; RBSP, rachis of basipterygium; SCA, scapula; SCL, supracleithrum. 134 PLATE 29 Fig. 101. Pelvic girdle of Colisa lalia. A. Dorsal aspect. B. Ventral aspect. Fig. 102. Ventral aspect of pelvic girdle of Trichogaster leeri. Abbreviations: BSP, basipterygium; CL, cleithrum; CO, coracoid; PCL, postclei- thrum; PL, processus lateralis; PMA, processus medialis anterior; PMAS, processus medialis accessorius; PME, processus medialis; PMP, processus medialis posterior; PTT, posttemporal; RAD, radial; RBSP, rachis of basipterygium; SCA, scapula; SCL, supracleithrum. 136 Size BELONTIIDAE ea ET a OF Pr ee EE oe vy TRICHOGASTERINAE / ie / / HELOSTOMATIDAE OSPHRONEMIDAE ANABANTIDAE 103 Fig. 103. The proposed phylogeny of the anabantoid families and subfamilies. TRICHOGASTER COLISA VPARASPHAERICHTHYS / SPHAERICHTHYS MALPULUTTA / , PAROSPHROMENUS } MACROPODUS HELOSTOMA TRICHOPSIS OSPHRONEMUS BELONTIA ANABAS SANDELIA CTENOPOMA ANCESTRAL ANABANTID STOCK 104 Fig. 104. The proposed phylogeny of the anabantoid genera. 139 INDEX Note: Since the 26 species and 14 genera of anabantoids are mentioned on numerous pages, entries in the following index refer only to important discussions and plates. Less important mentions of generic and specific names are entered collectively under family names. All page references to illustrations are restricted to specific names and are entered in boldface type. Taxonomic names appearing in tables, names of bony elements in the legends of figures, authors listed in the bibliography, and chapter titles are not included in this index. Acanthopteran characters: luciocepha- lids, 70 Acanthopteran fishes: jaw mechanism, 49 Acanthopterygii: branchiostegals, 57 Actinopteran fishes, 50 Adaptive correlations, 60 Adaptive features: Osphronemus, 65 Adaptive zones, 53 Adductor arcus palatini, 54 Adductor mandibulae muscles: eral, 49; Helostoma, 50 Aeoliscus, 51 Air-breathing, 69 Air-breathing capacity: Sandelia, 43 Air ventilation, 54 Ala laminaris externa: Anabantidae, 11; Belontiidae, 20; Helostoma- tidae, 31 All-or-none element, 66 All-or-none reaction: Helostoma, 66 Amia, 38, 70 Anabantidae: as ancestral stock, 38; composition, 37; diagnosis, 38; Gill's definition, 37; otoliths, 129: position among Percesoces, 37; summary, 73 Anabantids: primitive characters, 61 Anabantoidei: diagnosis, 73; as op- posed to Luciocephalus, 69; as opposed to Ophicephaliformes, 71, 72; relation to ophicephalids, 71, 72 Anabas: archaic characters, 42, 43; occurrence in Africa, 62; as opposed to Ctenopoma, 42, 62; physiological gen- 141 traits, 62, 63; relation to Cteno- poma, 42, 43; relation to Sandelia, 42; representative of Anabantidae, 37, 73; secondary division fish, 63; suprabranchial cavity, 54; type genus, 37 Anabas testudineus: as carnivore, 38; fossil record, 61; hyobranchium, 119; neurocranium, 87, 103, 109; oromandibular region, 113; pectoral and pelvic girdles, 131; salt toler- ance, 62; skull, 81, 95 Anal pterygiophores: Anabantoidei, 58 Ancestral anabantoids: origin, 61 Ancestral features: Belontia, 44 Anguilla, 70 Angular: Anabantidae, 10; Belon- tiidae, 19; Helostomatidae, 30, 41; Osphronemidae, 25 Anisomerism, 58 Antepenultimate vertebra, 32 Aplocheilus, 63 Archetype, 64 Arrested evolution: Osphronemus, 65 Articular facet: epiotic, Helostoma- tidae, 28; intercalary, 23; pre- frontal, 26; sphenotic, 14, 27; supratemporal, 14 Articular process: cleithrum, 20; epi- otic, 23; hyomandibular, 19, 24, 31, 38 Ascending process of angular: Belon- tiidae, 19, 39, 50; Helostomatidae, 31; Osphronemidae, 25 Helostoma, 66; 142 Ascending process of dentary: Ana- bantidae, 10; Belontiidae, 20, 50; Osphronemidae, 25 Ascending process of parasphenoid, 23 Ascending process of premaxillary: Anabantidae, 10, 38, 49; Belon- tiidae, 20, 39, 50; Osphronemidae, 220) Asteriscus, 129 Atrophy of fin rays, 59 Auricular process of opercular: Belon- tiidae, 18; Helostomatidae, 29; Osphronemidae, 24 Basapophyses, 32 Basibranchials: Anabantidae, 8; Belontiidae, 17; Helostomatidae, 29; Osphronemidae, 24 Basicranial region: Anabantidae, 8; Belontiidae, 16; Helostomatidae, 28; Osphronemidae, 23 Basihyal: Anabantidae, 8; Helostoma- tidae, 29 Basioccipital: Anabantidae, 8; Belon- tiidae, 16; MHelostomatidae, 28; ophicephalids, 72; Osphronemidae, 23 Basipterygia: Anabantidae, 11, 38, 39; Belontiidae, 21, 44, 58: Helo- stomatidae, 31; Ophicephaliformes, 72; Osphronemidae, 25, 40 Basisphenoid: Anabantidae, 5; Belon- tiidae, 13; Helostomatidae, 26; Osphronemidae, 22 Bay of Djakarta: occurrence of Ana- bas, 68 Belontia: ancestral features, 44; as new generic name, 40; relation to Malpulutta, 45; representative of Belontiidae, 37; representative of Belontiinae, 47; taxonomic position, 44; type genus of Belontiinae, 73 Belontia hasselti: hyobranchium, 128; neurocranium, 89, 105, 109; omniv- orous habits, 52; oromandibular region, 115; pectoral and pelvic girdles, 133; skull, 83, 97 Belontiidae: composition, 87; diag- nosis, 39; otoliths, 129; protrusility of jaws, 50; summary, 73 Belontiids: evolution, 48 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI Belontiinae: composition, 47; diag- nosis, 48; summary, 73 Betta: representative of Belontiidae, 37; representative of Macropodinae, 47, 73; taxonomic position, 44 Betta splendens: breeding behavior, 45; hyobranchium, 123; neuro- cranium, 89, 105, 111; oromandibu- lar region, 115; skull, 83, 97 Betta unimaculata: carnivorous habits, om Body form, 55, 60 Boleophthalmus, 70 Bradytely: Osphronemus, 65, 74 Branchiocranium, 2 Branchiostegal rays: Anabantidae, 9, 43; Belontiidae, 18, 48; Helostoma- tidae, 29; loss, 57; Osphronemidae, 24; phylogenetic trends, 56; Tricho- gasterinae, 49 Breeding behavior: Betta, 45 Brevoortia, 51 Buccopharyngeal cavity: Helostoma, 2 Bulla acustica: Anabantidae, 7 Caudal vertebra: definition, 32 Cauda of sacculith, 35 Ceratobranchial: | Anabantidae, 8; Belontiidae, 17; Helostomatidae, 29; Osphronemidae, 24 Ceratohyal: Anabantidae, 8; Belon- tiidae, 18: Helostomatidae, 29; Os- phronemidae, 24 Chaetodon chinensis, 45 Channa, 70 Channiformes: osteology, 71, 72 Cisjordan, 62 Clarias, 70 Clariidae: regression of suprabranchial organ, 44 Cleithral symphysis: Anabantidae, 11, 38, 39 Cleithrum: Anabantidae, 11; Belon- tiidae, 20; NHelostomatidae, 31; ophicephalids, 72 Clupea, 51 Colisa: as opposed to Trichogaster, 46; osteological characters, 46; representative of Belontiidae, 37; representative of Trichogasterinae, 47, 73; suprabranchial cavity, 55 INDEX Colisa fasciata: neurocranium, 107, 111; oromandibular region, 117; skull, 85, 101 Colisa lalia: hyobranchium, 126; neurocranium, 93; number of verte- brae, 57; pectoral and_ pelvic girdles, 135, 137 Colliculi of sacculith, 35; Anaban- tidae, 38, 39; Belontiidae, 39, 40 Condyle: exoccipital, 23; quadrate, 30 Congruence of functions, 60 Convergence: Luciocephalus and ana- bantoids, 69; relation to air-breath- ing, 70; teleost radiation, 67 Convergent adaptations: air-breath- ing, 70 Coracoid: Anabantidae, 11; Belon- tiidae, 21; Helostomatidae, 31 Cretaceous: occurrence of teleosts, 61 Ctenoid scales, 35, 36, 40 Ctenopoma: distribution, 64; opposed to Anabas, 42, 43, 62; opposed to Sandelia, 43; physiological traits, 62; representative of Anabantidae, 37, 73; stomach contents, 52 Ctenopoma multispinis: graded series, 64; hyobranchium, 119; neurocra- nium, 87, 103, 109; oromandibular region, 113; skull, 95; taxonomic position, 42 Ctenopoma muriei: graded series, 64; skull, 81, 95; suborbitals, 6 Cupanus: taxonomic position, 45 Cylic evolution: clariids, 44 Cycloid scales, 35 Dentary: Anabantidae, 10; Belon- tiidae, 20; Helostomatidae, 30, 41; Osphronemidae, 25 Dentition: correlation with diet, 53; phylogenetic trends, 50 Dependency of evolving structures, 61 Diagnosis: Belontiinae, 48; Macropo- dinae, 48; Trichogasterinae, 48 Dipnoi, 70 Doras, 70 Dorsolateral process of angular, 31, 50 Belon- Ectopterygoid: absence in 143 tiidae, 19; Anabantidae, 10, 38, 39; Belontiidae, 39; Helostoma- tidae, 30, 41, 42; loss, 57; Osphro- nemidae, 24, 40, 41 Electrophorus, 69 El Faiyum, 62 Entopterygoid:. Anabantidae, 10; Belontiidae, 19; Helostomatidae, 30; Osphronemidae, 24 Eocene: occurrence of teleosts, 61 Epaxial muscles: Anabantidae, 55; Belontiidae, 15; Helostomatidae, 26, 56; Osphronemidae, 22, 23, 56 Epibranchial: Anabantoidei, 71, 72: Belontiidae, 17; Helostomatidae, 29; Ophicephaliformes, 71, 72; Os- phronemidae, 24: Sandelia, 43; taxonomic value, 72 Epibranchial process: Belontiidae, 17; Osphronemidae, 24 Epihyal: Anabantidae, matidae, 29 Epiotic: Anabantidae, 7; Belontiidae, 15; Colisa, 47; Helostomatidae, 28, 41; Osphronemidae, 23; Tricho- gasterinae, 47, 49 Epiotic process: Belontiidae, 15; Helostomatidae, 28; Osphronemi- dae, 23; Sphaerichthys, 46 Epiphytes, 52 Epipleural ribs: Anabantoidei, 32; Belontia, 17, 89; Belontiinae, 48; Colisa, 47; Sphaerichthys, 46; Trichogaster, 47 Erythrean Rift Valley, 63 Erythrinus, 70 Ethmoid: Anabantidae, 5; Belon- tiidae, 12; Colisa, 47; Helostoma- tidae, 26; Osphronemidae, 22, 40, 41; Sphaerichthys, 46; Trichogaster, 47 Ethmoid process of palatine: Belon- tiidae, 19; Helostomatidae, 30; Lu- 8: Helosto- ciocephalus, 69; Osphronemidae, 24 Evolution: Anabantoidei, 64, 65; Belontiidae, 48, 66, 67; dentition, 51, 53; hyobranchial region, 538; pelvic girdle, 60; postcranial axial skeleton, 57; pterygiophores, 58; suborbital shelf, 56 144 COMPARATIVE OSTEOLOGY AND PHYLOGENY OF THE ANABANTOIDEI Evolutionary pattern: Belontiidae, 66; fresh-water fishes, 63; hyobranchial apparatus, 52 Evolutionary sequence: dentition, 51, 53 Evolutionary trend: pelvic fins, 59 Exoccipital: Anabantidae, 7; Belon- tiidae, 15; Helostomatidae, 27; Os- phronemidae, 23 Extrascapular: Anabantidae, 6, 38; Belontiidae, 39; Helostomatidae, Al; loss, 57 Facial nerve, 59 Feeding habits: Anabas, 51, 52; Belontiidae, 51, 52; Helostoma, 41; Macropodinae, 67; Trichogaster- inae, 67 Filter apparatus, 41 Fontanelle: Sandelia, 5 Foramen exoccipitale: Anabantidae, 7; Belontiidae, 15; Helostomatidae, 28; Osphronemidae, 23 Foramen magnum: Anabantidae, 38; Belontiidae, 39; Colisa, 47; Helo- stomatidae, 41; increase in size, 56; Osphronemidae, 40; Trichogas- terinae, 47, 49 Foramen of trigeminal nerve: Ana- bantidae, 6 Frontals: Anabantidae, 5; Belontiidae, 13; Helostomatidae, 26; Lucio- cephalus, 70; ophicephalids, 71, 72; Osphronemidae, 22 Functional complexes, 61 Functional correlations, 60 Genetic correlations, 60 Genotype, 61 Gill rakers, 52, 53 Glossopharyngeal nerve: dae, 7; Belontiidae, 15 Growth pattern, 60 Gular element, 69 Gymnarchus, 70 Anabanti- Habitat: Ctenopoma, 43; Sandelia, 48 Helostoma: all-or-none reaction, 66; arrested evolution, 66; jaw mech- anism, 50; planktonic feeding, 51; quantum evolution, 66; representa- tive of Helostomatidae, 37; supra- branchial cavity, 55; as type genus, Al, 73 Helostoma temmincki: hyobranchium, 121; neurocranium, 89, 103, 109; oromandibular region, 115; osteo- logical specializations, 65; otoliths, 129; pectoral and pelvic girdles, 133; skull, 83, 97 Helostomatidae: composition, 37; diagnosis, 41; family of Labyrin- thici, 37; quantum evolution, 66, 74; summary, 73 Helostomidae Jordan, 41 Hingelike joint of dentary, 30, 42 Hippocampus, 51 Horizontal skeletogenous septum, 32 Horizontal sulcus of otolith, 35 Horny teeth: Helostoma, 538 Hyobranchial apparatus: Anabanti- dae, 8; Belontiidae, 17; evolution- ary pattern, 52; Helostomatidae, 29; Osphronemidae, 24; position, 54 Hyomandibular: Anabantidae, 10; Belontiidae, 19; Helostomatidae, 31 Hypertrophy: first pelvic fin, 59 Hypobranchials: | Anabantidae, 8; Belontiidae, 17; Helostomatidae, 121; Osphronemidae, 24 Hypohyal: Anabantidae, 8; Helo- stomatidae, 29 Hypurals, 32 Indo-Australian Archipelago, 63 Intercalary: Anabantidae, 7; Belon- tiidae, 15; Helostomatidae, 27; Os- phronemidae, 23 Interdependency: evolving structures, 61; trends, 60 Interhyal: Anabantidae, 8; stomatidae, 29 Internal carotid artery: Belontiidae, Ney, MG Interneurals: Anabantoidei, 32; Co- lisa, 47; Trichogaster, 47 Interopercular: Anabantidae, 9, 38; Belontiidae, 18, 39; Helostomati- dae, 30; Osphronemidae, 24, 40, 41 Interorbital septum: Anabantidae, 5; Betta, 13, 44; Helostomatidae, 26; Osphronemidae, 23 Helo- INDEX Isospondylous characters, 70 Jaw mechanism: acanthopterans, 49 Labrus opercularis, 45 Labyrinth organ: taxonomic value, on, 12 Labyrinthici, 1, 37 Labyrinthine suprabranchial organ: Anabantidae, 8; Belontiidae, 17 Lachrymal: Anabantidae, 5; Belon- tiidae, 13; Betta, 44; Helostomati- dae, 26; Osphronemidae, 22 Lagena, 34 Lagenolith, 34, 36, 129 Lapillus, 129 Lates niloticus, 39 Lemuria, 63 Lepisosteus, 70 Levator arcus branchialis, 54 Locomotion: Betta, 58; Colisa, 58: Sphaerichthys, 58; Trichogaster, 58; Trichopsis, 58 Lower Miocene: Asia-Africa bridge, 62 Lower Oligocene: mammalian fossils, 62 Luciocephalidae: Boulenger’s defini- tion of family, 69; Gill’s definition, 37 Luciocephaloidei: Berg’s definition, 37, 69; opposed to Anabantoidei, 69 Luciocephalus: opposed to Anaban- toidei, 69; as an osphronemid, 37 Luciocephalus pulcher: otoliths, 129; resemblance to Anabantoidei, 70; subordinal rank, 387; taxonomic status, 69, 70 Macropodinae: composition, 47; diag- nosis, 48; summary, 73 Macropodus: osteological characters, 45; relation to Malpulutta, 45; representative of Belontiidae, 37; representative of Macropodinae, 73; suprabranchial cavity, 55 Macropodus chinensis: opposed to M. cupanus, 45; osteological charac- ters, 45 Macropodus cupanus, 74; neurocra- nium, 91; opposed to M. opercularis, 145 45; oromandibular region, 117; skull, 83 Macropodus_ opercularis, 74; hyo- branchium, 125; meristic variation, 58; neurocranium, 91, 105, 111; omnivorous habits, 52; opposed to M. cupanus, 45; osteological char- acters, 45; pectoral and_ pelvic girdles, 135; skull, 85, 99 Macropodus viridi-auratus, 45 Malpulutta: relation to Belontia, 45; relation to Macropodus, 45; rela- tion to Sphaerichthys, 45; repre- sentative of Belontiidae, 37; repre- sentative of Macropodinae, 47, 73; taxonomic position, 45 Marl shales, 61 Maxillary: Anabantidae, 10; Belon- tiidae, 20; Helostomatidae, 30; Os- phronemidae, 25 Meristic variation: Macropodus, 58 Mesocoracoid, 72 Metapterygoid: Anabantidae, 10, 38, 39, 71; Belontiidae, 19, 44; Belon- tiinae, 48; Helostomatidae, 30, 41; Ophicephaliformes, 71, 72; Osphro- nemidae, 25, 40 Miocene: Asia-Africa bridge, 62 Misgurnus, 70 Monopterus, 69 Morphological trends: correlated with size, 58; fin rays, 58; lateral line, 58; scales, 58; Trichogasterinae, 67 Mosaic evolution, 61 Myodome: Anabantidae, 6; Belon- tiidae, 14; Helostomatidae, 27, 28 Nasal region: Anabantidae, 5; Belon- tiidae, 11; Helostomatidae, 25; Os- phronemidae, 21 Nasals: Anabantidae, 5; Belontiidae, 11; Helostomatidae, 25; Osphro- nemidae, 21; Sandelia, 43; Sphae- richthys, 46; Trichopsis, 44 Naturphilosophen, 64 Neural arches, 31 Nomenclature: girdles, 2; otoliths, 4; postcranial axial skeleton, 2; skull bones, 2 Objectives of study, 1, 2 Oil containers in eggs, 45 146 COMPARATIVE OSTEOLOGY AND Olfactory foramen: Belontiidae, 12; Helostomatidae, 26; Osphronemi- dae, 22 Ontogenetic mechanisms, 60 Opercular: Anabantidae, 9, 38; Belon- tia, 44; Belontiidae, 18: Helosto- matidae, 29; Osphronemidae, 24 Opercular series: Anabantidae, 9; Belontiidae, 18; Helostomatidae, 29; Osphronemidae, 24 Ophicephalidae: as family of Labyrin- thici, 37 Ophicephalids: ordinal affiliation, 72: relation to anabantoids, 71, 72: taxonomic position, 72 Ophicephaliformes: convergence with anabantoids, 72; opposed to Perci- formes, 72; ordinal affiliation, 71, 74; osteology, 71, 72; resemblance to Perciformes, 72; suprabranchial organ, 71; taxonomic position, 1 Te 2 Opisthotic, 7 Orbital process of frontal: Anabanti- dae, 5; Belontiidae, 13; Betta, 13; Macropodus cupanus, 13 Orbital region: Belontiidae, 13; Helo- stomatidae, 26; Osphronemidae, 22 Orbital wing: Osphronemidae, 22, 25, 26 Orbitosphenoid, 72 > Oriental region: origin of anaban- toids, 61, 62 Oromandibular apparatus: — Lucio- cephalus, 70 Oromandibular region: Anabantidae, 9, 39; Belontiidae, 19, 39: Helo- stomatidae, 30; Osphronemidae, 24, 4] Osphronemidae: composition, 37; diagnosis, 40; as family of Labyrin- thici, 37; as family of Perciformes, 37; summary, 73 Osphronemus: adaptive features, 65; arrested evolution, 65; bradytely, 65; herbivorous habits, 51; repre- sentative of Osphronemidae, 37: suprabranchial organ, 55; tachytely, 65; type genus, 40, 73 Osphronemus goramy: fossil record, 61, 74; hyobranchium, 121; neuro- cranium, 87, 103, 109; occurrence PHYLOGENY OF THE ANABANTOIDEI in Upper Eocene, 65; opercular region, 113; oromandibular region, 113; otoliths, 129; pelvic girdle, 113; physiological traits, 63; pres- ence in Lower Eocene, 63; skull, 81 Ostium of sacculith, 35, 38, 39, 40 Otic bulla: Anabantoidei, 71; Ophi- cephaliformes, 71, 72 Otic region: Anabantidae, 6; Belon- tiidae, 13; Helostomatidae, 27; Osphronemidae, 22 Otoliths: Anabantoidei, 34, 35, 129; Luciocephalus, 70, 129; nomen- clature, 4 Owen, 64 Palatine: Anabantidae, 9; Belontiidae, 19; Helostomatidae, 30; Osphro- nemidae, 24 Palatine facet, 22 Paleocene, 62 Paleogeography, 74 Parallelism, 67, 70 Parasphaerichthys: opposed to Sphae- richthys, 47; representative of Be- lontiidae, 37; representative of Trichogasterinae, 47, 73; taxonomic position, 47 Parasphenoid: Anabantidae, 8, 38; Belontiidae, 16; Helostomatidae, 28, 41; Ophicephaliformes, 71, 72; Osphronemidae, 23 Parietals: Anabantidae, 6; Belontiidae, 14; Helostomatidae, 27; Osphro- nemidae, 23, 40 Parosphromenus: representative of Belontiidae, 37; representative of Macropodinae, 47, 73; taxonomic position, 45 Pelvic fin rays: Colisa, 47; function, 59; Trichogaster, 47 Pelvic girdle: Belontiinae, 58; in- ternal anatomy in Osphronemus, 59, 60; Ophicephaliformes, 72; phylogenetic trend, 58 Penultimate vertebra, 32 Percesoces, 1, 37 Perciformes, 1, 37, 67; opposed to Ophicephaliformes, 72 Percoidei, 67 Periophthalmus, 69, 70 INDEX Pharyngeal processes: Anabantoidei, 71; interdependency with supra- branchial cavity, 60; Ophicephali- formes, 71, 72 Pharyngeal process of basioccipital: Anabantidae, 8, 388; Belontiidae, 16, 39; Colisa, 47; Helostomatidae, 28; Osphronemidae, 24; Sphaerich- thys, 46; Trichogaster, 47 Pharyngeal process of parasphenoid: Belontiidae, 16, 39; Helostomati- dae, 28, 41, 42; Osphronemidae, 23; relation to body form, 55; re- lation to suprabranchial cavity, 54; Sphaerichthys, 46 Pharyngeal teeth: Anabantidae, 52; Belontiidae, 53; Helostomatidae, 53; Osphronemidae, 52 Pharyngobranchial: Anabantidae, 9; Belontiidae, 18; Helostomatidae, 29; Osphronemidae, 24 Phenotypical expression, 61 Phyletic evolution, 64; Belontiidae, 66 Phylogenetic trends, 49; body form, 55; branchiostegal rays, 56; denti- tion, 50; general, 60; pelvic girdle, 58; protrusility of jaws, 49; sub- orbital shelf, 56; summary, 74; suprabranchial cavity, 54 Phylogeny: anabantoid families and subfamilies, 138; anabantoid genera, 139; uniformity of trends, 60, 61 Pleistocene, 61 Pleural ribs, 32; Sphaerichthys, 46 Pliocene: Asia-Africa bridge, 62 Polyacanthidae, 37, 40 Polyacanthus hasselti: type genus, 40 Polyacanthus opercularis: synonym, 40 Polyisomeres, 57 Polyisomerism: anal pterygiophores, 58; vertebral column, 57 Polypterus, 70 Postcleithrum: Anabantidae, 11; Be- lontiidae, 20; Helostomatidae, 31 Posterior process of quadrate, 19, 25, 30 Postorbital process: Helostomatidae, 28; Osphronemidae, 22 Posttemporal: Anabantidae, 7, 43; 147 Belontiidae, 16; Helostomatidae, 28 Posttemporal fossa: Anabantidae, 7. 38; Belontiidae, 39; Helostomati- dae, 27, 28; Osphronemidae, 23; relation to body form, 55 Post-Tethys immigration, 62 Postzygapophyses, 31 Preadaptive polyphagous stage, 53 Predaceous behavior: Anabantidae, Bi Prefrontal-lateral ethmoid: Anabanti- dae, 5; Belontiidae, 12: Helosto- matidae, 25; Osphronemidae, 22 Premaxillary: Anabantidae, 10; Belon- tiidae, 20; Helostomatidae, 30, 41: Osphronemidae, 25 Preopercular: Anabantidae, 9; Belon- tiidae, 18, 44; Helostomatidae, 30; Osphronemidae, 24, 40 Prevomer: Anabantidae, 5, 38; Belon- tiidae, 13, 39; Helostomatidae, 26, 41, 42; Osphronemidae, 22, 40 Prezygapophyses, 31 Processus lateralis: Anabantidae, 11; Belontiidae, 21; Helostomatidae, 31; Osphronemidae, 25 Processus medialis; Anabantidae, 11; Belontiidae, 21; Helostomatidae, 31; Osphronemidae, 25 Processus medialis accessorius: Colisa and Trichogaster, 21, 59; Osphro- nemidae, 25 Processus medialis anterior: Anabanti- dae, 11; Belontiidae, 21, 58, 59; Helostomatidae, 31; Osphronemi- dae, 25 Processus medialis posterior: Anaban- tidae, 11; Belontiidae, 21; Helosto- matidae, 31; Osphronemidae, 25 Programme evolution, 61 Prootic: Anabantidae, 6; Belontiidae, 14; Helostomatidae, 27; Ophi- cephaliformes, 72; Osphronemidae, 23 Protractile jaw mechanism, 39 Protrusility of jaws, 39, 49, 50 Pterosphenoid: Anabantidae, 5; Be- lontiidae, 13; Belontiinae, 49; Helo- stomatidae, 26, 41; Osphronemi- dae, 22 Pterygiophores, 32; polyisomerism, 58 148 COMPARATIVE OSTEOLOGY AND Quadrate: Anabantidae, 10; Belon- tiidae, 19; Helostomatidae, 30; Os- phronemidae, 25 Quantum evolution: Helostoma, 66, 74 Quantum shifts, 67 Rachis of pelvic girdle: Anabantidae, 11; Belontia, 44; Belontiidae, 21; Colisa, 59; Helostomatidae, 31; Macropodinae, 58; Osphronemidae, 25; Trichogaster, 59; Trichogaster- inae, 58 Radials: Anabantidae, 11; Belon- tiidae, 21; Helostomatidae, 31; Sphaerichthys, 46 Rasbora, 63 Regression of suprabranchial organ, 44 Residual air, 55 Retroarticular: Anabantidae, 11; Be- lontiidae, 20; Helostomatidae, 31; Osphronemidae, 25 Reversed evolution, 58 Ribs, 32 Saccobranchus, 70 Saccular swelling, 7 Sacculith, 34, 35, 38, 39, 44, 69, 129 Sagitta, 129 Salt tolerance, 62 Sandelia: origin, 64; osteological char- acters, 64; relation to Anabas, 42; relation to Ctenopoma, 43; repre- sentative of Anabantidae, 37, 73; stomach contents, 52 Sandelia capensis: hyobranchium, 119; neurocranium, 87, 103, 109; oromandibular region, 1138; pec- toral and pelvic girdles, 131; skull, 81, 95 Scapula: Anabantidae, 11; Belon- tiidae, 21; Helostomatidae, 31 Schistic evolution, 65, 66 Secondary division fresh-water fish, 63 Secondary polyisomerism, 57 Selection forces on labyrinth organ, 72 Selection pressure, 66 Skull elements: loss, 57 Sliding stalk of premaxillaries, 49, 50 Sodium alizarin sulfonate, 2 PHYLOGENY OF THE ANABANTOIDEI South African fauna, 64 Sphaerichthys: as opposed to Para- sphaerichthys, 47; osteological characters, 46; relation to Malpu- lutta, 46; representative of Belon- tiidae, 37; representative of Tricho- gasterinae, 47, 73; taxonomic posi- tion, 46 Sphaerichthys osphromenoides: hyo- branchium, 125; neurocranium, 93, 107, 111; oromandibular region, 117; pectoral and pelvic girdles, 135; skull, 85, 99 Sphenotic: Anabantidae, 6; Belon- tiidae, 14; Helostomatidae, 27; Ophicephaliformes, 72; Osphronem- idae, 22 Subopercular: Anabantidae, 9; Belon- tiidae, 18; Helostomatidae, 29; Osphronemidae, 24, 40 Suborbitals: Anabantidae, 43; Belon- tiidae, 13; Helostomatidae, 26; Os- phronemidae, 22, 40, 41 Suborbital shelf: Anabantidae, 6; Belontiidae, 13; evolution, 56; func- tion, 56; Helostomatidae, 26; Os- phronemidae, 22, 40, 41, 56; phylo- genetic trend, 56; Trichogasterinae, 49 Superficial abductor muscles, 59 Suprabranchial cavity: Anabantidae, 54; Belontiidae, 17, 48; Belontiinae, 54; interdependence with pharyn- geal processes, 60; Macropodinae, 49, 54; phylogenetic trend, 54; Trichogasterinae, 54 Suprabranchial organ: Anabantoidei, 71, 74; Helostomatidae, 29; Lucio- cephalus, 69, 74; Ophicephali- formes, 71, 74; Osphronemidae, 24; Sandelia, 43, 70; taxonomic character, 69; taxonomic value, 70 Supracleithrum: Anabantidae, 11; Belontiidae, 20; Helostomatidae, 31 Supraoccipital: Anabantidae, 7; Be- lontiidae, 15; Helostomatidae, 28; Osphronemidae, 23 Supraoccipital crest: Anabantidae, 7; Belontiinae, 49; Helostomatidae, 28; Macropodinae, 49; Osphro- nemidae, 23 INDEX Supratemporal: Anabantidae, 6; Be- lontiidae, 14; Helostomatidae, 27; Osphronemidae, 22 Swim bladder: Anabantoidei, Luciocephalus, 69 Symplectic: Anabantidae, 10; Belon- tiidae, 19; Helostomatidae, 30; Os- phronemidae, 25 Symplectic process of hyomandibular, 10, 19, 30, 31, 38, 39, 40, 41, 44, 48 Syngnathoides, 51 69; Tachytely: general, 65; Osphronemus, 65, 74 Taste buds, 59 Teeth: Anabantoidei, 51; hyobran- chial apparatus, Anabantidae, 38, Belontiidae, 17, 18, Helostomati- dae, 41, Osphronemidae, 24, Tricho- gasterinae, 46; palatine, Ana- bantidae, 9, 43, Belontiidae, 39; parasphenoid, Anabantidae, 8, Be- lontiidae, 16, 17, 46, 49, Helosto- matidae, 28, Osphronemidae, 23; prevomer, Anabantidae, 5, 38, Be- lontiidae, 39; phylogenetic trends, 50 Tertiary: African invasion, 62 Tethys Sea, 62 Transjordan, 62 Transverse process of parasphenoid: Anabantidae, 8, 38; Belontiidae, 16, 39; Helostomatidae, 28, 41, 42 Trichogaster: opposed to Colisa, 46; osteological characters, 46; repre- sentative of Belontiidae, 37; repre- sentative of Trichogasterinae, 47, 73; suprabranchial cavity, 55 Trichogasterinae: composition, diagnosis, 48; summary, 73 AT: 149 Trichogaster leeri: neurocranium, 111; oromandibular region, 117; pelvic girdle, 137; skull, 101; stomach contents, 52 Trichogaster_ trichopterus: hyobran- chium, 127; neurocranium, 93, 107; skull, 85, 101; stomach contents, 52 Trichopsis: osteological characters, 44: representative of Belontiidae, 37; representative of Macropo- dinae, 47, 73; suprabranchial cav- ibyp 9 Trichopsis vittatus: hyobranchium, 123; neurocranium, 91, 105, 111; omnivorous habits, 52; oromandib- ular region, 115; pelvic and pec- toral girdles, 113; skull, 83, 99 Trigemino-facialis chamber, 14 Tympanum of foramen exoccipitale, 7 Ultimate vertebra, 32 Umbra, 70 Uniformity of trends, 60 Upper Cretaceous: teleost radiation, 62 Upper Eocene: mammalian fossils, 62 Urohyal: Anabantidae, 9; Belontiidae, 18; Helostomatidae, 29; Osphro- nemidae, 24 Utriculith, 34, 129 Utriculus, 34 Vagus nerve: Anabantidae, 7 Ventilation, 55 Vertebrae, 31, 38, 40, 57 Wallace’s line, 63 Zambezi River system, 64 weedeat * cai aaron = iy. 2b Pmt hen CF The Sher) oe { ‘ 7447) i > a _s } u = - — edi jk gare Oise Jatt : ‘Tanto oe nee ae Ptlis Jike’| : 7 : m) PES gtabtinniags Tel seul, ‘wend 0 7 eee Meee oh 22058 “0 wiree ~~el =~ - : a a SAS e, a MingtRes” bees +" Una »r ie.) " sf VE, ee ie a , > “rr se f Ap . ¥ . ‘ % va } / he Pn wey i> r ‘ ree a ¥ hoa 7 ’ los ; ‘ & j ar tea ee a y h i :? 71 P § : Lye @ 7 r f ' i _ . > 7 ; rl ce ft 7) pa ~ Ly ( 7 i indbrowatle 2 \ ~ ’ ; i Mir 7 ‘Wmrutire x - ’ > , j = r | a :: | sa) - x i . m ‘ > > boned f : - : = ae > y = =m a , ~ i = ia / = Z - f ‘ at g A . 2 \ 73 , — = 2 A b x t . ‘ - , | JEANNE (1959): The Morphology tomy of American Species of the saronius. 82 figs. No, 27. $3.00. GER, JAMES M. (1962) Taxonomy ‘Setaria (Gramineae) in North America. eae 6 pls. No. 29. $3.00. I OOF, HERBERT FREDERICK (1942): The Genus Conotrachelus Dejean (Coleoptera, culionidae) in the North Central Sheltie 9 pls. Vol. 19, No. 3. $1.50. * " SELANDER, RICHARD B. (1960): Bionomics, Systematics, and Phylogeny of Lytta, a M leloidae). 350 figs. No. 28. $4.50. i SMITH, ‘FRANK (1923): The Calciferous Glands of Lumbricidae and Diplocardia. 12 pls. Vol. 9, No. 1. $1. BD. 4 Mcteederss' WILHELM GERHARD (19380): Morphological Studies of the Genus ee 4 pls. Vol. 12, No. 1. $1.00. enus of Blister Beetles (Coleoptera, SPOONER, CHARLES STOCKMAN (1938): The Phylogeny of the Hemiptera, Based on a Study of the Head Capsule. 24 pls. 2 figs. Vol. 16, No. 3. $1.00. SPRAGUE, vicToR (1941): Studies on Gregarina blattarum with Particular Ref-_ erence to the Chromosome Cycle. 6 pls. 2 figs. Vol. 18, No. 2. $.75. STANNARD, LEWIS J., JR. (1957): The Phy- logeny and Classification of the North American Genera of the Suborder Tubu- lifera (Thysanoptera). 14 pls. No. 25. $2.50. STEVENS, FRANK LINCOLN (1924): Para- sitic Fungi from British Guiana and Trini- - dad. 19 pls. 1 map. Vol, 8, No. 3, $1.25. and RYAN, SISTER MARY HILAIRE (1939): The Microthyriaceae. Vol. 17, No. 2. $1.50. TEHON, LEO Roy (1935): A Monographic Rearrangement of Lophodermium. 5 pls. Vel. 18, No. 4. $2.00. ‘ VAN CLEAVE, HARLEY J. (1953): Acantho- cephala of North American Mammals. 13 pls. 10 figs. Vol. 23, Nos. 1-2. $4.00. YUNCKER, TRUMAN GEORGE (1921): Re- _ vision of the North American and West Indian Species of Cuscuta. 13 pls. Vol. 6, Vet 2-3. $2.00. U “eh " I j fa a oe lel ’ CO ee a iM; i\ A | ; ANY 10a Ay may ; | j NY Aba Vy pATITE A ‘ j WHE