BULLETIN of THE BIOLOGICAL SOCIETY or WASHINGTON 27 OCTOBER 2004 NUMBER 11 Editor: Richard v. Sternberg Review Editor: Lynne R. Parenti Reviewers: Gareth Nelson and Richard Winterbottom Copies available as the supply lasts from: The Custodian of Publications Biological Society of Washington National Museum of Natural History Smithsonian Institution Washington, D.C. 20560 (Cost $75.00 for the 2-volume set, including postage and handling) Cover illustration: Pholidichthys leucotaenia, dorsal gill-arch musculature (see Plate 169). “This whole book is but a draught—nay, but the draft of a draft.” Herman Melville, Moby Dick. STUDY OF THE DORSAL GILL-ARCH MUSCULATURE OF TELEOSTOME FISHES, WITH SPECIAL REFERENCE TO THE ACTINOPTERYGII Victor G. Springer and G. David Johnson Karolyn Darrow, Illustrator Appendix: Phylogenetic Analysis of 147 Families of Acanthomorph Fishes Based Primarily on Dorsal Gill-Arch Muscles and Skeleton Victor G. Springer and Thomas M. Orrell (VGS, GDJ, TMO), Division of Fishes MRC 159, Department of Zoology National Museum of Natural History, PO. Box 37012 Washington, D.C. 20013-7012, U.S.A. e-mail: Springer. Victor@nmnh.si.edu (KD) Department of Entomology MRC 105 National Museum of Natural History, PO. Box 37012 Washington, D.C. 20013-7012, U.S.A. e-mail: Darrow.Karolyn@nmnh.si.edu Abstract.—The dorsal gill-arch musculature (DGM), aspects of the asso- ciated skeleton, the transversus ventralis 4, and the semicircular ligament are described for many species in over 200 families and over 300 genera of teleostome fishes, and the DGM musculature of over 200 taxa is illustrated. A partially new system of DGM nomenclature is used. The transversus dor- salis, is shown to be a much more complex system than has been generally recognized. DGM data are variously analyzed and shown to be of importance for defining various currently recognized suprageneric pre-acanthomorph taxa. Among the many conclusions pertaining to acanthomorphs are: monophyly of Percopsiformes is hypothesized; Icosteidae, Menidae, and Centriscidae are probably more closely related to pre-percomorph groups than to percomorph groups; there is no basis for inclusion of Amarsipidae in the Stromateoidei; monophyly of the Polycentridae (Polycentrus, Polycentropsis, Afronandus, Monocirrhus) is corroborated based on a combination of gill-arch muscle and additional characters. A new ordinal-group name, Anabantomorpha, is erected to include the seven families with parasphenoid teeth: Nandidae, Badidae, Pristolepidae, Channidae, Anabantidae, Heleostomatidae, and Osphronemi- dae. Anabantoidei includes the last four of these families, which have supra- branchial organs. The superfamily name Labroidea is proposed to distinguish the unequivocally monophyletic group of families, Labridae, Odacidae, Scar- idae, from other families included in the suborder Labroidei. The first syna- pomorphy for the gobioid family Odontobutidae is hypothesized based on the position of the /evator internus 2 relative to the obliquus dorsalis. Certain gill-arch skeletal characters previously unrecognized or inadequately evalu- ated are reported and discussed (e.g., epibranchial-ceratobranchial accessory cartilages; relationship of epibranchials 5 and 4; epibranchial 4 flange; esoph- ageal raphe). A separately authored appendix provides a cladistic analysis of 168 taxa in 147 acanthomorph families based almost exclusively on DGM and gill-arch skeletal characters. Among the many results implied by the study are: mono- phyly of Percopsiformes is corroborated. Smegmamorpha Johnson and Pat- terson (1993) are polyphyletic, and the name is rejected for nomenclatural purposes. Its constituents comprise two or three not closely related clades: Mugilomorpha + Atherinomorpha (= Percesoces Cope, 1875), Gasterosteo- morpha, and, possibly, Centrisciformes (a pre-percomorph group, comprising only Centriscidae). Gasterosteomorpha includes a monophyletic Hypoptych- idae (Hypoptychus + Aulichthys), Elassomatidae, Aulorhynchidae, monophy- letic Synbranchiformes (Synbranchidae + Mastacembelidae), and Gasteros- teidae, thus corroborating various hypotheses of Johnson and Patterson (1993) and Johnson and Springer (1997). Labroids are monophyletic only with in- clusion of Pholidichthyidae, but the group remains supported only by phar- yngognath characters. Blennioidei are monophyletic and their intra-relation- ships resolved. Their closest relatives are, stepwise: Gobiesocidae, Draconet- tidae (+ Callionymidae, which were not included in the analysis), Dactylop- teridae. A new ordinal-group name, Benthomorpha, is proposed for this clade. Caproidae are polyphyletic; relationships of its two genera appear to be with tetraodontiforms on the one hand, and acanthuroids, on the other. As such, Johnson and Patterson’s (1993) hypothesis that caproid relationships are among percomorphs is corroborated. Sphyraenidae and Polynemidae form a monophyletic group (first proposed by Regan, 1912). CONTENTS ins teAuithomswerelacc me menE een see orice 1 Glupeidace ease ner 43 INtrOdUCHON cue ne ACO E OG el ane oe no eee 2 Gonorynchiformes ............. 44 Methods) 22... te S Ash see oehper svocelieeier: 2 (QUBWWCE® sscscccocvceecs 44 Wlaterilall!) 2 <:.\c/ceaiake irae recone et pesca ees Pa 3 Gonorynchidae .......... 45 Muscles and Skeletal Elements ................ 3 Cypriniformes ................. 46 IMMISCIS MAMNES soccccconccccdgcoeuuandgonoes 3 Gy punidacieee eee eee 46 Muscleity pes mysraer ese heer paciaier reer 4 (COMETEYSOMINES 565000000000 00c00 48 Origins and insertions ....................-. 4 Characidaeiiee sss ae 48 Anatomical orientation ...................-. 4 Distichodontidae ........ 49 TROYES WHMUTEIES Go0000ces0000000800000005 4 SHINMTAVIOTAINGS 5 o5cqccagccanvcccns 49 Acanthomorph accessory cartilages .......... 4 Diplomystidae ........... 49 Epibranchials 5 and 4 ...................... 5 Gymnotiformes ................ 51 Abbreviations and Definitions for Anatomical Gymnotidae ............. 51 StMGlUreS ase ceacasceenoelictatiy sea rsueraere 6 Salmonitonnes tsar ener Sl Classification) eet eee cece 16 Salmonidae ............. 51 Gnathostomatal Ayer oor eee ee 18 Retropinnidae ........... 53 MElEOStOmi Ass eee ic ees eM 18 Galaxaidacianeee eee eeeee 54 SANGO USAR cococcacvcosccc0s0c00KnGeEs 18 Osmendacyeeee eee 56 Coelacanthomorpha Argentiniformes ............... S// Coelacanthidae .......... 18 INGUNUIMVCEYS 550000000006 S7/ Dipnoi. Giese hearths us Ochs oid eee 19 Alepocephalidae ......... 3)7/ Ceratodontidae .......... 19 Platytroctidae ........... 58 ING NOP UMTAVAN 5ooccccccccg0beabcenGaKecs 21 ESocifonmesseenneeee ellen 59 Pre-Acanthomorpha ................-. 21 Esocidaciiamt aes eee a) Cladistial siyicdenc fehl s= daeeas 21 Wimbridaceeeee nee 60 Polypteridae ............ 21 SUOMUVIOVATNES cocococsaccvccenec 62 OWONEROSIE scccacnccccacnc0c0s 22 Diplophidae ............. 62 Acipenseridae ........... 22 Sternoptychidae ......... 63 Polyodontidae ........... 23 Gonostomatidae ......... 63 Ginglymodihee eee oe nee 24 Ateleopodiformes .............. 64 Lepisosteidae ........... 24 Ateleopodidae ........... 64 Halecomorphilipee eee 24 Aulopitornmes saps nee ene 65 INTIS 5050000000000000 24 ANUNIGDIGENS 5 56c600000000¢ 65 Osteoglossomorpha ............. 2D) Synodontidae ........... 66 Hiodontidae ............. 25 Chlorophthalmidae ....... 66 Notopteridae ............ 26 Myctophiformes ............... 67 Gymnarchidae ........... 2 Neoscopelidae ........... 67 Mormyridae ............ 27 Myctophidae ............ 68 Arapaimidae ............ 28 Results—Pre-Acanthomorpha .... 69 Pantodontidae ........... 30 Tables t= wee See sete ee 75-81 Osteoglossidae .......... 31 Acanthomorpha .................... 74 Elopomorpha .................. 32 Additional material ............. 74 Megalopidae ............ 32 ampridiionnesisseeaeeene reer 80 Blopiddeanee eee 33 Veliftendaciee ane 80 ANIGTHMNCENS 5 c5000000000006 34 Lampridae .............. 81 Notacanthidae ........... 36 Polymixiiformes ............... 82 Halosauridae ............ 37 Polymixiidae ............ 82 (COMBUCES oooccsccccon0e 37 Paracanthopterygil ................ 83 PNOAVIUUTICENS Gogccccocccnc 38 Percopsiformes ................ 83 Synaphobranchidae ...... 39 Aphredoderidae ......... 84 @lupeomorphayaeeee eee eee 39 IRercopsid acyee eee 84 Denticipitidae ........... 39 Amblyopsidae ........... 85 Pristigasteridae .......... 40 @phiditornmes essere 86 Eneraulidacaeeenene errr 41 Ophididaceasneeee eee 86 Chirocenthidaceae eee eer 42 Bythitidacne eee eeeeeeee 88 @arapidae™....<\0 sano 88 Gadilonnes#ey. oss seen nese ees 89 Ranicipitidaeyes eases ane a 89 Batrachoidiformes .............-- 90 Batrachoididae .......... 90 Bophiitornme sie eases eer 90 Chaunacidae ............ 90 ANcanthopteny oles snes 91 Stephanoberyciformes .......... 91 Melamphaidae .......... 91 Gibberichthyidae ........ 92 Stephanoberycidae ....... 92 Barbourisiidae ........... 93 Rondeletiidae ........... 93 Cetommmidaeeeaes see 94 Icosteifonrmesmeeeeee ee oe eee 94 Kcosteidaeien ase 94 MableskSiSAOw ya ssascse tae. 98-107 Zeiformes and Possible Relatives EAS PAE ck 8 A195 gk SORE 107 ZENON iee se eeae a oe sites 108 Oreosomatidae ......... 108 Rarazenigaeme eae eee 108 Zenmontidacweeeee eee 108 Grammicolepidae ....... 109 (CEVOMGMINOITAINES so 0n0ceueancacece 110 Gaproidaciaen eee 110 Tetraodontiformes ............. 111 Triacanthodidae ........ 111 IMI@IMNIIOMTEMES sccesacauacccdudse Li Menidae fa niasiecsee ae 112 IBSITYWEMIOMINES aca scncotansaocos 113 Trachichthyidae ........ 113 IBEIAKEMOS scoocadcaoces 114 Holocenthdaewsaee eee 115 Anomalopidae .......... 116 Rercomorp hale eee 117 Smegmamorpha ............ LA 7/ Gasterosteomorpha .......... 117 Centniscifommes 2.2... ..55- LZ CEMIISOCHO ssccscooces I 7/ Gasterosteiformes ......... 118 Gasterosteidae .......... 118 Hypoptychidae ......... 120 Aulorhynchidae ........ 121 Synbranchiformes ......... 122 Synbranchidae ......... 122 Mastacembelidae ....... 123 Elassomatiformes ......... 124 Elassomatidae .......... 124 Mugilomorpha .............. 124 IMME sosocogaacoce 124 Atherinomorpha ............ 126 Atheriniformes ........... 126 Atherinidae ............ 126 Bedotiidae ............. 126 Cyprinodontifomes ........ 127 Aplocheilidacwenesee see 127 Cyprinodontidae ........ 127 Belonitionness sees 128 Adrianichthyidae ....... 128 Belonidacmenme eee 129 Scomberesocidae ....... 130 Hemiramphidae ........ 131 IE OCOSIGRES 5550000050000 131 IREROUOMMMNES s5cagcgnvccgrooe 133 Acropomatidae ......... 133 Percichthyidae ......... 133 Leptobramidae ......... 134 Watldaer. si 1. i BAe 135 Centropomidae ......... 135 Centrarchidacieeereeoeee 136 Bathyclupeidae ......... 137 Symphysanodontidae .... 138 Epigonidae 25-5. .-.-.-- 139 IMKoTONVGETES 55G50cc0c0000 140 Semanidacmepee eae 140 utjanidaese seer 141 Hacmulidacwaeeseeee 142 IMenmnitdacwneae eee ae 142 Apogonidae ............ 143 Priacanthidae ........... 143 Ostracoberycidae ....... 144 Gihitidaciaeeee ene 144 Pemphendacieer eee 145 Glaucosomatidae ....... 145 Wactanidacimertce rer 146 Lateolabracidae ........ 147 SOHENCES sosacacascese 147 Rolynenidaceee ss aeeaener 148 Sillagimidacteeeeesee sere 149 IMMINICHS so saccccccnced 150 Centrogeniidae ......... 151 Ambassidae ............ 151 Caristidacieenanee eee: 152 Bramidaewaeneeee oe 153 Moxotidacte meee eee 154 IPSSO)NIGES soaccossnoce 155 Rencidacserye nit. fers 156 Cepolidaciensenee oe 157 Callanthtidae ........... 158 Genmeldacienee eee 159 Grammatidae ........... 160 Opistognathidae ........ 160 Pseudochromidae ....... 161 Leiognathidae .......... 162 Roly centhdacseen see: 163 Sphyraenidae .......... 166 Keuntidaers ies aceon 167 Ammodytidae .......... 168 Trachinidae ............ 168 Uranoscopidae ......... 169 Cheimarrichthyidae ..... 170 Scorpacnoid cine 170 vi Scorpaenidae ........... 171 Sebastidaenee eee eee 171 Platycephalidae ......... 172 Champsodontidae ....... 172 Hexagrammidae ........ 173 Anoplopomatidae ....... 175 Rhamphocottidae ....... 175 (COUTHCES socscoccecagc0e 176 Normanichthyidae ...... ei (CargmexornGle 5.oc0c0ca0ca00ce 177 Nematistiidae .......... 177 Caran cidacweene een eee 178 Rachycentridae ......... 178 Coryphaenidae ......... 179 Echeneidacsenaaeae eee: 180 Scombroidei ................- 181 Pomatomidae .......... 181 Scombrolabracidae ...... 182 Scombridae ............ 182 SyORMRONGIE 2 sccaccacnonccccoc 183 Nemipteridae ........... 183 Wethnnidaey sessn ese seee 184 Centracanthidae ........ 185 SOIMGAS 5co0ccccccccc0s 185 Girelloideiaeee eee 187 Girellidacwee esses nee 187 IMAI EY oooccocececc0c 188 Terapontidae ........... 188 Wabroideiiyys cece cecios cease 189 Cia soccocesocvcce 189 Pomacentridae ......... 191 Embiotocidae .......... 193 ILFINRONCER socoocooecodceuc 194 [Lala s5ccucccc00c000 196 Pholidichthyoidei ........... 199 Pholidichthyidae ........ 200 Acanthuroidei .............. 201 LUNAS sonocconn000s 201 Ephippidae ............ 202 Zanclhidacwnnnre een 202 Acanthuridae ........... 203 Anabantomorpha ............ 203 INETVENGEYS pcocccc0csccces 204 Badidaek = aanatnis cone 204 Pristolepidae ........... 205 Channidacseeeee sere eee 206 Anabantidae ........... 207 Stromateoidei .............. 207 Amarsipidae ........... 208 Centrolophidae ......... 209 Z@Oarcoldeim@ee Faeries ee eee 209 Bathymasteridae ........ 209 Zaproridae ............. 210 Pholidaenssse os ac. see 211 Stichacidae ean eeeeenere 211 Notothenioidei .............. 211 Bovichtidae ............ 211 Pseudaphritidae ........ 212 Dactylopteroidei ............ 213 Dactylopteridae ........ 213 Malacanthidae .......... 214 Callionymoidei ............. 215 Draconettidae .......... 215 Callionymidae .......... 216 Gobiesocidae .......... 218 Blennioidei=eaes--) aoe oe 218 Tripterygiidae .......... 219 IBISTNVICAS caoccocccecoce 220 Dactyloscopidae ........ 221 Clinidae (Myxodinae) ... 221 Clinidae (Clininae) ..... 222 Labrisomidae .......... 223 Chaenopsidae .......... 224 Gobioideiieeeeee eee eee eee 224 Rhyacichthyidae ........ 225 Odontobutidae ......... 226 Xenisthmidae .......... 228 Bleotnidacar ee eeeeneee 228 Microdesmidae ......... 229 Gobiidaé» :... 25545-55008 230 Pleuronectiformes ........... 233 Ipsettodidaeiee ener 233 AES 1@ 62 Wl cocococsace 194, 225 Acknowledementsineeene eee ene eee 235 Appendix: Phylogenetic Analysis of 147 Families of Acanthomorph Fishes Based Primarily on Gill-arch Muscles and Skeleton), cae eeeseae a oe ee era 237 Literature::Citedigateiieeh eae asssta cet 255 Plates (separate volume) FIRST AUTHOR’S PREFACE This study was initiated about 1996 as the primary effort of the first author (VGS), who had long been interested in determining the interrelationships of the acanthomorph family Pholidichthyidae. Springer and Freihofer (1976) last discussed the problematic inter- relationships of the then monospecific family (Springer and Larson, 1996, described a second Phol- idichthys species). Subsequently, Stiassny and Jensen (1987), expanding on the work of Kaufman and Liem (1982), hypothesized the composition and interrela- tionships of an acanthomorph suborder Labroidei. Stiassny and Jensen based their hypothesis on a rel- atively broad selection of acanthomorph fishes and almost exclusively on a limited number of gill-arch characters. Referring only to Springer and Freihofer’s (1976) study, they noted similarities of the gill-arch skeleton of Pholidichthys to that of the labroids. Johnson (1993:9—10) discussed errors of oversight and commission in Stiassny and Jensen’s (1987) study and noted, critically, that, other than characters associated with pharyngognathy, there was none that corroborated monophyly of the labroids. VGS re-ex- amined Pholidichthys in light of Stiassny and Jen- sen’s and Johnson’s studies, and noted problems with both, although he agreed with Johnson’s general crit- icism. As a result, VGS, with the assistance of GDJ, undertook to survey a wide variety of acanthomorph fishes to determine the distribution of the states of the muscle characters used by Stiassny and Jensen. VGS expanded the study to include a broad spectrum of non-acanthomorph fishes because of problems in determining muscle homologies among the acantho- morphs. Darrow joined the project as full-time illustrator in the fall of 1997 and remained on the project until mid-2000, after which she continued on contract and then as volunteer until late 2003, when all the gill- arch muscle illustrations were completed. VGS contracted with Tom Orrell to run PAUP pro- gram analyses of a limited number of acanthomorph taxa beginning in 2002, but in 2003 involved him in the preparation of the major cladistic analysis form- ing our co-authored Appendix to the present study. After essentially completing the actinopterygian pre-acanthomorph portion of the study, including an analysis of the data, and much of the descriptive por- tion of the acanthomorphs, joint efforts with GDJ ceased in early 2002 at the sole instigation of VGS. VGS is responsible for selecting most of the taxa used in the study, preparing a large majority of the dissections, all the descriptions, supervision of prep- aration of all the illustrations, the partially new sys- tem of nomenclature used to designate the muscles, the interpretation and results of the non-acantho- morph portion of the study, most of the acantho- morph interrelationships, discussions that are not out- growths of the cladistic analyses, all of the choices of taxa, characters, and character codes for the cla- distic analyses, and preparation of all preliminary and final drafts of the entire manuscript. J, VGS, there- fore, accept full responsibility for all errors, factual or otherwise, and eccentricities that this study em- bodies. On the other hand, I gladly share with my co-authors responsibility for any favorable aspects of the study. I especially want to thank them for their input: GDJ for his early encouragement and impor- tant suggestions for taxa to include, ready and com- prehensive knowledge of the existing classifications of many groups of fishes and their defining charac- ters, and his often constructive challenges, his critical reading of an early complete draft of the pre-acan- thomorph section of the actinopterygian portion of the study, early drafts of a large number of the acan- thomorph descriptions and discussions, commenting on a near final draft of the pre-Appendix portion of the manuscript, and very importantly, for bringing Karie Darrow to my attention; Karie Darrow for her dedication to the project, even after monetary com- pensation ceased, and the unstinting use of her great illustrative talents, as well as for the numerous oc- casions on which she caught my descriptive mistakes; and to Tom Orrell for his insightful and knowledge- able handling of the PAUP program used in the phy- logenetic analysis, important suggestions for the preparation and interpretation of the output, and pa- tience under stress of my importunities. In the pre-Appendix portion of the text that fol- lows, the editorial “we” and “‘our” are used to ac- cord with authorship, but their use is not intended to imply agreement by the second author. NO Introduction The present study has two main purposes. First, to provide an annotated, descriptive atlas of the dorsal gill-arch muscles of the extant fishes (coelacanths, lungfishes, actinopterygians) included in the Super- class Teleostomi (= Grade Teleostomi, in part, of Nelson, 1994:65) or Osteichthyes (= branch 8 of Gill and Mooi, 2002:fig. 2.2) with special reference to the Actinopterygii. Second, to determine how the char- acter states exhibited by these muscles accord with existing morphologically based phylogenetic classi- fications of the fishes, and to a much lesser extent, molecular-based classifications. Because our study initially involved two ventral gill-arch structures (notably, the semicircular liga- ment and the transversus ventralis 4), we also sur- veyed them. We variously include treatment of other ventral gill-arch muscles and muscles that span both the dorsal and ventral gill-arch elements. On the other hand, as a result of the way the study developed, we did not survey the protractor pectoralis, which is of- ten closely associated with the dorsal gill-arches. We regret this omission, but the muscle was destroyed during preparation of many of the gill-arches before we realized its potential importance (for an extensive survey of this muscle see Greenwood and Lauder 1981). In our attempts to provide accurate illustrations of the muscles and their attachments, we also attempted to apply the same attention to the dorsal gill-arch skeletal elements. We frequently illustrate and dis- cuss skeletal structures that were previously unre- ported or erroneously described. On the other hand, we do not address all the skeletal characters that have been used in previous classifications, only those that are obvious in the illustrations, have direct bearing on the position of muscle attachments, or that by chance came to our attention (e.g., whether infra- pharyngobranchial 2 is present or absent, and if pre- sent bears teeth or is edentate). In a study as broad as ours, keeping up with rel- evant literature was a problem. Although, we at- tempted to do this, we recognize the possibility that references will have been missed. Another problem is that of references that became available after our discussion and analyses were in an advanced state and to adjust for them would have required continual revision and delay of the study. As far as we are aware, we have included and adjusted for the litera- ture published through the end of 2002, and much of the literature of 2003. Methods Dissections were made and studied primarily using a Zeiss Operation Technoscope. Drawings were made using a Wild M-3 stereoscopic microscope with at- BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON tached camera lucida. As required, details were checked using a high resolution Leitz stereoscopic dissecting microscope. In general, gill arches were removed from speci- mens using the following procedure. First, the hyoid arches including the branchiostegals were either sep- arated from the membranes connecting them to the opercular series and the interhyal linking them to the hyomandibula and retained attached to the gill-arch- es, or, usually, the gill arches were released from the hyoid arches by separating the hypohyals from the basibranchials. The basihyal was then freed by mak- ing a semicircular incision around the floor of the mouth. Next, the gill filaments were stripped off while trying to assure that the external and posterior levators were not removed with them (levator exter- nus I and levator posterior, require the most care to avoid damage or removal). Tissue enclosing the gill- arch musculature laterally was then removed. Partic- ular care was necessary at this point as the levator externus 4, levator posterior, and protractor pecto- ralis in some forms (e.g., clupeomorphs, lampridi- forms) may be closely applied or imbedded in the tissue and would be damaged or removed with the tissue. A cut was then made across the pharyngeal roof just anterior to the gill arches and pharyngo- branchial 1, if present, and the latter was released from its attachment to the skull. Muscles, ligaments, bones, nerves, and blood vessels attaching the gill arches to the skull were carefully scraped or cut free. A cut was then made through the pre-pectoral area to free the ventral attachment of the gill arches to the body (cut passes through ventral aorta and, depend- ing on taxon, may slice through the urohyal). From this cut, a posterodorsally arching cut was made on each side of the specimen in the tissue containing the pharyngocleithrales and attaching the gill arches to the surface of the cleithrum. The sphincter esophagi was then severed and the retractor dorsales severed from their attachments to the vertebrae. Any attached tissues (viscera, nerves, etc.) were cut and the gill arches released. The variety of taxa dissected fre- quently required considerable ad hoc modifications to this general procedure. After removal, the gill arches were partially cleaned by picking away the most obvious blood ves- sels, nerves, viscera, and extraneous fatty and con- nective tissues. The gill arches were then stained. Early in the study we used a KOH-alizarin red-s so- lution to stain bone and an acetic acid-ethyl alcohol solution of alcian blue to stain cartilage (Dingerkus and Uhler 1977). We found, however, that the KOH- alizarin solution was destructive of the muscles and we substituted a non-destructive ETOH-alizarin so- lution (Springer and Johnson 2000) for it. Regardless of which alizarin solution was used, the muscles usu- ally acquired a pink color or, in the case of alcian, a NUMBER 11 blue-green color, enabling one to distinguish them more easily from one another or surrounding tissues. After staining, the muscles were further cleaned of extraneous tissues and described. Descriptions and illustrations were often done in stages. In the first stage, as much information as pos- sible was recorded without disturbing the muscles other than to truncate the levators, if they obscured the other muscles. The muscles were then drawn. In the next and subsequent stages, muscles were bisect- ed or removed in order to expose hidden muscles or muscle attachments, and at each stage, the drawing was modified, usually unilaterally, to show additional information. Levator muscles and the retractor dor- salis are truncated, without mention in most of the illustrations. Some muscle attachments were inferred without dissection by referring to cleared and stained prepa- rations. In so far as they were not removed during cleaning, ligaments and miscellaneous connective tis- sues are included in the illustrations and mentioned in the descriptive accounts, but the absence of such structures from the illustrations or descriptions does not necessarily imply that they are actually absent. We strove for clarity in the illustrations, at the same time attempting to portray the muscles as close- ly as possible to their actual appearance—bilateral asymmetry and anomalies were included. Photo- graphs would have been more accurate, perhaps, but much less clear. The plates, with a few exceptions noted in the plate legends, are based on single spec- imens. The specimen illustrated may not be typical of the taxon in every detail illustrated. For this rea- son, if the reader notes a difference between a char- acter as coded in the PAUP data matrix (Appendix, Table 12) and the representation of that character in the illustration, the description of the taxon should be read for explanation. All such conflicts, however, may not be explained, e.g., proportional or presence- absence characters that are distorted resulting from parallax or are obscured in the view illustrated. Finally, during the course of determining obscured muscles and skeletal elements, the muscles of many of the specimens were of necessity greatly damaged or removed and the specimens will be of limited or no use to future investigators. This is particularly true of specimens that were prepared early in the study using a KOH-alizarin solution to stain the bones, as the solution badly macerates the muscles, and its res- idue continues to do so over time. Material Institutional abbreviations denoting specimens are those proposed by Leviton et al. (1985) and Leviton and Gibbs (1988). Relevant study material is reported at the begin- Ww ning of each descriptive account. If an illustration is indicated, the first indicated specimen is usually the one illustrated. Occasionally, additional material ex- amined for only one or a few characters is cited in the text. A list of material (see Acanthomorpha sec- tion), specifically for acanthomorphs, not otherwise mentioned in the text, was examined for osteological information, and served as the source for information in Table 8. In general, small specimens in the range of 50-150 mm SL were selected for dissection. De- pending on the taxon, these may have been adults or juveniles (some taxa rarely attain a length of even 50 mm as adults; specimens longer than 150 mm were used for taxa with proportionally small heads, e.g., eel-like forms). For many taxa, only one specimen was available for study, but even for taxa where more specimens were available, only one specimen may have been studied if the dissection was successful (i.e., little or no damage). We recognize this defi- ciency, but in a survey of the magnitude of the pre- sent one, we had to limit the depth of our examina- tions: any taxon could have been the basis for an independent study, and we dissected about 500 spec- imens comprising 208 families and about 400 spe- cies. Muscles are highly variable in their expression, sometimes varying bilaterally in the same individual, between individuals of the same species, or ontoge- netically. Where our material and examinations per- mitted, and we considered the variation important, we discuss variation. Muscles and Skeletal Elements The muscles mentioned in the text and on the il- lustrations are listed with their definitions and the ab- breviations we use to represent them in the section “Abbreviations and Definitions for Anatomical Structures.” The skeletal elements are similarly in- cluded, but only some are defined. A discussion of epibranchial 5 (Eb5) is given below because the in- terpretation of the presence or absence of this ele- ment is important in deciding the attachment of ad- ductor 5, and we use the opportunity to note new phylogenetic inferences based on the relationship of Eb5 to Eb4, ceratobranchial 4 (Cb4), and Cb5 (see section below, ““Epibranchials 5 and 4’’). Muscle names.—There is a wealth of names avail- able for many gill-arch muscles, but relatively few are standardized. Winterbottom (1974b) valiantly and most recently attempted a synonymy of teleost fish musculature. We utilize many of the names he rec- ognized as senior synonyms and that are commonly employed in the literature (e.g., /evator externus, or external levator). However, we also use a few names he treated as synonyms, and for many muscles we devise our own names. For these last muscles, our intent is to have the name indicate the attachments of the muscle, e.g., musculus laminalis dentalis 5- ceratobranchialis 4 (M. UP5-Cb4), a muscle origi- nating on upper pharyngeal tooth plate 5 and insert- ing on ceratobranchial 4. Such names may be un- wieldy, but they are descriptive, and one rarely needs to refer to them other than by their abbreviations. There is no law of priority with regard to muscle names, and we find that despite Winterbottom’s at- tempt to standardize muscle names, complete stan- dardization in the literature has not occurred, e.g., names of ceratodontid muscles have not been stan- dardized. There is the additional problem, which we have not solved, of assuring homology of usage. Our nomenclature is based variably on morphological to- pology and/or homology. If the reader is in doubt of our usage from the context of our application or dis- cussion, it is probably safest to assume topology. Muscle types.—Aside from functional anatomical types (e.g., contractors, extensors), we recognize two general positional types of dorsal gill-arch muscles, those that are bilaterally paired, i.e., present on each side, and those that are transverse, extending from one side to the other. An interpretive problem devel- ops when the middle portion of a transverse muscle is lost, as often occurs in acanthomorphs (e.g., trans- versus pharyngobranchialis 2 of Pomacentridae, Plate 160; transversus epibranchialis 2 of Embioto- cidae, Plate 162.1); we have no evidence for the transverse fusion of a bilaterally paired muscle). Origins and insertions.—Levators originate on the cranium, or in the case of the levator posterior, may originate from the body musculature. Bilaterally paired muscles attaching pharyngobranchial elements to epibranchials and/or ceratobranchials are consid- ered to originate on the pharyngobranchial elements. Retractores dorsales are considered traditionally to originate on the vertebral column and insert on pha- ryngeal elements. Muscles attaching an epibranchial to another epibranchial (recti dorsales = RecD) on the same side of the gill arches are considered (tra- ditionally) to originate on the more posterior epi- branchial: RecD4 originates on epibranchial 4 and inserts on epibranchial 3. Anatomical orientation.—Anterior and posterior are defined by the dorsal mid-longitudinal axis of the fish that runs between the pharyngobranchials. Me- dial, or proximal, and distal, or lateral, are, in effect, positions relative to the pharyngobranchials. Anterior and posterior refer to the positions relative to the head and tail of a fish, but the medial angle of artic- ulation of epibranchials, and of the ceratobranchials with the epibranchials, often imposes an almost lon- gitudinal orientation on them—thus, the proximal and distal ends of the epibranchials deceptively ap- pear to be the anterior and posterior ends. The de- BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON scriptions are based on the anatomical position, not the deceptive appearances. The major axis of a pharyngeal element may be oriented almost perpendicularly, in which case its an- terior end may be described as dorsal and its dorsal surface described as posterior. Transverse muscles.—Interpretation of the individ- ual muscles comprising the transversus dorsalis fre- quently involved subjectivity because of the nature or degree of continuity between the various compo- nents. Some components were unambiguously defin- able, but others were not (e.g., what degree of sep- aration of the components of transversus pharyngo- branchialis 3-epibranchialis 4 (TPb3-Eb4) should exist before recognizing the components as separate muscles, TPb3 and TEb4: only complete disconti- nuity of the components; continuity, but by only a few muscle fibers, etc.?). For some taxa, where more than one specimen was dissected, the components might be continuous in one specimen and discontin- uous in another. The interpretations, therefore, con- tain a degree of subjectivity, which the illustrations reflect. Acanthomorph accessory cartilages.—Rosen (1984:3, 25, fig. 25a) first noted the presence of an accessory cartilage (AC) at the joint of an epibran- chial and ceratobranchial in the gill arches of an acanthomorph (the fourth arch of Acanthurus), in which he indicated that it was of unknown signifi- cance. Rosen and Patterson (1990:9, fig. 42a) next called attention to an acanthomorph AC (in Lobotes, also in the fourth arch) and again indicated that it was of unknown significance, neglecting to mention Rosen’s earlier finding. We know of no other reports of accessory cartilages at the Eb-Cb joints of acan- thomorphs, which is the only group in which they occur. Among acanthomorphs, Eb-Cb joint accessory cartilages are predominantly restricted to perciforms (Table 8). Although ACs may occur in any gill arch, they are most commonly associated with the fourth arch. In acanthomorphs having AC4, Ad5 usually at- taches to it. AC4 superficially resembles Eb5, which, as first noted by Baldwin and Johnson (1996), is restricted to pre-acanthomorphs (here further restricted to pre- Ctenosquamata, see Table 6). Based on examination of larvae of a few taxa (Osmeridae, Osmerus, 17 mm SL; Characidae, Corynopoma, >4 mm SL; Chloro- phthalmidae, Chlorophthalmus, 12-13 mm SL), Eb5 is autogenous early in ontogeny (but may fuse on- togenetically with Eb4 and/or Cb4). In contrast, the acanthomorph AC4 is always associated with the dis- tal (usually posterodistal) end of Cb4, and appears to bud off Cb4 relatively late in ontogeny. In larval Morone (Moronidae) as large as 14 mm SL (all gill-arch elements with substantial ossification and vertebral column fully differentiated and ossi- NUMBER 11 fied), AC4 is absent and there is no evidence of a cartilaginous projection from Cb4 from which AC4 might form. In a 28 mm SL specimen of Morone, there is a projection extending from the cartilaginous tip of Cb4, which, we presume is the precursor of the autogenous AC4 present in the much larger spec- imens of Morone (100 mm SL) used for the muscle descriptions. Among four cleared and stained speci- mens of Ambassis sp., USNM 218805, AC4 on both sides of two specimens, 30.7—35.4 mm SL, appear to be in the process of budding off; one specimen, 39.0, has autogenous AC4s on both sides, and one speci- men, 45.6 mm SL, has an autogenous AC4 on one side, and a bud on the other. In a specimen of Am- bassis buruensis, USNM 305331, 55 mm SL, a well- developed process extends posteriorly from the distal end of Cb4 with no evidence that budding off might occur. Although we did not investigate them, AC1—3 and ACS (the last known only in Lates, Latidae) probably also develop as buds off the distal ends of Cb1—3 and Cb5, with which they are very closely associated (CT surrounding the cartilaginous distal ends of the Cbs envelops the associated AC). There is a problem in polarizing the character state of AC4, but based on its appearance in Velifer, we arbitrarily treat its presence as an acanthomorph syn- apomorphy. Epibranchials 5 and 4.—TYhe occurrence and in- terpretation of Eb5 has received considerable atten- tion (Nelson 1967d; Greenwood and Rosen 1971; Rosen 1974; Fink and Fink 1996; Johnson and Pat- terson 1996, 1997). It has not been established, how- ever, that the element, which is always cartilaginous, is an epibranchial (Nelson 1969a:520, reported Eb5 was ossified on one side of one specimen of a gym- notid). Eb5, which is usually constrained as articulating with the distal end of Eb4 (e.g., Nelson 1969a:520), is reported to be lacking in all ctenosquamates (Bald- win and Johnson 1996:372). Eb5 varies from being completely autogenous to partially fused with the distal end of Eb4, to puta- tively, completely fused with the distal end of Eb4, or infrequently, as we believe, with the distal end of Cb4 (only Albula). We find that in most groups with an autogenous Eb5, it is more closely associated with the distal end of Cb4 than it is with Eb4, and it ap- pears to be fused with Cb4 in Albula based on com- parison with EbS and its association with Cb4 in Pterothrissus, also Albulidae). Except for Osmerus (Osmeridae), it is problematic if EbS is always present in taxa in which the element is interpreted as partially or completely fused to Eb4. The possible alternatives are that it has been com- pletely lost secondarily after fusion, or lost indepen- dently before fusion. Johnson and Patterson (1996: 275) observed that Eb5 is autogenous in O. mordax larvae, but becomes fused to Eb4 in later stages. Fink and Fink (1996:231) reported an autogenous Eb5 in a Gonorynchus larva, 18.5 mm SL, which is not pre- sent at later stages, but Johnson and Patterson (1997: 597), who examined Fink and Fink’s specimens, were unable to find this cartilage. We also examined Fink and Fink’s (1996:231) specimens (partially errone- ously cited; see Johnson and Patterson 1997:597, for correct citation) and confirm Johnson and Patterson’s findings. In the plesiomorphic clupeomorph, Denticeps, the autogenous Eb5 attaches ventrally to the dorsodistal surface of Cb4 and anteroventrally to the ventrodistal end of the rod-like Eb4. In clupeoids with an autog- enous Eb5, however, the element attaches ventrally to the dorsodistal end of Cb4 and articulates closely dorsally and ventrally, but not in between, with the vertically expanded mostly bony distal end of Eb4 (e.g., Dussumieria, Plate 29:B). Eb5 thus forms the cartilaginous distal border of a foramen through which the posteriormost efferent artery passes (Nel- son 1969a:520); the proximal border of the foramen is mostly bony. Based on this landmark foramen, Nelson considered that Eb5 is present in clupeo- morphs in which there is no joint line dorsally be- tween it and the dorsodistal end of Eb4, but there is one at the ventrodistal end (foramen complete), or the ventral portion of the foramen is open (foramen incomplete). Nelson did not specifically indicate the fusion of Eb5 with Eb4 in those clupeomorph taxa in which the foramen is complete or the foramen is completely surrounded by cartilage with no joint line dorsally or ventrally. One can reasonably assume (as did Rosen 1974), however, that based on the config- uration of the distal end of Eb4, that Eb5 is present and that it has fused dorsally and ventrally with Eb4. There also appears to be inferential support for fu- sion of Eb5 and Eb4 in certain salmoniforms. In sal- monids, Eb5 is either autogenous (e.g., Oncorhyn- chus, Plate 36) or, apparently, fused dorsally with Eb4 (e.g., Prosopium, foramen open ventrally; Ro- sen, 1974:fig. 9e). Johnson and Patterson (1997:596— 597) appear to accept EbS as present in the ostario- physan Gonorynchus based on the ventrally open fo- ramen in the elongate posterior cartilaginous exten- sion of Eb4. Circumstantial evidence based on the attachments of adductor 5 (Ad5) supports the probable fusion of Eb5 with Eb4 in pre-acanthomorphs: Ad5, which al- ways attaches to Cb5 at one end, almost always at- taches to an autogenous Eb5 at the other end, or to the distal end of Eb4 when EbS is putatively fused with Eb4 (Table 6). Probably extrapolating from the clupeomorph con- ditions in which EbS is fused dorsally to Eb4 and the foramen is open ventrally, Nelson (19674d:75, fig. 1d; 6 also our Plate 7) considered that Eb5 might be pre- sent in the plesiomorphic osteoglossomorph Hiodon. In other osteoglossomorphs, the state of the foramen is variable and, with the exception of Pantodon, mimics the states of the clupeoids. Only Pantodon, among the osteoglossomorphs, has an autogenous Eb5 (Plate 13B), but its articulations (one end with Cb4 and the other with an accessory cartilage attach- ing to Cb5) are different from those that might give rise to the putatively fused conditions seen in the oth- er osteoglossomorphs. Excluding Albula, we code three character states for Eb5 (Table 6): absent (0), autogenous (1), and putatively partially or completely fused with Eb4 (2). State 2 is interpreted based on the appearance of the distal end of Eb4 (presence of a complete or incom- plete foramen). If the Eb5 character states are applied to the pre-acanthomorph cladogram, it must be con- cluded parsimoniously that state 2 evolved indepen- dently in the Osteoglossomorpha and Clupeoidei; hence, there is no evidence for the existence of an independent or fused Eb5 at the base of the Osteo- glossomorpha (the autogenous EbS, and its articula- tions, in Pantodon is an autapomorphy). Eb5 autog- enous first appears as a basal synapomorphy in the Elopocephala. Perhaps, ontogenetic studies of the osteoglosso- morphs will reveal, as in Osmerus, that an autoge- nous Eb5 is present and becomes fused with Eb4 during development. If such is the case, however, the fusion of Eb5 to Eb4 would still have been achieved independently by the Clupeocephala. There is another possible solution to the problem: the existing cladogram is incorrect. If Osteoglosso- morpha is replaced by Elopomorpha and placed as the sister group of the Clupeomorpha, an autogenous Eb5 becomes a synapomorphy of the newly com- posed Teleostei, and a fused Eb4-Eb5 becomes a syn- apomorphy of the Clupeomorpha + Osteoglosso- morpha. The reason for suggesting these changes is that Arratia (1999), based on a limited number of recent taxa, but citing studies by other authors who used additional characters, proposed that Elopiformes are the sister group of all other Teleostei and that the Osteoglossomorpha are close to the Clupeomorpha. The distribution of suprapharyngobranchial 1 (SPb1) also makes more sense parsimoniously if the Elopo- morpha exchange places with the Osteoglossomorpha in the cladogram (see discussion of LI] in pre-acan- thomorph Results section). This extended discussion bears on character states for the attachment of Ad5. To differentiate states in which Ad5 attaches to an autogenous Eb5 from those in which it attaches to a putatively (partially or com- pletely) fused EbS with Eb4 or Eb5 with Cb4, we indicate the latter two skeletal conditions as ‘“Eb4*”’ or “Cb4*.” Ve indicate these abbreviations, where BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON applicable, in the discussions and on the illustrations of the Elopocephala, but not the Osteoglossomorpha. Abbreviations and Definitions for Anatomical Structures Note: the following definitions occasionally con- tain important caveats on the interpretation of certain muscles, e.g., ER, the esophageal raphe. *—_following a pre-acanthomorph Eb4 or Cb4 indi- cates putative fusion of Eb5 with Eb4 or Cb4. See section “Epibranchials 5 and 4” for discussion. AB—autogenous bone; tiny bone attached to tip of Eb4 levator process and dorsal end of Eb5; only in Cyprinidae. AC—accessory cartilage; mostly restricted to acan- thomorphs. Among pre-acanthomorphs having ACs (Polypterus, Atractosteus, Pantodon, Diplo- mystes, Galaxias), the ACs may occur in a variety of locations and may be normal or adventitious in the taxon in which they occur. Among acantho- morphs, ACs are common, but occur predomi- nantly at the joint between the distal ends of an Eb-Cb pair, hence ACI, AC2, etc, with AC4 the most common in occurrence. Except for the rarely occurring ACS, acanthomorph ACs occurring at positions other than at an Eb-Cb joint are not num- bered. See section ““Acanthomorph accessory car- tilages”’ for discussion. Ad—adductor, adductores. Adl, Ad2, etc., adductor of 1st arch, 2nd arch, etc; plural, Ads, Adls, etc.; muscle attaching Eb to Cb of a single arch, except Ad5, which attaches Cb5 variously to one or more of the following: Cb4, AC4, Eb4, or Eb5. Ad4 dorsal attachment is on dorsoposterior or ventral surface of Eb4, often beginning anterior to OP on Eb4 and usually extending laterally further than OP. When present, one or more of first three Ads may be completely obscured by, or fused with, an overlying gill-filament muscle (GFM, q.v.) of same arch. In general, Ads are better developed than GFMs. We occasionally had difficulty distinguish- ing Ads from GFMs in acanthomorphs, and sub- jectivity in deciding may have resulted in some erroneous decisions. Among, pre-acanthomorphs, Ads1—3 occur only in Polyodontidae, Notacanthidae, Cyprinidae, and possibly Anguillidae. They are variably present among acanthomorphs, most commonly among percomorphs. It is unlikely that Ad1—3 of pre- acanthomorphs and acanthomorphs are homo- logues. It is unlikely, furthermore, that the acan- thomorph Ads1—3, which are absent in basal acan- thomorphs, are serial homologues of Ad4 and Ad5 in acanthomorphs, which occur early in pre-acan- thomorph phylogeny. The identification of Ad5 in various non-perci- NUMBER 11 form acanthomorphs in which ER is also present can be problematic. In some cases, it appears that Ad5 is absent, in others Ad5 ends dorsally at a raphe with the ventral end of OP, and in others that Ad5 is present as the lateral portion of what oth- erwise appears to be OP. In pre-acanthomorphs, apparent fusion of Ad5 medially with OP ventro- laterally is common. Additional study of ER, Ad5, and OP is warranted. See also OP below and re- marks following Ad5 in description of Arapaima (Arapaimidae). Ad4'—adductor 4 primus, attaches to the anterodistal surfaces of Eb4 and Cb4 (additional to Ad4); only in Notacanthidae. Bb—basibranchial, Bb3, Bb4, etc.; unpaired ventral gill-arch skeletal element. Cb—ceratobranchial; plural, Cbs; also Cbl, Cbls, etc.; ventral gill-arch skeletal element. CPb—circumpharyngobranchialis; a sub-epithelial muscle first described, but not named, by Anker (1978:261) for a cichlid muscle, and apparently not reported subsequently. CPb is present in various perciform fishes and is often very well developed. It appears to originate from the SO longitudinal muscle layer, extending anteriorly, and surround- ing or only bordering, and attaching variously to Pb2, Pb3, and UP4. In some perciforms (e.g., cich- lids, pomacentrids) TPb2a, may represent a dis- junct portion of CPb. CT—connective tissue. Eb—epibranchial; plural, Ebs; also, Ebl, Ebls, Eb2, etc.; dorsal gill-arch skeletal element. Eb4* indicates putative fusion of Eb5 with Eb4; only in pre-acanthomorphs. Epibranchial flange—a dorsolateral or anterolateral extension of the dorsodistal bony edge of an epi- branchial such that it partly or completely “shields” the cartilaginous distal end. Present only in some percomorphs and usually restricted to Eb4, although flanges may be present on other Ebs. The flanges are much reduced in size in many taxa (and difficult to show on our illustrations). A well-de- veloped Eb4 flange is present, e.g., in all members of the Labroidei, Opistognathidae, Pseudochromi- dae, Grammatidae; variably developed in Plesiop- idae (well developed in Assessor and Paraple- siops; weakly developed in Trachinops and Belo- nepterygion; absent in Acanthoplesiops and Noto- graptus); well developed in all atherinomorphs except very weakly developed in belonids and ab- sent in scomberesocids. Also very weakly devel- oped in the mugilid, Agonostomus. In labroids, the cartilaginous distal end of Eb4 has been lost and the cartilaginous end of Cb4 attaches by a tendon to the ventral surface of the flange. EO—epibranchial organ; plural, EOs; in pre-acatho- morphs usually formed, at least in part, by mod- erate to extraordinary expansion of the cartilagi- nous distal end of Eb4; in acanthomorphs (only stromateoids), EO involves an out pouching of the esophagus and distal end of Eb4 is not involved. Certain anabantoid families (e.g., Channidae, An- abantidae) have a suprabranchial organ which in- volves modification of the first epibranchial. ER—esophageal raphe; a fine line of connective tis- sue or myoseptum usually dividing OP transverse- ly at about mid-level or demarcating the ventral end of OP and separating it from Ad5 and/or SO. Very common in pre-acanthomorphs, but frequent- ly difficult to decide the constitution of the muscle fibers ventral to ER: SO, OP, or Ad5. Relatively uncommon in acanthomorphs, but when present usually appears to separate OP ventrally from Ad5, resulting in OP attaching ventrally to Cb4 rather than CbS5 (its usual ventral attachment in acantho- morphs), and Ad5 attaching to Cb4 well medial to distal end of bone, rather than to the distal end. GC— gongyloid cartilage, first named by Di Dario (2002) and first described by him in print, but first noted by Nelson (1966a:157) in his Ph.D. disser- tation; present only in engrauloids, pristigasteroids, and Chanos (Chanidae). See discussions in Addi- tional remarks sections under Cetengraulis and Chanos. GFM; GFM1, 2, 3—gill filament muscle. Here con- sidered to be essentially the same as Winterbot- tom’s (1974b:260 and fig. 26c) interbranchiales abductores: “extrinsic [gill] filament muscles .. . connecting the bases of the oral filaments to the gill arch (cerato- or epibranchial [we would mod- ify this to cerato- and/or epibranchial]) ... may [also] become intimately associated with the gill rakers .. .”’ They are often inconspicuous, fine, and stringy and are frequently destroyed when strip- ping gill filaments from the gill-arches. Those of the second and third arches in acanthomorphs may extend dorsoanteriorly and attach to the posterior edge of the preceding arch or they may continue dorsomedially on the dorsal surfaces of Eb2 and Eb3, that of the second arch sometimes meeting the lateral end of TEb2. We report them only in some acanthomorphs, and only when they are con- spicuous or fused with an adductor (Ad). In some taxa, they are questionably distinct from Ads (q.v.); decision on assignment as GFM or Ad is somewhat arbitrary. Additional study of the acan- thomorph Ads and GFMs is desirable. Winterbottom (1974b:259—260 and fig. 26) also recognized interbranchiales adductores, muscles attaching to the gill filaments of both hemibranchs of a single gill arch. We do not report on these muscles. Hb—hypobranchial; Hb1, etc.; plural, Hbs: ventral gill-arch skeletal element. IAB—interarcual bone; a putatively ossified IAC, only in Synbranchidae and Carapidae. IAC—interarcual cartilage; usually an autogenous rod-like cartilage joining Eb1l uncinate process to Pb2; found only among acanthomorphs. IAC2—interarcual cartilage 2; autogenous cartilage joining Eb2 and Pb2; only in Menidae. Interbranchiales abductores, adductores—see GFM. LCb—levator ceratobranchialis (.e., LCb2, LCb4, LCb5) muscle originating on skull and inserting on a Cb; only in pre-acanthomorph Cyprinidae and acanthomorph Adrianichthyidae. See also remarks following LCbS5 in muscle description of Cyprini- dae for comment on homology. Not to be confused with LES of Dipnoi, which inserts on Cb5. LCb5A—levator ceratobranchialis 5 accessorius; muscle originating in supratemporal fossa of skull near origin of LCb5, wrapping medially first, then anteriorly around LCb5 and inserting in CT pad attached to anterolateral surface of Cb5; only in Cyprinidae. This muscle appears to be the same as Holstvoogd’s (1965:216, and fig. 12b) M. troch- learis, which Winterbottom (1974:253) synony- mized with LP. It is also the same as Winterbot- tom’s LP internus (internus and externus portions not labeled), as indicated in his fig. 22 (Cyprinus). Muscle re-named by us at suggestion of R. Win- terbottom, who correctly noted (in litt.) that tro- chelar most often refers to cranial nerve IV, which supplies the superior oblique muscle of the eye, hence, inappropriately applied to a gill arch leva- tor. lat—lateral. LE—levator externus or external levator; muscle originating on cranium and inserting on an Eb: LE1, LE2, etc.; plural LEs, LEIs, etc. With rare exception (Carapidae), LEs originate on the skull, typically in a cluster or continuous line, usually together with LIs; however, LE4 may be displaced posteriorly in some taxa, especially those with EOs (see also LP). LE3 always inserts on or close to the Eb3 uncinate process in elopocephalans (ig- noring those taxa in which the uncinate process is absent). According to Vanderwolle et al. (1998), all carapid levators originate on the medial surface of the hyomandibula. LE1’, LE2’, ete—levator externus I primus, etc.; the second of two LEls, etc., arbitrarily designated, presumably the result of the division of an LE. Levator process (on Eb4)—a cartilaginously tipped process, lateral or posterior to the uncinate process (q.v.) on which LE4 and/or LP usually inserts. Presence or absence of the process may be onto- genetically associated: process isolated from distal cartilaginous end of Eb, or, in acanthomorphs, car- tilage lost during ontogeny by osseous exclusion. In the absence cf a cartilage tip, the process is BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON arbitrarily considered absent. Johnson and Patterson (1996:272—275) discuss the confusion and phylogeny of uncinate and le- vator processes on Eb4 and note that an Eb4 un- cinate process “‘characterizes acanthomorphs,”’ and that “Loss of a separate levator process ap- pears to be a synapomorphy of Acanthopterygu, although it may occasionally occur secondarily within Percomorpha ...” The presence of an Eb4 levator process in percomorphs is more common than Johnson and Patterson implied. It is frequent- ly present in generally considered plesiomorphic percomorphs (e.g., Acropomatidae, Percichthyi- dae, Moronidae, Scorpaenidae, Epigonidae, Apo- gonidae, Serranidae, Lutjanidae, Priacanthidae, La- teolabrachidae etc.) as well as in some specialized members (e.g., Opistognathidae, Trachinidae). Ll—levator internus or internal levator; muscle orig- inating on cranium and inserting, variously, on Pb2, Pb3, Pb4, UP4, or UPS (LI1, LUPS, etc.; plu- ral, LIs, LI1s, etc.); LI1 inserts on Pb2 (and/or Pb3 in some acanthomorphs; especially Blennioidei, which lack Pb2); LI2 normally inserts on Pb3; LI3, absent in ctenosquamates, variously inserts on Pb3, Pb4, UP4, UPS; LI4, only in Diplomystidae, in- serts on UP4. Some confusion may arise when comparing our LIs with those mentioned in the literature, as many authors number the LI based on the Pb to which it attaches; hence our LI1 is often referred to as LI2 in the literature, our LI2 as LI3, and our LI3 as LI4. Lila, LIlp—levator internus 1 anterioris, levator in- ternus I posterioris; LI1 represented by two lon- gitudinally separated muscles, both inserting on Pb2; only in Myctophidae. LI1'—levator internus I primus; the posterior divi- sion of LI1 inserting on Pb3 anteriorly, often close- ly juxtaposed to LI1 insertion on Pb2; frequently present in, but not limited to, Gobioidei. LI3 (part)—a separate, probably anomalous, basal portion of LI3; only in Heterotis (Osteoglosso- morpha). LI3’—levator internus 3 primus; second of two LI3s inserting on Pb4; only in Searsia, (Platytroctidae). lig—ligament. LP—levator posterior (or levator posterioris); mostly restricted to acanthomorphs, but present in some clupeoids and ostariophysans. Muscle originating variously on skull or body musculature and usually inserting on Eb4 together with LE4. In some taxa, fused with LE4 or coalesced in a musculous and connective tissue sheet with LE4 and/or PP, which attaches along edges of gill arches 4 and 5, and the individual muscles are not clearly separable. When LP is clearly distinguished, its origin is pos- terior or posteromedial to LE4 origin and usually well removed from it. In taxa with a single levator NUMBER 11 muscle on Eb4 (including, however, LE4’) that originates with other LEs and/or LlIs, there is no problem identifying the muscle as LE4 (and LE4’) because LP never clusters with the other LEs. In pre-acanthomorph taxa that have the origin of the single levator on Eb4 well posterior to those of the other levators, one might be tempted to designate the muscle LP, but examination of the muscles in related forms invariably indicates that only LE4 is present (i.e., it joins LE cluster). Additionally, LE4 generally inclines anteriorly, whereas LP frequent- ly inclines medially or anteromedially. Only three acanthomorph taxa appear to have lost LE4 and retained LP: Pholidichthys (Pholidichthyidae), Spi- nachia (Gasterosteidae), and, possibly, Echenei- dae. med—medial. mid—middle. M.—musculus; muscle. M. Eb1-Cb1—M. epibranchialis 1-ceratobranchialis 1; muscle joining dorsomedial end of Eb1 with dorsoanterodistal end of Cbl (only in Calliony- midae). M. Eb1-IAC—M. epibranchialis 1-cartilago inter- arcualis; muscle originating on Eb1 and attaching to IAC (only in Adrianichthyidae). M. Eb4-F—M. epibranchialis 4 faucis; muscle orig- inating on Eb4 and meshing with SO in throat (Latin, faucis) region (only in Blenniidae). M. Intrb—M. intrabranchialis (pl. intrabranchiales), M. Intrb1, Intrb 2 etc; muscle present in the CT (variously termed a diaphragm or septum) between the hemibranchs of a single branchial arch, over- lain by the gill filaments, which must be scraped away to expose it. Known only for Chondrichthy- es, in which they have been termed interbranchi- ales (Marion, 1905:905 & figs. 7, 8, 12; Daniel, 1934:105 & fig. 108) or constrictor branchiales (Edgeworth, 1935:129), and Dipnoi, in which they have also been termed interbranchiales (Fiirbrin- ger, 1904:488) or constrictor branchiales (Edge- worth, 1935:129; Fox, 1965:490). Here renamed to avoid confusion with the ‘“‘interbranchiales” (which include interbranchiales adductores and abductores), originally named by Winterbottom (1974b:259 & fig. 26) for small teleostean muscles attaching to the gill filaments (see also GFM). Edgeworth (1935:129) erroneously reported M. Intrbs in acipenserids (see Additional remarks sec- tion under Acipenser ruthenus). M. Pb2-Eb1—M. pharyngobranchialis 2-epibran- chialis 1; muscle originating on Pb2 and inserting on Eb1; only in mormyrids and some anguilli- forms. M. Pb2-Eb2—M. pharyngobranchialis 2-epibran- chialis 2; muscle originating on Pb2 and inserting on Eb2. According to Winterbottom’s (1974b:253) definition of obliquui dorsales, any muscle origi- nating on a Pb and inserting on an Eb could be termed an OD, but to do so might cause confusion. Our M. Pb2-Eb2 has been designated OD2 by Endo (2002:101) for gadiforms, in which he con- sidered its presence a specialization. In pre-acan- thomorphs, OD2 is present only in the osteoglos- somorphs (Hiodon and Heterotis), where it origi- nates on Pb3. In most acanthomorphs, the muscles we treat as obliquui dorsales originate entirely or primarily on Pb3 and insert on Eb3 and/or Eb4. To distinguish the acanthomorph OD2 (in Brotula, which also has M. Pb2-Eb2), we elected to denominate the “OD” originating on Pb2 as a new muscle: M. Pb2-Eb2; likewise, we designate other ““ODs” as M. Pbs- Ebs to avoid confusion. See also OD. M. Pb3-Cb5—M. pharyngobranchialis 3-ceratobran- chialis 5; muscle originating on Pb3 and inserting on Cb5 (only in Sparidae and Centracanthidae). M. Pb3-Eb1—M. pharyngobranchialis 3-epibran- chialis 1; muscle originating on Pb3 and inserting on Eb1 (only in Callionymidae). M. Pb3-Eb2—M. pharyngobranchialis 3-epibran- chialis 2; muscle originating on Pb3 and inserting on Eb2 (only in Pomatomidae). M. Pb3-Eb3-Eb4—M. pharyngobranchialis3-epi- branchialis 3-epibranchialis 4; muscle originating on Pb3 and inserting on Eb3 and Eb4 (only in gobud Gnatholepis). M. Pb3-Eb3—M. pharyngobranchialis 3-epibran- chialis 3; muscle originating on Pb3 and inserting on Eb3 (among pre-acanthomorphs, only in No- vumbra, Umbridae; among acanthomorphs, at least in some gobioids). M. Pb3-Eb3-Eb2—M. pharyngobranchialis 3-epi- branchialis 3-epibranchialis 2; short muscle orig- inating on Pb3 and inserting on Eb3 and Eb2 (only in Gymnarchidae). M. Pb3-Eb4-Eb2-Cb3—M. pharyngobranchialis 3- epibranchialis 4-epibranchialis 2-ceratobranchial- is 3; muscle originating on Pb3 and inserting on Eb4, Eb2, and Cb3 (only in Callionymus). M. Pb3p—M. pharyngobranchialis3 posterior; short cone-like muscle attaching anteriorly to Pb3 and inserting, apparently without attaching, into a con- cavity at the anterior end of the first vertebra. Function problematic; found only in Hemiramphi- dae and Exocoetidae. M. Pb3-Pb4-Eb2—M. pharyngobranchialis 3-phar- yngobranchialis 4-epibranchialis 2—muscle orig- inating on Pb3 and Pb4 and inserting on Eb2; only in the osteoglossomorph Gymnarchidae. M. Pb3-UP4—M. pharyngobranchialis 3-laminalis dentalis 4; muscle attaching to Pb3 and UP4; only in Embiotocidae. M. Pb4-Eb2—M. pharyngobranchialis 4-epibran- 10 chialis 2; muscle originating on Pb4 and inserting on Eb2; only in notopteroids. M. SO-Pb2—M. sphinctoris esophagi-pharyngo- branchialis 2; muscle originating on each side as an anterior extension of the transverse layer of the sphincter oesophagi, becoming discrete anteriorly as it extends along medial side of Pbs and inserts on Pb2 and, variously, Eb4; noted in leiognathids, but probably more widely distributed. M. SO-Pb3—M. sphinctoris esophagi-pharyngo- branchialis 3; muscle originating on each side as an anterior extension of the dorsal SO longitudinal muscle layer, becoming discrete anteriorly and in- serting on Pb3. More common than noted in the descriptions or on the plates as it was identified late in the study (e.g., Ditropichthys, Cetomimidae, Plate 74). M. SO-Pb4—M. sphinctoris esophagi-pharyngo- branchialis 4; dorsolateral extension of SO that at- taches to Pb4, only in Psenopsis, Centrolophidae. M. SPb2-Eb2—M. suprapharyngobranchialis 2-epi- branchialis 2; interrupted portion of LE2 originat- ing ventrolaterally on SPb2 and inserting on Eb2; only in Acipenseridae. M. SPb2L—M. suprapharyngobranchialis 2 lateral- is; interrupted portion of LE2 originating on skull and inserting on SPb2 dorsolaterally; only in Aci- penseridae. M. SPb2-LE1—M. suprapharyngobranchialis 2-le- vator externus 1; probably a component of LE2 originating from CT on medial surface of LE] and inserting anteriorly on mid-medial surface of SPb2; only in Acipenseridae. M. SPb2Ma—M. suprapharyngobranchialis 2 medi- alis anterioris; interrupted portion of LE2 origi- nating on skull and inserting on SPb2 anterome- dially; only in Acipenseridae. M. SPb2Mp—M. suprapharyngobranchialis 2 me- dialis posterioris; interrupted portion of LE2 orig- inating on skull and inserting on SPb2 postero- medially; only in Acipenseridae. M. TEb2-Pb2—M. transversus epibranchialis—2 pharyngobranchialis 2. A part of TD arising from ventral surface of TEb2 and inserting on Pb2; not homologous with TPb2, which arises dorsal or an- terior to TEb2; only in Cepolidae. M. UP4-Eb2—M. laminalis dentalis 4-epibranchialis 2; muscle originating on UP4 and inserting on Eb2; only in Albula. M. UP4-Eb4—WM. laminalis dentalis 4-epibranchialis 4; muscle originating on UP4 and inserting on Eb4; only in Congridae. M. UP4-Eb5-Cb4—M. laminalis dentalis 4-epibran- chialis5-ceratobranchialis 4 (not illustrated); mus- cle originating on UP4 and inserting on Eb5-Cb4 joint; only in Megalopidae. M. UPS5-Cb4—M. laminalis dentalis 5-ceratobran- BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON chialis 4; muscle originating on UPS and inserting on Cb4; very similar to M. UP5-Cb4-Eb5; only in Albulidae and Gonostomatidae. M. UP5-Cb4-Eb5—M. laminalis dentalis 5-cerato- branchialis 4-epibranchialis 5; muscle originating on UPS and inserting at inner angle of joint formed by Cb4 and Eb5; only in characoids, but very sim- ilar to M. UPS-Cb4. MPb1—mediopharyngobranchial; cartilage articulat- ing posteriorly with anterior or medial end of Eb1; may comprise single, medial element; present only in Chanidae, Gonorhynchidae, and some Clupeo- idei. ObV3—obliquus ventralis 3 (not illustrated); ventral gill-arch muscle attaching to Hb3 and Cb3; may also insert on SCL. OD—(plural, ODs) obliquus dorsalis (obliquui dor- sales). We follow convention in describing these muscles as originating on Pbs and inserting on Ebs. Gareth Nelson (commenting on a draft of the MS) noted that, functionally, it is more accurate to reverse the origin-insertion designations. ODs usually originate on Pb3, but origin may include Pbl (only Diplomystes, Diplomystidae) and Pb4 (only pre-acanthomorphs) and Pb2 (only acanthomorphs). Designation derives from Eb on which muscle inserts. See M. Pb2-Eb2 for discus- sion of why this muscle is not considered to be an OD. OD2—origin on Pb3; only in osteoglossomorphs. OD3—insertion on Eb3 includes uncinate process, if present. OD3'—origin on Pb3 with, ventral to, or lateral to OD3 or OD3-—4, becomes ventral to them pos- teriorly, and attaches on Eb3 dorsally ventral or medial to uncinate process; except for Oncor- hynchus, present only in acanthomorphs. OD3—4—a complete or almost complete fusion of OD3 and OD4, essentially restricted to acantho- morphs, in which origin is usually restricted to Pb3 and insertions usually on Eb3 and Eb4 bony surfaces supporting cartilaginous tips of unci- nate processes. OD4—in pre-acanthomorphs originates on Pb3, Pb3 and Pb4, or Pb4; in acanthomorphs origi- nates almost exclusively on Pb3; insertion on Eb4 includes bony surface supporting cartilage tip of uncinate process, if present. OD4yv—small ventral branch of OD4; only in Het- eropriacanthus (Priacanthidae). OD4'— originates on Pb4 in all pre-acanthomorphs except Megalops and Brycon, in which it orig- inates on Pb3. Among acanthomorphs, OD4’ is only present in percopsiforms, in which it orig- inates dorsal to OD4 or OD3-—4 on Pb3, extends posteriorly dorsal to them and inserts on the Eb4 levator process. NUMBER 11 OP—obliquus posterioris; highly variable muscle, sometimes in as many as four parts, attaching dor- sally to the posterior surface of Eb4, usually me- dial to dorsal attachment of Ad4, but often almost completely overlapping Ad4 posteriorly. In most pre-acanthomorphs, OP is usually interrupted transversely at mid-length by ER, and in most of the few acanthomorphs that have it, ER usually separates the ventral end of OP from Ad5 (see dis- cussion in Ad). OP medially is frequently insepa- rable from SO or ventrolaterally from Ad5. In some pre-acanthomorph (e.g., Amia) and most acanthomorph taxa, OP is continuous, uninterrupt- ed by ER, and attaches ventrally to Cb5 near at- tachment of Ad5, although OP ventromedial edge may join a restricted raphe with Ad5 posterodis- tally. Aerts (1982) reported that OP in cichlids com- prises three separate sections: medial, central (here termed middle), and lateral, and that LE4 fuses with the middle OP section to form a continuous muscle extending from the origin of LE4 to the attachment of the OP middle section to Cb5. This combined muscle has been called a “sling” by Lauder and Liem (1983:171) and Stiassny and Jen- sen (1987:284), and it also occurs in (but is not limited to) labrids (broad sense), embiotocids, and, perhaps (our opinion) pomacentrids, and may in- clude participation by LP. Most acanthomorphs only give evidence of having one OP section, most probably the middle section; we may have missed the divisions early in our work and further study is desirable. OP’—obliquus posterioris primus; slender muscle (possibly anomalous) originating on Eb4 levator process and joining ER with OP; only in Albuli- dae. PP—protractor pectoralis; muscle of pectoral girdle, occasionally illustrated and/or discussed, but only when included in a CT sheet also containing, and usually not clearly distinguishable from, LE4 and/ or LP. Greenwood and Lauder (1981) provide an extensive survey of this muscle in fishes. Pb—pharyngobranchial (commonly truncated spell- ing for “infrapharyngobranchial’’; used for con- venience); Pb1, Pb2, etc.; dorsal gill-arch skeletal element. PC—pharyngoclavicularis, -es (or pharyngocleith- ralis, -es), ventral gill-arch muscle, usually two on each side, but only one present on each side in eels and pre-halecostomes, but that of pre-halecostomes may be divided. See also PCE, PCI below. PCa, PCp—pharyngoclavicularis anterioris, -poster- ioris; divisions of the single PC of the pre-hale- costome Polypterus. PCE, PCl—pharyngoclavicularis externus, -internus (of Winterbottom, 1974b:267); ventral gill-arch 11 muscles attaching Cb5 to the cleithrum; both pres- ent on each side in most halecostome actinopter- ygians; one or both frequently illustrated in our plates but usually not discussed (absence in illus- trations not intended to imply actual absence). PCI is illustrated or described in all acanthomorph taxa in which the muscle attachment includes the distal end of Cb5 and/or joins a raphe with OP ventrally. PrO—protractor pharyngeus, anteriorly inclined le- vator-like muscle originating on ventral cranial surface, extending posteriorly, and inserting on dorsal non-musculous esophageal connective tis- sue; present only in Neoceratodus (Dipnoi). RCb5E—retractor ceratobranchialis 5 externus; muscle originating on ventral basioccipital process and inserting dorsolaterally on Cb5; only in Cy- prinidae. RCb5I—retractor ceratobranchialis 5 internus; mus- cle originating as CT along dorsolateral surface of vertical SO fold abutting basioccipital process and inserting by long tendon on CT pad attaching to Cb5; only in Cyprinidae. RCb5T—retractor ceratobranchialis 5 transversus; muscle originating medially from CT and SO, and inserting on dorsolateral margin of Cb5; only in Cyprinidae. Appears to be the same as retractor pharyngeus superioris of Winterbottom, 1974b: 258; fig. 22b). Only in Cyprinidae. RD—retractor dorsalis; muscle usually originating on anterior vertebrae and inserting variously on one or more of Pb3, Pb4, UP4, UP5, and Eb4. Origin and insertion usually not described by us. RDs may insert anteriorly or posteriorly on Pbs, and the difference is probably important. RDs may be unpaired, branch only at beginning of insertion, comprise a bilateral pair (one RD on each side), a bilateral pair and smaller unpaired median member (RD’), or vertical pair of muscles on each side. RD varies from being incorporated almost entirely within the SO (ventral to the circular or transverse muscle layer) to being entirely external to SO (see also SOD). RD’ indicates either the unpaired median muscle be- tween the individual RDs of a bilateral pair or the dorsal muscle when RD consists of vertical pair of muscles on each side (see RD). RecCb—rectus ceratobranchialis; short muscle con- necting distal ends of two successive Cbs; only in Callionymidae. RecCom—rectus communis, a ventral gill-arch mus- cle infrequently and only incidentally appearing in the illustrations; not discussed in descriptions. RecD—,rectus dorsalis (plural recti dorsales); muscle typically joining epibranchial on one side with epi- branchial immediately anterior; RecD2, RecD3, RecD4, number derives from posterior epibranchi- al, i.e., RecD2 originates on Eb2 and inserts on PCE PCI Fig. 1. Eb1; RecD1, however, originates on Eb! and prob- ably inserts on skull or peters out in skin that roofs mouth; RecD5 (only Callionymus) origin includes Cb5, insertion includes various skeletal elements anteriorly. We have applied RecD to a variety of problematic muscles found in clearly unrelated taxa (e.g., Menidae, Callionymidae, Cyprinidae, Anguillidae). See also discussion following RecDs in Callionymus. RecV4—-~rectus ventralis 4 (Fig. 1); ventral gill-arch muscle attaching Cb4 to Hb3 and/or SCL. SCL—semicircular ligament (Fig. 1); anteriorly open U-shaped ligament attaching anteriorly to the ven- tromedial ends of Cb3 and Cb4 and, often, mid- posteriorly to ventral surface of cartilaginous pos- terior end of Bb3, which, in acanthomorphs, is of- BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON Ventral view of gill-arches of Pempheris schomburgkii, USNM 318588 to show semicircular ligament (SCL). Both RecComs greatly truncated; left-side PCE almost completely removed; basihyal removed. Ventrally elongate cartilaginous tip of posterior end of Bb3 joins SCL mid-posteriorly. Photograph extensively retouched. ten elongate and ventrally recurved. The ventral aorta divides into left and right branches, which pass anteriorly on either side of the Bb3 attach- ment. ObV3 and RecV4 usually attach to SCL. The attachment of SCL to Bb3 is often obscured in ventral view, and the connection is easily broken during dissection or in trying to determine if it is attached to Bb3. Several character states for the Bb3 attachment to SCL were apparent in our dis- sections, but because of intermediates, we ana- lyzed SCL only for presence or absence. When we are certain of our observation, we report when SCL is attached to Bb3. In the descriptions, where SCL is merely described as present, we are uncer- tain as to whether it was free or attached to Bb3. Stiassny (1992:269—271) discussed and illustrated NUMBER 11 SCL. Her statement that SCL is an “‘acanthomorph innovation”’ is incorrect as SCL is present in sev- eral pre-acanthomorphs (Table 1; SCL of Novum- bra is especially similar to that of acanthomorphs). SL—standard length; all specimen lengths are SL, unless indicated otherwise. Sling—see OP. SO—sphincter esophagi; as generally recognized in the literature. SOD—sphincter esophagi division; a narrow to broad band of SO transverse, or circular, muscle sepa- rated dorsally from the remainder of the transverse layer and passing dorsal to RDs. We imprecisely restrict SOD to the condition in which RD extends noticeably anteriorly external to SOD before en- tering the transverse muscle layer (thus excluding the Aulopiformes and Ateleopodiformes as having SOD). SOD cannot be present if RD is absent, but SOD is not always present when RD is present. Although apparent in some of the acanthomorph illustrations, we may have failed to record the pres- ence of a fine, often inconspicuous mid-ventral branch of SOD that separates the left and right RDs. SPb—Suprapharyngobranchial; dorsal gill-arch skel- etal element of endochondral origin articulating with the cranium and, normally, with Ebl (SPb1) or Eb2 (SPb2); present only in Latimeria and some actinopterans (i.e., Chondrostei, Ginglymodi, Ami- idae, Elopiformes, Albuliformes, and Platytrocti- dae). TD—transversus dorsalis; transverse muscles attach- ing the gill-arch elements on one side with those on the other; not labeled as such on plates. Com- prises TDA and TDP and their components. TDA and TDP muscles may be continuous and on the same level, or, most commonly in acanthomorphs TDA is somewhat dorsal to the level of TDP. TDA may be broadly or narrowly continuous with TDP or completely separate. TDP muscles may be con- tinuous posteriorly with SO or SOD. Same names for component muscles of TDP or TDA reported in different taxa do not necessarily imply homol- ogy; likewise, different names in different taxa may obscure homology. TDA—transversus dorsalis anterior; transverse mus- cles attaching to the anterior skeletal elements: Pb2, Eb1, Eb2 (e.g., TEb2; Pb3 attachments in acanthomorphs generally not reported in muscle names for TDA muscles); not labeled as such on plates. In acanthomorphs, the medial portion of a TDA muscle may be lost or replaced by tendinous tissue or a thick CT pad, resulting in a pair of muscles, each of which may attach secondarily to an additional skeletal element, e.g., interrupted TEb2, might attach to Pb3 as well as to Eb2. TDP—transversus dorsalis posterior; transverse 13 muscles, attaching to the posterior skeletal ele- ments: Eb3, Eb4, Pb4, UP4 and/or to Pb3 in the area joining these elements (e.g., TPb3-Eb3); not labeled as such on plates. TD plexus—slender muscle branches joined to a me- dian CT sheet dorsal to TD; branches attach to Eb1l, Eb2, and Eb3; only in Acanthurus (Acan- thuridae). TEb1—transversus epibranchialis 1; essentially re- stricted to Labridae, Odacidae, Scaridae. TEb1-Eb2—transversus epibranchialis 1-epibran- chialis 2; only in Pantodon (Osteoglossoidei). TEb2—transversus epibranchialis 2. Muscle may ap- pear to comprise two more-or-less fused segments, giving impression of twisting (see especially Go- biidae) as they pass between levators (usually LI1 and LI2) to insert on Eb2. Borden (1999) differentiated a muscle he termed TD2 from another he termed OD2 in Naso (Acan- thuridae) on the basis that TD lacks a mid-line ra- phe and OD2 has one. Although usually present, the presence or absence of a raphe and its extent when present are highly variable, and we recognize a single muscle, TEb2, for Borden’s TD2 and OD2. See also M. TEb2-Pb2. TEb2a—transversus epibranchialis 2 anterioris; in pre-acanthomorphs only in Maurolicus (Stomiifor- mes), variously in acanthomorphs (e.g., Pseuda- pocryptes, Gobiidae; labroids). TEb2p—transversus epibranchialis 2 posterioris; in pre-acanthomorphs only in Maurolicus (Stomiifor- mes), variously in acanthomorphs (e.g., Pseuda- pocryptes, Gobiidae; Dicrolene, Ophidiidae). TEb2v—transversus epibranchialis 2 ventralis; only in Diplophos (Stomiiformes). TEb2-Eb1—transversus epibranchialis 2-epibran- chialis 1; only in Beryx (Berycidae), not to be con- fused with TEb1-Eb2. TEb3—transversus epibranchialis 3; present only in acanthomorphs. Except for the pre-acanthomorph engraulid genus Coilia, attachment of TD to Eb3 alone or together with another skeletal element, is restricted to Acanthomorphs, and is a synapomor- phy of the group. TEb3-Eb4—transversus epibranchialis 3—epibran- chialis 4. TEb4—transversus epibranchialis 4. May be discrete or continuous anteriorly and/or posteriorly with other muscles. TPb1-2-3-Eb1-2—transversus pharyngobranchialis 1,2,3-epibranchialis 1,2; only in Diplomystes (Di- plomystidae). TPb2—transversus pharyngobranchialis 2. In pre- acanthomorphs and primitive acanthomorphs, this is frequently a band-like muscle anterior to TEb2, if latter is present, and usually attaches to both Pb2s. In acanthomorphs, beginning with paracan- thops and zeoids, TPb2 lies partly or entirely dor- sal to TEb2 and fuses partly or entirely ventrally with TEb2, and, except for its loss or the loss of TEb2, rarely if ever has any other topographical position. TPb2 may also be pad-like or consist of a bi-lateral muscle pair usually joined by CT; may comprise a medially open semicircular ribbon of muscle on each side, which may join only TEb2 antero- and posteromedially, or may attach ante- riorly only to IAC, Pb2, or Pb3. Occasionally, it may be present only unilaterally and vestigially as a semicircular ribbon. Recognition of TPb2, when not attached to Pb2 (e.g., Rhamphocottus, Moro- ne), is based on the configuration of the muscle, which appears the same in other taxa in which it attaches to Pb2. For these reasons, we ignored the presence of an attachment to Pb2 as the defining factor in the identification of TPb2. TPb2 deserves more study than we were able to devote to it, and our designations of the various states of the muscle as TPb2 or some variety of it, possibly obfuscates its various homologies. In most acanthomorphs, our TPb2 is the same muscle Anker (1978) designated as the “‘m. cranio- pharyngobranchialis 2,” which designation has been followed by most recent authors, and espe- cially Stiassny and Jensen (1987). The muscle that the latter authors treat as the transversus pharyn- gobranchialis 2, we nominate as TPb2a, as it ap- pears to have a different history from our TPb2. TPb2’—transversus pharyngobranchialis 2 primus; a separate dorsoanterior portion of TPb2 (e.g., Ho- plostethus, Trachichthyidae) or separate muscle be- tween TPb2 and TPb3 (e.g., Tylosurus, Belonidae). TPb2-Pb2a—transversus pharyngobranchialis 2- pharyngobranchialis 2 anterioris. Putative fusion of TPb2 and TPb2a (only in Labroides, Labridae, but see TPb2-Pb2a-Pb3). TPb2a—transversus pharyngobranchialis anterioris (see also CPb and discussion of TPb2 above). Most prominently present in labroids, atherino- morphs, Pholidichthys. TPb2d—transversus pharyngobranchialis dorsalis. Name applied to muscle of variable structure dor- sal to and continuous with TPb2 or TPb2v in a few acanthomorphs (e.g., Aphredoderus, Aphre- doderidae; Agonostomus, Mugilidae). In Agonos- tomus, it is flat, pad-like and attaches to JAC and is, perhaps, also represented by anterior portion of TPb2 in moronids and mullids. TPb2p—transversus pharyngobranchialis posterior- is; a posterior separation of TPb2. TPb2v—transversus pharyngobranchialis ventralis; muscle joining ventral surfaces of Pb2s (e.g., Hem- iramphidae). The designation might be applied to TPb2a of Pholidichthys, but it appears that Pb2 has rotated so that its anterior surface lies ventral; the BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON muscle is more similar to TPb2a of atherino- morphs than it is to TPb2v of atherinomorphs. TPb2-Pb2a-Pb3—Transversus pharyngobranchialis 2-pharyngobranchialis 2 anterioris-pharyngo- branchialis 3. Compound muscle comprising a fu- sion of TPb2, TPb2a, and TPb3 (only in Sympho- dus, Labridae). TPb2-Pb3a—transversus pharyngobranchialis 2- pharyngobranchialis 3 anterioris; present only in some aulopiforms. TPb2-Pb3-Eb4—transversus pharyngobranchialis 2- pharyngobranchialis 3-epibranchialis 4; only in Gymnarchus (Osteoglossomorpha). TPb2-Pb3-Pb4-Eb4—transversus pharyngobran- chialis 2-pharyngobranchialis 3-epibranchialis 4; only in Lovettia (Galaxiidae). TPb3—transversus pharyngobranchialis 3; common- ly present in a large variety of fishes. TPb3-Eb3—transversus pharyngobranchialis 3-epi- branchialis 3; only and commonly present in acan- thomorphs (see also TEb3). TPb3-Eb3-Eb4—transversus pharyngobranchialis 3- epibranchialis 3-epibranchialis 4; only in acantho- morphs (see also TEb3). TPb3-Eb4—transversus pharyngobranchialis 3-epi- branchialis 4. TPb3-Eb4-UP4-UP5—transversus pharyngobran- chialis 3-epibranchialis 4-laminalis dentalis 4- laminalis dentalis 5; only in Anguilla (Anguilli- dae). TPb3a—transversus pharyngobranchialis 3 anterior- is; only in pre-acanthomorphs. TPb3a-Eb2—transversus pharyngobranchialis 3 an- terioris-epibranchialis 2; only in a few pre-acan- thomorphs. TPb3p—transversus pharyngobranchialis 3 poster- ioris, only in pre-acanthomorphs. TPb3p-Pb4—transversus pharyngobranchialis 3 pos- terioris-pharyngobranchialis 4; only in a few pre- acanthomorphs. TPb3-Pb4-Eb3—transversus pharyngobranchialis 3- pharyngobranchialis 4-epibranchialis 3; only in acanthomorphs. TPb3p-Pb4-Eb3-Eb4—transversus pharyngobran- chialis 3 posterioris-pharyngobranchialis 4-epi- branchialis 3-epibranchialis 4; only in veliferids and some girellids. TPb3’—transversus pharyngobranchialis 3 primus; posteriormost of three TPb3s (after TPb3a, TPb3p); only in Aulopus (Aulopidae). TPb3-Pb4—transversus pharyngobranchialis 3- pharyngobranchialis 4; in various pre-acantho- morphs, but only in Bovichtus (Bovichtidae) among the acanthomorphs. TPb3-Pb4-Eb4—transversus pharyngobranchialis 3- pharyngobranchialis 4-epibranchialis 4; present NUMBER 11 only in a few acanthomorphs, Psenopsis (Centro- lophidae), Tetracentrum (Ambassidae). TPb3p-Pb4-Eb4—transversus pharyngobranchialis 3 posterioris-pharyngobranchialis 4-epibranchialis 4; only in pre-acanthomorphs, e.g., Searsia (Pla- tytroctidae). TPb3-UP3-UP4—transversus pharyngobranchialis 3-laminalis dentalis 3-laminalis dentalis 4; only in the eel Synaphobranchus (Synaphobranchidae). TPb3-UP4—transversus pharyngobranchialis 3-lam- inalis dentalis 4; present Xenocephalus (Uranos- copidae) and some gobioids. TPb3p-UP4— present only in Albula (Albulidae). TPb3-UP4-Eb4—transversus pharyngobranchialis 3-laminalis dentalis 4-epibranchialis 4; present only in one species of Channa (Channidae). TPb4—transversus pharyngobranchialis 4; present in a variety of pre-acanthomorphs, but only in Pseudaphritis (Pseudaphritidae) among the acatho- morphs. TPb4a—transversus pharyngobranchialis 4 anterior- is; an almost completely separate anterior section of TPb4. TPb4-Eb3—transversus pharyngobranchialis 4-epi- branchialis 3; only in Coilia (Clupeoidea) and Nemipterus (Nemipteridae). TPb4-Eb4—transversus pharyngobranchialis 4-epi- branchialis 4; present only in pre-acanthomorphs. TUP4—transversus laminalis pharyngobranchialis 4; only in gobioids Pseudapocryptes and Ptereleo- tris. TUP4a—transversus laminalis dentalis 4 anterioris; only in Diplomystes (Diplomystidae). TUP4p—transversus laminalis dentalis 4 posterioris; only in Diplomystes (Diplomystidae). TUPS5—transversus laminalis dentalis 5; muscle con- necting UP5s; only in Stomiiformes. TL—total length. TV4—transversus ventralis 4 (Fig. 1); ventral gill- arch muscle connecting Cb4 on one side to Cb4 on the other; dorsally interrupted in some acantho- morphs (e.g., labroids) and attaching also to CbS. TV5—transversus ventralis 5; ventral gill-arch mus- cle connecting Cb5 on one side to Cb5 on the oth- er; occasionally illustrated, but not described. Uncinate process (pertaining to Ebs1—4; occasionally refers also to Pb2, which may have a distinct pro- cess that articulates directly or indirectly with Eb] or Pb3)—defined for our purposes as a cartilage- tipped process. Hence, if the cartilage tip is absent the process is considered absent; however, with re- gard to Eb1, Eb3, and Eb4, if the bony support is present but the cartilage tip is absent, the uncinate process is frequently described as “all bony,” but not differentiated from absent in the cladistic anal- ysis (Appendix). In acanthomorphs an uncinate process is fre- 15 quently present on Eb1, usually well medial to the lateral end of the bone, and in percomorphs it often articulates with the lateral end of IAC. A cartilage- tipped uncinate process that articulates with Pb3 may also be present on Eb2, but it appears to be restricted mainly to some pre-acanthomorphs. Among the Acanthomorpha, an Eb2 uncinate pro- cess has evolved independently in Hemiramphus (Hemiramphidae, in which the process does not articulate with another skeletal element), and, var- iably, in Pholidichthys (Pholidichthyidae), in which it articulates with Pb2. Rosen and Patterson (1990:figs. 49c, 50c), probably erroneously, illus- trated a posteriorly directed Eb2 uncinate process in the acanthomorph genera Peprilus (Stromatei- dae) and Trichiurus (Trichiuridae). We have ex- amined a cleared and stained T. lepturus and find that it has a bony process on Eb2, to which LE2 probably inserts, the usual condition in acantho- morphs. Many cichlids have a separate, expanded cartilaginous process on the anteromedial edge of Eb2 which extends anteriorly ventral to Ebl, and which we do not consider to be an uncinate pro- cess. An uncinate process on Eb3 appears first in os- teoglossomorphs (Hiodon) followed by one on Eb4 in elopiforms (Megalops). Tightly juxtaposed uncinate processes on Eb3 and Eb4 are first present in some aulopiforms, but appear to be an acantho- morph synapomorphy, with occasional reversions. The acanthomorph Eb4 levator process is usu- ally lateral or posterior to and well separated from the uncinate process. LE4 may insert directly on the levator process or close to it. When only one process, uncinate or levator, is present on Eb4, there is a problem in deciding which it is. If the process is tightly joined to the Eb3 uncinate pro- cess, there is usually little question that it is an uncinate process. If the process is well separated from the Eb3 uncinate process, it is usually not possible to decide, unless LE4 or LP inserts on it, in which case it almost certainly is the levator pro- cess (among acanthomorphs, LE4 inserts on the uncinate process only in a few lampridiforms). The uncinate process is absent in nemipterids, and ‘in Lethrinus and Gnathodentex [both Leth- rinidae] it ... has become closely associated with the levator process ... and has lost a close asso- ciation with the uncinate process of the third epi- branchial” (Carpenter and Johnson, 2002:120). Early ontogenetic stages, therefore, may also offer a solution to the problem of identification of a car- tilage-tipped process on Eb4 when only one such process is present. UP—upper pharyngeal tooth plate, e.g., UP4, UPS. Sarcopterygii Teleostomi or Osteichthyes Gnathostomata Fig. 2. Cladogram of major groups of extant Gnathostomatan fishes (derived essentially from Nelson, 1994:inside front cover) treated in the present study. Classification The classifications of fishes followed in our study are given in Figs. 2—4. For the most part, these are compilations of current hypotheses of relationships from which the fossil taxa have been removed. The cladogram in Fig. 2 presents the entire higher clas- sification of taxa that include organisms generally called fishes. Fig. 3 illustrates the classification of the Actinopterygii we examined, but provides detailed branching only for the pre-Acanthomorpha. The acanthomorph classification (Fig. 4) details the major primitive clades, but generalizes the terminal clade, Percomorpha; see also Footnote 1, page 18). The following discussion briefly examines the lit- erature or other information on which the classifica- tions are based. The classification of the recent Gnathostomata in Fig. 2 derives essentially from Nelson (1994), and is generally accepted in some form (usually with more detail) in most, if not all, current general classifica- tions of fishes (e.g., Long 1995:27, Maisey 1996:11, Gill and Mooi 2002:19). Pre-acanthomorph classification.—The five bas- almost clades of the Actinopterygii (Cladistia, Actin- opteri, Neopterygii, Halecostomi, and Teleostei) and their inter-relationships that we follow were first hy- pothesized by Patterson (1982:253; N.B., most recent authors overlook this reference, in which Patterson first included the Cladistia in the Actinopterygii, and placed it as the basalmost clade). Patterson (1994), who did not cite his 1982 study, concluded that the fossil, anatomical (extant taxa), and molecular evi- dence supporting monophyly of the Actinopteri con- flicts, and that the Actinopteri must be considered to comprise a polytomy of Ginglimodi, Halecomorphi, and Teleostei. The anatomical evidence (Patterson BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON 1994:68) based on extant taxa, however, supported his earlier (1982) hypothesized relationships: (Ging- limodi (Halecomorphi, Teleostei)), which we employ. Gardiner et al. (1996) reported a similar conflict, with morphological characters (including those of fossils) indicating that Ginglimodi and Halecomorphi are paraphyletic and molecular characters indicating that they are monophyletic. There are two main opposing classifications for the basal clades of the Teleostei. Patterson and Rosen (1977) hypothesized the arrangement that we use in our cladogram. Arratia (1999) provided a radically different set of inter-relationships. Arratia based her hypothesis on 14 extant and 34 fossil taxa, using 196 characters, of which only three (her character num- bers 165, 167, 190) can be considered as non-oste- ological. Character number 165, presence of adipose fin, appears only as a synapomorphy for two recent salmonids; 167, coiling direction of intestine, appears only as a synapomorphy for two recent osteoglos- somorphs; and 190, presence of nasal sacs, supports monophyly of 12 of the 14 extant taxa and 12 of the 14 fossil taxa she examined (excludes Amia and Lep- isosteus) and also supports monophyly of Esox and Umbra, the only two esociforms she examined. The major difference between Arratia’s classification and that in our Fig. 3, is that her Elops and Megalops replace the Osteoglossomorpha as the sister group of all other teleosts, and the osteoglossomorphs are made the sister group of the remaining teleosts. (For possible evidence of a closer relationship of Osteo- glossomorpha to Elopomorpha, see “‘bilaterally paired Pb muscles” in Results section of pre-acan- thomorphs. For possible evidence for support of a closer relationship of Osteoglossomorpha to Clupeo- morpha see discussion of EbS and Eb4 in “Muscles and Skeletal Elements.”’) The classification of the Osteoglossomorpha is ex- tracted from Taverne (1998:figs. 21—22; timing pre- cluded accommodation of Hilton’s (2003) rearrange- ment of the taxa in our Osteoglossiformes). The Elo- pomorpha is slightly modified from Forey et al. (1996:fig. 2) to make the Anguilliformes a polytomy. Forey et al. studied only the Anguillidae of the three anguilliform families we include. There is little sup- port available for the sister group of the Elopiformes, comprising the Albuliformes, Notacanthiformes, and Anguilliformes. Nelson (1973:347) proposed An- guillomorpha for this group (noted as a new super- order only in the combined index of Greenwood et al., 1973:520). The branchings of Clupeocephala into Otocephala and Euteleostei, of Otocephala into Ostariophysi and Clupeomorpha, of Euteleostei into Neognathi and Protacanthopterygii, of Neognathi into Neoteleostei and Esociformes, and all the branches of the Prota- canthopterygii (Argentiformes, Salmoniformes, Os- NUMBER 11 Ctenosquamata Eurypterygii Neoteleostei Stomiiformes Euteleostei Argentiniformes Protacanthoptery gil Osmeroidei Myctophiformes Aulopiformes Ateleopodiformes Osmeroidea Galaxioidea 17 Acanthomorpha Diaphus : ae anyctus ~} Myctophidae Neoscopelus Neoscopelidae Chlorophthalmus — Chlorophthalmidae Saurida Synodontidae Aulopus Aulopidae Ateleopus Ateleopodidae Gonostoma Gonostomatidae Maurolicus Sternoptychidae Diplophos Diplophidae Umbra Dallia + Umbridae Novumbra Esox Esocidae Searsia Platytroctidae Alepocephalus Alepocephalidae Argentina Argentinidae Mallotus qe @enisiaae Hypomesus Lovettia Lepidogalaxias Galaxiidae Galaxias Retropinna Retropinnidae Coregonus Thymallus $f sumone Oncorhynchus Gymnotus Gymnotidae Diplomystes Diplomystidae Xenocharax Distichodontidae Brycon Characidae Zacco Ae Opsariichthys _)pGyprmateae Gonorynchus Gonorynchidae Chanos Chanidae Clupea x SBIR Pee —}Clupeidae Chirocentrus Chirocentridae Coilia . Cetengraulis + Ener cH Illisha Pristigasteridae Denticeps Denticipitidae Synaphobranchus Synaphobranchidae Anguilla Anguillidae Conger Congridae Aldrovandia Halosauridae Notacanthus Notacanthidae Pterothrissus ; Albula Jpsitlchs Elops Elopidae Megalops Megalopidae Scleropages : aah cee ~ } Osteoglossidae Pantodon Pantodontidae Arapaima aa Hees _} Arapaimidae Mormyrus : Petrocephalus —}Monnyridae Gymnarchus Gymnarchidae Notopterus Notopteridae Hiodon Hiodontidae Amia Amiidae Atractosteus Lepisosteidae Polyodon Polyodontidae Acipenser Acipenseridae Polypterus Polypteridae Salmoniformes Clupeocephala Salmonoidei Gymnotiformes Siluriformes F if Gonhisi Characiformes Ostariophysi Cypriniformes Gonorynchiformes Otocephala Clupeoidea Clupeoidei |Engrauloidea Pristigasteroidea Clupeomorpha Ei eoe ple Denticipitoidei Anguilliformes Anguillomorpha Notacanthiformes Elopomorpha Albuliformes Elopiformes Teleostei Osteoglossoidei Osteoglossiformes Ost ha ere Epelossououp Notopteroidei Halecostomi N = Hiodontiformes copletyeu Halecomorphi Actinopteri Ging! : Acti ii P soees Ces ateegeg] Cladistia Fig. 3. Cladogram of the Actinopterygii (compiled from literature) with particular reference to pre-Acanthomorpha included in present study. 18 meroidei, Salmonoidei, Osmerioidea, Galaxioidea) are from Johnson and Patterson (1996:315—316). The branching of the Esociformes is taken from Nelson (1972:25) and Wilson and Veilleux (1982) and sup- ported by Johnson and Patterson (1996:314) and San- ford (2000:214, 218, 219). Lopez et al. (2000), based on molecular evidence, however, arrived at a differ- ent set of relationships: (Umbra (Dallia (Esox, No- vumbra))). The branching of the Clupeomorpha. Greenwood (1968) hypothesized Denticipitoidei as the sister group of the Clupeoidei (= all other Clupeomorpha), and its position has persisted to the present. The un- resolved trichotomy of the Clupeoidei (Clupeoidea, Engrauloidea, Pristigasteroidea) was last hypothe- sized by Grande (1985). The branching of the Ostar- iophysi was hypothesized by Fink and Fink (1996: 210-211). With the exception of the monofamilial Ateleo- podiformes, the branching of Neoteleostei to Acan- thomorpha was first proposed by Rosen (1973). Ro- sen included the ateleopodiforms as acanthomorphs, but Olney et al. (1993:155) demonstrated they are not acanthomorphs and placed them as an unresolved tri- chotomy with Eurypterygii and Stomiuformes. The stomiiform branches were hypothesized by Harold (1998:fig. 4). The aulopiform branches were hypoth- esized by Baldwin and Johnson (1996:359). Acanthomorpha_classification—The general ar- rangement of the higher acanthomorph groups is that of Johnson and Patterson (1993:fig. 24; modified slightly in our Fig. 4). We arbitrarily recognize many groups of families included in their Percomorpha as separate branches (with unresolved interrelation- ships) of a polytomous Percomorpha bush, which in- cludes a polyphyletic Perciformes and polytomous Smegmamorpha as one of its several branches. Cur- rently, there is no basis for separating Percomorpha from Perciformes. ! ' Johnson and Patterson (1993:591—592), in an attempt to con- serve the names Percomorpha and Perciformes, defined a mono- phyletic Percomorpha as a clade including, among other groups, the Perciformes. They illustrated their conclusions in their fig. 11b (polychotomous smegmamorphs as sister group of Perciformes), fig. 18 right (polychotomous Smegmamorpha as sister group of a polychotomy comprising Dactylopteriformes, Scorpaeniformes, Perciformes, Pleuronectiformes, Tetraodontiformes), and fig. 24 (monophyletic Percomorpha comprising two clades, Smegmamor- pha and “Perciformes, etc.” To summarize in their own words .. . we know of no sound characters justifying a pre-perciform position for Scorpaeniformes, and this emphasizes the tenuity of the dis- tinction between our Smegmamorpha ... and the Perciformes... . This raises the embarrassing or mortifying possibility that we should wind up a volume on percomorph phylogeny by concluding that the group does not exist. The alternative is to save the Per- comorpha by expanding it to include fishes that were originally excluded from it, the atherinomorphs. We believe that there is a monophyletic group comprising ‘‘perciforms and their immediate relatives’ and our smegmamorphs. That group can be character- BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON Gnathostomata (= Chondrichthyes + Telostom1) Telostomi (= Sarcopterygu + Actinopterygii) Remarks. Gill-arch levators are absent in Chon- drichthyes (Edgeworth, 1935:140, = his Elasmobran- chii; Greenwood and Lauder, 1981:222), but are pres- ent in basal Sarcopterygii and Actinopterygii; there- fore, the presence of levators is a synapomorphy of the Teleostomi. Sarcopterygii (= Coelacanthidae + Dipnot1) COELACANTHIDAE Latimeria chalumnae Smith. Remarks. Information on Latimeria is based here primarily on that reported by Millot and Anthony (1958), translated a bit freely from the original French. Muscle-bone abbreviations and remarks are ours. Description. LEI originates as fibers on apex of the processus post-oticus, near point of articulation of Pbl, and [continues] along posterior three-fourths of exterior border of Pbl. Remarks. Nelson (1968b:fig. 5A) interpreted Mil- lot and Anthony’s Pb1l as a probable fusion of Pb1 and SPbl, but he (1969b:487 and fig. 1) later indi- cated it as ““PH1” (= Pbl), noting that the Pbs of Latimeria “‘are difficult to compare with those of oth- er fishes.”” Evidence possibly supporting the element as either Pbl or Pb1+SPbl1 is that LEI attaches to Pbl in the basalmost actinopterygian, Polypterus, and to SPb1 in the relatively plesiomorphic Acipen- ser, Chondrostei. Unfortunately, the basalmost extant dipnoan, Neoceratodus, lacks both Pbs and SPbs. Nelson (1968b:fig. 5a, 1969b:fig. 1) interpreted Mil- lot and Anthony’s Pb2 as SPb2, their Eb2 as Pb2, and their Eb4+5 as Eb4. LE2, LE3, LE4 fuse at their cranial origin, which occupies the posterior surface of the processus post- oticus and extends to inferior border of the supratem- poral. LE2 insertion begins on posterior extremity of Pb2 [= Nelson’s SPb2; see remarks following de- scription of LE1]. It soon divides and ends capping the posterior epiphysis of Cb2. LE3 and LE4 are unit- ized by a series of apomorphies ... and we propose to name it Percomorpha. Johnson and Patterson, could not offer support for a monophyletic “‘Perciformes, etc.” and their Smegmamorpha should have been treated as just another “immediate relative” of the ‘Perciformes etc.”’ polychotomy. Nelson (1994:253) failed to note this discrepancy when he essentially reproduced Johnson and Patterson’s fig. 18 right, as well as taking unexplained, although probably correct, issue with the monophyly of the Smegmamorpha. NUMBER 11 S \F; Ses S G G ES S EF ee fom om LK WF LH OP RO QI OY” KEP FL ss SRS SS SC NS EC ee LOS LS SS FE EE ES SE PT Kr Se LT SF & ed as a bulky fleshy body obliquely inclined poster- oventrally. Near their movable insertions, the fibers divide and extend separately to posterior epiphyses of Cb3 and Cb4. LES is noticeable for its restricted mass and at- tachments. It originates as three weak tendons on ex- ternal surface of anocleithrum. The three tendons continue as brushlike muscles, fusing into a small, transversely flat muscle, which extends anteroven- trally toward the posterior end of Cb5, to which it is attached only by very poorly differentiated fibrous tissue. Remarks. The origin of this muscle on a cleithral element and its insertion on Cb5 suggest that it is probably homologous with PP (which Millot and An- thony do not mention) in Dipnoi and Actinopterygii. It does not appear to be homologous with LES of Neoceratodus (Dipnoi), which originates on the cra- nium and inserts in the dorsal esophageal CT just posterior to the gill arches. Millot et al. (1978) expanded on the anatomy of Latimeria. They (1978:fig. 7; plate 22) provided a drawing and photograph of a cross-section through the gill arches. They did not mention the presence of any muscles, but clearly, there were no M. Intrbs present in the illustrations. The absence of M. Intrbs in Latimeria is an indication that their absences in ACANTHOMORPHA PERCOMORPHA EUACANTHOPTERYGII UNNAMED CLADE ACANTHOPTERYGII HOLACANTHOPTERYGII EUACANTHOPTERYGII Cladogram of major groups of Acanthomorpha (modified from Johnson & Patterson (1993:fig. 24). Actinopterygii and Coelacanthidae represent inde- pendent losses (homoplasies). Ads absent. SO composition unrecorded. Remarks. Millot and Anthony (1958) do not men- tion which SO muscle layers are present, but Millot et al. (1978:plate 4), without comment about the mus- cle layers, provided a cross section of the esopha- gous, which indicates only a single (transverse or cir- cular) muscle layer. RDs absent. Dipnoi Remarks. Only the generally considered least spe- cialized family, Ceratodontidae, is discussed below. CERATODONTIDAE Neoceratodus forsteri (Krefft), AMS 1.40438001, 200 mm TL. Plates 1.1, 1.2 Description. LE1 mainly on dorsoposterolateralmost surface of Cb1 with very minor tendinous attachment extending from ventromedialmost edge of muscle to Eb1 dor- soposterolaterally. See remarks following LE4. 20 LE2 on posterolateral edge of Eb2-Cb2 joint. See remarks following LE4. LE3 originates with LE4 and M. Intrbr3 on cra- nium; inserts on posterolateral edge of Eb3-Cb3 joint. See remarks following LE4. LE4 originates with LE3 and M. Intrb3 on crani- um; inserts on posterolateral edge of Eb4-Cb4 joint. Remarks. LE1—4 are all relatively short and, ac- cording to Edgeworth (1935:129) and Fox (1965: 490), who studied larvae, originate separately on the ventral surface of the auditory capsule. Greenwood and Lauder (1981:fig. 2) illustrated LE3 and LE4 originating from a common stalk. Fox (1965:fig. 8), based on a 34.5 mm larva (his largest specimen), illustrated the first three levators as originating sep- arately from the others, and the fourth as originating with a muscle he recognized as the 6th levator, which we believe is PP. Fiirbringer (1904:489) described the insertions of the four anteriormost levators as follows (translated from German, substituting our muscle terminology): LE1 on Cbl1 and posterior end of Ebl. LE2 with approximately equal parts on Cb2 and posterior bor- der of Eb2. LE3 and LE4 insertions reduced to a minimum on [their respective] Cbs; insertions are al- most exclusively on posterior border of [their respec- tive] Ebs. Edgeworth (1935:129) described them as inserting “into the dorsal ends of the lower segments of the [branchial] bars,” and Fox (1965:490), essen- tially agreed, “insert on the tops of ceratobranchiale 1—4, respectively, laterally to the epibranchiale.” LES originates on the ventrolateral surface of the auditory capsule posterior to the origin of LE4, and, superficially, appears to be an anterior continuation of the PP origin. LES has a distinct, moderately long, tendinous stem that inserts into the non-musculous esophageal tissue, where that tissue first constricts posterior to the branchial arches, and at the dorsoan- teriormost edge of the musculous SO. At the LES insertion, SO is continuous only around the ventral surface of the esophageal tissue. LE5 can be consid- ered to be closely associated with Cb5 only by a few fine, tendinous strands extending from the tendinous stem of the right-side LES to the PP in the vicinity of the posterior end of Cb5. Remarks. Fiirbringer (1904) and Edgeworth (1935) recognized only four LEs in Neoceratodus, whereas Fox (1965) recognized six. We are uncertain as to the exact homologies of Fox’s LES and LE6. Fox (1965: 490) stated that, ““Levators 5 and 6, which behind merge ventrally with the coracobranchialis and pos- terior transversus musculature, may include a portion homologous with the dilator laryngeus and the be- ginning of the cucullaris muscle [= our PP]. . .” Fox (1965:fig 7) indicated the presence of only LES in his larval 27 mm specimen, in which it is quite long. Dorsally it closely approaches or attaches to the pos- BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON terior surface of the auditory capsule; anteriorly at mid-length, it closely approaches or attaches to the posterior (or distal) end of Cb5; posteroventrally it is free, but clearly directed toward the clavicle or cleith- rum. In a 20 mm specimen, Edgeworth (1935:figs. 43 and 43a) labels this muscle constrictor branchialis 5 and shows it extending toward the clavicle and im- pinging on, or fusing with, coracobranchialis 5 pos- teriorly before attaching to the clavicle. In Fox’s next, and last, stage larva, 34.5 mm, LES is no longer in contact with the cranium, having lost much of its dorsal extent, but maintaining its mid-length attach- ment to the distal end of Cb5. It lies along (fuses with?) the anterior or anterolateral surface of a very large LE6, which makes its first appearance and more-or-less duplicates the relative length and posi- tion of LES in Fox’s 27 mm larva. As Fox suggested, LE6 probably represents an early stage in the for- mation of the PP, but unless LE5 loses its identity and fuses with PP in specimens larger than our 200 mm specimen, we do not believe it is part of PP. PP originates on ventrolateral surface of the au- ditory capsule beginning musculously as an apparent, narrow, posteriorly continuous extension of the LES origin, but the origin becomes tendinous posteriorly; PP rapidly expands, becoming fan-like as it extends ventrally, attaching anteriorly along posterior (or lat- eral) border of Cb5, and reaching ventrally to the subarcualis rectus on Cb5 (Edgeworth, 1935:fig. 43b; Wiley, 1979:fig. 3c) and coracobranchialis 4 on Cb5 (Edgeworth, 1935:fig. 43b); posteroventrally, it in- serts broadly on clavicle with and posterior to the coracobranchiales. (See remarks following LES.) M. Intrb1—4 origins are about same size as those of LEs. Intrb1 and 2 originate on cranium near ori- gins of LE] and 2; Intrb3 originates together with combined origins of LE3 and 4; Intrb4 originates posterior to the previous. Ventral to their origins, the fibers of each M. Intrb gradually separate in a single plane on the branchial septum, ultimately branching into about 8 filaments, each of which is separated from an adjacent filament by a narrow space; each muscle arches posteriorly, then ventrally; the fila- ments re-unite anteroventrally and attenuate, finally continuing anteriorly as a long, fine tendon, which inserts in CT near the anteroventral end of each mus- cle’s respective Cb. Remarks. M. Intrbs are known only in lungfish and elasmobranchs; thus constituting a possible synapo- morphy contra-indicating a sister-group relationship between lungfish and sarcopterygians. PrO levator-like, oriented almost horizontally, originating on ventral surface of cranium a little me- dial to medial end of Eb4, extending posteromedially, and inserting on non-musculous esophageal tissue posterior to branchial arches. Remarks. We have not found this muscle men- NUMBER 11 tioned in previous studies; it is tempting to interpret it as another form of RD, which otherwise makes its first appearance in basal Neopterygii. SO consists only of transverse muscle layer. Ads absent. RDs absent (but see remarks following PrO). Additional remarks. Huxley (1876:27) illustrated the gill arches of Neoceratodus. On page 37 he de- scribed them, recognizing anterior and posterior un- paired “‘mesobranchials”’ and another cartilage: “close to the ventral end of the fifth arch [= Cb5], was a small nodule of cartilage, which is probably a rudimentary sixth arch ...” In our specimen, we found the anterior mesobranchial, which lies medially between the ventral ends of Cb2 and Cb3 on each side. Huxley illustrated its position as extending an- teriorly from the ventral end of Cb2 to a point an- terior to the ventral end of Cb1. His mesobranchial 1 is undoubtedly a basibranchial. On only the right side of our specimen there is an unpaired cartilage between the ventral ends of Cb3 and Cb4. On the left side of our specimen, the ventral end of Cb5 appears to have become deeply notched sub-terminally, but continuous posteriorly with the remainder of the Cb. A small autogenous plug of cartilage fills the gap between the two continuous parts. It appears that ventral fragmentation of the Cbs may be common in Neoceratodus. Actinopterygii Pre-Acanthomorpha (Cladistia—M yctophiformes) Cladistia POLY PTERIDAE Polypterus ornatipinnis Boulenger, USNM 164514, 170 mm TL. Plate 2 Description. LE1 on dorsoposterior edge of Pb1 and lateral por- tion of mostly cartilaginous Eb1 (small ossification center present in each Eb1). Remarks. The only actinopterygians in which a le- vator inserts in whole or in part on Pbl are Polyp- terus and some osteoglossiforms (Petrocephalus, Os- teoglossum, Scleropages). LE2 on dorsolateral surface of cartilaginous Eb2. LE3 on dorsal surface of cartilaginous Eb3. LE4 ventromedially on CT just lateral to SO, ven- trolaterally on cartilaginous distal end of Cb4 and greatly reduced cartilage, which is here designated Eb4, and ventroanteriorly on distal end of Eb3 (see additional remarks at end of description). LP absent. LI1-3 absent. 21 TD absent; SO joined dorsally to gill arches by CT, possibly resulting from loss or extreme reduction of Eb4. (TD is also absent in Dipnoi and probably coelacanths.) OD3 and OD4 absent (OD first appears in Neop- terygii, Ginglymodi). OP absent, possibly concomitant with loss of Cb5. Ad1-—3 and Ad5 absent. Remarks. If it was present in early polypterid phy- logeny, Ad5, which normally attaches Cb5 to the fourth arch, was probably lost with the loss of Cb5 (see also additional remarks). Ad4 is questionably represented by muscle joining distal end of Cb4 and reduced Eb4 with distal end of Eb3; muscle is adjacent to and questionably contin- uous with LE4. Remarks. Ad4 normally attaches Eb4 to Cb4 in other actinopterygians, but with the considerable re- duction of Eb4, Ad4 may have shifted its dorsal at- tachment to Eb3. The presence of Ads, which are absent in Chondrichthyes and Sarcopterygii (Dipnoi + Coelacanthidae), is a synapomorphy of the Actin- opterygil. RD absent. SO longitudinal muscle layer absent. Remarks. SO comprises a single (circular or trans- verse) muscle layer. SO extends anterodorsally only to a dorsally projected horizontal joining the distal ends of Cb4s. As noted by Edgeworth (1935:167), SO in Polypterus ‘“‘does not attach to any branchial bars,’ and as such, is unique among the Actinopter- ygii. Similar conditions pertaining to SO exist in the primitive dipnoan, Neoceratodus, in which, however, SO begins relatively even more posteriorly. Additional remarks. SCL absent, TV4 absent. Prominent ligament, originating on parasphenoid, in- serts on distal end of Cbl. Jollie (1984:fig. 17b) illustrated the gill arches of Polypterus, purportedly based on Allis (1922:pl. 8, fig. 17b). Jollie, however, greatly increased the length of the ventral cartilaginous portion of Allis’s Eb1, interpreted the cartilage as Eb1, and labeled the two bony arms extending dorsally from the cartilage as ““SPb”’ and “‘IPb.’”’ We believe Jollie’s changes were unwarranted. Polypterus is unusual in having only four gill arch- es. Britz and Johnson (2003) discuss the two hypoth- eses concerning the homology of the missing arch (whether it is the fourth or fifth) and strongly support the argument that the missing arch comprises the fifth ceratobranchials. Allis (1922:232 et seq.) reported that: the first gill arch comprises Pb1 (which articulates with Eb1) Eb1, Cb1, and Hb1; the second arch comprises Pb2 with a fused, reduced Eb2, Cb2, and Hb2; the third arch comprises Pb3 with a fused, reduced Eb3, Cb3, and Hb3; the fourth arch comprises only Cb4. We agree with Allis on the composition of arch 1 and arch 4, considering that the vestigial Eb4 is variably absent. For arches 2 and 3, we recognize Allis’s Pbs as Ebs; for arch 4, the vestigial Eb4 may be absent or easily overlooked, Allis gave no evidence for his assump- tion that fusion had occurred between an Eb and a Pb. Allis (1922:233) mentioned that an autogenous bit of cartilage is interposed between the [first] epibran- chial and ceratobranchial of some specimens, but not others. This cartilage was also reported by van Wijhe (1882:257 and pl. 5, fig. 7), who identified it as an epibranchial, but which we term an accessory carti- lage (AC). The AC (Plate 2A) is present in our spec- imen, and in an 85 mm SL cleared and stained spec- imen of Polypterus senegalus Cuvier (USNM 229760), but not in a 132 mm SL specimen from the same lot. The smaller of the two specimens also has an AC on the third arch of one side. We note that accessory cartilages are also present on the first and second arches of the lepisosteid Atractosteus (Plate SA). Neither Allis nor van Wijhe, noted another, small, autogenous cartilage attached to the distal end of Cb4 (Plate 2B), which is present in P. ornatipinnis and the smaller, but not larger, specimen of P. senegalus. Although Britz and Johnson (2003:499) noted the presence of this cartilage they refrained from identi- fying it. They cite literature in which the element has been identified variously as Eb4 or an epipharyngo- branchial. Wiley (1979:161) discussed the homologies of the pharyngoclaviculares (PCs) in actinopterygians. He noted that polypterids, chondrosteans, and gars, which compose the pre-halecostomes, have a single PC and that halecostomes have PCI and PCE. Allis (1922:259) described PC in Polypterus as having a single origin on the cleithrum, but dividing into two parts with separate insertions, which he did not name, on Cb4, and which Wiley did not equate with the halecostome PCI and PCE. One can infer from Wi- ley’s discussion, however, that the PCs (or divisions thereof) of the pre-halecostomes could be homolo- gous with the halecostome PCI. We assign PCp to the more posterior insertion and PCa to the well-sep- arated more anterior insertion of PC in Polypterus. Wiley’s hypothesis that [no portion of] the pre-hal- ecostome PC is homologous with PCE of halecos- tomes, is based on two considerations. First, Wiley (1979:161) credits Edgeworth (1911) with finding that PCs are derived from the circular [our trans- verse] muscle layer of the oesophagus (among actin- opterygians, Edgeworth’s work is based on the pre- halecostomes Polypterus, Acipenser, Lepisosteus, and the halecostome Amia; Wiley accepted the presence of PCI and PCE in Amia). Second, Wiley dissected the halecostome Hiodon, which has PCI and PCE, BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON and found that the muscle fibers of PCE are contin- uous with SO longitudinal fibers. He postulated therefrom, that the PCE of halecostomes could not be homologous with the PC of pre-halecostomes, pu- tatively derived from the SO circular layer. We cannot find any indication that Edgeworth (1911), who studied the development of PCs, report- ed that they are derived from the SO circular muscle layer. In fact, Edgeworth (1911:232) concluded that “the coraco-branchiales [= our pharyngoclavicula- res] of Elasmobranchs, Teleostomi, and Dipnoi ... are derived from the ventral end of one or more bran- chial myotomes, i.e., are of cranial origin.” Edge- worth (1935:155) reaffirmed his earlier finding, stat- ing that in Dipnoi and Teleostomi (he modified his earlier finding for elasmobranchs), “they are devel- oped from the outer ends of the Transversi ventrales just after these have grown inwards from the ventral ends of the branchial muscle-plates.”” Even so, Wi- ley’s hypothesis that PCE of Hiodon (and other Te- leostei) is not homologous with PCE of Polypterus is probably viable. Pre-halecostomes lack the SO lon- gitudinal layer (the layer is a synapomorphy of hal- ecostomes, see Results section following pre-acan- thomorph section of this study), so that any muscle derived from the halecostome SO longitudinal mus- cle layer cannot have a homologue in a pre-halecos- tome. Chondrostei ACIPENSERIDAE Acipenser ruthenus Linnaeus, USNM 62372, ca. 260 mm TL. Plate 3 Description. Remarks. Muscles are complex, difficult to char- acterize, generally not so discrete as we describe. It is possible that considerable intraspecific variation occurs; however, there was essential agreement be- tween the right and left sides of our specimen. Mar- inelli and Strenger (1973:fig. 24) present a lateral view of the gill-arch musculature of A. ruthenus that agrees with our illustration (Plate 3B), insofar as the muscles they illustrate (they do not present a dorsal view of this musculature). LE1 ventrally on Eb1, dorsoanteriorly on SPb1 and cranium; broad raphe on medial surface joined by anterior end of M. SPb2-LE1. LE2 complex, comprising five parts: M. SPb2-Eb2 on SPb2 ventrolaterally and Eb2 dorsally. M. SPb2L origin on cranium, insertion on SPb2 dorsoanteriorly. M. SPb2Ma origin on cranium, insertion anteriorly on mid-medial surface of SPb2. NUMBER 11 M. SPb2Mp origin on cranium, insertion posteri- orly on mid-medial surface of SPb2 and raphe with anterormedial fibers of LE3. M. SPb2-LE1 origin from CT on medial surface of LE1, insertion anteriorly on mid-medial surface of SPb2 ventral to SPb2Ma. LE3 origin dorsally on cranium and SPb2 posterior surface; insertion on dorsal surface of Eb3; lateral- most fibers of left LE3 continuous with fibers of RecD4 (aberrant?). LE4 on Eb4 lateral to and continuous with RecD4 posterolaterally. LP absent. LI1—3 absent. TD represented by TEb4, which is posteromedially continuous with SO. OD3 and OD4 absent. OP absent. RecD4 joins dorsolateral surfaces of Eb3 and Eb4, continuous with LE4 ventrally. Ad1-—3 absent (see remarks following Ad4). Ad4 on Eb4 dorsoposteriorly and Cb4 posteriorly, continuous ventromedially with SO. Remarks. Edgeworth (1935:131) reported that Ad1-—4 are present in A. sturio, but that only Ad4 is present in A. ruthenus, A. fulvescens, and Scaphir- hynchus. Ad5 dorsally on lateral process at distal end of Cb4, ventrally on Cb5 distal end. RD absent. SO comprises only transverse muscle layer. Remarks. Wiley (1976:30—32) reported that there are two muscle layers lining the buccal cavity of Po- lyodon and Acipenser, an inner longitudinal layer overlain by a circular layer, and that these layers are undifferentiated from the same muscle layers lining the esophagus. We have examined both genera and find there is no muscle dorsally in the area between the gill arches. The SO in Acipenser and Polyodon comprises a single muscle layer: circular (or trans- verse). Additional remarks. SCL absent. TV4 absent. Edgeworth (1935:129) stated that constrictor bran- chiales (also known as interbranchiales in elasmo- branchs (Daniel, 1934:105—106, 149) and lungfishes (Fiirbringer 1904:488) are present in acipenserids. He (1935:130) described these as “narrow muscle bands external to the [epibranchials and ceratobranchials]. In adult stages they lie in shallow grooves in the epi- and kerato-branchialia. In Acipenser their ventral ends run into the outer ends of the Obliqui ventralies i, 11 and iii and Transversus ventralis iv. In Scaphir- hynchus they are overlapped by the outer end of these muscles.” Edgeworth (1935:figs. 224b and 231) il- lustrated these putative muscles in a sagittal section of a 32 mm specimen of A. ruthenus and in a ventral view of the branchial muscles and arches of Sca- i) es) phirhynchus platyrhynchus {= S. platorynchus], size not given. We did not find the muscles in A. ruthenus (a second specimen, USNM 64607, 295 mm TL was examined for these muscles), and they were neither reported nor illustrated by Marinelli and Strenger (1973) in their comprehensive anatomical study of A. ruthenus. We found only nerves and blood vessels coursing along the grooves of the epi- and cerato- branchials (see also Marinellei and Strenger, 1973: fig. 240). We did not examine Scaphirhynchus. POLYODONTIDAE Polyodon spathula (Walbaum), USNM 101093, 217 mm TL. Plate 4 Description. LEI on dorsal edge of Eb!l near medial end and by tough CT on posterior edge of SPb1. LE2 on dorsal edge of Eb2 near medial end and by tough CT on posterior edge of SPb2. LE3 on Eb3 uncinate process, continuous with LE4 above insertion. LE4 on Eb4 dorsal edge somewhat distal to medial end, continuous with LE3 above insertion. LP absent. LI1-—3 absent. TD consists of TEb4, which is continuous and un- differentiated from SO. OD3 and 4 absent. OP absent. RecD4, small, anteriorly on dorsodistal edge of Eb3, posteriorly by tendon to Eb4 at insertion of LE4. Ad1—4, each dorsally, broadly on posterior surface of respective Eb, ventrally, narrowly on anterior sur- face of respective Cb just medial to inner angle formed by Eb-Cb joint. Ad5 dorsally on ventrodistalmost surface of Eb4, ventrally on dorsal surface of Cb5. Remarks. Edgeworth (1935:131) considered our Ad5 to be a 5th levator. RD absent. SO continuous with TEb4; comprises only trans- verse muscle layer (see remarks under SO in Acipen- ser). Additional remarks. SCL absent. TV4 absent. Dan- forth (1913) described the musculature of Polyodon. Our findings are in essential agreement. He reported that the four levators arise from a continuous sheet and separate as they proceed toward their insertions. He worked on specimens much larger (ca. 1 m) than the one we describe, in which LE3 and LE4 are scarcely separate at their insertions. The difference noted may be ontogenetic. Ginglymodi LEPISOSTEIDAE Atractosteus tropicus Gill, USNM 120715, 415 mm TL. Plate 5 Description. LE1 on dorsal surface of cartilaginous distal end of Eb1. LE2 on Eb2 (both removed when gill arches were removed; position in illustration is approximate). LE3 on dorsal surface of cartilaginous distal end of Eb3. Remarks. Short ligament (not illustrated) extends posteriorly from base of LE3 and inserts on anterior margin of cartilaginous distal end of Eb4. LE4 on dorsal surface of cartilaginous distal end of Eb4. LP absent. LlIlon dorsal bony surface of Pb2 and medial sur- face of SPbl. LI2 mostly on cartilage at ventral junction (not shown) of bony and cartilaginous portions of Pb3 proximal to anteriormost Pb3 teeth. LI3 absent (Pb4 absent). TD comprises only TPb3, which is continuous an- teriorly but originates posteromedially from raphes with SO and RDs and anastomoses ventrally with CT underlying SO. Laterally, TPb3 attaches and sur- rounds Pb3 dorsolateral process (see also RD) and is continuous with anterior end of OD4. OD3 absent. OD4 originates on dorsal end of Pb3 process and inserts along anterodorsal edge of medial end of Eb4. OP absent or indistinguishable medially from Ad4 and/or SO. Muscle laterally in this region joins Eb4 and Cb4 medial to joint of these two elements. Ad1-3 absent. Adé4 absent or indistinguishable medially from OP and/or SO (see OP and Ad5). Ad5 on posterodistal end of Cb4 and dorsodistal end of Cb5, inseparable medially from SO; some fi- bers on Cb4 questionably represent Ad4. Remarks. The muscles in this area are surrounded and penetrated with tough connective tissue, and the delineation of the muscles is not as clear as they ap- pear in Plate 5. RD inserts anteriorly on raphe that joins SO, Pb3, and TPb3 in region at and ventral to posterior end of TPb3. Remarks. Wiley (1976:32) stated that the RDs of lepisosteids share muscle fibers with the circular muscle layer of the esophagous [SO]. We note that SO does not have a longitudinal muscle fiber layer in Atractosteus and that RD fibers are not continuous with the SO circular muscle fibers, although the fibers BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON of both RD and SO are permeated with Goined by) anastomosing connective tissue. (See discussion of RD in Results section of pre-acanthomorphs.) SO attaches to Pb3 anteriorly and is free from Eb4, but dorsolaterally is inseparable from OP and/or Ad4 on Eb4; comprises only transverse muscle layer. Additional remarks. SCL absent. TV4 absent. Wi- ley (1976:32), erroneously reported TV4 present in lepisosteids, and concluded that its presence was a synapomorphy of the Neopterygii. His fig. 14, how- ever, accurately shows TV4 is absent in lepisosteids. TV4 first appears in Amia and is a synapomorphy of the Halecostomi. Pb2 is well removed from contact with Pb3. There are two cartilaginous SPbs, one each on the uncinate processes of Eb] and Eb2. There is a pair of wedge-shaped ACs between the anterodistal and posterodistal ends of Ebl and Eb2 and their re- spective Cbs. See also references to Wiley (1976) in Additional remarks under Amia calva. Halecomorphi AMIIDAE Amia calva Linnaeus, USNM 230909, 120 mm TL. Plate 6 Description. Remarks. Allis (1897) and Holstvoogd (1965) de- scribed and illustrated the dorsal gill-arch muscula- ture of A. calva; however, we find the illustrations and descriptions wanting. LE1 on dorsoposterior edge of Eb1 just lateral to proximal cartilaginous end. LE2 on dorsomedial cartilaginous process (end) of Eb2, which articulates with Pb3 (end ventromedially articulates with Pb2). LE3 on posteriorly extending portion of Eb3 car- tilaginous medial end. LE4 on both cartilage and bone at distal end of Eb4. LP absent. LI1 on anterior cartilaginous tip of Pb2. LI2 on cartilaginous portion of Pb3 dorsal to un- coalesced tooth plates. LI3 absent (Pb4 absent). TD comprises only TPb3, which attaches to an- terolateral portion of Pb3, with some posteromedial fibers on right side continuous with right-side RD. OD3 absent. OD4 relatively short; origin on anteromedial car- tilaginous portion of Pb3, insertion on medial end of Eb4, posteromedially joining small raphe with dor- somedial end of OP. OP on Eb4 dorsomedially, on Cb5 distally, contin- uous with SO ventromedially, overlaps ventral end of Ad5; ER absent. Ad1-—3 absent. NUMBER 11 Ad4 on dorsoposterior surface of Eb4 and dorsal surface of Cb4. Ad5 on distal cartilaginous ends of Cb4 and Cb5. RD completely separate from SO; with separate dorsal and ventral attachments on Pb3 ventral to TPb3; few or no muscle fibers continuous with TPb3 on left side, several fibers continuous with TPb3 pos- teriorly on right side. (See discussion of RD in Re- sults section of pre-acanthomorphs.) SO on Pb3 posteriorly; does not attach to Eb4; longitudinal muscle fibers sparsely distributed within SO, roughly paralleling mucosal folds, extend ante- riorly to posterior margins of posteriormost upper pharyngeal tooth patches. Additional remarks. SCL absent. TV4 free from Cb5s. Pb1 positioned horizontally, in line with Eb1. We agree with Wiley’s (1976:30—31) conclusions, based on Amia and lepisosteids, which have only OD4, that TD and OD muscles are apomorphic for Neopterygii; however, OD3 is a synapomorphy of the Teleostei. Osteoglossomorpha HIODONTIDAE Hiodon alosoides (Rafinesque), USNM 350554, 2 specimens, 102—130 mm. Plate 7A, B Additional material. @ = Hiodon tergisus Lesueur, USNM 266581, 82.1 mm; USNM 342742, 154 mm; USNM 350555, 163 mm. Plate 7C Description. LE1 finely, tendinously on cartilaginous medial end of Ebl. LE2 on cartilaginous medial end of Eb2. LE3 on cartilaginous tip of Eb3 uncinate process. LE4 dorsodistally on Eb4. Remarks. LE4 origin is well separated posteriorly from the linearly contiguous or clustered origins of the other levators. As such, LE4 in Hiodon could be interpreted as LP (Winterbottom, 1974b:footnote p. 252), which, in fishes with both LE4 and LP, always originates posterior to the origins of the other leva- tors. In other osteoglossomorphs, except the highly specialized Heterotis, the origin of the levator insert- ing dorsodistally on Eb4 is linearly contiguous with the other levators, and there is no levator inserting posterior to it. Greenwood and Lauder (1981:226) opined that LE4 in Hiodon has been displaced pos- teriorly because of “‘the large swimbladder extension in the otic region.” The interpretation of the poste- riormost levator as LE4 might be contraindicated by the condition of the levators in one of the three spec- imens of H. tergisus. The disposition of the levators in two of the specimens of H. tergisus we examined 25 is similar to that of both specimens of H. alosoides (Plate 7A). In the third specimen of H. tergisus (USNM 342742), however, there is, bilaterally, a slender, additional levator tendinously inserting on Eb4 well anterior to the typical posteriorly inserting LE4 (Plate 7C). The additional muscle is probably anomalous. Our observations indicate that LP is pres- ent in pre-acanthomorphs only among some of the fishes belonging to the Otocephala (some clupeo- morphs and ostariophysans), but it is usually present in acanthomorphs. LP absent (see remarks under LE4). LI1 on bony dorsal surface of Pb2. LI2 (see Remarks under LI3). LI3 (see Remarks). Remarks. There appears to be only one other LI besides LI1. In the smaller specimen of H. alosoides, a few fibers of the second LI insert posteriorly on Pb3, and the remainder insert on Pb4. In the larger specimen, the second LI inserts extensively on both Pb3 (posteriorly) and Pb4 (anteriorly). In two of the three specimens of H. tergisus, the second LI has minor insertion posteriorly on Pb3 and the remainder on Pb4. In the third specimen, the second LI has completely separate insertions on Pb3 and Pb4, with the more extensive insertion on Pb4. It is not possible to decide if the second LI represents: LI2 inserting partially on Pb4, fused LI2 and LI3, or LI3 with par- tial insertion on Pb3. Nelson (1969b:18) indicated the presence of only two LIs in osteoglossomorphs, but did not specify their insertions. We find only one LI in Petrocephalus, Mormyrus, and Gymnarchus (L11), and only two in Arapaima (LI1, LI2). Notopterus, Heterotis, Osteoglossum, Scleropages, and Pantodon all have three (LI1-—3). TD is broad, undifferentiated, consists of TPb3- Pb4-Eb4, continuous posteriorly with SO, from which it is differentiated only by divergence of mus- cle fibers at posterior attachment to Eb4. OD2 absent. @ Present in two smaller specimens (Plate 7C) in which it is small, originates on Pb3 near posterolateral origin of OD4, and inserts jointly with LE2 on Eb2 cartilaginous medial end. Remarks. Among pre-acanthomorphs, OD2 is oth- erwise known only in Heterotis, which, together with Hiodon, also lacks TEb2. The bones joined by the bilateral pair of OD2 muscles are the same as those joined by TEb2 (or TEb2 complex) of other osteo- glossomorphs (Arapaima, Osteoglossum, Scleropa- ges, Pantodon). TEb2 in these fishes, however, is a transverse muscle connecting the two sides of the gill arches. It is possible that OD2 is a modified TEb2. OD3 origin on Pb3 lateral to and continuous with OD4 origin, insertion dorsally on Eb3 uncinate pro- cess just ventral to LE3 insertion. OD4 origin on Pb3 medial to and continuous with OD3 origin, insertion on anteromedialmost edge of 26 Eb4 coincident with anterior attachments of Ad4 and OP. RecD4 origin on Eb4 dorsally anterolateral to OD4 insertion; insertion split, dorsally on dorsoposterior surface and edge of Eb3 uncinate process and ven- trally on medial end of Eb3 (not illustrated), vari- ously fusing with LE3 basally. Remarks. Nelson (1967a:281; 1967c) used the term obliquus inferior [of the dorsal gill-arch mus- culature] for a muscle joining two adjacent epibran- chials in eels. Winterbottom (1974b:259) noted that obliquus inferior is a muscle of the eye and that the term also had been applied to various other muscles. He, therefore, coined the name recti dorsales (sing., rectus dorsalis) for longitudinal muscles joining ad- jacent epibranchials “‘to reflect their analogous po- sition to that of the recti ventrales.”” RecDs occur in a diverse variety of fishes (e.g., osteoglossomorphs, anguilliforms, callionymids, etc.), and are probably homologous only within restricted groups of taxa. OP a narrow strap, originating on posteromedial margin of Eb4 coincident with ventral portion of OD4 insertion, ending posteroventrally at ER, which is joined ventrally by Ad5 and SO; second, even slen- derer portion of OP may be present, originating sep- arately somewhat ventral to narrow portion, inserting together with narrow portion. @ Originating laterally with Ad4 attachment; not distinguishable from SO. Remarks. Below ER, it is not possible to decide if OP is continuous laterally with Ad5 or is replaced by a medially expanded Ad5. Ad1-—3 absent. Adé4 origin on Eb4 dorsomedially, coincident with OD4 insertion, insertion on dorsoposterior surface of Cb4. Ad5 on Eb4 dorsoposteriorly and on Cb5 laterally, joining ER medially with SO (see also remarks under Ad5 in description of Arapaima). RD absent. SO longitudinal muscle layer thick dorsally and ventrally, thin to almost absent laterally, extends an- teriorly only to horizontal at posterior margin of TPb4-Eb4. Additional remarks. SCL absent. TV4 free from Cb5s. The distinct crossed anterior ends of Pb3s are dorsal to the crossed anterior ends of Pb2s, and the complex is completely encompassed in a ball of thick connective tissue (not shown) that attaches to the ventral surface of the cranium. NOTOPTERIDAE Notopterus notopterus Pallas, USNM 344674, ca. 127 mm; USNM 191464, ca. 143 mm. Plate 8 BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON Description. LEI on dorsal Eb1 bony surface just lateral to car- tilaginous medial end; dorsoanteriorly, fibers on me- dial surface mesh with fibers on lateral surface of LE2. LE2 dorsally on Eb2 cartilaginous and bony sur- faces just lateral to joint with Pb3; dorsoanteriorly, fibers on medial surface mesh with fibers on lateral surface of LE3’ and fibers on lateral surface mesh with fibers on medial surface of LE1. LE3 on tip of Eb3 uncinate process jointly with “3” attachment of OD3-—4. LE3’ on dorsal surface of medial end of Eb3; dor- soanteriorly, fibers on lateral muscle surface mesh with fibers on medial muscle surface of LE2, and fibers on medial surface mesh with fibers on lateral surface of LI3. See also ““Additional remarks” be- low. LE4 on cartilaginous dorsoposterior end of Eb4. LP absent. LI1 on bony dorsal surface of Pb2 just lateral to medial cartilaginous end. LI2 on dorsal surfaces of Pb3 and Pb4; insertion paralleling and distinct from LI3 insertion. LI3 on dorsal surface of cartilaginous Pb4, paral- leling and distinct from portion of LI2 insertion on Pb4, also a few fibers dorsally on medial end of Eb3; dorsoanteriorly, fibers on lateral surface mesh with fibers on medial surface of LE2. TD broad, undifferentiated, consists of TPb3-Pb4- Eb4. In smaller specimen, TD overlies and is free from anterior end of SO. In larger specimen, TD is continuous posteriorly with SO, from which it is dif- ferentiated only by divergence of muscle fibers at posterior attachment to Eb4. OD3-—4 origin on Pb3 bony portion adjacent to an- terior cartilaginous tip, posteriorly muscle divides longitudinally with branch inserting on cartilaginous tip of Eb3 uncinate process and branch dorsally on cartilaginous distal end of Eb4 where it forms lateral half of raphe with dorsal end of Ad4 and is partially continuous with LE4 insertion. OD4 origin beginning on dorsoposterior bony por- tion of Pb3 medial to LI2 insertion, continuing along most of length of cartilaginous Pb4, and inserting on Eb4 dorsodistally and forming medial half of raphe with Ad4. RecD4 a sliver of muscle tendinously attached an- teriorly to dorsal tip of Eb3 uncinate process; passes posteriorly between Eb4 portion of OD3—4 and OD4, conforming with surface of OD4, and joins raphe with Ad4 along with and posterior to OD4. OP absent or fused indistinguishably with SO; ER present at about level of dorsalmost attachment of Ads. Ad1-—3 absent. Ad4 on EB4 dorsoposteriorly, joined there by ra- NUMBER 11 phe with OD3—4, OD4, and RecD4; on Cb4 dorsal surface anterior to Eb4-Cb4 joint. Ad5 on posterior cartilaginous distal end of Eb4 and posteromedial surface of Cb5, mostly undiffer- entiated from SO laterally. RD absent. SO longitudinal muscle fibers thickest dorsally, ex- tending anteriorly to anteromedial end of Pb3. Additional remarks. SCL absent. TV4 free from Cb5s. Holstvoogdt (1965:fig. 5) illustrated the leva- tors of the notopterid Xenomystus nigri. He found only six, four of which appear to be equivalent to our LE1, LE2, LE4, and LI1. The equivalents of the other two levators, which he labels as L.II EX. and L.II.INT. are unclear, but possibly represent our LI2 and LI3, which, if true, would indicate that Xeno- mystus appears to lack LE3 and LE3’. GYMNARCHIDAE Gymnarchus niloticus Cuvier, USNM 319410, ca. 315 mm TL. Plate 9 Description. Remarks. The dorsal gill-arch musculature and skeleton are considerably reduced and muscle fusions are evident. Because of this, it is simpler to describe much of the musculature in narrative form. The only levator that is clearly present is LI, which is on Pb2 posteromedially. A large, long mus- cle, or muscle complex, probably incorporating LE1 and/or RecD1 (“LE1 complex” on Plate 9A), ex- tends well anteriorly from the anterior edge of Eb1. The dorsal surface of the muscle attaches along the ventral surface of the cranium. A long muscle, pos- sibly comprising LE2 anteriorly, is incorporated dor- sally in the LE] complex. As this questionable LE2 extends posteriorly, it attaches musculously to the posterior bony surface of Ebl, then tendinously to the bony surface of Eb2 (incorporating a RecD2?), continues musculously (as RecD3?) posteriorly, and forms tendinous anterior and posterior attachments to Eb3 and the anterior bony edge of Eb4. A short mus- cle (M. Pb3-Eb3-Eb2) connects the posteromedial bony dorsal surface of Eb2 with Pb3 and Eb3, where the latter two elements meet. Another muscle, RecD4, joins the dorsoposterior bony surface of Eb3 with the dorsal bony surface of Eb4. TD comprises TPb2-Pb3-Eb4, originates anteriorly at about level of Pb2s as sparse fibers on surface of CT; anterolaterally, fibers appear to form part of me- dial edge of LE! muscle complex. TD becomes sheet-like posteriorly beginning at about level of Pb3s, and is undifferentiated posteriorly from SO. OD3 absent. OD4 absent. OP absent; presence or absence of ER unclear. Ad1-3 absent. Ad4 on ventral surface of Eb4 and dorsal surface of Cb4. Ad5 dorsally attaches to tendon extending laterally from Eb4, and ventrally to posterodistal end of CbS5; a unique, laterally extending strap of SO arises from anterior surface of tendon and sandwiches Ad5 be- tween it and another SO strap arising from Cb5 area ventrally. RD absent. SO longitudinal muscle layer comprising sparsely distributed fiber bundles in “‘cottony”” matrix, ex- tending anteriorly to anterior end of TD. Additional remarks. SCL absent. TV4 free from Cb5s. Pb1 and Pb4 are absent. Tiny AC present only on left side at posteromedial tip of Eb1. MORMYRIDAE Petrocephalus tenuicauda (Steindachner), USNM 118801, 92.3 mm. Plate 10 Additional material. @ = Mormyrus longirostris Pe- ters, USNM 261878, 225 mm. Description. LEI on Eb] anterior and posterior edges (bridging channel that carries blood vessels and nerves) with some fibers also on Pb1; fuses posteromedially with LE2. @ Unclear if LE2 is represented among fusions. LE2 on Eb2 posteromedially, fuses laterally with LE! and medially with M. Pb3-Pb4-Eb2, which may represent a modified LI2 and/or LI3. On left side, a muscle originates narrowly and separately on crani- um and inserts narrowly, tendinously on Eb2 among fibers of broad, fused complex LE2. This muscle is also labeled LE2 on Plate 10. @ Unclear if LE2 is present, but if so, is incorporated in RecD2. LE3 on all bony Eb3 uncinate process just lateral to RecD3 and OD3 attachments. @ Extremely long and variable in width along its length, changing al- most to thin tendon in places. LE4 absent. LP absent. LI1 on dorsomedial surface of Pb2. LI2 see LE2. I) See IL. M. Pb3-Pb4-Eb2 originates on posterodorsal sur- face of Pb3 and dorsal surface of greatly reduced Pb4, and inserts on Eb2 posteromedially; anteriorly, fuses with lateral portion of LE2. @ LE2 portion questionably present. TD apparently undivided, consisting of TPb3-Pb4- Eb4, more-or-less continuous posteriorly with SO. @ Comprises TPb3 and questionably TPb4-TEb4. TPb3 dense band of muscle fibers attached to Pb3 dor- soanteriorly, abruptly changing posteriorly to loose 28 strands of diagonal muscle fibers attaching to Pb4, and these continuing posteriorly as dense network of fibers attaching to Eb4 posteromedially and continu- ing posteriorly as SO. OD3 tendinously on Pb2 (rather than Pb3 as in most fishes), and tendinously on all bony Eb3 unci- nate process. On right side only, there is a possibly anomalous strap of muscle extending anteriorly from Eb3 uncinate process that fuses anteriorly with LE1- LE2 complex; strap lies dorsal to M. Pb3-Pb4-Eb2. @ Strap absent. OD4 absent. M. Pb2-Eb1 narrowly on posteromedial edge of Eb1 (ventral to anteriorly extending, fused LI12+3? and LE2) and broadly on anterodorsal surface of Pb2. @ Originates on dorsoanterior surface of Pb3, with slender tendon on medial surface of muscle expand- ing anteriorly and joining CT covering Pb2-Pb3 joint. RecD2 broadly on dorsal surface of Eb2 ventral to LE2, and narrowly on posteromedial corner of Eb] ventral to LE2. ® Presence of LE2 questionable. RecD3 on all bony uncinate process of Eb3 and dorsal surface of medial cartilaginous end of Eb2. @ Asymmetrical; left side with cluster of small muscles originating on Pb3, Pb4, and Eb4 medially, and in- serting on Eb2 anteromedially; right side with two small muscles, one from Eb2 anteromedially to Eb4 anteromedially, the other from Eb2 anteromedially to Eb3 dorsomedially. OP questionably present. Remarks. ER, often denoting the ventral end of OP, when OP is present, is irregularly indicated on both sides of SO at about the level of the distal end of Eb4. On the left side there is a narrow strap of muscle that originates on Eb4 and ends at the raphe. On the right side there is no strap. @ Many raphe-like CT intrusions in SO dorsolaterally; no defined strap-like muscle; muscle fibers complexly oriented. Differenc- es from Petrocephalus possibly due to much larger size of specimen. Ad1-—3 absent. Ad4 dorsally on Eb4 dorsoanteriorly and ventrally on Cb4. Ad5 dorsally on dorsoposterodistalmost cartilagi- nous tip of Eb4 and ventrally on Cb5; fuses medially with SO. RD absent. SO longitudinal fibers essentially restricted to dor- sal arc of SO ventral to transverse fibers, few if any longitudinal fibers ventrally. @ Fibers sparsely dis- tributed dorsally, few if any laterally or ventrally. Additional remarks. SCL absent. TV4 free from Cb5s. Pb4 is greatly reduced and offers little surface for attachment of a levator. This could explain the absence of LI3, if present ancestrally. Bishai (1967:20—21) described and illustrated the dorsal gill-arch musculature of Mormyrus caschive BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON Linnaeus. Using terminology different from ours, he reported the presence of LE1—4, LI2 and 3 (on Pb2 and 3, respectively, which would equal our LI] and LI2), OD3, and a muscle that we would term OD3’ (the two originating on separate parts of Pb3 and in- serting together on bony Eb3 uncinate process), RecD2 and 3, and Ad4. He did not mention fusion of any muscles, and his illustrations do not indicate the presence of cartilage. It is possible that his LI3 may insert partially on Pb4. Nelson (1969b:18) reported on the dorsal gill-arch muscles of Mormyrus ovis, in which he recognized only LE1 and 2, LI1, OD3, OD4, and RecD 2 and 3 (no mention of Ads). Although Nelson mentioned Bishai’s study, he made no comment on the differ- ences between his findings and those of Bishai. We believe there is considerable room for differences in interpretation of the muscles, as well as the possibil- ity that no two specimens of the same species (or two sides of a specimen) are more than generally similar. Our remarks undoubtedly also apply to our speci- mens of Petrocephalus and Mormyrus. ARAPAIMIDAE Heterotis niloticus (Cuvier), USNM 303214, 237 mm. Plate 11 Description. Remarks. Heterotis is unique among osteoglossi- forms we examined in having an epibranchial organ. The organ is formed by the complex coiling and ex- tensive development of what is probably the dorsal cartilaginous margin of Eb4, and, at least, includes that margin. The thin muscle branches attaching var- iously to the surface of the cartilage are difficult to homologize with reasonable certainty. Adding to the difficulty are bilateral asymmetry of these muscles, a sheet of muscles lining the internal surface of the cartilaginous coil, and an extensive network of in- vasive nerves supplying the organ internally and ex- ternally. We have not indicated nerves in other illus- trations, but in Plate 11A, we show a large nerve (identification not determined), which superficially resembles a muscle. The nerve penetrates the dor- somedial side of the cartilaginous coil. LEI on cartilaginous and bony dorsomedial end of Eb1, fusing dorsally with anteroventral, longitudinal fibers of LE2. Short ligament (not illustrated) joins dorsomedial surface of cartilaginous medial end of Eb1 to bony dorsal surface of anterior end of Pb2. LE2 on dorsalmost prominence of cartilaginous medial end of Eb2; muscle fusing ventrally with dor- sal, longitudinal fibers of LE]. RecD2 parallels LE2 ventrally, with ribbon of CT (not illustrated), which connects cartilaginous processes of Eb1 and Eb2, ly- ing laterally and attaching to both muscles. Short lig- NUMBER 11 ament (not illustrated) joins dorsomedial surface of cartilaginous medial end of Eb2 to dorsal bony sur- face of anterior end of Pb3. LE3 tendinously on cartilaginous tip of Eb3 un- cinate process; muscle thin, weak, posterodorsally di- rected, applied medially to lateral surface of epibran- chial organ; origin not noted, but apparently not at- taching directly to cranium; muscle possibly absent on right side, or lost during dissection. LE4 on dorsal surface of epibranchial organ, con- forming with curvature of organ; possibly absent on right side or lost during dissection. LP absent. LI1 on Pb2 mid-dorsally. LI2 on bony Pb3 dorsal surface; right side with slender posterior branch inserting anteriorly on dorsal surface of cartilaginous Pb4; branch absent on left side. LI3 attached to fascia covering surface of cartilag- inous Pb4; anterior, elevated portion of muscle inter- rupted at attachment, continuing posteriorly as CT and then small portion of muscle applied to posterior surface of Pb4 (muscle uninterrupted on right side); CT at base of elevated portion gives rise to posterior attachment of M. Pb4-Eb2. LI3 passes anteriorly lat- eral to LI2. TD possibly absent anteriorly (see OD2), compris- es TPb3-Pb4 (on posterior Pb3 segment and, possi- bly, also on Eb4), and is undifferentiated posteriorly from SO. OD2 originates on dorsal surface of autogenous cartilaginous Pb3 anterior segment (Nelson, 1968a: 268) anterior to OD3 origin, and inserts on cartilag- inous medial end of Eb2 dorsally. Medial fibers of right-side OD2 overlap anterior end of left-side OD2 and attach on dorsoanterior end of left-side Pb3 car- tilaginous segment. Anterior ends of Pb3 cartilagi- nous segments and OD2 origins are enveloped in tough CT pad. Remarks. OD2 is known otherwise in pre-acantho- morphs only in some specimens of Hiodon tergisus. See remarks under OD2 in description of Hiodon. OD3 originates on dorsal surface of Pb3 anterior segment and inserts by long tendon on tip of elongate cartilaginous cap of Eb3 uncinate process. Right-side origin is ventral to OD2 origin; left-side origin begins at posterior end of OD2 origin. OD4 absent on left side; questionably represented on right side by thin, elongate muscle originating on Pb3 just anterior and medial to LI2 insertion and in- serting on posterior (= medial in view) cartilaginous wall of EO (Eb4). OP absent; ER absent. RecD2 on posterior surface of dorsal prominence of medial cartilaginous end of Eb! and anterior sur- face of dorsal prominence of medial cartilaginous end of Eb2, parallels LE2 ventrally, with CT ribbon (not 29 illustrated) connecting cartilaginous processes of Eb1 and Eb2 passing laterally and attaching to both mus- cles. Left-side RecD2 comprises a vertical pair of muscles, each member of which attaches tendinously to Eb2; right-side RecD2 has vertical series of three separate, but contiguous musculous attachments to Eb2. RecD3 absent. RecD4 absent, but long, broad ligament arising from Eb3 uncinate process and lateral edge of Pb4 inserts on dorsal bony surface of Eb4. M. Pb4-Eb2 attaches to posterior surface of dor- somedial cartilaginous extension of medial end of Eb2 and inserts by flat tendon on Pb4 ventral to in- sertion of elevated portion of LI3 (q.v.). Ad1-—3 absent. Ad4 fan-like, attaching inner cartilaginous ventro- posterolateral surface of EO (Eb4) to bony medial surface of posterior end of Cb4; dorsally, begins at lower internal rim of large foramen in anterior sur- face (lateral in view) of cartilaginous posterior end of Eb4 (Plate 11B) and expands anterodorsally end- ing as fine sheet of CT attaching to medial surface of dorsoposterior bony flange of Eb4. Large, fan- shaped gill-filament muscle (not illustrated) on pos- terolateral bony surface of Eb4 matches somewhat the shape of Ad4 on internal surface. On left side, an apparently anomalous slip of Ad4, which attaches dorsal to remainder of muscle, is visible through the Eb4 foramen, after removal of fine sheath of muscle lining inner wall of epibranchial organ. Ad5 small, horizontal muscle joining posteroven- tral cartilaginous surface of EO (Eb4) to dorsopos- terior cartilaginous surface of Cb5. RD absent. SO longitudinal muscle fibers questionably absent, sparse if present. Additional remarks. SCL absent. TV4 free from Cb5s. Glottis present. Arapaima gigas (Cuvier), USNM 177528, 2 speci- mens, ca. 135-139 mm. Plate 12 Description. LE1 on Eb! dorsoposterior edge just lateral to me- dial cartilaginous end. LE2 absent. LE3 absent. LE4 on dorsoposterior edge of distal cartilaginous end of Eb4. LP absent. LI1 on Pb2 dorsoposteriorly. LI2 on dorsal surface of Pb3 posteromedially. LI3 absent. Remarks. M.Pb4-Eb2 conceivably represents LI3 30 that has shifted its origin to Eb2. (See also remarks under LI3 in Hiodon). TD comprises two parts: TPb3-Eb2 and TPb3- Pb4-Eb4. TPb3-Eb2 a very thin sheet of muscle at- taching to dorsal surface of Pb3 anterior cartilaginous segment (Nelson 1968a:267—268) and cartilaginous processes of Pb3 (posterior bony segment) and Eb2, where these processes join, and is anterodorsal to TPb3-Pb4-Eb4, which attaches to medial edges of Pb3, Pb4, and Eb4. TEb2 is dorsal to anterior attach- ments of OD3 and OD4. TPb3-Pb4-Eb4 is undiffer- entiated posteriorly from SO. OD3 on autogenous anterior Pb3 cartilaginous seg- ment and Eb3 uncinate process. OD4 anteriorly on Pb3 bony portion, posteriorly on dorsomedial edge of Eb4, fusing posteriorly with RecD4. Remarks. Our interpretation of OD4 and RecD4 (q.v.) may be in error and the muscle we indicate as RecD4 may be the Eb3 branch of an OD3—4, a mus- cle otherwise limited to Notopterus and Pantodon among Osteoglossomorpha. Pantodon lacks a sepa- rate OD4. Only Notopterus among the Osteoglosso- morpha has OD4, OD3-—4, and RecD4, and Notop- terus lacks OD3, which is present in all other Osteo- glossomorpha except Pantodon and the highly spe- cialized Gymnarchus, which lacks OD completely. Our general observation among Neopterygii, is that, excluding Notopterus, OD4 or OD3—4 may be pres- ent, but not both. Very few neopterygian taxa (Gym- narchus; the ostariophysan Chanos) lack an OD, a lack we consider autapomorphic for each of these taxa. OP absent; ER absent. Remarks. There is considerable CT permeating and obscuring the nature of the area that might be defined as OP. RecD2 on cartilaginous posterior edge of Eb] me- dial end and bony posterior edge of Eb2 medial end; muscle strands continuous posteriorly with M. Pb4- Eb2. RecD3 absent. Remarks. Muscle reported to be present by Nelson (1969b:18), but he may have considered it the same as the muscle we designate M.Pb4-Eb2 (q.v.). We use RecD to indicate a muscle that connects two succes- sive epibranchials. RecD4 on posterior edge of cartilaginous tip of Eb3 uncinate process, fusing posteriorly with poste- rior end of OD4. Possibly absent, see remarks under OD4. M. Pb4-Eb2 on Pb4 dorsally and posteromedial cartilaginous edge of Eb2, muscle strands continuous anteriorly with RecD2 (see also remarks under LI3). Ad1-—3 absent. Ad4 broadly on broad posterolateral surface of BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON Eb4 and narrowly on dorsodistal surface of Cb4 me- dial to Eb4-Cb4 joint (not visible in illustrations). Ad5 slender, from outer edge of dorsoposterior- most distal cartilaginous tip of Eb4 to posterior sur- face of cartilaginous distal end of CbS. Remarks. Winterbottom (1974b:254—255) dis- cussed the problem of what to call a muscle attaching Eb4 to Cb5 when there is only one such muscle. Both OP and Ad5 may exhibit these attachments (Ad5 may attach Cb5 to Eb4, Eb4*, Eb5, or Cb4, but almost always attaches to EbS5 when EbS is present (or to Eb4* when this state replaces an autogenous Eb5), exceptions: cyprinids and Searsia, Argentiniformes). OP is usually attached well medial to the distal end of Eb4, whereas Ad5 attaches to the distal end (es- pecially when Eb4* is present). Winterbottom rec- ommends using Ad5 for the single muscle, and we concur in the present case, especially as the appear- ance of the single muscle is more like an Ad5 than an OP, and OP appears to be present in most other osteoglossomorphs (see Hiodon alosoides). RD absent. SO longitudinal muscle fibers questionably absent, sparse if present. Additional remarks. SCL absent. TV4 free. Small AC attached to dorsolateral tip of Cb5 (not reported by Nelson, 1968a), found otherwise only in Panto- don, thus providing evidence supporting the clade comprising these two genera and Heterotis, which lacks the AC. Arapaima and Heterotis are the only osteoglossiforms we examined that have a glottis (opens dorsally into an air sac), at least one that is so close to the dorsal gill arches. The air sac is joined broadly to the SO, but was not illustrated. PANTODONTIDAE Pantodon buchholzi + Peters, USNM 303224 (ca. 80 mm), USNM 353993 (2:50.4—59.2 mm), USNM 355626 (72 mm). Plate 13 Description (composite). LE1 absent. LE2 on dorsoposterior edge of Eb2 somewhat lat- eral to medial end, meets attachment of TEb1-Eb2 on Eb2. LE3 absent. LE4 on dorsoposterior edge of Eb4 somewhat me- dial to lateral end, meets attachment of OD3—4 on Eb4. LP absent. LI1 primarily dorsally on Pb2, wrapping around cartilaginous medial end of Eb1 with slight insertion on medialmost end of Eb1. LI2 on dorsolateral bony surface of Pb3, insertion posteriorly continuous with that of LI3 Eb4. LI3 anteriorly on posterolateral cartilaginous end NUMBER 11 of Pb3, but mainly on dorsolateral surface of UPS (few fibers on cartilaginous anteromedial end of Eb4 in One specimen), insertion anteriorly continuous with that of LI2, posteriorly continuous with small group of SO muscle fibers. TD comprises TEb1-Eb2 and TPb3-Eb4. TEbI1- Eb2 divides briefly laterally with short branch attach- ing to posterior edge of medial end of Eb1, and long branch attaching along posterior edge of medial half of Eb2; continuous posteriorly by few crossing di- agonal muscle strands with TPb3-Eb4. TPb3-Eb4 at- taches to medial edges and surfaces of Pb3 and Eb4 and is continuous posteriorly (undifferentiated) with SO. M. Pb2-Eb2 originates on ventroanterior surface of Pb2 (few weak muscle strands may attach on ventro- medial end of Eb1) and flares posteriorly as it inserts along dorsoanterior edge of medial half of Eb2. OD3-—4 origin on Pb3 dorsal surface, divides pos- teriorly with slender branch inserting on Eb3 mid- posterolaterally and much broader branch inserting on Eb4 mid-dorsolaterally together with insertion of RecD4. Remarks. Muscle is similar to that of Notopterus, q-V. RecD4 originates ventrally on posteromedial edge of Eb3 and inserts dorsally on Eb4 together with Eb4 branch of OD3-4. OP questionably absent; however, possibly repre- sented by a thin, narrow band of muscle originating on Eb4 near medial end of Ad4 and scarcely sepa- rable medially from OP, fans out ventrolaterally, passing anterior to Ad5, Eb5, and AC, and attaches to Cb5 at and anterior to ventral attachment of Ad5. ER absent. Ad1-—3 absent. Ad4 dorsally on posterolateral surface of Eb4, ven- trally on Cb4 dorsal surface at anterior angle formed by Eb4-Cb4 joint. Ad5 dorsally on posterolateral surface of Eb5, ven- trally on posterolateral surface of Cb5 together with lateral attachment of TV5. SO longitudinal muscle fibers questionably pres- ent. SOD broad. RDs separate, on posterior ends of Pb3 and UPS, which is ventral to Pb3 posteriorly and Eb4 anteri- orly. Remarks. RD is autapomorphic in Pantodon. Additional remarks. SCL absent. TV4 free from Cb5s. Pb1 and Pb4 absent (we presume that UP4 is also absent and that the large toothplate ventral to Pb3 and Eb4 is UP5). Eb5 and accessory cartilage (AC) present: Eb5 transverse, joined laterally and more-or-less equally to posterodistal ends of Eb4 and Cb4 and medially to dorsal end of AC; AC joined ventrally to dorsodistal end of Cb5. The orientation 31 of Eb5 is duplicated only in Diplomystes, Ostario- physi, which lacks AC. Among osteoglossomorphs, AC is found otherwise only in Arapaima, thus sup- porting the clade comprising these two genera and Heterotis, which lacks AC. OSTEOGLOSSIDAE Osteoglossum bicirrhosum Cuvier, USNM 315447, 210 mm; USNM 198123, 59.6 mm. Plate 14 Additional material. @ = Scleropages jardini (Sa- ville-Kent), USNM 217049, 215 mm. Description. Remarks. Because of muscle fusions, our interpre- tations of LE1, LE2, RecD2, RecD4, and some other problematic muscles of Osteoglossum are, in part, subjective, and are described together. The smaller specimen appeared to differ in no significant way from the larger specimen. LE1 is a thick ribbon of muscle originating on cra- nium inseparably from dorsal end of Pbl, and in- serting along most of the dorsal surface of elongate, cartilaginous Pb! and dorsomedial end of Eb1. Dor- somedial portion of LE1 insertion on Eb1 joining ra- phe with anterior end of RecD2, which originates on Eb2 dorsomedially. RecD2 broadens considerably medially, forming what we consider a separate mus- cle, M. Pb2-Eb2, and attaching anteriorly to dorsal surface of broad cartilaginous anterior end of Pb2 just ventral to insertion of LI (anterior end of Pb2 spat- ulate, becoming almost vertical medial to attachment of M. Pb2-Eb2). At attachment on Eb2, RecD2 joins complex raphe with LE2 insertion, LI3 ventrolateral surface, and anterior end of M. Pb4-Eb2 (described below following OD4). LE2 and LI3 weakly (thinly) continuous dorsal to raphe on Eb2, with LI3 con- tinuing posteroventrally to its insertion. Anterior por- tion of LI3 insertion on Pb3 just lateral and parallel to LI2 insertion; posterior portion of insertion on car- tilaginous Pb4, fusing and attaching inextricably with origin of M. Pb4-Eb2. Anterior end of RecD4 and posterior end of OD3 meet in raphe and attach on medial surface of cartilaginous tip of Eb3 uncinate process. RecD4 attaches posteriorly to mid-dorsal cartilaginous edge of Eb4 anterior to insertion of LE4. @ Anterodorsal portion of LE] extends dorsal to dorsal end of Pb1 and originates separately on cra- nium. RecD2 not expanded medially—M. Pb2-Eb2 absent—not attaching to Pb2. LE2 overlaps LI3 lat- erally, but both muscles are clearly separate anterior to their attachments to Eb2. RecD4 absent. LE3 on cartilaginous dorsal tip of Eb3 uncinate process. LE4 broadly along dorsodistal cartilaginous sur- face of Eb4. LP absent. LI1 on lateral surface of vertical expansion (Nel- son, 1968a:266) of cartilaginous anterior end of Pb2. @ LI1 in deep excavation on lateral surface of ver- tical expansion of Pb2. LI2 on dorsal bony surface of Pb3; medially, in- sertion parallels and is inseparable from lateral at- tachment of TD; laterally, LI2 insertion parallels and is medial to anterior portion of LI3 insertion (the two muscles are thin and “‘plastered”’ together); posteri- orly, insertion impinges on anterior attachment of OD3, which is on posterior cartilaginous end of Pb3. @® Insertion does not impinge on OD3 anteriorly. LI3 (see initial paragraph under Description above). TD comprises two portions: short TEb2 and long TPb3-Pb4-Eb4 that is posteriorly scarcely separable from SO. TEb2 is broad, somewhat triangular, with partial median raphe extending posteriorly from mus- cle’s anterior apex; muscle attaches to dorsomedial cartilaginous end of Eb2, where it is inextricably meshed with complex attachments of LE2, LI3, RecD2, and M. Pb4-Eb2. TEb2 is dorsal to and dis- continuous from remainder of TD. @ An apparently anomalus strand of muscle is continuous from pos- terior margin of right side of TEb2 and posteromedial side of OD3. OD3 on dorsoanterior cartilaginous surface of Pb3 (excluding small, separate anterior cartilaginous seg- ment of Pb3; see Additional remarks, below) and in- serts on raphe with RecD4 on medial side of dorsal end of Eb3 uncinate process. OD4 anteriorly ventral to OD3, originates on an- teriormost bony surface of Pb3 medial to LI2 inser- tion, and inserts on dorsomedial cartilaginous surface of Eb4 just anteroventral to attachment of RecD4. M. Pb4-Eb2 on Pb4 dorsolaterally, extending an- terolaterally and inserting on dorsomedialmost carti- laginous edge of Eb2. OP probably represented by strap of muscle at- taching dorsally to posterior ventromedial surface of Eb4 ventral to Eb4 portion of TD, extending ven- trally to ER at about level of mid-distal end of Eb4. @ ER particularly well developed. Ad1-—3 absent. Ad4 broadly on posterior surface of cartilaginous distal end of Eb4, ventrally, narrowly on dorsodistal bony surface of Cb4. Ad5 barely separable from SO medially, on pos- terodistal cartilaginous edge of Eb4 and posterodistal surface of Cb5. RD absent. SO longitudinal muscle fibers in thin layer sur- rounding esophagus. @) Longitudinal fibers in thick BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON layer dorsally and ventrally, thinner to almost inter- rupted mid-laterally. Additional remarks. SCL absent. TV4 free from Cb5s. Pb3 with small, autogenous anterior segment; segment absent in Scleropages. Nelson (1968a:268) believed Pb3 cartilage segment among Teleostei to be restricted to Arapaima and Heterotis, in both of which it is relatively large. He considered its pres- ence an indication that the two genera are closely related. The absence of the segment in Scleropages and in some Osteoglossiformes with reduced gill- arch skeletons (e.g., Pantodon, Gymnarchus), might be secondary. Elopomorpha: Elopiformes MEGALOPIDAE Megalops cyprinoides (Broussonet), USNM 350458, 133 mm. Plate 15 Additional material. ®@ = Megalops atlanticus Valen- ciennes, USNM 303317, 146 mm. Description. Remarks. Holstvoogd (1965: especially figs. 3a and 3b) illustrated and briefly discussed the dorsal gill-arch muscles of M. cyprinoides. Although our findings generally agree with his, we find his illus- trations difficult to interpret, and he indicated the presence of only two obliqui dorsales, apparently missing the muscle we identify as OD4. LE1 on Eb! just lateral to cartilaginous tip of un- cinate process. LE2 on dorsoposterior edge of Eb2 just anterolat- eral to SPb2 and medial end of Eb2. LE3 on Eb3 uncinate process, which articulates with Eb4 uncinate process. LE4 on dorsodistalmost end of Eb4. LP absent. LI1 on Pb2 anterodorsally and on SPb1 anterior surface. L]2 absent. LI3 on Pb4 dorsolaterally. TD comprises TPb3a and TPb3p-Pb4-Eb4. TPb3a is on Pb3 at and anterior to uncinate process; muscle narrows anteriorly and joins CT between anterior ends of Pb3s; posteriorly, TPb3a overlies anterior- most end of TPb3p-Pb4-Eb4 and is continuous with that muscle by slender muscle strand. ® TPb3a en- tirely anterior to TPb3p-Pb4-Eb4, continuous poste- riorly with remainder of TD by broad muscle strand. OD3 originates on Pb3 ventral to TPb3, partially divided longitudinally, inserts on Eb3 uncinate pro- cess. OD4 originates on Pb3 with and ventral to OD3, and inserts by long tendon anteriorly on Eb4 uncinate process lateral to OD4' insertion. NUMBER 11 OD4' originates on Pb3 dorsoposteriorly and on Pb4 dorsoanteriorly and inserts on Eb4 uncinate pro- cess medial to OD4 insertion. OP attaches dorsally on Eb4 posteromedial surface and joins ER ventrally, undifferentiated from SO ventral to ER. M. UP4-Eb5-Cb4 (not illustrated) questionably distinct muscle strap anterior to (overlain posteriorly by) OP, with dorsoanterior fibers attaching to lateral edge of UP4 and posteroventrally tendinously attach- ing to junction of Eb5 and Cb4. The alternative to considering M. UP4-Eb5-Cb4 as a distinct muscle is that it is a slightly differentiated SO portion. Ad1-—3 absent. Ad 4 broadly dorsally on posterodistal margin of Eb4, ventrally on Cb4 posteromedially (not visible in lateral view) medial to internal angle formed by Eb4- Cb4 joint. Ad5 complex; mainly attaching dorsally to Eb5 posterior surface, but tendinously bound also to pos- terior end of Cb4; muscle fibers shift directions (featherlike) along mid-axis as muscle extends ven- trally, with lateral fibers attaching to Cb5 medially and continuing anteriorly along most of ventral length of Cb5, then changing to broad membranous sheath medial to PCI and PCE (PCI not visible in lateral view, Plate 15C); medial fibers join raphe with TVS. RD absent. SO longitudinal muscle layer restricted to isolated area delimited posteriorly by horizontal between dis- tal ends of Eb4s and anteriorly by horizontal between mid-lengths of Pb4s. Additional remarks. SCL questionable, similar to that of Elops (see Additional remarks under Elops saurus for discussion of SCL). TV4 free from Cb5s. Large, tough CT pad covers dorsoanterior surfaces of Pbls and Pb2s. EbS5 attaches to posterodistal end of Cb4. Johnson and Patterson (1996:273) reported the presence of a small interarcual cartilage between the uncinate processes of Eb3 and Eb4, in Megalops and Elops. We did not find this cartilage in either specimen of the two Megalops species we examined, nor in one cleared and double stained specimen each of Megalops cyprinoides (USNM 173580) and M. at- lanticus (USNM 357435), but we reconfirmed its presence in Johnson and Patterson’s specimen of M. atlanticus (their specimen of M. cyprinoides was un- available). The presence of the structure is variable and probably of little use for establishing familial in- terrelationships. In Elops (q.v.), the putative rod-like interarcual cartilage is not autogenous, but is an ex- tension of the medial cartilage tip of Eb4. We did find a small cartilage between the tips of the uncinate processes of Eb3 and Eb4 of the albulid Prerothrissus (Plate 18), but not in Albula, its sister group. 33 ELOPIDAE Elops saurus Linnaeus, USNM 121694, 126 mm. Plate 16 Description. LE1 on Eb! near dorsomedial end and ventral to articulation with SPb1. LE2 on dorsomedial end of Eb2 and entire anterior surface of SPb2; LI2 and LI3 meet LE2 on SPb2. LE3 on dorsal tip of Eb3 uncinate process, joined there by OD3 insertion. LE4 attaches by long tendon to dorsodistalmost end of Eb4. LP absent. LI1 on Pb2 dorsoanteriorly, attaching to entire me- dial surface of SPb2; muscle looping anteromedially from posterior attachment to SPbl, forming two sheet-like layers with less extensive posterior layer incompletely overlapping surface of more extensive anterior layer. Remarks. Unlike Megalops, there appears to be no separate origin of LI1 on the braincase, its origin be- ing confined to SPbl. LI2 on Pb3 posterolaterally and SPb2 medial sur- face dorsally. LI3 on lateral edge of Pb4 at junction with UP4, anterior edge attached to SPb2. TD comprises TPb3 and TPb4-Eb4. TPb3 has two sections: smaller dorsoanterior section dorsally on Pb3 dorsolateral process, overlies anterior end of OD3; larger posterior section on Pb3 lateral edge, attaching posterolaterally along common line with LI2, overlies anterior end of OD4, and is posteriorly continuous by slender, diagonal muscle strand with SO. Remarks. Winterbottom (1974b:256 and fig. 18) stated that TD is absent in Elops because the trans- verse muscles are continuous with SO. We do not accept that continuation with SO is sufficient basis for rejecting the presence of a transversus muscle. Even if one rejects our TPb4-Eb4 as a transversus, TD is clearly indicated by TPb3 in Elops. OD3 anteriorly on posterior edge of Pb3 dorsolat- eral process ventral to TPb3 anterolateral attachment; posteriorly on dorsal tip of Eb3 uncinate process to- gether with LE3. OD4 anteriorly on posteromedial surface of Pb3 and anteromedial surface of Pb4, posteriorly on an- terolateral surface of Eb4. OP dorsally on Eb4 posteriorly, ventromedially joins ER. Ad1-—3 absent. Ad4 dorsally broadly on posteriormost surface of Eb4, ventrally on posteriormost end of Cb4 lateral to Eb5 ventrally. Ad5 dorsally on medial junction of Cb4 and Eb5, curving first ventromedially around posterior end of 34 Cb5, then laterally, attaching along lateral surface of Cb5. RDs absent. SO longitudinal muscle layer restricted to isolated area delimited posteriorly by horizontal between dis- tal ends of Eb4s and anteriorly by horizontal at an- terior end of TPb4-Eb4. Additional remarks. TV4 free from Cb5s. Presence of SCL questionable: Hb3 posteroventral flange is straight anteriorly, curves medially posteriorly, and is attached tightly to posteroventral cartilaginous pro- cess of Bb3, thus forming with contralateral Hb3 an anteriorly open semicircle, the edge of which is lined with CT. ObV3, on each side, attaches along the lat- eral edge of the straight portion of the Hb3 flange and is musculously continuous medially with its re- spective RecV4, which attaches to the posterior edge of the curved portion of the flange, but not to the cartilaginous tip of Bb3. There is no free ligamentous portion along the semicircle, but if some portion were free, we would interpret the semicircular ligament as being present. (VB. There is no autogenous ball of cartilage ventral to the posteroventral tip of Bb3, as is present in Albula.) Megalops (both species) is sim- ilar to Elops, but Hb3s do not curve medially pos- teriorly, the CT lining the semicircle is slightly loose, and RecV4 is attached to the posterior margin of the posteroventral cartilaginous process of Bb3. Nelson (1968b:fig. 6a) indicated the presence of an interarcual cartilage between the uncinate process of Eb3 and the proximal end of Eb4. Johnson and Patterson (1996:273) claimed to have confirmed this. Our observations differ. The “‘interarcual cartilage” of our specimen, and Johnson and Patterson’s, on re- examination, is a non-autogenous, elongate cartilag- inous extension of the dorsomedial cartilaginous head of Eb4. The extension attaches to the tip of the Eb3 uncinate process. Conceivably, this cartilaginous ex- tension could bud off in large specimens, and such should be examined to determine if this occurs. See also Additional remarks under Megalopidae. Elopomorpha: Anguillomorpha ALBULIDAE Albula vulpes (Linnaeus)?, USNM 247511, 127 mm Si Plate 17 Description. LEI broadly on Ebl, attaches to skull separately from other LEs, incorporates tendon dorsally. LE2 broadly on Eb2, incorporates tendon dorsally. LE3 on Eb3 uncinate process ventral to cartilagi- nous tip. LE4 on anterior surface of levator process of Eb4 just ventral to cartilaginous edge. BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON LP absent. LI1 on Pb2 anteriorly, comprises two ventrally continuous sections; anterior section shorter than pos- terior section, attached all along medial surface of cartilaginous SPb1. Remarks. See remarks under LI] in Aldrovandia (Halosauridae) description. LI2 on dorsolateral bony portion of Pb3. LI3 on dorsolateral edges of UP4 and UPS, inser- tion continuous medially with attachment of M. UP4- Eb2. TD comprises two, more-or-less continuous sec- tions, TPb3a, which attaches along anterior margin of Pb3 dorsal process, and TPb3p-UP4, which attach- es laterally to medial edges of Pb3 and UP4, abutting OD4 ventromedial margin; separable posterolaterally from SO by slight change in orientation of muscle fibers, but continuous posteriorly otherwise with SO. OD3 long, slender, originates on bony portion of Pb3 dorsal process and inserts on cartilaginous tip of Eb3 uncinate process; anomalous slip of muscle aris- es from origin of left-side OD3, passes through TPb3 and inserts on cartilaginous tip of Eb3 uncinate pro- cess. OD4 large, vertically oriented; ventromedially abutting TPb3; origin curving medially, attaching continuously from posterodorsal edge of Pb3, across medial cartilaginous process of Eb3, over Pb4 to base of Eb4; muscle twisting on itself dorsally (clockwise on right side, counterclockwise on left), almost di- visible into two parts, inserting on dorsomedial (un- cinate) process of Eb4. Remarks. Wiley (1976:fig. 13c) believed this mus- cle to be a transversus dorsalis posterior and made no mention of the muscle we treat as TPb3-UP4. Whatever the homology may be, we do not believe our OD4 is a homologue of a transversus dorsalis, which we consider to be a muscle joining the two sides of the gill arches. A vertically oriented OD4 is known only in albulids. OP dorsally on posterior surface of Eb4 medial to uncinate process, where it overlaps M. UP5-Cb4, ventrolaterally joining ER, below which Ad5 and SO appear to be confluent. OP’ slender, originating on dorsoposterior bony surface of Eb4 uncinate process (attachment covered by OD4) and inserting tendinously at ER. Remarks. Although present on both sides of our specimen, verification in other specimens is needed. M. UP4-Eb2 origin on UP4 continuous with me- dial end of insertion of LI3; insertion on cartilaginous cap of Eb2 dorsomedial (uncinate?) process. Remarks. Similar muscle, M. Pb4-Eb2 present in Aldrovandia affinis (Halosauridae). M. UP5-Cb4 dorsally with some fibers attaching to Eb4, but main portion of muscle continuing an- teriorly ventral to Eb4 and attaching to dorsal surface NUMBER 11 of UP5; posteroventrally attaching to anterior surface of fingerlike (fused Eb5) posterior cartilaginous end of Cb4. Remarks. M. UP5-Cb4 is known elsewhere only in Pterothrissus. Ad1-—3 absent. Ad4 large, prominent, on posterior surface of Eb4 levator process and posterior end of Cb4. Ad5 undifferentiated medially from SO, joined dorsomedially to ER; attaches dorsolaterally to car- tilaginous rod-like posterodistal end of Cb4 and ven- trally along lateral half of posterior surface of Cb5. Remarks. Albula lacks Eb5; however, the confor- mation of the cartilaginous distal end of Cb4 appears similar to the combined distal end of Cb4 and Eb5 in Pterothrissus, presumably indicating fusion of these two elements in Al/bula. The possibility of such fusion is indicated by our specimen of the anguillo- morph Notacanthus (Notacanthidae), in which EbS is autogenous on one side and, based on appearance, fused with Eb4 on the other. SO longitudinal layer appears to be restricted, at least dorsally, to isolated area ventral to TPb3 (not present immediately posterior to TPb3). RD absent. Additional remarks. SCL attaches mid-dorsally to small, autogenous cartilaginous ball, which attaches in turn to ventral surface of cartilaginous tip of pos- terior end of Bb3 (anterior end of RecV4 attaches to SCL; ObV3 attaches medially only to Hb3 flange continuous with SCL). A relatively larger, somewhat cone-shaped autogenous cartilage is present in a smaller cleared-and-stained specimen examined (au- togenous cartilaginous element attached to ventro- posterior tip of Bb3 is known otherwise only in some acanthomorphs, e.g., the melamphaid Poromitra cap- ito, q.V., which lacks SCL). The autogenous cartilage was not mentioned by Nelson (1969a, fig. 7a). TV4 is free from Cb5s, but is attached to ventral surface of Bb4. Pb1 is oriented horizontally—in line with Eb1. SPb2 is absent; cartilaginous SPb1 articulates with medial end of Eb1 (as opposed to ossified SPb1 in Elops and Megalops, both of which have cartilag- inous SPb2s). Pterothrissus gissu Hilgendorf, FRSKU 22120, 132 mm. Plate 18 Description. LE1 on Eb1 mid-dorsoposteriorly and on lateral surface of SPb1 dorsally ventral to origin of muscle. LE2 on anterior surface of Eb2 uncinate process, originates by long tendon (all other levators originate musculously). LE3 on anterior surface of Eb3 uncinate process. 35 LE4 on dorsal cartilaginous tip of Eb4 levator pro- cess. LP absent. LI1 on dorsal surface of Pb2 and medial surface of SPbl. Remarks. See remarks under LI1 in Aldrovandia (Halosauridae) description. LI2 on posteromedial surface of Pb3 ventral to OD4. LI3 on dorsolateral edges of UP4 and UPS. TD comprises TPb3 and TPb4. TPb3 divided into anterior and posterior sections by OD3, which orig- inates on Pb3 ventral to anterior section; anterior sec- tion attaches to anterolateralmost surface of Pb3 un- cinate process; posterior section attaches to uncinate process just ventral to anterior section and is contin- uous posteriorly by diagonal muscle strand with TPb4. TPb4 attaches to dorsomedial cartilaginous edge of Pb4 along line with OD4 origin. OD3 origin on dorsoanterior bony Pb3 surface be- low TPb3 anterior section; insertion on dorsoanter- iormost edge of tip of Eb3 uncinate process. OD4 large, vertically oriented; anteromedially, muscle originates continuously along bony dorso- medial margin of Pb3 and medial margin of Pb4 at junction of cartilage and bony UP4; dorsolaterally, muscle divides into two sections: posterior section inserts on Eb4 uncinate process and small AC at tip of process, anterior section extends posteriorly pass- ing ventral to insertion of posterior section and in- serts broadly on anterior surface of Eb4 levator pro- cess. Remarks. A vertically oriented OD4 is known only in albulids. OP dorsally on posteromedial surface of Eb4, ven- trally ending at ER. M. UP5-Cb4 dorsally with some fibers attaching to Eb4, but main portion of muscle continuing an- teriorly ventral to Eb4 and attaching to dorsal surface of UP5; ventrally attaching broadly to posterodistal end of Cb4 with minor attachment to Eb5 medial surface. Remarks. M. UP5-Cb4 is known elsewhere only in Albula. Ad1-—3 absent. Ad4 large, prominent, dorsally on posterior surface of Eb4 levator process, ventrally on Cb4 dorsopos- terior surface anterior to inner angle formed by Eb4- Cb4 joint. Ad5 on Cb4 posterodistal end, wraps medially around distal end of Cb5 and attaches to Cb5 pos- teroventral surface beginning at about mid-length, joined dorsomedially to ER and ventromedially by raphe with TVS. SO longitudinal muscle layer attaches anteriorly to Pb3. RD absent. 36 Additional remarks. SCL attached mid-dorsally to posteroventral cartilaginous tip of Bb3. TV4 free from Cb5s, but mid-dorsal surface attaches to Bb4 ventral surface. Small, autogenous ball of cartilage (AC) present between dorsalmost cartilaginous tips of Eb3 and Eb4 uncinate processes. Another AC at inner angle formed by Eb1-Cb1 joints, supports gill raker at joint. Large, more-or-less vertically oriented Eb5 attaches to posteromedialmost tip of Cb4. Eb5 appears to be represented by non-autogenous process at posteromedialmost end of Cb4 in A/bula. Pb1 car- tilaginous (ossified in Albula). NOTACANTHIDAE Notacanthus chemnitzi Bloch, USNM 214342, ca. 340 mm. Plate 19 Description. LE1 on mid-posterior edge of Eb1 just lateral to uncinate process. Remarks. Slender ligament, originating on skull, inserts on medial end of Eb1; similar ligament inserts on medial end of Pb2 in related Aldrovandia (Halo- sauridae). LE2 on Eb2 uncinate process. LE3 on cartilaginous tip of Eb3 uncinate process; muscle present and well developed only on left side of illustrated specimen (but presence indicated in il- lustration based on USNM 44246, examined in situ, in which LE3 is present on both sides). LE 4 absent (both specimens). LP absent. LI1 on dorsal surface of Pb2. LI2 on dorsal surface of posterolateral cartilagi- nous process of Pb3. LI3 absent (Pb4 absent). TD comprises three continuous portions that are demarcated only laterally: TEb2, TPb3, and TEb4. TEb2 attaches to anterior surface of medialmost end of Eb2, dividing laterally as it passes over anterolat- eral cartilaginous process of Pb3, to which a few strands of muscle also attach; TPb3 attaches to dor- solateral surface of Pb3; TEb4 attaches to postero- medial edge of Eb4 and is broadly continuous pos- teriorly with SO. OD3 absent. OD4 with split origin: on Pb3 dorsally ventral to TEb2, and on tip of Eb2 uncinate process; origins fuse posteriorly and insert on Eb4 dorsodistally. Remarks. Origin on Eb2 is unique. OP distinct, ribbon-like, originating dorsally on posteromedial surface of Eb4 and extending poster- oventrally, then curving anteriorly and joining con- tralateral OP in medial raphe, anterior point of which joins medial raphe of TV5. As such, OP does BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON not attach ventrally to a gill-arch element. ER is absent. Ad1-—3 present. Ad1 relatively large, inseparable from gill-filament muscle, which appears to overlie it dorsolaterally, but is continuous with it ventrome- dially. Ad2 and 3 well developed, with well-devel- oped gill-filament muscle portions dorsoanteriorly. Lateral portions of gill-filament muscles of Ad2 and 3 were destroyed during dissection and their full ex- tent is unknown (see also remarks under Ad1—3 in Oncorhynchus (Salmoniformes). Ad4 relatively large, dorsally on long dorsoposter- ior edge of Eb4, ventrally on much of posterodorsal surface of Cb4. Ad4’', small (smaller than Ad3), vertical muscle band, dorsally on anterodistal surface of Eb4, ven- trally on bony anterolateral surface of Cb4; appar- ently not associated with gill-filament muscle. Remarks. Ad4’ appears to be an autapomorphy and was not observed elsewhere in this study. It differs from Ad4 in attaching to the anterior surface of Cb4, whereas Ad4 in pre-acanthomorphs, except Poly- odon, attaches to the dorsoposterior surface of Cb4 just medial to internal angle formed byEb4-Cb4 joint. In Polyodon, Ad4 attaches to posterior surface of Eb4 and anterior surface of Cb4. Ad5 enveloped in tough CT (removed in Plate 19); on left side attaches dorsally to Eb5, which is at- tached to distalmost surfaces of Eb4 and Cb4; on right side attaches dorsally to cartilaginous Eb4 pro- cess that has shape and position of separate Eb5 on left side; relatively long and free as it wraps around Cb5 and attaches well medially on CbS. Remarks. In a cleared and stained specimen (USNM 214339), Eb5 is present on both sides and it is equally associated with the cartilaginous ends of Eb4 and Cb4. Cb5 bears simple gill rakers but no teeth. RD absent. SO longitudinal muscle layer present (not illus- trated), spongy, extends anteriorly beyond horizontal between medial ends of Ebl1s. Additional remarks. SCL absent. TV4 free from Cb5s, but attached to mid-anteroventral surface of Bb4. Pbl absent, apparently replaced functionally by elongate cartilaginous end of Eb1. Pharyngobranchial toothplates absent. Bb1 bears edentulous tooth plate mid-dorsally, not reported by McDowell (1973:132) in Notacanthus. Tough, slender ligament on antero- lateralmost edge of anterior cartilaginous end of Eb] (originates on ventral margin of opercular bone Mc- Dowell, 1973:132). Slender ligament (not illustrated) originates on skull separately from clustered origins of LEs and LIs and inserts on Pb3 medial to Pb3 origin of OD4. NUMBER 11 HALOSAURIDAE Aldrovandia affinis (Giinther), USNM 319707, ca. 425 mm TL. Plate 20 Description. LE1 on base of Eb! uncinate process. LE 2 on anterior surface of Eb2 uncinate process along and ventral to posterior cartilaginous tip of pro- cess, which also has separate cartilaginous dorsoan- terior tip (see M. Pb4-Eb2). LE3 on cartilaginous tip of Eb3 uncinate process at attachment of Eb3 branch of OD3—4. LE4 on cartilaginous tip of Eb4 levator process. Remarks. Slender ligament, originating on skull, also inserts on levator process. Similar ligament also present in closely related Notacanthus (Notacanthi- dae), which lacks LE4. LP absent. LI1 inserts over most of dorsoanterior surface of Pb2 and on cartilaginous tip of Ebl uncinate process. Remarks. Long, ribbon-like ligament, originating on skull, inserts on medial tip of Pb2. Similar liga- ment, inserting, however, on medial end of Eb1, pres- ent in related Notacanthus (Notacanthidae). The unusual partial insertion of LI1 to include the medial end of Ebl may be the result of the loss of SPb1. In some other elopomorphs (Elops, Megalops, Albula, and Pterothrissus), SPb1 is present and at- taches on the dorsomedial end of Eb1 (which pre- sumably divides to become the uncinate process), and LI1 expands ventrally to almost completely envelop SPb1 in its insertion, which is otherwise on Pb2. The insertion on SPb1 just misses including the edge of Eb1, so that it would be no great change for LI1 to have its insertion extended (as in Aldrovandia) to in- clude the tip of the uncinate process if SPbl were lost. Compare LI1 in Aldrovandia with LI1 in Elops, Megalops, Albula, and Pterothrissus. LI2 on dorsolateral edge of Pb3. LI3 broad based, anteriorly on cartilaginous Pb4 at junction of cartilage with UP4, continuing poste- riorly onto edge of UPS, the anterior edge of which lies under posterior end of Pb4. M. Pb4-UP5-Eb?2 origin on posterolateral corner of Pb4 continuous anteriorly with insertion of LI3, pos- teriorly on dorsolateral surface of UPS (not illustrat- ed); insertion on osseous dorsal edge of Eb2 uncinate process. TD comprising continuous TPb3-Pb4-Eb4 with ad- ditional, roughly trapezoidal TPb3 section arising mid-dorsally from TPb4 area; anterior fibers of sep- arate section, attaching to and wrapping anteriorly around cartilage-tipped lateral process of Pb3; TD completely undifferentiated posteriorly from SO. OD3-4 origin broadly continuous on posterior edge of Pb3 and anterior end of Pb4, dividing pos- 37 teriorly with one section inserting on Eb3 uncinate process and other section inserting on Eb4 anterolat- erally. OP dorsally on ventral surface and posteromedial edge of Eb4; ventrally on small Eb5 attached to Cb4 and dorsodistal cartilaginous end of Cb4. ER absent. Ad1-—3 absent. Ad4 on Eb4 dorsoposteriorly and dorsoposterior bony surface of Cb4. Ad5 dorsally on Cb4 distally and tiny Eb5, which articulates ventrally with Cb4 and is attached by lig- ament to Eb4 levator process; free ventrally for short distance posterior to attachment on mid-ventromedial surface of Cb5. RD absent. SO longitudinal muscle layer (not illustrated) ex- tends anteriorly at least to horizontal between medial ends of Ebls. SO fibers also attach to UPS, which is ventral to, but separated by SO and OP muscle fibers from medial arm of Eb4. Additional remarks. SCL absent. TV4 free from Cb5s. Ligament connects posterior bony surface of Eb2 uncinate process to anterodistal cartilaginous end of Eb3; another connects posterior bony surface of Eb3 uncinate process to anterodistal cartilaginous end of Eb4; another attaches tip of Eb4 uncinate process to Eb5, which, unusually, attaches to Cb4 rather than Eb4, and continues dorsoposteriorly. Ribbon-like lig- ament inserts on cartilaginous tip of medial process of Pb2; its origin was not recorded, but a similar ligament in Notacanthus attaches to the cranium. Pb2 lacks tooth plate. CONGRIDAE Conger cinereus Riippell, USNM 115969, 345 mm. Plate 21 Description. Remarks. Nelson (1967c) described and illustrated the gill-arch musculature of Conger marginatus (cur- rently considered a junior synonym of C. cinereus). Insofar as Nelson’s and our findings can be com- pared, there is much similarity, but some differences mentioned below suggest further study is indicated. Although there is a question about the homology of the pharyngeal tooth plates, we use Nelson’s (1966b) terminology for convenience. LEI tendinously on Eb! mid-dorsoposteriorly. LE2 tendinously on Eb2 mid-dorsoposteriorly. LE3 on cartilaginous tips of joined Eb3 and Eb4 uncinate processes. LE4 finely, tendinously on Eb4 mid-dorsally. LP absent. LI1 on cartilaginous medial end of Pb2. LI2 on Pb3 laterally, UP3 dorsolaterally, and UP4 dorsoanterolaterally. LI3 absent (Pb4 absent). TD thin, sheet-like, overlies SO posteriorly, but scarcely separable from SO, comprises TPb3-Eb4, at- taching dorsally on Pb3 at and along line of attach- ment of OD4, and on Eb4 anteromedially. OD3 absent. OD4 origin on Pb3 medial to LI2 and lateral to line of attachment of TPb3-Eb4, insertion along en- tire length of Eb4 uncinate process. Remarks. The tips of the uncinate processes of Eb3 and Eb4 are tightly bound together, and OD4 is pos- sibly minutely attached to the cartilaginous tip of Eb3 uncinate process. In contrast to our findings, Nelson (1967c:349) reported that the superior oblique (= our OD) in C. marginata attaches Pb3 only to Eb3. If it is determined that the OD of C. cinereus lacks a “‘true’’ connection to Eb3, the lack is autapomorphic. An OD attachment to Eb3 defines the Teleostei, and there are exceedingly few other pre-acanthomorph taxa that lack the attachment (i.e., Notacanthus, Gon- ostoma, Maurolicus). RecD2 on ventroanteriormost surface of Eb2 and mid-posterior edge of Eb]. RecD3 on ventroanteriormost surface of Eb3 and mid-posterior edge of Eb2. M. Pb2-Eb1 on Pb2 dorsoposteriorly and Eb] me- dial cartilaginous tip. Remarks. Nelson (1967c:349 and fig. 2) termed this muscle obliquus inferior accessorius. M. UP4-Eb4 on UP4 dorsoposterolaterally, poste- riorly becoming broad, thin CT sheet, which attaches to ventrolateral margin of Eb4 (at angle of Eb4-Cb4 joint) and extends medially, becoming incorporated in SO. Remarks. Nelson (1967c:349) termed this muscle “retractor dorsalis.” Although appropriately named functionally, the muscle does not originate on the vertebral column and cannot be considered even ho- moplastically as RD. Nelson (1966a:123), however, reported that [exceptionally] among the several eels he examined, which also included Conger and An- guilla, but not Synaphobranchus, ““RD” in the [spe- cialized] Muraeninae, has become attached second- arily to the vertebrae. OP strap of muscle attaching dorsally to Eb4 pos- teromedially, slightly distinguished by denser fibers from SO medially, ending ventrally at ER, which is joined ventrally by Ad5; ventromedially partially overlain by SO fibers. Ad1-—3 absent. Ad4 dorsally on Eb4 posteroventral surface and ventrally on Cb4 posterodorsal surface. Ad5 joins distal end of Cb5 to bony sub-distal end of Cb4, dorsomedially joining ER. RD absent. See SO. SO thin longitudinal muscle layer surrounds eso- phagous and attaches to toothplates and Pb3. BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON Additional remarks. SCL absent. TV4 free from Cb5s. Uncinate process present on Eb4. Pb1 absent. ANGUILLIDAE Anguilla rostrata (Lesueur), USNM 340815, 228 mm, USNM 190998, 328 mm. Plate 22 Description. Remarks. All levators extend lateral to LI1, except LE4, which passes medial to LI1. LI1 extends dor- solaterally, all other levators extend anteriorly. Nel- son (1967c) described and illustrated the gill-arch musculature of Anguilla rostrata. There is much sim- ilarity between his and our findings, but a few dif- ferences exist. Although there is a question about the homology of the pharyngeal tooth plates, we use Nelson’s (1966b) terminology for convenience. LEI! on dorsal surface of Eb] about half length distally. LE2 on dorsal surface of Eb2 about half length distally. LE3 finely tendinously on dorsoanterior surface of bony Eb3 uncinate process, just ventral to OD3 in- sertion. LE4 very slender, on Eb4 mid-dorsally just lateral to bony uncinate process, which is tightly joined to bony Eb3 uncinate process. LP absent. LI1 broad, thin, on Pb2 dorsal surface; insertion just medial to and paralleling M. Pb2-Eb1. LI2 on UP3 dorsolaterally. Remarks. Nelson (1967c:349) did not describe the insertion of LI2 in either Anguilla or Conger, but reported that the muscles of Anguilla are “rather sim- ilar’ to those of Conger. In Conger, LI2 inserts on Pb3, UP3, and UP4. LIB absent (Pb4 absent). TD comprises TPb3-UP3-UP4-Eb4, on medial margin of Pb3, just barely on posteromedial edge of UP3, on dorsomedial surface of UP4, continuing un- interrupted posteriorly as SO, and medial tip of Eb4. OD3 origin on dorsoanterior end of Pb3, continu- ous with OD4 origin, insertion on bony Eb3 uncinate process dorsal to LE3. Remarks. Nelson (1967c:fig. 4) indicated that OD4 attaches to a cartilage tipped Eb4 levator process. In both our specimens, the tips of the levator processes on Eb3 and Eb4 are bony and although the two pro- cesses are tightly bound together by CT, OD3 attach- es only to Eb3. OD4 origin on Pb3 dorsoposteriorly, insertion on Eb4 anteromedial edge. RecD1 very long, on Eb! anterodistal tip, anteri- orly ends tendinously in CT of roof of mouth, par- allels LEs. NUMBER 11 Remarks. Ordinarily we would consider this mus- cle as LE1’, but its serial position, like that of RecD2 and RecD3, supports Nelson’s (1967c) assignment (our RecD is the same as Nelson’s obliquus inferior accessorius). With the questionable exception of the osteoglossiform Gymnarchus, we know of no other genus in which RecD1 occurs. RecD2 short, posteriorly on anterodistal end of Eb2, joined along anterior edge by well-developed portion of GFM, on Eb! posterolaterally posterior to LE1 insertion. RecD3, short, on anterodistal end of Eb3, joined along anterior edge by well-developed portion of GFM; on Eb2 posterolaterally posterior to LE2 in- sertion. M. Pb2-Eb1 small, hidden by LI1, anteriorly on medial tip of Eb1, posteriorly on lateral edge of Pb2, parallels lateral side of LI1 insertion. OP dorsally on Eb4 dorsomedially, ventrally joins ER with SO and Ad5, not continuous below ER, which extends tendinously laterally and attaches to Cb4 distally together with dorsal end of Ad5. Ad1—3 absent. Ad4 dorsally on posterior edge of Eb4, ventrally on Cb4 anterior to Eb4-Cb4 joint. Ad5 dorsally on ER with lateral tendinous attach- ment to Cb4 posterodistal end; ventrally on poster- odistal half of Cb5. RD absent. See SO below. SO longitudinal muscle layer surrounds esopha- gous, dorsally fibers attach to tooth plates and medial margin of Pb3. Remarks. Nelson (1967c) considered this portion of the longitudinal muscle layer as RD. Additional remarks. SCL absent. TV4 free from Cb5s. Pb1 absent. SYNAPHOBRANCHIDAE Synaphobranchus sp., USNM 316662, ca. 340 mm. Plate 23 Remarks. Although there is a question about the homology of the pharyngeal tooth plates, we use Nel- son’s (1966b) terminology for convenience. Description. LE1 broadly on Eb! dorsally beginning at about mid-length of bone and extending laterally to small posterodistal flange-like process. LE2 on small posterodistal Eb2 flange-like pro- cess. LE3 on posterior flange-like Eb3 process at OD3-— 4 insertion. LE4 on Eb4 dorsally slightly distal to mid-length. LP absent. LI! on Pb2 anterolaterally. LI2 inserts posteriorly on Pb3 dorsoposteriorly, 39 UP3 dorsomedially, and on anterodorsal edge of UP4, and inserts anteriorly on dorsoposterior edge of Eb2. LI3 absent (Pb4 absent). TD comprises undifferentiated layer of muscle, TPb3-UP3-UP4, attaching laterally to Pb3, UP3 dor- somedially, and on most of dorsal surface of UP4. OD3-—4 origin on anterior end of Pb3, insertion on Eb3 dorsally at and posterior to insertion of LE3 and on anteromedial edge of Eb4 beginning well lateral to joint with Pb3 and ending just proximal to point opposite insertion of LE3. M. Pb2-Eb2 attached to ventral surfaces of Pb2 and anterior end of Eb2. RecD2 on proximal two-thirds of anterior edge of Eb2 and on dorsomedial tip of Eb1. OP questionably present as strap of muscle attach- ing dorsally to posteromedial surface of UP4 and ending ventrally at ER, which is at level of Cb4 and extends laterally as slender tendon to posterodistal end of Cb5; inseparable medially, and ventral to ER, from SO. Remarks. The shift of the dorsal attachment of OP from Eb4 to UP4 is unique among the taxa we ex- amined. Nelson (1967c) considered this muscle sec- tion as RD. Ad1-—3 absent. Ad4, very large, dorsally on most of Eb4 posterior surface, ventrally on Cb4 dorsal surface anterior to internal angle formed by Eb4-Cb4 joint. Ad5 absent. RD absent. See SO. SO thin, longitudinal muscle layer extending an- teriorly to horizontal between anterior tips of Pb3s, surrounding esophagous, attaching to toothplates and Pb3. Additional remarks. SCL absent. TV4 free from Cb5s. Pb2 reduced, cartilaginous. Pb1 absent. Clupeomorpha DENTICIPITIDAE Denticeps clupeoides Clausen, USNM 358795, 36.4 mm, and BMNH uncataloged, 2 specimens, 30.0— 30.6 mm. Plate 24 Description. Remarks. All levators except LI1 extend anteriorly, horizontally, at almost 180° angle and form a thin, posteriorly concave fan of overlapping muscles that is closely applied to, and originates on, the convex surface of the otic bulla. LI1 extends laterally at about 90° angle from other levators, is medially con- cave, and conforms with the laterally convex surface of the otic bulla on which it originates. 40 LE1 on minute cartilaginous tip of poorly devel- oped Eb! uncinate process. LE2 on minute cartilaginous tip of poorly devel- oped Eb2 uncinate process. LE3 on minute cartilaginous tip of well-developed Eb3 uncinate process. LE4 on cartilaginous dorsomedial end of Eb4 un- cinate process, there meeting OD4 insertion. LP absent. Remarks. Greenwood and Lauder (1981:226) were uncertain that LP was absent in Denticeps because of the small size (ca. 50 mm SL) of their specimens. We agree that size was a problem for dissection and illustration, but are confident that LP is absent. LI1 on Pb2 posterolaterally (see also remarks at beginning of description above). LI2 on Pb3 posterolaterally. LI3 on cartilaginous Pb4 (not illustrated) postero- laterally. TD comprises TEb2 and TPb3-Pb4-Eb4. TEb2 at- taches to Eb2 dorsomedially. TPb3-Pb4-Eb4 on Pb3 and Pb4 at and along part of OD4 origin, becoming wider posteriorly and attaching on dorsal surface of Eb4 medial to uncinate process. OD3 origin on Pb3 dorsal surface ventral to TEb2, continuous with OD4 origin, and insertion on Eb3 uncinate process. OD4 originates on Pb3 dorsoposterior surface ven- tral to TEb2, and on Pb4 dorsal surface at and medial to Pb4 portion of TPb3-Pb4-Eb4, and inserts on an- teromedialmost surface of Eb4 uncinate process. OP slender strap of muscle, dorsally on postero- medial edge of Eb4 uncinate process, overlapping much of Ad4 posterior surface, ending ventrally at ER, which is dorsal termination of major portion of Ad5 (unless continuation is interpreted as OP fused laterally with ADS). Ad1-—3 absent. Ad4 dorsally on dorsoposterior edge of Eb4 un- cinate process continuing to distal end of Eb4, ven- trally on Cb4 dorsal edge anterior to internal angle formed by Eb4-Cb4 joint. Ad5 dorsally on posterior surface of Eb5 and at ER ventral to OP, medially continuous with SO, ven- trally on Cb5 posterior surface. SO longitudinal muscle layer thickest dorsally and ventrally; anterior extent undetermined. RDs absent. Additional remarks. SCL absent. TV4 free from Cb5s. Eb5 present (not previously reported, although possibly implied by discussions in Johnson and Pat- terson, 1996). A single large UP present. PRISTIGASTERIDAE Ilisha africana (Bloch), USNM 357405, 120 mm. Plate 25 BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON Description. LE1 slender, weak, on dorsal bony edge of Eb1 uncinate process just distal to cartilaginous tip. LE2 slender, weak, on dorsal bony edge of Eb2 uncinate process just distal to cartilaginous tip. LE3 on tip of all bony Eb3 uncinate process, at and anteromedial to LE3’ insertion, larger and angled dorsoanteriorly about 10° lower than LE3’. LE3’ on tip of all bony Eb3 uncinate process at and posterior to LE3 insertion, slenderer and angled dorsoanteriorly about 10° higher than LE3. LE4 broad, thin, on bony dorsoposterior edge of Eb4* coincident with posterior edge of OD4 inser- tion; questionably comprising two very thin sections, anterior and posterior, on right side, but not on left side. LP broad, thin, on dorsodistal cartilaginous edge of Eb4*, commencing at lateralmost edge of LE4 in- sertion, partially coincident with Ad4 dorsal attach- ment. LI1 on Pb2 dorsal surface medial to uncinate pro- cess. LI2 on Pb3 dorsoposteriorly. LI3 on Pb4 (not illustrated) dorsoposteriorly. TD comprises TPb3-Pb4-Eb4™, slightly partitioned laterally at posterolateral origin of OD4 on Pb3 and at attachment to medial end of Eb4*; on right side only, joining ER with OP ventrally; continuation pos- teriorly with SO marked by change in muscle fiber direction. OD3 origin on Pb3 dorsal surface anterior to un- cinate process and on medial edge of uncinate pro- cess; insertion on dorsomedial edge of Eb3 bony un- cinate process. OD4 origin begins on Pb3 among TD fibers, pass- ing dorsal to most of Pb3 attachment of TD, and continuing on Pb4 dorsoanteriorly; insertion broadly dorsally on Eb4* anterior surface ventral to LE4 in- sertion. OP strap of muscle originating on Eb4* postero- medial to LE4 insertion, and extending ventrally to ER; undifferentiated from SO ventral to ER. Ad1-—3 absent. Ad4 dorsoposteriorly on Eb4*, ventrally on Cb4 at anterior angle formed by Eb4*-Cb4 joint. Ad5 dorsally on distal end of Eb4*, ventrally on Cb5 distal end, inseparable mid-medially from SO. RD absent. SO longitudinal muscle layer spongy, circumeso- phageal, thickest dorsally, extending dorsoanteriorly to below anterior end of TD, but becoming extremely attenuated anterior to horizontal between medial ends of Eb4*s. Additional remarks. SCL absent. TV4 free from Cb5s, attached dorsally to ventral surface of Bb3. Tiny MSPb1 anterior to anterior end of each Eb1. GC attached ventrally to anterior ends of Pb2s, which NUMBER 11 are slightly dorsal to anterior ends of Pb3s (see also Additional remarks under Cetengraulis, and Chanos). ENGRAULIDAE Cetengraulis edentulus (Cuvier), USNM 186377, 2 specimens, 118-124 mm. Plates 26.1, 26.2 Description. Remarks. All levators except LE] pass medial to LI1. LE2—4 extend horizontally anteriorly at approx- imately 180° from insertions, LE4’ extends dorsoan- teriorly about 15° from insertion, LI is more-or-less vertical, and LI2 and LI3 are angled dorsoanteriorly about 45°. LE1 thin, slender, on bony process at distal edge of base of Eb1 uncinate process; originates somewhat more anteriorly on skull than other LEs. LE2 on tip of Eb2 uncinate process. LE3 tendinously on Eb3 uncinate process ventral to OD3 insertion. LE4 on medial edge of Eb4* levator process dorsal to insertion of OD4; originates as long tendon. LP slender, membranously attached along lateral edge of LE4, inserting with LE4 insertion; originates as very long, slender tendon. LI1 broad, thin, convex medially (medial surface conforming with surface of cranium), inserting broadly on dorsal surface of Pb2. LI2 on posterior end of Pb3 dorsal surface. LI3 inserting as long, slender tendon on medial edge of Pb4 dorsoanteriorly. TD comprises TPb3, TPb4, and TEb4*. TPb3 sep- arate muscle attaching to dorsolateral edge of Pb3 beginning just posterior to base of uncinate process, continuous posteriorly by diagonal strand of muscle inserting on Pb4 anteriorly in one specimen and at mid-length on other. Anteriorly, TPb4 comprises loose strands of fibers, which attach on each Pb4 dor- somedially, and is continuous posteriorly with slen- der TEb4*, which attaches to ventromedial ends of Eb4*s and is continuous posteriorly with SO+EO. OD3 slender, originates on Pb3 uncinate process ventromedial to dorsal tip and inserts tendinously on Eb3 uncinate process just ventral to tip and dorsal to LE3. OD4 very weak, almost thread-like, originates by short, fine tendon attached to tiny (almost pinpoint) bony area on dorsomedial edge of otherwise cartilag- inous Pb4; inserts on ventromedial edge of Eb4*. OP unclear if absent or included in complex sur- face muscles on EO. Ad1-3 absent. Ad4 two widely separated dorsal attachments pos- teriorly on Eb4* levator process, uniting ventrally in slender attachment (not visible in lateral view) on Cb4 at innermost angle formed by Eb4*-Cb4 joint. 41 Ad5 questionably included in complex surface muscles arising from EO and Cb5 and attaching an- terolaterally to Eb4*. Remarks. There is a posteroventral, rod-like car- tilaginous continuation of the dorsodistal end of Eb4* that articulates ventrally with the distal end of CbS. Nelson (1967d:fig. 2j) indicated that this extension represents a partially fused Eb5 in Engraulis and we agree. RDs absent. SO longitudinal muscle layer spongy, circumeso- phageal, thickest dorsally, becoming extremely atten- uated anterior to EO, and extending between UP4s to below anterior end of TPb4. Additional remarks. SCL absent. TV4 anterior to Cb5s (condition relative to Bb complex not exam- ined). Elongate anterior cartilaginous tips of Eb1s not segmented as in Coilia. UP4 edentate, extending dor- sally as thin plate. Short, cylindrical muscle-like lig- ament joining ventroposterior surface of Ebl unci- nate process to ventrolateral surface of Pb2 uncinate process. Another similar ligament joining ventropos- terior surface of Eb2 uncinate process to ventrolateral surface of Pb3 uncinate process. The area between Pb2s and anterior ends of Pb3s is covered by a leathery CT sheet. The sheet attaches to the dorsalmost ends of the Pb2 uncinate processes, but the main part of the sheet lies well ventral to the dorsalmost ends, at about the level of the anterior ends of the Pb3s. The sheet apparently conforms with the ventral surface of the cranium. Mid-ventrally, the sheet incorporates a small, seed-shaped cartilage, which Di Dario (2002:500) identifies as GC, an ele- ment he first named and described in a printed jour- nal (but see GC in Abbreviations and Definitions sec- tion): ““[the engrauloid] Cetengraulis ... has a typi- cal gongyloid cartilage in all aspects [our italics] ex- cept that it is markedly rod-shaped ... this shape is hypothesized as convergent to that of the [clupeid second] mediopharyngobranchial,’” which Di Dario found in three of the 20 clupeid genera he examined. Di Dario followed up by allowing that in the ‘‘ab- sence of simultaneous occurrence of the gongyloid cartilage and the second mediopharyngobranchial in any individual, the possibility of their homology can- not be conclusively discarded.”’ We find that the pu- tative GC in our specimen of Cetengraulis (if present, it was destroyed during dissection in the other spec- imen) is quite dissimilar in shape, relative size, po- sition, and in being incorporated in a CT sheet, from that of other clupeoid taxa (engrauloids, pristigaster- oids) as described and illustrated by Di Dario and that we have examined. Because most (8 of 10) of the other engrauloids Di Dario (2002: table 1) ex- amined have a GC that is typical and the other two lack GC, we think that the element in question in Cetengraulis is probably a modified (reduced) GC, 42 intermediate between character states of “full” de- velopment and complete absence. Coilia neglecta Whitehead, USNM 357380, 2 spec- imens, 138-161 mm. Plate 27 Description. Remarks. All levators besides LI] pass medial to LI1. LI1 is more-or-less vertical, following around the convex surface of the prootic. LI2 and LI3 pass dorsoanteriorly at about 60° and 45° angles and are “laminated” against the dorsomedial surface of LI1. LE2—4 extend anteriorly at almost 180° from their insertions. LEI absent. LE2, very fine, narrowly on bony Eb2 uncinate process just distal to medial head of Eb2, which ar- ticulates with Pb2 uncinate process. LE3 on dorsoanterior surface of Eb3 uncinate pro- cess just ventral to cartilaginous tip. LE4 narrowly on dorsalmost edge of Eb4* just dorsal to OD4 insertion. LP absent. LI1 very broad, thin, on Pb2 dorsomedial margin, which is tightly bound and ventral to Pb3 slender anterior process. LI2 thin, on Pb3 (not shown) dorsal surface pos- terolaterally. LI3 thin, on Pb4 dorsolaterally. TD comprises TPb3, TPb4-Eb3, TEb4*. TPb3, most distinct of the three, is broadly on Pb3 postero- laterally anteroventral to OD3 origin, and is posteri- orly dorsal to OD4 origin and continuous by loose diagonal strands of muscle with TPb4-Eb3. TPb4- Eb3 is on Pb4 dorsolaterally just medial to LI3 in- sertion and medial end of Eb3 (in smaller specimen, there is slight separation of Pb4 and Eb3 sections, hence these could be accorded separate names). TPb4-Eb3 is broadly continuous posteriorly with TEb4*, which is on Eb4* dorsalmost edge ventral to OD4 insertion. OD3 on bony dorsal surface of Pb3 uncinate pro- cess, extending a little anteriorly onto slender anterior process of Pb3, and is anteriorly dorsal to TPb3 at- tachment; insertion on dorsomedial edge of Eb3 un- cinate process. OD4 origin mainly on Pb3 dorsoposteriorly, with few strands on dorsoanteriormost surface of Pb4; in- sertion on dorsomedialmost edge of Eb4* ventral to LE4 insertion. OP consists of strands of muscle attaching dorsally to Eb4* posteromedial surface, ventrally joining ER at mid-level of cartilaginous distal end of Eb4*; ven- tral to ER, apparently comprises Ad5 and SO. Ad1-—3 absent. BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON Ad4 on Eb4* dorsoposteriorly lateral to OP, dor- sally with broad medial section separated by space from slender lateral section (nerve passes through space), ventrally on Cb4 dorsal surface medial to Eb4*-Cb4 joint. Ad5 continuous medially with SO, dorsolaterally on distal cartilaginous end of Eb4*, ventrally on pos- terodistal end of Cb5, medially not separable from SO. RD absent. SO longitudinal muscle layer thin, spongy, circu- mesophageal, evenly distributed, extending dorsoan- teriorly much attenuated at least to anterior end of TPb3 (possibly continuing over CT anteriorly). Additional remarks. SCL absent. TV4 free from Cb5s, but slight attachment dorsally to ventral sur- face of cartilaginous Bb copula. There is an autoge- nous cartilage, (Plate 27A, MPb1) anterior to carti- laginous anterior tip of each Eb1 (see additional re- marks under Dussumieria, Clupeidae). Ventral end of Pbl attaches to dorsal surface of this cartilage. GC absent. CHIROCENTRIDAE Chirocentrus dorab (Forsskal), USNM 283241, 195 mm; USNM 283242, 144 mm. Plates 28.1, 28.2 Description. Remarks. All levators except LE1 and the muscle sheath termed LP, extend medial to LI1. LI1 is angled dorsoanteriorly at about 50—60°; the LP sheath com- prises a perpendicular pyramidal section overlying or meshed with a horizontal section; the other levators are all angled horizontally at about 180°. LE1 on tip of Eb! uncinate process. LE2 on tip of Eb2 uncinate process. LE3 on medial edge of distal tip of Eb3 uncinate process. LE4 on dorsomedialmost edge of Eb4* levator process. LP thin, fanlike sheet of muscle fibers attached along much of lateral edge of LE4 and dorsomedial edge of Eb4*, dorsalmost fibers almost perpendicular to those of LE4. Remarks. Greenwood and Lauder (1981:226) be- lieved LP was absent in C. dorab “unless, atypically, it is closely associated with the 4th external levator ... We find LP is typically closely associated with LE4 in clupeoids and characoids. LI1 posteromedially on Pb2 dorsal surface. LI2 absent. LI3 on lateral edge of Pb4 lateral to TPb4; appears to have two slightly separate origins. TD comprises three parts: TPb3, TPb4, TEb4*. TPb3 on Pb3 uncinate process and Pb3 dorsolaterally posterior to uncinate process, broadest centrally, dor- NUMBER 11 sal to OD4 origin, continuous posteriorly by few muscle strands with TPb4 in smaller specimen, not continuous in larger specimen. TPb4 broad dorsally, narrow at attachment to Pb4 just medial to LI3 in- sertion, continuous by diagonal muscle strap with TEb4*. TEb4* on Eb4* medially ventral to OP, con- tinuous with SO posteriorly, but distinguishable from SO by change in direction of muscle fibers. OD3 very small, slender, origin on Pb3 uncinate process just lateral to TPb3 attachment, insertion on posterior surface of Eb3 uncinate process ventral to LE3 insertion. OD4 massive, origin beginning on broad dorso- posterior bony surface of Pb3 and extending onto dorsoanterior surface of Pb4, insertion on anterome- dial surface of Eb4* levator process, joining raphe with OP. OP a muscle strap and filaments attaching poste- riorly on dorsomedial surface of Eb4* levator process and medial arm of Eb4*, joins raphe dorsally with OD4, inserting ventromedially among SO fibers and ventrolaterally on posterodistal surface of Eb4*. Ad1-—3 absent. Ad4 dorsomedially on posterior surface of Eb4* levator process, ventrally narrowly on Cb4 dorsal surface just medial to inner angle formed by Eb4*- Cb4 joint. Ad5 dorsally, narrowly on Eb4* ventrodistally; ventrally, broadly on posterior surface of Cb5; com- pletely separate from SO medially. RD absent. SO longitudinal muscle layer circumesophageal, evenly distributed, extending dorsoanteriorly as sheet to below TPb4, thence as sparse filaments to below posterior end of TPb3. Additional remarks. SCL absent. TV4 free from Cb5s. ER absent (a chirocentrid apomorphy?). CLUPEIDAE Dussumieria acuta Valenciennes, USNM 296827, 3 specimens, 121—123 mm. Plate 29 Additional material. @ = Clupea harengus Linnaeus, USNM 325930, 106 mm; USNM 349799, 112 mm. Description. LE1 on cartilaginous tip of Eb1 uncinate process. LE2 on cartilaginous tip of Eb2 uncinate process. LE3 on cartilaginous tip of Eb3 uncinate process together with LE3’ insertion, angled dorsoanteriorly similarly to LE2, attaching to cranium near LE2 at- tachment. LE3’ inserting on cartilaginous tip of Eb3 uncinate process together with LE3 insertion, almost horizon- 43 tal, extending anteriorly medial to LE1-—3, attaching to cranium near cranial attachment of Pb1. LE4 narrowly inserted on dorsoanteriormost edge of broad Eb4 levator process, continuous ventrolat- erally with abruptly thinner, sheet-like LP, which is also distinguishable by abrupt change in inclination of muscle fibers. @ On Eb4*; gradual change in mus- cle fiber inclination between LE4 and continuous, thin LP. LP thin, sheetlike, inserted broadly along dorsal- most edge of Eb4 levator process (lateral to insertion of LE4, but posterior given orientation of Eb4) and Eb5, continuous anteroventrally with LE4, but distin- guishable by thinness and abrupt change in inclina- tion of muscle fibers. @ broadly on Eb4*, continuous with LE4 along most of lateral edge of LE4, mainly distinguishable from LE4 by abrupt thinness of mus- cle-fiber sheet. Remarks. Winterbottom (1974b:252) discussed LE4 in clupeids concluding that it was a “moot point” whether the sheet of thin muscle continuing posteriorly from LE4 could be interpreted solely as LE4 or as LP. If not as LP, he believed that LP could have evolved from a clupeid-like condition. Winter- bottom (1974b:fig. 24) provided a generalized illus- tration of the levators in Clupea harnengus and did not differentiate an LP. Greenwood and Lauder (1981:215) disagreed with Winterbottom’s interpre- tation of LE4 in clupeids, **. . . we would identify the usually thin, sheet-like but somewhat expanded mus- cle lying ventral to [the protractor pectoralis] as the levator posterior muscle and not, as he does, a muscle composed entirely of the expanded 4th levator exter- nus ... even in Clupea harengus.”’ Greenwood and Lauder believed that, ““Winterbottom included the posterior levator, the 4th levator externus, and some non-muscular tissue lying above and between these muscles, in the muscle he identified as the 4th levator externus.”’ Our observations on Dussumieria acuta support Greenwood and Lauder’s general interpreta- tion of LE4 and LP in clupeids, but appear to support Winterbottom’s for C. harengus, in which the differ- entiation, other than a change in thickness, is unap- parent. Among pre-acanthomorphs, LP (in any form) is present only in otocephalans, and the muscle cannot be considered homologous with that of acantho- morphs. LI1 on Pb2 dorsal surface posteromedially. @ On medial margin of Pb2. LI2 on posterodorsal surface of Pb3, attaching to Eb2 uncinate process as muscle extends anterodor- sally to origin; free from Pb1; joined by raphe with LI3 on Eb2 uncinate process (see LI3). @ At origin, attaches jointly to Pb1 and cranium; completely free from Eb2 and LI3. LI3 on Pb4 dorsolaterally and dorsal edge of UP4 44 (not illustrated), fans out broadly anterodorsally, mid- anteriorly forms partial raphe with LI2 at attachment to Eb2 uncinate process. @ On dorsal surface of car- tilaginous Pb4 (UP4 absent) at and medial to TD at- tachment to Pb4. TD comprises TPb3-Pb4 and TEb4. TPb3-Pb4, ex- tensive, begins anteriorly at about mid-length of Pb3s, continues posteriorly along Pb3s and Pb4s me- dial to, and continuous by raphe with, OD3 and OD4 origins, continuous posteriorly by diagonal muscle strands with TEb4. TEb4 on Eb4 dorsomedially, con- tinuous posteriorly with SO by diagonal muscle strands. @ TD comprises TPb3 and TPb4-Eb4*. TPb3 short, begins just posterior to Pb3 uncinate process (well posterior to OD3origin), overlaps OD4 origin, posteriorly continuous by sparse muscle fibers with TPb4-Eb4*. TPb4-Eb4* attaches to Pb4 posterolat- erally and ventromedial edge of Eb4* levator pro- cess, undifferentiated posteriorly from SO. OD3 short, origin on Pb3 uncinate process (which joins Eb2 uncinate process), insertion on Eb3 unci- nate process ventral to insertions of LE3 and LE3’. @ Long, origin on Pb3 anterior to uncinate process, fibers attach to Eb2 uncinate process as muscle pass- es posteriorly to insertion on Eb3 uncinate process. OD4 origin on lateral surface of Pb3, medially joining raphe with TPb3-Pb4, faning out posteriorly and inserting on dorsoanterior surface of Eb4 levator process, obscures insertion of OD4’ on Eb4. @ Origin on posterolateralmost edge of Pb3, insertion on an- terodorsalmost medial edge of Eb4* levator process; raphe absent. OD4’ origin at junction of Pb4 and UP4, extends posteriorly mostly ventral to OD4, fans out posteri- orly and inserts along broad anterior surface of Eb4 levator process medial to OD4 insertion. @ Origin on Pb4 dorsoposteriorly, extends posteriorly (without fanning out) ventral to OD4 and inserts on antero- medial edge of Eb4* levator process just ventral to OD4; OD4 and OD4' more or less fused at Eb4*. OP (not labelled), questionable area of muscle, continuous with SO, originating on Eb4 posterome- dially ventral to TEb4 and joining ER posterolater- ally. @ Same as Dussumieria, but substitute Eb4* and TEb4*; even less distinguishable in Clupea). Ad1-—3 absent. Ad4 dorsally on dorsoposterior edge of Eb4 leva- tor process and ventrally on Cb4, forming anterior wall of ““pocket”’ separating more-or-less vertical, an- terolateral continuation of SO-OP complex (see OP above), which forms posterior wall of pocket (not illustrated). @ Substitute Eb4* levator process. Ad5 forms slender tendinous attachment dorsally to large, cartilaginous Eb5; attaches musculously to Cb5 distalmeost end: undifferentiated medially from BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON SO ventral to ER. @ Musculously attached dorsally to Eb4* distal end. SO with conspicuous posteromedially curving ER on each side, sheet of CT (not illustrated) arises from entire length of ER and extends posteriorly (attach- ment not traced). Longitudinal muscle layer begins as isolated dorsal patch at about horizontal through posterior ends of ERs (longitudinal fibers absent pos- terior to patch) and extends anteriorly to below TEb4 and attaches to mid-medial side of Pb4. @ ER very reduced, present on only one side of one of two spec- imens examined. RD absent. Additional remarks. SCL absent. TV4 free from Cb5, but attached mid-dorsally to Bb3. Small, un- paired, roughly U-shaped, non-staining, questionably cartilaginous pad attached to anterior ends of Pb2s dorsally. In attachments and position, U-shaped pad is faintly similar to GC of non-clupeid clupeoids. MPbI1 absent. Large EbS present, articulating dorsal- ly with dorsodistal end of Eb4, and ventrally with small ventrodistal end of Eb4 and broad distal end of Cb4. @ U-shaped pad absent; small, cartilaginous MPb1 present just anterior to anterior tip of each Ebl (MPb1 reported as absent in Clupea by Nelson, 1967b:391, who, p. 392, implied that the unpaired MPb1- anterior to the tips of Ebls in many clupeids, may have originated as segmentation of the cartilag- inous tips of Ebls). Eb5 absent or, by comparison with Dussumieria, fused with Eb4* dorsodistal end. Nelson (1967d:fig. 2b) illustrated Eb4* of Clupea harengus and labeled the dorsally continuous, but ventrally separate, posterior cartilaginous end of Eb4 as EbS5. He (1967b:fig. 2d) illustrated the Eb5 portion of the distal end of Eb4 in C. harengus as fused ven- trally and separate dorsally and did not label it as separate from the remainder of Eb4. On both sides of both our specimens of Clupea, the ventral end is fused indistinguishably with the remainder of Eb4. Gonorynchiformes CHANIDAE Chanos chanos (Forsskal), USNM 173572, 140 mm, USNM 347538, 63.3 mm. Plate 30 Description. LE1 on tip of Eb! uncinate process. LE2 on tip of Eb2 uncinate process. LE3 absent. LE4 originating as long tendon among other LE origins, passing medial to all other LEs, except LP, at about 180° angle, and inserting on tip of Eb4 un- cinate process in larger specimen (smaller specimen lacks distinct uncinate process, and LE4 inserts on NUMBER 11 small, dorsally raised cartilaginous process of medial head of Eb4). Remarks. In almost all other pre-acanthomorph taxa with LE4, it originates on the dorsal margin of Eb4 (no levator process present) or on or near the LE4 levator process (including those taxa in which EbS is fused with the dorsodistal end of Eb4). John- son and Patterson (1996:273) reported that Chanos lacks an uncinate process, quite possibly based on examination of a small, early ontogenetic stage spec- imen (they did not list the size of the specimens they examined; see also illustrations of Chanos gill arches in Johnson and Patterson, 1996). There is a bony ridge separating the cartilaginous tip of the uncinate process from the cartilaginous proximal head of Eb1 in our larger specimen (obscured by LE4 insertion in Plate 30). The reasons we indicate the process as an uncinate process is that it appears to have developed as a separation of the medial head of Eb4 (Johnson and Patterson, 1996:273), and a posterior (actually lateral) levator process is present. The alternative, that two Eb4 levator processes are present is also possible, and would be uniquely synapomorphic for Chanos. LP tiny, inserting on LE4 levator process well pos- terior to uncinate process; origin not recorded. Remarks. Greenwood and Lauder (1981:228) wrote that LP appeared to be absent in Chanos. The relatively small size of the muscle may have caused them to overlook it, or its presence may be variable. The fragile levator process was damaged on both sides of our smaller specimen during dissection and it was not possible to determine if LP was present. LI1 on Pb2 bony surface dorsoposteriorly. LI2 on Pb3 bony surface dorsoposteriorly. LI3 on Pb4 dorsoanterolaterally. TD comprises TPb3 and TEb4. TPb3 a broad band of muscle attaching to Pb3 bony dorsal surface, with slender, diagonal muscle strap extending from pos- terolateral end of left side and attaching to dorsal surface of Pb4 (diagonal strap absent in smaller spec- imen). TPb3 well separated from and not continuous with TEb4. TEb4 more extensive than TPb3, attach- ing to Eb4 long ventrolateral arm, continuous pos- teriorly with modified SO muscles. OD3 absent. OD4 absent. OP indistinguishable, if present, obscured by EO. See also Ad5. Ad1-3 absent. Ad4 dorsally broadly on dorsoposterior edge of long ventrolateral arm of Eb4, narrowing consider- ably ventrally and attaching to Cb4 just medial to inner angle formed by Eb4-Cbé4 joint. Ad5 questionably represented by sheet of muscle attaching dorsally to large EbS5 cartilaginous plate and ventrally to distal cartilaginous process at end of Cb5. 45 Also possibly represented by muscle strap (Plate 30B, not labeled) extending from EbS dorsally to ra- phe (ER?, not labeled) on posterior surface of EO and undifferentiated from SO. Muscle sheet and/or strap possibly including OP. SO longitudinal muscle layer circumesophageal in straight portion of esophagus (not illustrated) imme- diately posterior to EO; presence of SO longitudinal muscle fibers in EO problematic; fibers absent ante- rior to EO. RD absent. Additional remarks. SCL absent. TV4 well anterior to, and free from, anterior to tips of Cb5s. Pb2 and Pb3 edentulous, UP4 and UP5 absent. Eb] with au- togenous cartilaginous segment of anterior tip (MPb1) present; see discussion in additional remarks under Gonorynchus (Gonorynchidae). Large, unpaired, elongate-ovate cartilage, tenta- tively identified as GC, present overlying anterior tips of Pb2s and Pb3s, surrounded by CT, and attaching mid-ventroanteriorly to medialmost ends of MPbls. In so attaching, Chanos’s GC differs, perhaps, from that reported by Di Dario (2002), who first described GC. He reported that GC occurs only in most en- grauloids and pristigasteroids among “remaining clu- peiforms [= our Clupeomorpha] and basal teleoce- phalans [= Recent Teleostei],’”” but did not mention its attachments. We find that it attaches to the anterior ends of the Pb2s (our observation based on pristi- gasteroid //isha). Di Dario’s comparative material in- cluded two specimens of Chanos, and it would ap- pear that if his specimens exhibited a GC-like ele- ment, he would have discussed it. His overlooking GC in Chanos is understandable. We failed to extract it in one of our two dissections of Chanos gill arches, and it appears that it has never been mentioned in any previously published study that treated the gill arches of the genus (but described by Nelson, 1966a: 157, in his dissertation). Given the cladistic interrelationhips illustrated in Fig. 3, GC in Chanos and GC in Clupeoidei are most parsimoniously interpreted as homoplasies, although the element may represent retention of a basal oto- cephalan character that has been lost independently several times. GONORYNCHIDAE Gonorynchus moseleyi Jordan and Snyder, USNM 354590, 231 mm. Plate 31 Additional material. Gonorynchus forsteri Ogilby, USNM 353921, 90.6 mm. Description. Remarks. There is no substantive difference in the musculature of the two species. 46 LE1 on dorsoposterior edge of medialmost bony portion of Eb1, which lacks uncinate process. LE2 on dorsoposterior surface of bony and carti- laginous medial end of Eb2, which lacks uncinate process. LE3 minute, on cartilaginous tip of Eb3 uncinate process. LE4 reduced, on dorsomedial edge of expanded dorsolateral cartilaginous margin of Eb4*. See also remarks following LP. LP absent. Remarks. Greenwood and Lauder (1981:228) re- ported that LP is present in Gonorynchus. It is pos- sible that they identified the muscle we believe to be LE4, of which they made no mention, as LP. Unless a specimen is found that has both LE4 and LP, the identification of the muscle in question may remain unresolved. LlIlon posteromedial surface of Pb2. LI2 on posterodorsal cartilaginous surface of Pb3. LI3 on posterolateral surface of cartilaginous Pb4. TD comprises TPb3, TPb4, and TEb4*. TPb3 at- taches over most of posterior half of surface of Pb3, well separated and discontinuous from TPb4. TPb4 attaches to anterodorsal surface of Pb4 and is contin- uous ventroposteriorly with TEb4*. TEb4* very broad, attaches to medial edge of surface of Eb4* medial arm dorsal to attachment of Ad4, and is con- tinuous posteriorly with greatly expanded SO as it forms EO. OD3 small, originates tendinously on bony dorsal surface of Pb3 uncinate process at lateral edge of TPb3 and inserts by slender tendon on dorsoanterior margin of cartilaginous tip of Eb3 uncinate process. OD4 small, originates on anterolateral edge of Pb4 and inserts by slender tendon on dorsoanteriormost cartilaginous edge of Eb4*. OP absent, perhaps highly modified by complex of CT and nerves in area between epibranchial organ laterally and esophagus medially. Ad1-—3 absent. Ad4 on mid-posterior bony surface of Eb4* and bony dorsal surface of Cb4 medial to inner angle of Eb4*-Cb4 joint. Ad5 apparently absent. GFM? muscle with small portion on dorsomedial cartilaginous surface of Eb4* and long portion on dorsoposterior surface of AC (= acc of Johnson and Patterson, 1997:fig. 2b) joining Eb4* and Cb5S. Inter- pretation highly questionable: are there gill filaments attached to AC? If not, muscle may represent sepa- rate portion of SO. RDs absent. SO longitudinal muscle layer (not illustrated) cir- cumesophageal in esophagus leading into EO, con- tinuing dorsoanteriorly as sparse fibers to at least be- low TPb3 © :d probably further. BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON Additional remarks. SCL absent. TV4 tripartite, di- vided longitudinally, with each part attaching medi- ally to ventral surface of posterior, cartilaginous bas- ibranchial copula to which Cb3 and Cb4 attach me- dially. Triangular-like cartilaginous element attaches to anterior tip of each Eb1, which Johnson and Pat- terson (1997:596) consider a mediosuprapharyngo- branchial (MPb1). We note that the anteromedialmost point of each MPb1 is minutely ossified where the elements of the two sides meet. Similar minute os- sifications occur posteromedially where each MPb1 meets the anterior cartilaginous tip of its respective Pb2. A strong, slender ligament joins posterior edge of cartilaginous tip of Eb3 uncinate process with an- terior bony edge of Eb4*. Pb2 and Pb3 edentate; UP4 and UP5 absent. Cypriniformes CYPRINIDAE Zacco platypus, USNM 336890, 2 specimens, 84— 101 mm; 3 cleared and counterstained, 28—82 mm. Plates 32.1, 32.2 Additional material. ®@ = Opsariichthys bidens Giin- ther, USNM 112443, 135 mm. Description. Remarks. Many of the muscles described below appear to be unique to cypriniforms. Our attempts to homologize the names applied to these muscles, re- ferring to Takahashi (1925), Holtsvoogdt (1965), and Winterbottom (1974b), were only partly successful because of unclear descriptions and illustrations. We use some of these authors’ names for these muscles, but have introduced our own terminology for others. Homologies among many of the dorsal gill-arch mus- cles of otophysans remain to be elucidated. LE1 on mid-posterodorsal surface of Eb1. LE2 on mid-posterodorsal surface of Eb2. LE3 on base of all bony Eb3 uncinate process. LE4 on cartilaginous tip of Eb4 levator process and tiny, horizontally oriented autogenous bone (AB, Plate 32.2) attached to posterior surface of cartilagi- nous tip of Eb4 levator process (AB also joined by CT to pad-like insertion of LCb5A; dorsal end of Eb5 attached to AB). LP absent. Remarks. Winterbottom (1974b:252—253 and fig. 22) treats our LCb5 as his LP externus and our LCbSA as his LP internus. LCb5 inserts on Cb5 and LCb5A inserts on a CT pad attached to Cb5. The muscle we define as LP inserts invariably on Eb4, or joins LE4 in a combined insertion on Eb4. Further- more, with the exception of Chanos, in all pre-acan- thomorphs with LP, LP inserts on or together with LE4. In Chanos, LP inserts on Eb4 well posterior to LE4. We, therefore, infer that LP is absent in cypri- NUMBER 11 nids. If one or the other of these two muscles, LCb5A or LCbS5, however, is derivative of the levator pos- terior, it would indicate that LP is a synapomorphy of the Otocephala. LI1 on posterior surface of Pb2 dorsolaterally, near joint with Eb2. LI2 on Pb3 mid-laterally and medial end of Eb3, which joins Pb3. @ On Pb3 mid- to posterolaterally and Eb3 and Eb4 medial ends, which join Pb3 (in- sertion occupying same area as combined LI2 and LI2’ of Z. platypus). Remarks. Takahashi (1925:41) reported that LI2 has only one insertion in Z. platypus (and O. unci- rostris); the muscle is apparently variable within the genus; see also remarks under LI2’. LI2’ on posterolateral cartilaginous edge of Pb3 and anteromedial cartilaginous and bony edges of Eb4, spanning joint between Pb3 and Eb4. (see LI2). @ Not present; see LI2. Remarks. Pb4 is absent in both Z. platypus and O. bidens, but is present in other cyprinids; e.g., New World Notropis hudsonius (USNM 315400, cleared and stained), Old World Abbotina (USNM uncat., cleared and stained). With the exception of the cyp- rinid genus Pseudogobio, which has three LIs (= our LI1, 2, 2’), Takahashi (1925:41—42) found only two LIs in cyprinoids, including Opsariichthys uncirostris and Zacco platypus. He mentions, however, that the second LI of Cyprinus carpio has two “‘caudae”’ [or- igins?]. His descriptions (p. 42) of the three LIs in cobitoids indicate very similar states to those we found for the three LIs in Zacco platypus. LI3 absent. LI4 (see LCb4). LCb4 slender, originating on exoccipital and in- serting by long tendon on Cb4 among medial fibers of Ad4. Remarks. Holstvoogd (1965:fig. 12b) termed this muscle Jevator IV internus in the cyprinid Leuciscus (and also for a muscle that we term LP attaching to Eb4 in characiforms). We reserve LI for muscles in- serting on pharyngobranchial elements. LCb4 occurs only in cypriniforms, possibly only cyprinids. LCb4, however, may represent a different character state for LI4, which is found only in Diplomystes (Diplomys- tidae), in which it inserts on UP4. LCb5 massive, on dorsoposterior surface of hy- pertrophied Cb5, originating in subtemporal fossa with two levels of attachment: dorsoanteriorly mainly or entirely on pterotic; ventroposteriorly mainly or entirely on exoccipital; found only in cypriniforms. See remarks following LCbSA. Remarks. Winterbottom (1974b:252—253) denoted this muscle as his LP externus (see remarks following LP above). LCb5A small, originating in supratemporal fossa together with ventroposterior level of origin of LCb5 47 (q.v.), wrapping medially, then anteriorly around LCb5 and inserting in thick CT pad strongly attached dorsally to anterolateral surface of Cb5 (pad also joined anteriorly by CT to tiny AB attached to tip of Eb4 levator process). Remarks. LCb5A appears to be the same as the internus branch of the levator posterior of Winter- bottom (1974b:fig. 22b; indicated by the left line leading from his label L.POST to the muscle; the right line leads to the externus branch, which equals our LCbS). TD absent (unique among Halecostomi); possibly replaced by thick SO section extending anteriorly ventral to pharyngobranchials. OD3 on Pb3 bony dorsal surface anterolaterally and tip of bony Eb3 uncinate process. OD4 on Pb3 bony surface posterolaterally and bony tip of Eb4 uncinate process. OP absent. RecD2 posteriorly on anteroventral surface of me- dial half of Eb2 and ventrolateral bony surface of Pb2, anteriorly on mid-ventromedial bony surface of Eb1. RecD3 posteriorly on anteroventral surface of me- dial half of Eb3 and ventrolateral surface of Pb3, an- teriorly on mid-ventromedial bony surface of Eb2. RecD4 posteriorly on anteromedial surface of Eb4, anteriorly on bony posteromedial surface of Eb3. Ad1—4 each attaching to ventral surface of its re- spective Eb and dorsal surface of its respective Cb at the internal angle formed by the two bones. Remarks. Among pre-acanthomorphs, only poly- odontids and cyprinids have Ad1-—3, and their attach- ments in the two groups are different and different from those of acanthomorphs having Ad1-3. Ad5 fan-like, with broad anterior end on dorso- posterior margin of Eb4 levator process and narrow posterior end on dorsolateral surface of Cb5. RCbST originates medially from CT and SO, well separated from contralateral RCbST, and inserts lat- erally on dorsolateral margin of Cb5. Remarks. Holstvoogd (1965:fig. 12a) called this muscle transversus ventralis posterior. We think the name misleading as transverses ventrales are other- wise applied to ventral gill-arch muscles. Winterbot- tom (1974:258; fig. 22b) identified this muscle a the retractor pharyngeus superioris. RCbS5SE massive, originates on ventral basioccipital process posterior to and partially dorsal to RCbSI, and inserts on Cb5 dorsolaterally. Remarks. Winterbottom (1974b:fig. 22) named this muscle retractor pharyngeus inferioris. RCbSI relatively slender, originates as fine line of CT along dorsolateral surface of a vertical SO fold medially abutting a ventral basioccipital process and inserts by long tendon anteriorly on thick CT pad to which LCbSA (q.v.) attaches (pad attaches to Cb5). 48 Remarks. The vertical SO fold may be a separate muscle. It is slightly disjunct anteriorly from a pos- teriorly ovoid area of SO muscles, which are anter- oventral to the ventral basioccipital process. A pair (one on each side) of muscle extensions continues a short distance anteriorly from the ovoid area, and these and the ovoid area muscles are covered by a tough fascia. The fascia divides and continues ante- riorly on each side forming a pad covering the sur- face of the anterodorsal ends of Pb2 and Pb3. An- teriorly from pair of muscle extensions, the fascia forms long, slender tendons that attach to the medial surface of Pb3. Winterbottom (1974b) did not men- tion RCbSI. SO longitudinal muscle layer absent at and poste- rior to horizontal joining posteriormost edges of Cb5s; very thick longitudinal layer begins below transverse layer anterior to horizontal and extends ventrally below gill arches well anterior to horizontal at anterior margins of Eb1ls (roofs much of oral cav- ity); transverse muscle layer absent anterior to anter- iormost external extent visible in Plate 32.1A (ante- rior portion of longitudinal muscle layer not overlain by transverse muscle. RD absent. Additional remarks. SCL absent. TV4 divided (in- terrupted), attached medially to each Cb5. Slender, almost threadlike cartilaginous EbS (may be in 1 to 3 pieces) attaches ventrally to dorsoposterior tip of Cb4 and dorsally to tiny autogenous bony element (AB) at posterodorsal tip of Eb4 levator process. Pb1, Pb4, UP4, and UPS absent. Characiformes Howes (1976) treated cranial muscles of certain characiform fishes, essentially avoiding the dorsal gill-arch muscles except for mention of Kampf’s (1961) brief treatment of of these muscles in Hydro- cyon forskali (= Hydrocynus forskalii, Alestiidae) and Winterbottom’s (1974b) general study of fish musculature, which included reference to Brycon guatemalensis (Characidae). In apparent justification, Howes (p. 219-220) stated, ““The arrangement [of the branchial muscles] in the Cynodontini is basically as in Brycon, and a provisional survey of branchial arch myology in . . . various characoid families (pers. obs.) suggests relative uniformity throughout the group. However, some [unspecified] specializations have been found in those taxa with epibranchial or- gans (Chilodus, Anodus).”’ There may be relatively more variation among characiform gill-arch muscu- lature than Howes opined. Kampf (1961:436, figs. 29-20), using terminology different from ours, reported that TD included at- tachment to Pb5, LE3 absent, and Ad1-—3 present, character states not present in the taxa we examined. BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON In the case of Pb5, we find that ventral strands of SO muscle attach to the medial surfaces of UP4 and UPS. We arbitrarily did not treat the latter as part of TD, because these tooth plates extend posteriorly into the esophagous. Kampf’s Ad1—3 may equal our GFM1— 3. But none of the three taxa we examined lacked LE3. CHARACIDAE Brycon guatemalensis Regan, USNM 114526, 99.0 mm SL. Plate 33 Additional material. @ = Brycon melanopterus (Cope), USNM 307072, 78.4 mm SL. Description. Winterbottom (1974b:fig. 20) presented a lateral view of the dorsal gill-arch muscles of B. guatema- lensis, but some muscles are obscured. LE1 on dorsal edge of Eb1 just lateral to cartilage tip of uncinate process; origin slightly dorsal to Pb1 articulation with skull. @ On Eb! uncinate process just ventrolateral to cartilaginous tip. LE2 on and just ventral to cartilaginous tip of Eb2 uncinate process. @ On Eb2 just lateral to uncinate process. LE3 on cartilaginous tip of Eb3 uncinate process. LE4 on dorsalmost tip of Eb4 levator process and tendinous base of LP. LP attached tendinously to dorsalmost tip of Eb4 levator process. LI1 on dorsoposterior surface of Pb2. LI2 on dorsal surface of Pb3 posterolaterally. LI3 on dorsoposterior surface of cartilaginous Pb4. TD comprises TPb3a-Eb2, TPb3p, and TPb4-Eb4 (see also discussion following Characiformes for pos- sibly excluded TD muscle). TPb3a-Eb2 with sparse muscle strands attaching to Pb3 dorsoanteriorly, but mainly attaching to cartilaginous tip of Eb2 uncinate process laterally, posteriorly continuous by diagonal muscle strap with TPb3p. TPb3p on Pb3 posterolat- erally, posteriorly continuous by diagonal muscle slips with TPb4-Eb4. TPb4-Eb4 broadly on Pb4 and narrowly on Eb4 medially, broadly continuous pos- teriorly with SO. OD3 origin on Pb3 anteriorly ventral to TPb3a- Eb2, extends posteriorly dorsal to TPb3p, and inserts on cartilaginous tip of Eb3 uncinate process medial to insertion of LE3. OD4 origin mostly on Pb4 anteriorly, slightly on Pb3 at joint with Pb4, muscle fans out posteriorly and inserts along most of lateral surface of broad levator process of Eb4. OD4’ long, slender; tendinous origin on Pb3 ven- tral to OD3 origin, joins OD4 insertion ventrally on anterolateral face of Eb4. @ OD4’ absent. NUMBER 11 OP a strap of muscle originating on posteromedial surface of Eb4, poorly separated from SO, ending ventrally at ER; muscle fibers continuing ventrally from ER include Ad5 and SO. M. UP5-Cb4-Eb5 (not illustrated), origin on lateral margin of UPS, insertion at inner angle formed by Cb4 and EbS. Remarks. M. UP5-Cb4-Eb5 appears to be a vari- ation of Kampf’s (1961:436, fig. 30, lower right) ob- liquus dorsalis inferior in Hydrocyon forskali (= Hy- drocynus forskallii (Cuvier)). Kampf described it, however, as connecting Pb5 (= our UP5?) with Cb5. Winterbottom (1974b) did not mention obliquus dor- salis inferior among his muscle synonymies. The al- bulids are the only other fishes we examined that have a possibly equivalent muscle, which we indicate as M. UP5-Cb4. Ad1-—3 absent. Ad4 on Eb4 dorsoposteriorly and Cb4 dorsal sur- face medial to inner angle of Eb4-Cb4 joint. Ad5 on Eb5 mid-posteriorly and Cb5 posteriorly, with thin dorsolateral tendinous extension (not illus- trated) crossing Eb5 dorsally to distalmost bony end of Eb4; joins ER below OP laterally and SO medi- ally, merges with SO medially and TV5 ventroanter- iorly. SO longitudinal muscle layer surrounds esophagus, thick dorsally, very thin elsewhere; fibers attaching anteriorly along medial edge of large UP4; fibers ex- tend anteriorly to below anterior end of TPb4-Eb4. @ Layer moderately evenly distributed around esoph- agus. RDs absent. Additional remarks. SCL absent. TV4 free from Cb5s. Slender ligament from cartilaginous tip of Eb4 levator process to Eb5 dorsally. Eb5 ventrally on dis- talmost end of Cb4. DISTICHODONTIDAE Xenocharax spilurus Giinther, USNM 227093, 125 mm. Not illustrated Description. LE1 broadly on Eb! uncinate process beginning a little lateral to cartilaginous tip, origin joins Pb1 ar- ticulation with skull; short, band-like tendon joins or- igin with Pb1 just ventrolateral to dorsal cartilage tip. LE2 on Eb2 just lateral to tip of uncinate process. LE3 on anteromedial edge of Eb3 uncinate pro- cess, there joining OD3 insertion. LE4 on dorsalmost edge of Eb4 levator process, joining raphes ventrolaterally with LP insertion and Ad4 dorsally. LP on Eb4, joining raphes ventromedially with LE4 insertion and Ad4 dorsally. LI1 on dorsomedial surface of Pb2. 49 LI2 on dorsal surface of Pb3 posterolaterally. LI3 dorsoposterolaterally on Pb4. TD comprises TPb3a-Eb2, TPb3p, TPb4, and TEb4. TPb3a-Eb2 on Pb3 dorsally beginning at an- terior edge of bony surface and extending posteriorly, and on cartilaginous tip of Eb2 uncinate process dor- sally, posteroventrally continuous by diagonal muscle slip with TPb3p anteriorly. TPb3p on Pb3 postero- laterally, beginning ventral to TEb2, and just failing to meet LI2 insertion anteriorly; posteriorly contin- uous by diagonal muscle strand with TPb4. TPb4 broadly on Pb4, sharply separated posteriorly from TEb4. TEb4 on Eb4 anteriorly ventral to insertions of OD3-—4 and OD4’. OD3-—4 origin on Pb3 anteriorly, divides posteri- orly with insertions on tip of Eb3 uncinate process medial to LE3 insertion, and on Eb4 dorsolaterally. OD4' origin on Pb4 anteriorly joining raphe dor- soanteromedially with OD3—4 origin posteriorly; in- sertion on Eb4 ventral to OD3—4 insertion. OP a muscle strap originating on Eb4 posterome- dially, joining TEb4 ventrolaterally and irregular ER dorsally, which is joined ventrally by Ad5 (ER con- tinues short distance medially into SO). OP unclearly differentiated from SO medially. M. UP5-Cb4-Eb5 origin on lateral margin of UP5, insertion at inner angle formed by Cb4 and Eb5. (Also see remarks following description M. UP5- Cb4-Eb5 in Brycon). Ad1-—3 absent. Ad4 well developed on Eb4 dorsoposteriorly be- ginning medially ventral to uncinate process and ex- tending laterally to end of bone, joining raphes dor- sally with LE4 and LP insertions; ventrally very nar- rowly on Cb4 just medial to Eb4-Cb4 joint. Ad5 bulbous, broadly on Eb5 posteriorly and Cb5 posteriorly, with thin dorsolateral tendinous exten- sion sheathing Eb5 dorsally and attaching to distal- most bony end of Eb4; muscle joins ER ventral to OP and merges medially with SO; ventrally joins TV5 posterolaterally. SO longitudinal muscle layer thick dorsally, very thin or absent elsewhere, begins at about horizontal connecting distal ends of Eb4s and extends anteriorly to anterior end of TEb4. RDs absent. Additional remarks. SCL absent. TV4 free from Cb5s. Slender ligament joins cartilaginous tip of Eb4 levator process to Eb5 dorsally, cartilage ventrally on posterodistalmost end of Cb4. Siluriformes DIPLOMYSTIDAE Diplomystes chilensis (Molina)?, USNM 259097, 119 mm. Plate 34 50 Description. LEI! on Eb! dorsoposteriorly. LE2 on Eb2 dorsoposteriorly. LE3 absent. LE4 absent, but see remarks under LI4. LP absent (see remarks under LI4). LI1 on posterior margin of small CT pad, which impinges on ventral cranial surface and incorporates: medial ends of PbI—3 and Eb1; autogenous ball of cartilage that articulates with medial ends of Eb1 and Pb2; and a few smaller autogenous cartilages; inser- tion splits OD3 origin. LI2 inserts mainly on dorsoanterior edge of large UP4, with minor insertion on anterolateral cartilagi- nous edge of Pb4. Remarks. Insertion of LI2 on Pb4 (or UP4) is un- common among fishes we examined (Table 2). LI3 medially on lateral cartilaginous edge of Pb4, ventrolaterally on UP4 dorsoposteriorly, and Eb4 me- dial cartilaginous cap at Eb4-Pb4 joint. Remarks. Insertion including Eb4 is unique among fishes we examined. LI4 originates on pterotic and inserts mainly on the dorsoposterior surface of UP4, with some fibers inserting on cartilaginous joint formed by Pb4 with Eb4. Remarks. Takahashi (1925:67), using different ter- minology, considered this muscle to represent RD in siluriforms, in which it originates variously on the pterotic or supraclavicle (? = posttemporal-supra- cleithrum) and inserts variously on the “‘posterior pharyngobranchial” or Eb4 (Diplomystes was not among his material). Takahashi did not distinguish Pb4 from its toothplate, and we are uncertain of the exact location of the insertion in the taxa he listed. The origin of LI4 would seem to exclude interpre- tation of LI4 as RD, which usually originates on the anterior vertebrae, or possibly the basioccipital. Hol- stvoogd (1965:215 and fig. 10) termed this muscle simply “Levator IV” in the clariid siluriform Clar- ias. Winterbottom (1974b:253 and fig. 21) question- ably designated the muscle LP in Diplomystes. Among pre-acanthomorphs, LP occurs only among Otocephala (clupeomorphs and ostariophysans), which might be evidence that the muscle we desig- nate LI4 is a modified LE4 that has shifted its inser- tion from Eb4 to UP4 (the slight partial insertion in- cluding the medial end of Eb4 providing evidence). LP, otherwise, invariably inserts on Eb4 together with LE4, which Diplomystes otherwise lacks. TD comprises four sections (a—d), of which the first is least differentiated. Anteriormost section, a, TPb1-Pb2-Pb3-Eb1-Eb2 (includes only anteriormost tip of Pb3) underlies pair of CT pads (see LI1) and is posteriorly continuous with next section, b, TPb3- Pb4, which joins medial bony edges of Pb3 and Pb4 and overlies tanterior end of next section, c, TUP4a. BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON Laterally, TUP4a attaches to dorsomedial surface (ledge) of UP4, and is continuous posteriorly with d, TUP4p, which curves anteriorly and inserts along dorsolateral surface of UP4; TUP4p is continuous posteriorly with SO. OD3 has a tripartite origin: dorsolateral origin (a) on Pb1 and posterolateralmost margin of CT pad (see LI1), including medial tip of Ebl, separated by in- sertion of LI] from medial origin (b) on posterome- dial margin of CT pad; these two origins, separated by LI1 insertion, meet and fuse posteriorly (obscured by recumbent LI1); ventrolateral origin (c) a short branch on Eb2, fuses with ventral surface of dorso- lateral origin; complex inserts on anterior surface of Eb3 uncinate process. OD4 origin on dorsoposterior surface of Pb3, ven- tral to OD3, continuing onto dorsoanteromedial sur- face of Pb4; insertion on anterior surface of Eb4 un- cinate (levator?) process. Remarks. Arratia (1987:11, 41) reported that Di- plomystes lacks an uncinate process on Eb4, although mentioning that Eb4 “has a short lateral projection which I do not consider an uncinate process.” She made no mention of a levator process. We find the uncinate (or levator) process present and well devel- oped. The uncinate process on Eb3 is armlike, where- as that of Eb4 arises vertically from the horizontal as a broad, dorsally obtuse flange with a distinct carti- laginous cap. OP dorsally on posteromedial surface of Eb4, ven- trolaterally on medial end of Eb5, ventromedially joining raphe (ER?) with SO, continuing ventrally below raphe and attaching to bony medial surface of Cb5; ventrolaterally overlapping medial portion of Ad5. RDs absent (see LI4 above). SO longitudinal muscle band circumesophageal, thick dorsally and ventrally, thin laterally, extending anteriorly to below anterior end of TPb3-Pb4. Ad1-—3 absent. Ad4 dorsally on posterodorsal surface of Eb4 un- cinate (levator?) process, ventrally on Cb4 dorsal sur- face. Ad5 dorsally on posteromedial surface of Cb4 and ventral surface of Eb5; ventrally on dorsoposterior edge of Cb5, extending medially under OP. Additional remarks. SCL absent. TV4 free from Cb5s. Pb4 partially ossified dorsomedially. UP5 ab- sent. Eb5 on posterodistalmost end of Cb4, horizon- tally oriented, a position duplicated only in Pantodon (Osteoglossomorpha). Small, sesamoid bone (basi- hyal?) heavily enveloped in CT present between an- terior hypohyals (presumably, ventral hypohyals be- cause they articulate with the urohyal; Arratia, 1987: fig 27, indicates these are dorsal hypohyals, whereas, Azpelicueta, 1994:figs. 6b and 11, correctly indicates they are ventral hypohyals). de Pinna (E-mail, Mar— NUMBER 11 Apr 1999) informs us that cartilaginously tipped Eb3 and Eb4 uncinate processes occur only in Diplomys- tes among siluriforms, and that the basihyal and Bb1 are absent in all catfishes. Gymnotiformes GYMNOTIDAE Gymnotus carapo Linnaeus, USNM 260272, 320 mm WL, Plate 35 Description. Remarks. Hoz and Chardon (1984) reported on the gill-arch musculature of the gymnotid Sternopygus macrurus, which is very similar to that of Gymnotus carapo. We remark on the differences. LE! on Eb! dorsoposteriorly near distal end. LE2 posteriorly on Eb2 dorsoposterior edge just lateral to LI2 insertion, and anteriorly on dorsopos- terior edge of Ebl medial to LE]! insertion. Remarks. Hoz and Chardon (1984) do not mention an attachment to Eb1. Verification in other specimens is desirable. LE3 on bony Eb3 uncinate process just lateral to OD3 insertion. LE4 on levator process at dorsodistal end of Eb4. LP absent. LI1 broadly, tendinously on Pb2 dorsoanteriorly and narrowly on posteromedial end of Eb! dorsally. Remarks. Hoz and Chardon (1984) indicate inser- tion only on Pb2. LI2 primarily on dorsoanteriormost surface of me- dial end of Eb3, barely continuing onto adjacent la- teralmost edge of Pb3, secondarily on dorsoposter- iormost edge of Eb2 immediately medial to LE2 in- sertion. Remarks. Identification of this muscle as LI2 is problematic because of its primary insertions on Ebs rather than Pb3. Hoz and Chardon (1984) indicate insertion is only on Eb3. LI3 on Pb4 and UP4 dorsolaterally. Remarks. Hoz and Chardon (1984) report insertion is only on Eb4. Their illustration (fig. 19b) is unclear, but the external position of the muscle appears sim- ilar to that of LI3 in G. carapo. We believe they intended to report the insertion as Pb4, a lapsus sim- ilar to their listing (p. 44) of LE4 as a [second] LE3, but labeling the muscle LE4 on their fig. 19b. TD comprises TPb2-Pb3-Pb4 and TEb4. TPb2- Pb3-Pb4 attaches to medial edge of Pb2 on one side and to CT on other, medial edge of Pb3 (or slightly medial to edge) along OD4 origin, and Pb4 medially, and is narrowly continuous posteriorly with TEb4. TEb4 on posteromedial edge of Eb4, posteriorly con- tinuous with SO. Remarks. Hoz and Chardon (1984) divide TD into 51 two parts, one attaching to Pb2 and Pb3 and one at- taching to Pb4, with no mention of an attachment to Eb4. OD3 relatively small, originating on Pb3 dorsoan- teriorly and Eb2 dorsomedialmost end, and inserting on medialmost end of bony Eb3 uncinate process. OD4 relatively large, originating on Pb3 and Pb4 dorsally and inserting on Eb4 dorsolaterally. OP absent. Ad1-—3 absent. Ad4 dorsally broadly on ventrolateral surface of Eb4 and ventrally broadly on anterolateral surface of Cb4 medial to Eb4-Cb4 joint. Ad5 ventrally broadly on posterolateralmost sur- face of Cb5, just impinging on lateralmost end of TV5, and dorsally narrowly on posterodistal surface of EbS. RDs absent. SO longitudinal fibers circumesophageal, begin- ning posterior to horizontal connecting distal ends of Cb5s, forming very thick section dorsally in area be- tween distal ends of Eb4s, and essentially absent an- terior to horizontal through posterior end of TEb4. Additional remarks. SCL absent. TV4 continuous from one side to the other, forming median raphe and attaching tightly to dorsoanteriormost tips of Cb5s. UPS absent. Salmoniformes SALMONIDAE Oncorhynchus mykiss (Walbaum), USNM 351540, 121 mm, USNM 333092, 91.8 mm. Plate 36 Description. LE1 on tip of Eb1 uncinate process. LE2 on tip of Eb2 uncinate process. LE3 on tip of Eb3 uncinate process just dorsal to OD3 insertion. LE4 on dorsolateralmost surface of Eb4. LP absent. LI1 broadly on Pb2 dorsally. LI2 on Pb3 dorsoposteriorly. LI3 insertion enveloping Pb4 dorsal and anterior surfaces and just attaching to dorsoanterior edge of UPS. TD a broad sheet of muscle comprising TPb3a and TPb3-Pb4-Eb4 (Pb3 portion slightly differentiated laterally from remainder). TPb3a short, laterally on Pb3 uncinate process, fusing medially with OD3 an- teroventral surface dorsal to OD3’. TPb3-Pb4-Eb4 anteriorly on Pb3 ventral to OD3’, forming raphe with OD4 origin, on Pb4 medial to LI3 insertion, and on medial end of Eb4, posteriorly continuous with, but noticeably demarcated from SO. OD3 anteriorly dorsal to anterolateral portion of n bo TPb3-Pb4-Eb4; dorsoanteriorly on Pb3 near anterior end; joined to opposite OD3 by median raphe; an- other raphe lateral to median raphe setting off small wedge-like anterior muscle section; anteroventral sur- face joined by medial end of TPb3a, possibly evi- denced by dorsolateral OD3 raphe; insertion on Eb3 uncinate process ventral to LE3 insertion. OD3’ completely ventral to OD3, origin on Pb3 posteromedial to OD3 origin and anterior to TPb3- Pb4-Eb4; fuses with ventral surface of OD3 at about mid-length of OD3. OD4 origin on Pb3 dorsoposterior surface and Pb4 dorsoanterior surface, forming raphe with TPb3-Pb4- Eb4 medially, inserting on dorsal surface of Eb4 me- dial to LE4 insertion, forming partial raphe with OP and/or Ad4 dorsally; some muscle strands continuous with OP and Ad4. OP dorsally on Eb4, ventrally joining ER at about mid-level of Eb5, below which OP is undifferentiated from SO (see also OD4). ER in smaller specimen better developed, extends medially from mid-level of EbS5 for width of OP and then turns relatively sharply posteriorly for about same distance on SO. Ad1-—3 absent. Remarks. Dietz (1912:fig. 2; p. 19), illustrated and reported Ads1l—4 in Salmo salar Linnaeus as follows (our translation): “This system of weak rudiments is present on the anterior four gill arches as a border of short, obliquely placed fibers along the exterior of the epibranchials and the proximal greater part of the cer- tatobranchials. Only against the inside of the fourth gill arch is it well developed in form . . .”” Although, we did not examine S. salar, it is clear from his de- scription that Dietz identified the gill filament mus- cles on the first three arches as adductors. Ad4 dorsally on Eb4, ventrally on Cb4 medial to inner angle formed by Eb4-Cb4 joint (see also OD4). Ad5 dorsally on mid-posterior surface of Eb5, and ventrally on. Cb5 distal end. RD absent. SO longitudinal muscle layer absent posterior to horizontal between distal ends of Cb5s, present as isolated, very thick band dorsally anterior to (and above) horizontal connecting distal ends of Cb5s and continuing anteriorly only to about horizontal con- necting anterior ends of Pb4s. Additional remarks. SCL attached mid-dorsally to cartilaginous posteroventral tip of Bb3. TV4 free from Cb5s. UP4 absent. Thymallus arcticus (Pallas), USNM 179764, 153 mm. Plate 37 Description. LE1 on cartilage tip of Eb1 uncinate process. LE2 on cartilage tip of Eb2 uncinate process dor- solaterally. BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON LE3 on cartilaginous tip of Eb3 uncinate process. LE4 on dorsalmost edge of distal end of Eb4, be- tween dorsal edge of OD4 insertion and dorsal at- tachment of Ad4. LP absent. LI1 on mid-dorsal bony surface of Pb2. LI2 relatively small; on left-side muscle almost in- separable from LI3 dorsal to LI2 insertion, which is on posterodorsal bony surface of Pb3; right-side LI2 and LI3 completely separate, insertions as on left side. LI3 on left side with two slightly separate inser- tions, anterior insertion on Pb4 dorsoanteriorly, pos- terior insertion on posterolateral edge of Pb4, con- tinuing posteroventrally and attaching tendinously to dorsoposterolateral edge of UP5; on right side with two completely separate insertions (essentially two separate muscles), anterior muscle on dorsoanterior surface of Pb4, posterior muscle on dorsolateral edge of UPS. Remarks. About one-fifth of left-side UPS is ven- tral to Pb4 (remainder is ventral to Eb4), whereas on right side almost one-half is ventral to Pb4. TD comprises TPb3 and TPb4-Eb4. TPb3 on Pb3 uncinate process joining raphe with anterior portion of OD3 and OD4 origins on right side and OD4 or- igin on left side; fibers changing from transverse to crisscross about halfway along raphe, then continuing as transverse fibers of TPb4-Eb4. TPb4-Eb4 on dor- soposterior surface of Pb4 and dorsomedial surface of Eb4, continuous with SO posteriorly. OD3 strap-like, much smaller than OD4; left-side OD3 origin on medial edge of cartilage tip of Pb3 uncinate process ventral to OD4; right-side OD3 or igin on dorsolateral surface of Pb3 uncinate process anterior to OD4 origin, but continuous posteriorly with it; insertion on both sides identical, on medial edge of cartilaginous tip of Eb3 uncinate process. OD4 massive; left-side OD4 completely dorsal to OD3, origin begins on lateral edge of Pb3 uncinate process and continues posteriorly onto Pb4, joins ra- phe with TPb3; right-side OD4 origin in line and posterior to OD3 origin, joins raphe with TPb3; mus- cle on both sides identical, insert on dorsal surface of Eb4 medial to LE4 insertion. OP muscle strap beginning medially on posterior surface of Eb4 and extending laterally short distance to dorsomedial edge of Ad4, barely separable medi- ally from SO, ventrally joining ER; ventral to ER indistinguishable from SO laterally and Ad5S medi- ally. Ad1-—3 absent. Ad4 dorsally on posterolateral surface of Eb4 un- cinate process, ventrally on Cb4 dorsal surface just medial to inner angle formed by Eb4-Cb4 joint. Ad5 dorsally on Eb5S mid-posterolateral surface, ventrally on Cb5 posterolaterally beginning at distal NUMBER 11 tip and extending slightly medially, medially mostly inseparable from OP. RD absent. SO longitudinal muscle layer absent posterior to horizontal between distal ends of Cb5s, present as isolated, very thick band dorsally anterior to (and above) horizontal connecting distal ends of Cb5s and continuing anteriorly only to about horizontal con- necting anterior ends of Pb4s. Additional remarks. SCL dorsomedially attached to posteroventral cartilaginous end of Bb3. TV4 free from Cb5s. UP4 absent. adductors. Coregonus artedi Lesueur, USNM 117488, 2 speci- mens, 132—140 mm SL. Plate 38 Description. LE1, small, on tip of Eb1l uncinate process. LE2 small, on tip of Eb2 uncinate process. LE3 on tip of Eb3 uncinate process dorsolaterally. LE4 tendinously on dorsalmost edge of distal end of Eb4. LP absent. LI1 on Pb2 dorsoposteriorly. LI2 on Pb3 dorsoposteriorly. LI3 on Pb4 dorsally. TD comprises three parts, a—c: a, TPb3a, raised dorsally (dorsal to OD3 origin), attaching to anterior bony surface of Pb3, including anterior bony edge of Pb3 uncinate process, posteriorly continuous (discon- tinuous in smaller specimen) by diagonal muscle strand with more ventral TPb3p (4), which attaches to Pb3 bony surface dorsoposteriorly; TPb3p contin- uous posteriorly by broad diagonal muscle strap with c, TPb4-Eb4, which attaches to dorsomedial edges of Pb4 and Eb4 and is broadly continuous by diagonal band of muscle with SO. OD3 origin divided; main portion on dorsoanterior bony surface of Pb3 ventral to TPb3a; separate, ven- trolateral smaller portion (not illustrated) on posterior edge of Pb3 uncinate process; insertion on Eb3 un- cinate process dorsomedially. Remarks. Smaller specimen has undivided origin on both sides. OD4 origin on Pb3 dorsoposteriorly and Pb4 dor- soanteriorly, insertion on dorsal edge of Eb4 well me- dial to LE4 insertion. OP dorsally on Eb4 medial to Ad4 dorsal attach- ment; ventrally joins ER at about level of dorsal at- tachment of Ad5 to EbS, and is undifferentiated from SO ventral to ER. Ad1-3 absent. Ad4 dorsally, broadly on Eb4 posterolaterally be- 55) ginning lateral to OP attachment, ventrally on Cb4 medial to inner angle formed by Eb4-Cb4 joint. Ad5 on mid-posterior surface of Eb5 dorsally, and Cb5 distal end ventrally. RD absent. SO longitudinal muscle layer absent posterior to horizontal between distal ends of Cb5s, present ven- trally and as isolated, very thick band dorsally ante- rior to (and above) horizontal connecting distal ends of Cb5s and continuing anteriorly only to about hor- izontal connecting anterior ends of Pb4s. Additional remarks. SCL present. TV4 complex, attached by tendons to Cb5. RETROPINNIDAE Retropinna osmeroides Hector, USNM 304419, 106 mm. Plate 39 Description. LE] finely, tendinously on cartilaginous tip of Eb1 uncinate process. LE2 finely, tendinously on cartilaginous tip of Eb2 uncinate process. LE3 on tip of Eb3 uncinate process. LE4 on dorsodistal edge of Eb4*. LP absent. LI1 on Pb2 dorsoposteromedially ventral to ante- rior end of Pb3. LI2 on Pb3 dorsoposteriorly (well separated from LI3 insertion on Pb4). Remarks. Right-side LI2 probably anomalous, has slender separate portion inserting on Pb4 anteriorly, and well separated from main LI2 and LI3 insertions. LI3 on Pb4 dorsoposteriorly. TD comprises TPb3a and TPb3p-Pb4-Eb4*. TPb3a lies dorsal to remainder of TD, comprises loose strands of muscle connecting bony lateral edges of Pb3 uncinate process dorsal to origin of OD3, con- nected posteriorly by fine muscle strand to TPb3p- Pb4-Eb4*. TPb3p-Pb4-Eb4* broad, compact, atta- ches to Pb3 dorsal surface dorsoposterolateral to OD4 origin, and to medial surfaces of Pb4 and Eb4*. OD3 origin on Pb3 dorsoanteriorly ventral to TPb3a, insertion on tip of Eb3 uncinate process ven- tromedial to LE3 insertion. OD4 origin on Pb3 posterodorsal bony surface ventral to posterior margin of TPb3p portion of TPb3p-Pb4-Eb4*, insertion mostly on Eb4* dorso- lateral edge proximal to LE4 insertion. OD4’ fine, originating on Pb4 dorsal surface com- pletely ventral to OD4; originating fibers pass through TPb3p-Pb4-Eb4*; muscle fuses with OD4 ventral surface well anterior to OD4 insertion. Remarks. Muscle is possibly anomalous; confir- mation of its presence in other specimens is desirable. OP comprises two or three muscle strands origi- 54 nating on posteromedial surface of Eb4* dorsal to SO attachment to Eb4*, joining ER dorsally, undiffer- entiated from SO ventral to ER. Ad1-—3 absent. Ad4 dorsally on posterior surface of Eb4*, ven- trally on Cb4 dorsal surface just medial to inner angle formed by Eb4*-Cb4 joint. Ad5 dorsally on posterodistal surface of cartilagi- nous Eb4* posterodistal hook-like process, ventrally on posterodistal surface of Cb5, medially continuous with SO or OP?. RD absent. SO longitudinal muscle layer absent or represented at most by few sparsely distributed fibers dorsally and ventrally posterior to horizontal through distal ends of Cb5s, dorsally becoming thick, then attenu- ating, and extending anteriorly to anterior end of TD. Additional remarks. SCL absent. TV4 absent. GALAXIDAE Galaxias auratus Johnston, USNM 344888, 85.6 mm. Plate 40 Description. LEI on mid-posterior edge of Eb1, attaches to dor- sal tip of Pb1 at origin. LE2 on mid-posterior edge of Eb2. LE3 tiny, jointly with OD3 insertion on cartilagi- nous tip and lateral edge of Eb3 uncinate process. Remarks. LE3 reduced relative to LE1 and LE2, similar to Lovettia. LE4 dorsodistally on Eb4* with posterodistal strand extending ventrally and attaching to ventro- posterior surface of Eb4* distal end. LP absent. LI1 on Pb2 dorsal surface anteromedially. LI2 on Pb3 dorsal surface posteromedially. LI3 on Pb4 dorsal surface. TD undifferentiated sheet of muscle comprising TPb3-Pb4-Eb4*, attaching to Pb3 medial to OD or- igins, Pb4 medial to LI3 insertion, and medial end of Eb4*. OD3 and OD4 are bilaterally asymmetrical. OD3 originates dorsally on cartilaginous anterior end of Pb3 on both sides. Left-side origin well separated anteriorly from OD4 origin on Pb3 bony dorsal sur- face, insertion on Eb3 cartilage-tipped uncinate pro- cess medial to and together with LE3; however, strand of muscle at anterior end of OD4 origin (well separated posteriorly from OD3 origin) also inserts with OD3. Right-side OD3 origin and insertion sim- ilar to left-side, but strand of muscle originates at posterior end of origin and inserts on Eb4* together with OD4 insertion. Right-side OD3 origin, except for strand, also well separated from OD4 origin. OD4 origin on both sides, except for strand on left BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON side, from multiple separate muscle strands on Pb3 bony dorsal surface, combining to insert on Eb4* dorsodistal end medial to LE4 insertion. OP muscle band originating dorsally on dorsome- dial edge of Eb4*, questionably separate dorsome- dially from SO, joining ER dorsally at mid-posterior level of Eb4* distal end; undifferentiated from SO ventral to ER. Remarks. Configuration of OP, Ad5, SO, and ER similar to that of Lovettia. Ad1-—3 absent. Ad4 dorsally, broadly on posterior edge of Eb4*, ventrally on Cb4 medial to inner angle formed by Eb4*-Cb4 joint. Ad5 broadly on posterodistal end of Cb5, dorsally narrowly attaching to mid-posterodistal end of Eb4*, medially indistinguishable from SO. RDs absent. SO longitudinal muscle layer absent posterior to horizontal between distal ends of Cb5s, present as isolated thick muscle band dorsally anterior to (and above) horizontal connecting distal ends of Cb5s, and continuing anteriorly only to about horizontal con- necting medial ends of Eb4s*. Additional remarks. SCL absent. TV4 attached to Cb5. Several tiny accessory cartilages on each side of gill arches. Small cartilage just anterior to anterior cartilaginous tip of Pb3 on both sides (possibly rep- resenting a segmentation of the anterior tips of Pb3). Anteriormost AC on left side between Eb1 cartilag- inous medial end and anterior cartilaginous end of Pb2 just anterior to bony portion; anteriormost on right side between medialmost tip of Eb1 and adja- cent cartilaginous edge of Pb2; posteriormost carti- lage on left side filamentous, parallels posteromedial edge of Pb2; posteriormost (not illustrated) on right side, very tiny, between cartilaginous tips of Pb3-Eb3 joint. Eb1l and Eb2 lack uncinate processes. Lepidogalaxias salamandroides, USNM 358461, 43.7 mm; Murdoch University, 47.3 mm. Plate 41 Description. Remarks. The muscles are highly modified and confusing. Our interpretations of TPb3, OD3, OD4, and OD4’ are provisional. LE1 on Eb! bony surface dorsodistally. LE2 on Eb2 at base and lateral to uncinate process. LE3 on dorsodistalmost end of Eb3 and ligament (not shown) extending from insertion to distal end of Cb4. LE4 on anterodistal surface of Eb4, muscle in cross section V-shaped with apex directed medially. LP absent. LI1 on dorsoposterior surface of Pb2. NUMBER 11 LI2 absent. LI3 unusually large, on Pb4 dorsally, insertion ending posteriorly at margin of Pb4 with UPS? (not shown). Remarks. In the larger specimen, LI3 also appears to be attached to UP5?, but damage during dissection leaves this observation problematic. TD appears to comprise only broad TPb3, which attaches on Pb3 dorsally, joining raphe anterolaterally with OD3, extending posteriorly as complexly ori- ented fibers, interrupted by irregular, somewhat me- dian raphe, and joining another raphe on each side medial to LI3. OD4' appears to originate at latter raphe. OD3 passes lateral to LI3, origin on Pb3 anterior end, joining raphe there with TPb3, insertion on Eb3 dorsodistally. OD4 a muscle strap originating on Eb3 along ra- phe with OD3 and attaching to Eb4 dorsodistally be- tween insertion of LE4 and OD4’. Remarks. This muscle could be interpreted as a RecD4, but other than its origin on Eb3, an apparent slight shift laterally from the highly reduced, tiny, rod-like, edentate Pb3, its insertion is fairly typical of OD4, and the relationship with OD3 is reminiscent of the more-or-less in-line origins of OD3 and OD4 in other pre-acanthomorphs. OD4’ appears to originate at a raphe with TPb3 and, on one side, RD well medial to LI3, and is in- terrupted by another raphe immediately medial to LI3 before attaching to dorsal surface of Eb4 just medial to OD4. OD4' lies dorsal to unnamed band of muscle attaching anteriorly to posteromedial surface of Eb4 and joining raphe posteriorly with RD. We were un- able to determine if OD4’ attaches to Pb4 or UPS. OP dorsally on Eb4 posterior edge medial to Ad4 dorsally, muscle ending ventrally at ER, which con- tinues laterally along dorsal end of Ad5, becoming tendinous and attaching to Eb4 posteroventrally and Cb4 posteriorly (ER tendinous attachments to Eb4 and Cb4 (obscured on Plate 41); OP undifferentiated from SO medially. Ad1-—3 absent. Ad4 dorsally on Eb4 posterolaterally, ventrally on Cb4 medial to Eb4-Cb4 joint. Ad5 on Cb5 posterodistally, joining ER dorsally, and forming tendinous attachment to Eb4 and Cb4. SO longitudinal fibers absent anterior to horizontal through posterior ends of Eb4s, unless represented by RDs; condition of specimen precluded determination of posterior extent, unless indicated by posterior ends of RDs. SOD absent. RDs completely external to SO, broadly inserted on raphe with unnamed muscle band on one side and with muscle band and OD4’ on other side. Additional remarks. SCL absent. TV4 divided, free 55 from Cb5, medial end of each division attaching to ventral surface of bony Bb4. Eb1 and Eb3 lack un- cinate processes. UP4 questionably absent. Eb4 and Cb4 appear to have unmodified distal cartilaginous ends (see also Rosen, 1974:fig.15B), with no trace of EbS. Lovettia sealei (Johnston), USNM 358617, 2 speci- mens, 50.2—54.2 mm. Plate 42 Description based primarily on smaller specimen. LE1 on Eb! dorsoposterior edge somewhat medial to mid-length of bone. LE2 weakly developed (easily overlooked or de- stroyed), on Eb2 dorsoposterior edge somewhat me- dial to mdi-length of bone. LE3 greatly reduced, on finger-like dorsal exten- sion (modified, displaced uncinate process?) of car- tilaginous distal tip of Eb3. Remarks. LE3 reduced relative to LEI and LE2, similar to Galaxias LE2. LE4 on Eb4* dorsodistally. LP absent. LI1 on Pb2 dorsomedially. LI2 on Pb3 dorsoposteriorly. LI3 on Pb4 dorsoposteriorly. TD not clearly partitioned, comprises TPb2-Pb3- Pb4-Eb4*, beginning as few muscle strands on Pb2, continuing posteriorly onto Pb3 and Pb4, and medial ends of Eb4*, slightly differentiated posteriorly from SO. OD3 origin on Pb3 dorsoanteriorly, insertion on dorsomedialmost edge of finger-like extension of car- tilaginous distal tip of Eb3. OD4 on Pb3 dorsoposteriorly and Pb4 dorsoanter- iorly, with slight separation of muscle between Pbs; insertion on dorsomedialmost cartilaginous edge of distal end of Eb4*. Remarks. Left-side OD4 of larger specimen has split origins on Pb3. OP relatively small strap of muscle dorsally on posterior surface near medial end of Eb4*, joining ER ventrally, not distinguished from Ad5 or from SO below ER. Remarks. Configuration of OP, Ad5, SO, and ER similar to that of Galaxias. Ad1-—3 absent. Ad4 on dorsoposterior surface of Eb4*, ventrally on Cb4 anterior to inner angle formed by Eb4* and Cb4. Ad5 on mid-distal end of Eb4* and posterodistal end of Cb5, not separable medially from SO and/or OP. SO, unable to determine by gross dissection if lon- gitudinal muscle layer is present. 56 RDs absent. Additional remarks. SCL absent. TV4 divided, free from Cb5, medial end of each side attached to ventral surface of Bb3. Eb1, Eb2, and Eb3 uncinate pro- cesses absent. UP4 and UPS absent. OSMERIDAE Hypomesus pretiosus (Girard), 2 specimens, USNM 357404, 135 mm; USNM 70839, 172 mm. Plate 43 Additional material. @ = Mallotus villosus, USNM 104700, 148 mm. Description. Remarks. The levators of the smaller Hypomesus specimen have relatively long tendinous origins, whereas the origins of the larger specimen (and those of Mallotus) are relatively short and mostly muscu- lous. LE1 slender, on dorsodistal edge of Eb1 uncinate process. LE2 slender, on dorsal tip of Eb2 uncinate process. LE3 on dorsal tip of Eb3 uncinate process. LE3’, thin, applied closely to anterior surface of LE3, insertion on Eb3 uncinate process just ventral to LE3 insertion. Remarks. The flat “anterior” surface of LE3 is positioned parallel to the body midline, as is usual in fishes, and the flat surface of LE3’ is applied (“‘lam- inated’’) to the flat surface of LE3. The muscles orig- inate together and their distinction might escape ca- sual examination. LE4 originates in cluster with other LEs; inserts on dorsalmost edge of Eb4* levator process. LE4’ very slender, closely applied for almost entire length to distal edge of LE4, origin not diverging much from that of LE4, insertion along edge of LE4. ® Origin diverging slightly but noticeably from that of LE4, but attached for most of length to distal edge of LE4, becoming tendinous toward insertion. Remarks. Not considered LP because origin is with clustered LEs. LP absent. LI1 on dorsoposterior surface of Pb2. LI2 on dorsomedial surface of Pb3 posterior end, at and lateral to attachment of TPb3p. LI3 on Pb4 dorsal surface at and lateral to TPb4- Eb4* attachment. TD comprises three parts: TPb3a-Eb2, TPb3p, and TPb4-Eb4* (described separately for Mallotus). TPb3a-Eb2 most distinct, dorsally situated relative to remainder of TD, with lateral attachments to joined tips of Pb3 and Eb2 uncinate processes, and crossed lateral attachments to dorsoanterior Pb3 bony surfac- es dorsal to OD3 origins, completely separate from TPb3p in larger specimen, continuous posteriorly by BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON diagonal fiber bundle with TPb3p in smaller speci- men. TPb3p on Pb3 dorsal surface posterior to un- cinate process and medial to LI2 insertion, continu- ous by diagonal muscle fibers with TPb4-Eb4*. TPb4-Eb4* on Pb4 dorsolaterally medial to LI3, and on dorsomedial edge of Eb4, posteriorly almost in- separable posteriorly from SO. @ Comprises TPb3a, TPb3p, and TPb4-Eb4*. TPb3a on Pb3 uncinate process laterally, ventrome- dially attached to CT between well-separated anterior ends of Pb3s (Pb3s of Hypomesus almost impinge along medial edges), no crossed lateral attachments to Pb3s dorsoanteriorly dorsal to OD3 origin, dor- sally situated relative to remainder of TD, continuous posteriorly with TPb3p posteriorly. TPb3p on Pb3 laterally posterior to uncinate process, continuous posteriorly with TPb4-Eb4a*. TPb4-Eb4* on Pb4 dorsolaterally and Eb4* medially, posteriorly distin- guished from SO by change in muscle fiber direction. OD3 origin on dorsoanterior bony surface of Pb3 ventral to TPb3a-Eb2, insertion on Eb3 uncinate pro- cess medial to LE3 and LE3’ insertions. @ Origin ventral to TPb3a. OD4 origin on posteriormost end of Pb3 and dor- soanterior surface of Pb4, insertion on medial edge of Eb4*. @ OD4 origin on Pb3 dorsoposteriorly and OD4’ on Pb4 dorsoanteriorly, muscles joining just posterior to OD4' origin, but somewhat (artificially?) separable for entire length, with OD4 inserting on Eb4* dorsoanteriorly and OD4’ inserting on medial edge of Eb4* ventromedial to OD4. OD4’ absent. @ Anteriorly on Pb4 dorsoanteriorly and posteriorly on Eb4* medial edge; just posterior to origin, joins OD4, but is somewhat separable (ar- tificially?) for entire length. OP strap of muscle dorsally on dorsomedial edge of Eb4*, partly dorsal to attachment of TPb4-Eb4* to Eb4*; ventrally joins ER with SO; indistinguish- able ventral to ER. Ad1-3 absent. Ad4 dorsally on dorsoposterior edge of Eb4*, ven- trally on Cb4 just medial to inner angle formed by Eb4*-Cb4 joint. Ad5 on posterodistal end of Eb4* and posterodistal end of Cb5, weakly separable from SO (or OP?) me- dially. ® On posteriorly extending finger-like carti- laginous extension of Eb4* distal end and posterodis- tal end of Cb5. RDs absent. SO longitudinal fibers begin well posterior to hor- izontal through posterior ends of Cb5s and continue anteriorly to horizontal between anterior ends of Pb4s (posterior to TPb3p). Additional remarks. SCL attached posteromedially to ventroposterior cartilaginous tip of Bb3—ligament almost circular with massive muscle attachments; @ SCL absent. TV4 free from Cb5s, divided at mid- NUMBER 11 line, attaching dorsally to ventral surface of Bb series (specific element not determined). Smaller specimen with two tiny ACs just anterior to anteroventral sur- face of Pbl on left side, one on right side; larger specimen with one AC (not illustrated) on right side only. Johnson and Patterson (1996:277) discuss prob- lems concerning the identification of the upper pha- ryngeal tooth plate, or plates, as representing UP4 or UPS, but appear to favor identifying it as UPS. The posterior end of Eb4 has the foramen for the fifth efferent artery bordered completely by bone an- teriorly and by cartilage posteriorly in the larger specimen of Hypomesus; the cartilaginous portion is interrupted in the middle in the smaller specimen (see Rosen, 1974:fig. 16D). The posterior end of Eb4 in the specimen of Mallotus is similar to that of the larger specimen of Mallotus except that there is a very fine joint line (both right and left arches) be- tween the dorsomedial end of ““Eb5” and the dor- solateral cartilaginous end of Eb4. The condition in our experience is exceptional. Eb5 usually appears to fuse to Eb4 dorsally first, and ventrally last (see com- ments on clupeomorphs by Rosen, 1974:280). Rosen (1974:284), however, states that the opposite is true for osmeroids: fusion takes place ventrally first; how- ever, he provided no evidence for this. Based on his illustrations of the osmeroid taxa (his fig. 16), all have either a free Eb5 or a fused EbS that can be interpreted alternatively as having had Eb5 fused completely with Eb4 and subsequently variously and partially lost, with no unequivocal indication whether fusion took place dorsally or ventrally first. Argentiniformes ARGENTINIDAE Argentina brucei Cohen and Atsaides, USNM 238005, 99.5 mm. Plate 44 Description. LE1 on Eb! at and just lateral to base of uncinate process. LE2 on dorsolateral edge of Eb2 uncinate process. LE3 on dorsolateral edge of Eb3 uncinate process. LE4 on dorsolateralmost tip of Eb4 levator process just dorsal to insertion of OD4 anteriorly and dorsal attachment of Ad4 posteriorly. LP absent. LI1 on Pb2 mid-dorsally. LI2 on Pb3 dorsoposteriormost bony surface, be- ginning just lateral to OD4 origin. LI3 on Pb4 dorsoposteriorly. TD long, broad, essentially undifferentiated, com- prising TPb3-Pb4-Eb4, which attaches along dorso- medial surfaces of Pb3 and Pb4 and meeting inser- tions of OD3 and OD4, and with narrow attachment 57 to posteromedial edge of Eb4; posteriorly continuous with SO. OD3 origin mid-dorsally on Pb3 just medial to LI2 insertion, insertion on Eb3 uncinate process ven- troanterior to LE3 insertion. OD4 origin broadly on Pb3 beginning at posterior end of OD3 origin, insertion on Eb4 levator process ventroanterior to LE4 insertion. OP strap of muscle overlying SO dorsolaterally, dorsally on Eb4 posteromedial surface, ending ven- trally at ER (right side ER very weak), scarcely sep- arable ventrally from Ad5 and SO. Ad1-—3 absent. Ad4 dorsally on posterodorsal surface of Eb4, ven- trally on dorsal surface of Cb4 anterior to inner angle formed by Eb4-Cb4 joint. Ad5 dorsally on mid-distal surface of Eb5, ven- trally on posterodistal bonysurface of Cb5 ventrally, forming raphe with posterolateral end of TV5, me- dially continuous with SO. SO with strap of muscle arising laterally, extending dorsally, then ventrally and attaching to anterior sur- face of Eb5 to form wall of pocket lined anteriorly by Ad4; continuous ventrolaterally with Ad5 (or ?OP). Longitudinal fibers commencing well posterior to horizontal between distal ends of Cb5s and con- tinuing anteriorly, thinly to horizontal connecting me- dial ends of Eb3. RD absent. Additional remarks. SCL absent. TV4 free from Cb5s. Eb5 large, on dorsodistal end of Cb4, articu- lating ventroanteriorly with posterodistal end of Eb4, attached dorsally by long, slender ligament to Eb4 levator process. Accessory cartilage on dorsodistal end of Cb5, surrounded by tubular CT, which is at- tached to mid-posterior surface of Eb5. ALEPOCEPHALIDAE Alepocephalus tenebrosus Gilbert, USNM 215582, 141 mm. Plate 45 Description. Remarks. LE] and LIs together surround and at- tach to dorsal end of very long, slender SPb1 and to cranium; LI] also attaches along dorsomedial surface of SPbl. LEI slender, on dorsolateral edge of Eb! uncinate process. LE2 slender, on dorsodistal edge of cartilaginous medial end of Eb2. Remarks. Medial cartilaginous end of Eb2 expand- ed dorsally, and dorsally does not articulate with Pb3; Eb2 uncinate process absent, unless modified dorsal expansion of medial end represents shift in position of uncinate process. LE3 slender, on cartilaginous tip of Eb3 uncinate process. LE4 very slender, inserts finely, tendinously on mid-dorsal cartilaginous edge of distal end of Eb4. LP absent. LI1 broadly on bony dorsal surface of Pb2, dor- sally on SPb1 beginning a little ventral to attachment to skull and continuing to muscle origin. LI2 slender, on dorsal surface of bony and carti- laginous posterior end of Pb3. LI3 from single origin, spreads ventrally, becom- ing fan-like and inserting on Pb4 as group of sepa- rated muscle strands; on right side only, anteriormost muscle strand is also attached to cartilaginous medial tip of Eb3. TD comprises short TPb3a, much longer TPb3p- Pb4, and short TEb4. TPb3a attaches to bony dorsal surface of Pb3 anterior processes and is noticeably distinct, although broadly continuous posteriorly with TPb3p-Pb4. TPb3p-Pb4 attaches along the dorso- medial margins of Pb3 and Pb4 and is broadly con- tinuous posteriorly with TEb4, although separated from TEb4 by a posterolateral notch in the muscles at the posterior end of Pb4 and anterior end of Eb4. TEb4 attaches along anteromedial margin of Eb4 and becomes continuous posteriorly with SO by broad area of crisscrossing muscle fibers. OD3 slender, origin on medial surface of Pb3 un- cinate process, with fibers extending anteriorly and joining posterolateral edge of TPb3; insertion on dor- soanteriormost edge of Eb3 uncinate process. OD4 slender, origin on posteroventral edge of Pb3 uncinate process ventral to OD3 origin; tendinous in- sertion on dorsoanterior cartilaginous edge of Eb4, continuous at insertion with OD4’ insertion. OD4' slender, origin on anterodorsal surface of Pb4; insertion on Eb4 continuous with OD4 inser- tion. OP absent, but possibly represented by stringy muscle strands attaching to posteromedial surface of Eb4; strands become complexly meshed posterov- entrally with SO muscle strands (complexity not il- lustrated). ER absent, but possibly represented by area where problematic OP muscle strands mesh with SO strands. Ad1-—3 absent. Ad4 on posterior surface of dorsodistal end of Eb4 and dorsoposterior surface of Cb4 at internal angle formed by Eb4-Cb4 joint. Ad5 attaching cartilaginous Cb5 distal end poste- riorly to ventral margin of large Eb5. SO longitudinal fibers loosely incorporated in fluffy brown matrix beginning well posterior to hor- izontal between posterior ends of Cb5s, and decreas- ing in density abruptly at horizontal between anterior ends of Eb4s; very thin matrix, possibly including BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON sparse muscle fibers, continuing into pharyngeal area anterior to gill arches. RDs absent. Additional remarks. SCL absent. TV4 free from Cb5s. Cartilaginous end of Pb1 ventromedial process articulating with anterior end of Pb2, ventrolateral cartilaginous end articulating with anterior cartilagi- nous end of Ebl. Pb1 and tips of Pb2 and Pb3 en- veloped in loose CT pad (not illustrated); pad at- tached posteriorly to long, slender ligament, which joins contralateral ligament at attachment to cranium. Large EbS present posteriorly, articulating ventral- ly with dorsal edge of posterior ends of Cb4 and Eb4; large AC articulating ventrally with Cb5 and Eb5 me- dially; right side only with two additional tiny ac- cessory cartilages laterally on Eb5. SPb1 present. There is great similarity in the shapes and “‘lay- out’”’ of the muscles of Alepocephalus and Searsia, especially when compared with their sister group, Ar- gentina. PLATYTROCHTIDAE Searsia koefoedi Parr, USNM 206873, 117 mm. Plate 46 Description. Remarks. LE] and LIs surround and attach to SPb1 ventral to origins on cranium (all levators orig- inate together on cranium). LEI slender, inserts by slender tendon on dorso- posterior edge of Eb! lateral to uncinate process. LE2 absent. LE3 slender, on cartilaginous tip of Eb3 uncinate process. LE4 very slender, finely, tendinously on mid-dor- sal cartilaginous edge of Eb4. LP absent. LI1 broadly on dorsomedial surface of bony por- tion of Pb2, dorsally on SPb1 near insertion on skull; posteroventral fibers of left-side LI] attach (anoma- lously) to lateral edge of Pb3. LI2 on posterodorsal surface of Pb3; right side with few fibers attaching to anteriormost surface of Pb4, dorsally difficult to separate from LI3’. LI3 on Pb4 posterolaterally; lightly attached to me- dial surface of SPb2. LI3’ slender, but moderately broadly on Pb4 dor- soanteriorly lateral to origin of OD4’; right side LI3’ both anterior and lateral to origin of OD4’, dorsal fibers not clearly separable from LI2. TD comprises short TPb3a, which attaches to bony Pb3 anterior process, and much longer TPb3p-Pb4- Eb4, which attaches along dorsomedial edges of Pb3, Pb4, and Eb4, and is broadly continuous with SO posteriorly. OD3 origin on medial surface of Pb3 uncinate pro- cess; insertion on dorsomedialmost edge of Eb3 un- NUMBER 11 cinate process; left side OD3 lies dorsal to and almost completely hides OD4; right-side hides all but anter- olateralmost OD4. OD4 origin with and ventral to OD3 origin on Pb3 uncinate process; insertion on dorsomedial cartilagi- nous edge of Eb4, continuous posteriorly with pos- terior portion of OD4’. OD4’ origin on anterodorsal surface of Pb4; left side fuses with OD4 at about mid-length; right side fuses with OD4 just anterior to insertion on Eb4. OP absent, but possibly represented by stringy muscle strands attaching to posteromedial surface of Eb4; strands become complexly meshed posteroven- trally with SO muscle strands (complexity not illus- trated). ER absent, but possibly represented by area where problematic OP muscle strands mesh with SO strands. Ad1-—3 absent. Ad4 on dorsomedial surface of Eb4 and postero- dorsal surface of Cb4 at inner angle formed by Eb4- Cb4 joint. Ad5 narrowly on Eb4 ventrodistally and broadly on Cb5 ventrodistally; right side with few fibers at- taching to AC. SO longitudinal fibers loosely incorporated in fluffy brown matrix beginning well posterior to hor- izontal between posterior ends of Cb5s, decreasing in density and ending abruptly, or nearly so, at about horizontal between posterior thirds of Pb4s (matrix, at least, continuing anteriorly into pharyngeal area anterior to gill arches). RDs absent. Additional remarks. SCL present with dorsomedial CT pad attaching to Bb3 posteroventral tip. TV4 free from Cb5s. Pbl with cartilaginous end of ventro- medial process articulating with anterior tip of Pb2; ventrolateral process with two separated cartilaginous tips, articulating with anterior cartilaginous tip of Eb1. Pb1 and tips of Pb2 and Pb3 enveloped in tough CT pad. Eb5 articulating laterally with Eb4 and me- dially with AC, which articulates ventrally with Cb5. Cartilaginous SPb1 on dorsal cartilaginous tip of Eb1 uncinate process, SPb2 on dorsal cartilaginous tip of Eb2 uncinate process. Johnson and Patterson (1996:275) reported that S. koefoidi has only SPb1. Their report was based on SIO 77-38 and 77-53 (Johnson and Patterson, 1996:323), collected in the Banda and Sulu seas, whereas our specimen was col- lected in the northeastern Atlantic (type locality, however, is Bahamas). Matsui and Rosenblatt (1987: 73) reported that, “the width of the frontals differs between S. koefoedi of the 3 ocean regions.” It ap- pears likely that more than one species is involved. There is great similarity in the shapes and “lay- out” of the muscles of Alepocephalus and Searsia, especially when compared with their sister group, A7- gentina. 59 Esociformes ESOCIDAE Esox niger Lesueur, USNM 355803, ca. 235 mm. Plate 47 Additional material. @ = USNM68894, 87.5 mm. Esox lucius Linnaeus, Remarks. Poor condition of E. /ucius specimen provided only limited amount of information, but no noteworthy differences were observed in the muscles of the two species. Description. LE] broadly, fibrotendinously on cartilaginous cap of Eb1l uncinate process. LE2 on tip of Eb2 uncinate process. LE3 on tip of Eb3 uncinate process. LE4 absent. LP absent. LI1 massive, broadly on Pb2 bony surface dorso- medially. LI2 broadly on Pb3 bony surface dorsoposteriorly. LI3 on Pb4 dorsomedial surface. TD comprises TPb3a and TPb3p-Pb4. TPb3a at- taches along lateral edge of anterior half of Pb3, is dorsal to origin of OD3, and posteriorly continuous by diagonal strand of muscle with TPb3p-Pb4. TPb3p-Pb4 attaches to Pb3 and Pb4 dorsal surfaces medial to insertions of LI2 and LI3, joins dorsolateral raphe with OD4 origin, with which TPb3p-Pb4 is otherwise continuous, and is also continuous poste- riorly with SO. OD3 origin on dorsal surface of Pb3 ventromedial to attachment of TPb3a, insertion on Eb3 uncinate process proximal to LE3 insertion. OD4 absent. See remarks following OD4’. OD4’ originates dorsally from raphe with TPb3p- Pb4 and ventrally from Pb4 dorsal surface, and in- serts on Eb4 uncinate process. Remarks. OD4' of Esox is similar to OD4' of the putatively related Novumbra, particularly in that both have their main origins on Pb4 rather than Pb3. No- vumbra has OD3—4 fused or separate (OD3 and OD4) as well as OD4’. Johnson and Patterson (1996:275) argued that eso- coids may have an uncinate process, but not a levator process. As far as we know, an Eb4 levator process never articulates with the Eb3 uncinate process, whereas, the Eb4 uncinate process usually, if not al- ways, does, thus supporting the Johnson and Patter- son’s interpretation of the process in Esox. The pres- ence of an Eb4 uncinate process, and its articulation with the Eb3 uncinate process, in Esox is either an autapomorphy of Esox, or evidence that its closest relationships are with the basal acanthomorphs, in which these states first occur. 60 OP indistinct, probably represented dorsally by muscle attaching to posterior surface of Eb4 begin- ning ventral to uncinate process and extending me- dially, and ending ventrally at ER on Cb4, there joined by Ad5 dorsally. This configuration OP, ER, and Ad5 also appears in Saurida (Synodontidae) and some acanthomorphs, e.g., Polymixia. Ad1-—3 absent (condition of GFMs not deter- mined). Ad4 dorsally appears to attach along entire poste- rior surface of Eb4, although medial half probably represents OP, which ventrally joins ER with Ad5. Ad5 on Cb5 distally and bony posterior surface of Cb4, where it joins ER dorsally with presumable OP ventrally. RDs absent. SO longitudinal muscle layer originates well pos- terior to gill arches and extends anteriorly to hori- zontal at anterior ends of Pb4s, abruptly decreasing or vanishing anterior to horizontal. Additional remarks. SCL attached dorsomedially to posteroventral cartilaginous end of Bb3. TV4 free from Cb5s (completely anterior to anterior tips of Cb5s). UPS absent. Eb 5 absent. UMBRIDAE Novumbra hubbsi, USNM 357403, 2 specimens, 57.2-59.7 mm. Plate 48 Description. Remarks. In general the two dissected gill arches were very similar, but because of small size and fra- gility, both our dissections involved damage. The de- scription and illustrations are mostly based on the larger specimen, but the posterior view is based on the smaller specimen. LE! on Eb! uncinate process at and just ventro- lateral to cartilaginous tip. LE2 on Eb2 uncinate process at and, variably, just ventrolateral to cartilaginous tip. LE3 on tip Eb3 uncinate process dorsoanteriorly. LE4 weak, on Eb4 dorsoposteriorly near distal end, ventral fibers continuous with Ad4 dorsoposter- iorly. Remarks. Johnson and Patterson (1996:273) re- ported that Novumbra lacks LE4. The muscle is pres- ent in both of our specimens. Because LE4 is poorly developed and specimens are small, the muscle could have been easily lost during dissection. LP absent. LI1 dorsomedially on bony surface of Pb2. LI2 on Pb3 dorsoposteriorly. LI3 on Pb4 dorsolaterally meeting junction of Pb4 with UP4. TD complex. Distinct TPb3a anteriorly, attaching to Pb3 uncinate process anteriorly, continuous pos- BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON teriorly by crisscrossing muscle bands (both speci- mens) with TPb3p. TPb3p irregular, attaching to dor- soposteromedial surface of Pb3, continuous posteri- orly with TPb4 by continuations of crisscrossing muscle bands, which meet raphe with OD4’ posteri- orly. TPb4 attaching to Pb4, muscle continuous with SO posteriorly. OD3-—4 comprises a fused single muscle in the larger specimen (illustrated) and separate muscles (OD3 and OD4) in the smaller specimen. The fused muscle in the larger specimen can be artificially sep- arated, with the OD3 portion much more slender than the OD4 portion (relative sizes not apparent in illus- tration). The fused muscle divides slightly at its in- sertions on Eb3 and Eb4, which are similar to the insertions of the two muscles in the smaller speci- men. In the smaller specimen, the two muscles are of about equal size and are described as folows: OD3 origin anteriorly on Pb3 dorsal surface ven- tral to TPb3a, insertion on medial edge of Eb3 un- cinate process. OD4 origin on Pb3 dorsal surface posterior to and continuous with OD3 origin, insertion on dorsome- dial all bony process of Eb4, fuses with insertion of OD4’. OD4’ origin dorsally from raphe with crisscrossing muscle bands of TPb3p and mainly ventrally on dor- sal surface of Pb4, insertion on dorsomedial bony process of Eb4, fuses with insertion of 4 portion of OD3-—4. Remarks. OD4’ of Novumbra is similar to OD4' of the putatively related Esox, particularly in that both have their main origins on Pb4 rather than Pb3. There is also a similarity with OD4’ of Lepidoglaxias (Plate 41). OP on mid-dorsoposterior margin of Eb4, ending ventrally at ER on Cb4 with Ad5. M. Pb3-Eb3 small, attached to dorsoposterior edge of Pb3 uncinate process and dorsomedial edge of Eb3 uncinate process near LE3 insertion. Ad1-—3 absent, but GFM2 and 3 unusual, superfi- cially like RecDs. GFM2 originates on dorsoanterior surface of Eb2 uncinate process and inserts on dor- soposterior edge of Eb1 uncinate process, where it is continuous with LE1 basally (not continuous basally with LE1 in smaller specimen). GFM3 (not illustrat- ed) slender muscle seen only on right side of smaller specimen, originates on anterior margin of Eb3 and inserts on posterior surface of Eb2 uncinate process. Remarks. Similar GFM2 and 3 are known among pre-acanthomorphs otherwise only in Dallia (Umbri- dae), which has a modified GFM1. GFM1 in Novum- bra was destroyed before it was realized that its char- acter state might have importance. Ad4 on Eb4 dorsoposteriorly and Cb4 dorsally just medial to Eb4-Cb4 joint, just meeting and joining ER at dorsodistalmost end of Ad5. NUMBER 11 Ad5 on posterodistal surface of Cb4, joining ER with ventral end of OP, and on dorsoposterior portion of Cb5, forming raphe ventrally with TVS. SO longitudinal muscle fibers begin well posterior to horizontal between posterior ends of Cb5s and ex- tend anteriorly to below mid-posterior margin of TPb3a. RDs absent. Additional remarks. SCL attached mid-dorsally to posteroventrally extending cartilaginous end of Bb3. TV4 free from Cb5s. Eb4 uncinate process bony. UPS5 absent. Eb5 absent. Dallia pectoralis Bean, USNM 111643, 107 mm. Not illustrated Description. LEI on anterior surface of Eb! uncinate process. LE2 on anterior surface of Eb2 uncinate process. LE3 dorsoanteriorly on Eb3 uncinate process. LE4 mostly on Eb4 bony dorsolateral surface, ex- tending onto cartilaginous distal end. LP absent. LI1 on dorsomedial bony surface of Pb2. LI2 on Pb3 dorsally. LI3 on Pb4 dorsally. TD comprising TPb3 and TPb4. TPb3 shorter and wider of the two transverses, attaching to Pb3 dor- soanterolaterally; on left side attachment is dorsal to OD3 and OD4 origins; on right-side, posterolateral edge forms raphe with OD3 and OD4 origins and with slender muscle (anomalous?) extending laterally and inserting on Eb2 uncinate process; continuous broadly posteriorly with anterior end of TPb4. TPb4 on dorsal surface of Pb4 medial to LI3 insertion, con- tinuous posteriorly with SO, but noticeably distin- guished by change from horizontal (TPb4) to curving muscle fibers (SO). OD3 origin on Pb3 (see TPb3 above), insertion on dorsomedial edge of Eb3 uncinate process. OD4 origin on Pb3 (see TPb3 above), inserting laterally on Eb4 dorsomedial edge, extending medi- ally along raphe with dorsal end of OP (left-side OD4 posterolaterally and OP dorsally forming free strap of muscle—not attached ventrally; right-side raphe almost completely attached to dorsal edge of Eb4). OP dorsally joining raphe with OD4 (q.v.), ven- trally on posterodistal end of Cb5, forming raphe (not ER, unless highly modified) posteroventrally with Ad5, posterolaterally overlapping Ad4 and apparent- ly fusing with Ad4 (q.v.) ventroanteriorly on Cb5. Ad1-—3 absent, but GFM1-—3 unusual; GFM2 and GFM3 RecD-like, similar condition observed only in Novumbra. GFM1 originates by long, slender tendon attaching to anterior surface of cartilaginous medial end of Eb1, becoming musculous in area anterior to 61 uncinate process, and covering (attaching to) most of dorsal surface of Eb1 lateral to distal end of Ebl1, there becoming slenderer and extending around Eb1- Cb1 joint and continuing along dorsoanterior edge of Cb1. GFM2 inserts by short, fine tendon to posterior edge of Eb1 ventral to uncinate process, muscle is over CT between Eb1 and Eb2 and attaches to an- terior edge of Eb2. GFM3 similar to GFM2, insertion on Eb2 and origin on Eb?. Ad4 dorsally broadly on Eb4 beginning anterior to lateral end of OP and extending to end of bone; an- teroventrally on Cb4 dorsal surface medial to Eb4- Cb4 joint; posteroventrally attaching to cartilaginous finger-like process extending medially from poster- odistal end of Cb4, there joining raphe with Ad5, and continuing ventrally and (unusually) attaching to Cb5 anterior to OP, with which it appears to fuse. Ad5 very short, joins dorsodistal end of Cb5 to posterodistal end of Cb4, joins raphes with OP and Ad4. SO beginning well posterior to horizontal through posterior ends of Eb4s and extending anteriorly to posterior margin of TPb3. RDs absent. Additional remarks. SCL interrupted mid-posteri- orly by attachment to posteroventrally extending car- tilaginous end of Bb3. TV4 attached to Cb5s, except for free short area at anterior end of muscle. UP5 absent. Eb5 absent. Umbra pygmaea (DeKay), USNM 343617, 61.1 mm. Plate 49 Description. LE1 on Eb!1 uncinate process. LE2 on Eb2 uncinate process. LE3 on Eb3 uncinate process; right side LE3 with muscle slip inserting on anteromedial edge of Eb4. LE4 on cartilaginous dorsodistalmost edge of LE4. LP absent. LI1 on Pb2 dorsoposteromedially. LI2 on Pb3 bony dorsal surface; posterior edge continuous with anterior edge of LI3, but slight break at anteroventralmost insertion of LI3 on Pb4. LI3 on Pb4 dorsally (see also LI2). TD comprises single continuous sheet of muscle, TPb3-Pb4-Eb4, continuous posteriorly with SO. OD3 origin on Pb3 anterodorsal surface continu- ous with OD4 origin, insertion on medial edge of Eb3 uncinate process ventral to OD4. OD4 origin on Pb3 anterodorsal surface immedi- ately at and posterior to OD3 origin, forming raphe medially with TPb3-Pb4-Eb4, insertion on dorso- medial edge of Eb4 at and just anterior to dorsal at- tachments of Ad4 and OP. OP separate band of muscle dorsally, attaching to 62 dorsomedial edge of Eb4 just medial to dorsal at- tachment of Ad4, just posterior to OD4 insertion, and partially overlying (posterior to) SO attachment to posterior surface of Eb4; joins ER ventrally at about mid-level of cartilaginous Eb5, below which OP is inseparable from SO. Ad1—3 absent (condition of GFMs not deter- mined). Ad4 dorsally on dorsoposterior edge of Eb4 ven- tral to LE4 insertion and lateral to OP attachment, ventrally on Cb4 at and medial to inner angle formed by Eb4-Cb4. Ad5 on distal surface of Eb5 and dorsodistalmost end of Cb5. SO longitudinal muscle fibers begin well posterior to horizontal at posterior ends of Cb4s and extend anteriorly only to horizontal between mid-length of Pb4s. RD absent. Additional remarks. SCL attached mid-posteriorly to ventral surface of cartilaginous posterior end of Bb3. TV4 questionably absent (rare condition), but an obliquus ventralis-like muscle attaches ventro- medial surface of each Cb4 to ventral surface of Bb4 cartilage. UP5 absent. Eb5 present. Stomiiformes DIPLOPHIDAE Diplophos taenia Giinther, USNM 206614, 190 mm. Plate 50 Description. LEI! very fine, on small bony process on lateral margin of Ebl uncinate process. LE2 very fine, on small bony process ventrolateral to expanded cartilaginous dorsomedial end of Eb2. LE3 slender, on dorsomedial bony edge of Eb3 uncinate process on and with OD3 insertion. LE4 originates in cluster with other LEs; inserts tendinously on dorsal tip of Eb4 levator process. LP absent. LI1 on most of length of dorsomedial edge of Pb2. LI2 dorsally on posterolateral edge of Pb3. LI3 on dorsolateral edge of UP5, meeting TUPS5 laterally. Remarks. A slight portion of the posterior end of UPS slips under the medial end of Eb4. We follow Johnson (1992) in recognizing the single tooth plate in stomiiforms as UP5. However, its close association with Pb4 suggests that it could be UP4. Further in- vestigation of the homology of this tooth plate is de- sirable. TD comprises TEb2, TEb2v, TPb3, and TUPS. TEb2 broadest, on Eb2 dorsoanteriorly, attaching to expanded cartilaginous medial end of Eb2 as muscle passes over it, continuous posteriorly by diagonal BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON strand of muscle with TPb3. TEb2v thin muscle tightly attached to CT covering anterior ends of Pb3s, lies anteroventral to TEb2, with thin, slender pos- terolaterally extending straps that attach to antero- medial surface of expanded cartilaginous medial end of Eb2 (verification of presence in other specimens desirable). TPb3 attaches dorsally on posterolateral edge of Pb3 with and medial to LI2 insertion, con- tinuous posteriorly by diagonal strand of muscle with TUPS. TUPS5 attaches to UPS anteromedial to LI3 insertion, is continuous posteriorly by diagonal strand of muscle with SOD. OD3 origin ventral to TEb2 beginning on Pb3 dor- sally a short distance posterior to anterior cartilagi- nous tip of Pb3 and continuing uninterrupted poste- riorly as OD4 origin to point a little posterior to be- ginning of cartilaginous posterior end of Pb3; OD3 separates dorsolaterally from OD4 shortly posterior to combined origins and inserts on Eb3 uncinate pro- cess anteriorly. OD4 origin on Pb3 continuous with OD3 origin, joins OD4’ posterolaterally, and insert together on anterior surface of Eb4 levator process. OD4’ originates on Pb4 anteriorly, beginning at edge joining Pb3, joins OD4 posteromedially, and to- gether insert on Eb4 levator process anteriorly. OP dorsally on Eb4 posteroventrally beginning near medial end and extending laterally a short dis- tance, ventrally on Cb5 slightly posteromedial to dis- tal end; ER absent. Ad1-—3 absent. Ad4 dorsally on Eb4 posteriorly beginning later- ally at Eb4 levator process and extending medially about half distance to medial end of bone; ventrally narrowly on Cb4 dorsoposteriorly medial to inner an- gle formed by Eb4-Cb4 joint; medial edge of muscle tendinous. Ad5 dorsally on Eb4 dorsodistally, ventrally on Eb5 (not illustrated) and posterodistally on CbS, join- ing raphe with TV5. Remarks. Ad5 completely occludes the tiny Eb5 from view. Fink and Weitzman (1982), using single stained material, did not report the existence of Eb5 in Diplophos taenia (their fig. 10 is based on a spec- imen from the same lot as ours). Baldwin and John- son (1996:372), using double-stained material, re- ported that Eb5 is present in Diplophos taenia. RDs separate by space equal to about diameter of one RD, muscle inserts on dorsomedial edge of UPS at junction with Pb4. SO longitudinal muscle fibers questionably pres- ent; requires histological examination. SOD present. Additional remarks. SCL absent. TV4 free from Cb5s. Identity of single UP questionable (see remarks following LI3). NUMBER 11 STERNOPTY CHIDAE Maurolicus muelleri (Gmelin), USNM 323033, 2 specimens, 54.7—55.7 mm. Plate 51 Description. Remarks. The muscles are thin and tiny and it is possible that alternative descriptions we provide in remarks following LE3, LE4, and OD4 may apply. LE1 very fine, inserts by extremely fine tendon on mid-dorsoposterior edge of Eb! lateral to uncinate process. LE2 very fine, inserts by extremely fine tendon on mid-dorsoposterior edge of Eb2. LE3 on tip of Eb3 uncinate process. Remarks. Possibly comprising a bilateral pair of closely applied muscles (LE3, LE3’), based on what appears to be a divided origin (attempts to separate muscle into two natural parts were inconclusive). LE4 originates with cluster of other LEs; inserts on tip of Eb4 levator process. Remarks. Possibly comprising a bilateral pair of closely applied muscles (LE4, LE4’), based on what appears to be a divided origin (attempts to separate muscle into two natural parts were inconclusive). LP absent. LI1 on Pb2 dorsomedial surface posterior to un- cinate process. LI2 on dorsoposterior surface of Pb3. LI3 on UPS dorsally at junction with cartilaginous Pb4. Remarks. Johnson (1992:fig. 3c) illustrated the up- per pharyngeal bones of Maurolicus and the inser- tions of LII—3 and RD. See also remarks regarding UPS following LI3 in Diplophos, Diplophidae. TD comprises TEb2a, TEb2p, and TPb3-Pb4 (larg- er specimen) or TPb4 (smaller specimen). TEb2a overlies bony anterior ends of Pb3 and attaches an- teriorly on dorsomedial surface of Eb2, separated from TEb2p by depression. TEb2p on dorsoposterior surface of Eb2, extending slightly further distally than TEb2a, overlies OD4 origins. TPb3-Pb4 slightly posteroventral and unconnected to TEb2p, attaches to dorsomedial edges of Pb3 and Pb4 and is indistin- guishable posteriorly from SOD; TPb4 is separated from TEb2p by a relatively large gap in smaller spec- imen. Remarks. Separation of TEb2 into TEb2a and TEb2p is problematic. OD3 absent. Remarks. Absence of an OD attachment to Eb3, either as OD3 or OD3—4, is an uncommon speciali- zation within the Euteleostei, and occurs only in some Stomiformes (Table 1) and ateleopodids. It may contribute to resolution of the Neoteleostei tri- chotomy (Fig. 3). OD4—4’ origin ventral to TEb2p, on most of dor- 63 soposterior surface of Pb3 and all of dorsal surface on Pb4, insertion broadly on dorsolateral surface of Eb4 levator process. Remarks. There is a definite separation of the or- igins on Pb3 and Pb4 in the larger specimen, but little or none in the smaller specimen. OD4’, originating on Pb4, is present in other stomiiforms we examined, and we infer that OD in Maurolicus represents a fu- sion of OD4 and OD4’. OP absent, ER absent. Ad1-—3 absent. Ad4 dorsally broadly on posterodorsal surface of Eb4 ventral to LE4, ventrally narrowly on Cb4 dorsal surface just medial to inner angle formed by to Eb4- Cb4 joint. Ad5 dorsally on posterodistal end of Eb4, ventrally on dorsodistal surface of Cb5, joining raphe with TVS (not illustrated) ventroanteriorly in smaller spec- imen, but not in larger specimen. RDs separated by distance greater than one RD, on UPS at junction with Pb4 (see remarks under LI3). SO longitudinal muscle fibers questionably pres- ent, requires histological verification. SOD present, but broadly continuous with TPb3- Pb4 anteriorly. Additional remarks. SCL absent. TV4 free from Cb5s. Identity of single UP questionable. GONOSTOMATIDAE Gonostoma elongatum Giinther, USNM 330309, 162 mm. Plates 52.1, 52.2 Description. LE1 very fine, inserts by long thread-like tendon to dorsal edge of Eb1 lateral to base of uncinate pro- cess. LE2 absent. Remarks. Absence of LE2 rare in fishes. LE3 slender, inserts by very long tendon to carti- laginous tip of Eb3 uncinate process. LE4 relatively large, originates in cluster with oth- er LEs, inserts by tendon on dorsomedial cartilagi- nous end of Eb4 levator process. LP absent. LI1 massive, on almost entire medial edge of Pb2, lateral edge joins or is formed by tendon, which in- serts on Eb! uncinate process. LI2 on posterolateral edge of Pb3. LI3 on lateral surface of UPS. Remarks. See also remarks regarding UP5 follow- ing LI3 in Diplophos, Diplophidae. TD comprises TEb2 anteriorly and well separated TUPS posteriorly. TEb2 attaches broadly on Eb2 an- teromedially, giving rise dorsoanteriorly on each side to short muscle lamina that attaches to tendon in- serting on Ebl uncinate process, joins raphe with 64 OD4 origin beginning at tendon and extending entire length of OD4 origin. TUP5 a small muscle strap attaching to lateralmost edge of UPS. OD3 absent (see remarks under OD3 in Mauroli- cus, Sternoptychidae). OD4 originating dorsally at raphe with TEb2 and ventrally on Pb3 dorsal surface, inserting on anterior surface of Eb4 levator process. Remarks. Left-side OD4 appears to have two al- most continuous origins, and the muscle divides at about mid-length with the ventral branch attaching separately to Eb4 ventral to OD4’. OD4' originating on dorsal surface of Pb4 and in- serting on medial edge of Eb4 levator process. M. UP5-Cb4 of left side originating by long ten- don on medial edge of UPS close to Pb4-Eb4 joint and inserting by long tendon on posterodistalmost cartilaginous surface of Cb4 close to tiny sesamoid bone (not illustrated) on medial side of Eb4-Cb4 joint; muscle on right side originating as pair of slen- der tendons, one on medial edge of UPS close to Pb4- Eb4 joint and the other on ventromedialmost bony edge of Eb4, tendons uniting just before inserting as on left side. Remarks. The tiny sesamoid bone was not seen on either side of two cleared and stained specimens, ca. 130 mm (USNM 327091). OP dorsally on ventral surface of thick medial arm of Eb4, ventrally at and medial to Ad5 insertion on Cb5, interrupted by ER at about mid-length, but clearly continuous ventral to ER. Remarks. Other than Gonostoma, the dorsal at- tachment of OP is always on the dorsoposterior sur- face or posteroventral edge of Eb4. Ad1-3 absent. Ad4 broadly on dorsoposterior edge of Eb4 levator process, narrowing almost to point at tendinous in- sertion on posteromedial end of Cb4. Ad5 on Eb4 near dorsodistal end and Cb5 dorso- distal end; ligament runs along lateral edge of Ad5. RDs appressed medially, insert on posteroventro- medial surface of UPS. SO longitudinal fibers appear to begin at about horizontal joining Eb4 uncinate processes and end anteriorly at attachments to medial surfaces of UP5s. SOD a thin strap arising from SO and medial edge of ER and extending dorsoanteriorly along dorsopos- terior surface of Eb4 from one side to other. Additional remarks. SCL absent. TV4 free from Cb5s. Identity of single UP questionable. Ateleopodiformes ATELEOPODIDAE Ateleopus loppei (Roule), USNM 349721, ca. 350 mm TL, Eastern Atlantic. Plate 53 BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON Additional material. @ = Ateleopus sp., USNM 359636, ca. 300 mm TL, Caribbean, only partial left-side dorsal gill arches (information provided for all muscles for which information is ascertain- able). ® = Ateleopodidae, USNM 361124, Vityaz Cruise station 2560, W. Indian Ocean, small spec- imen, only incomplete left-side dorsal gill arches (information provided below for all muscles for which information is ascertainable). Description. LE1 on Eb! posterodorsal surface distally, com- prises two separate elements oriented longitudinally on each side, right-side with additional element me- dial to paired elements. ® Comprises single element on each side. LE2 on Eb2 posterodorsal surface distally. © Same. LE3 slender, on connective tissue between Eb3 and Eb4 distally. ® Same. LE4 on Eb4 dorsodistally. ® Same. LP absent. @ Same. LI1 on Pb2 dorsally. ® Same. LI2 on Pb3 bony surface dorsally, fuses dorsally with LI3 ventral to origin. ® Same. LI3 left side: on cartilaginous posterior end of Pb3 laterally, fuses dorsally with LI2 ventral to origin; right side: on bony and cartilaginous posterior end of Pb3 laterally, fuses dorsally with LI2 ventral to ori- gin. ® @ on Pb3 bony surface dorsoposteriorly and anterolateral surface of UPS, fuses dorsally with LI2 ventral to origin. Remarks. Johnson (1992:11), who did not examine ateleopodids, stated that LI3 inserts on Pb4 in all non-neotelosts but among neoteleosts, shifts its in- sertion to UPS in stomiiforms and aulopiforms; LI3 and UPS are absent in ctenosquamates. Olney et al. (1993:154), based on a specimen of Ateleopus japonicus Bleeker, reported that LI2 inserts on Pb3 and LI3 inserts on UPS. Ateleopodids lack Pb4, and so the insertion of LI3 would be expected to occur on one or both, Pb3 and UPS. Whether the insertion of LI3 on some portion of Pb3 in USNM 349721 is typical or atypical for that species, does not bear on the problematic phylogenetic position of ateleopodids. TD comprises TEb2 and TPb3. TEb2 attaches to Eb2 dorsoposterior margin and is posteriorly contin- uous with TPb3, which attaches to Pb3 dorsomedial margin and is continuous posteriorly with SO. @ Same. OD3 absent. ® Same. OD4 origin on Pb3 dorsally near anteriormost tip, insertion on Eb4 dorsal bony surface just medial to LE4 insertion. @ Same. Remarks. Absence of Pb4 in ateleopodids presents difficulties in inferring synapomorphic states for the NUMBER 11 origin of OD4. See synapomorphies (34) (35) in Re- sults section. OP distinct dorsally as fan of muscle inserting on Eb4 posteromedially, extending ventrally and fusing with SO medially, becoming continuous ventrolater- ally with AD5 at posteriormost ventromedial edge of Cb4, where both attach to Cb4 by CT. ER absent. Ad1-—3 absent. Adé4 posteroventrally on Eb4 lateral to OP (not vis- ible in dorsal view), ventrally on dorsodistal cartilag- inous end of Cb4 just medial to Eb4-Cb4 joint. Ad5 dorsolaterally on posteromedialmost end of Cb4, ventrally on posterolateral end of Cb5, medially inseparable from OP (see OP above). RD broad, thick, dorosomedian sheet completely included in SO, divides anteriorly at horizontal join- ing Eb4s at mid-length, attaches to UPS and almost entire medial surface of Pb3. SO longitudinal fibers very sparse, restricted to ventralmost area of SO. Additional remarks. SCL absent. TV4 free from Cb5s. Pb1 absent. Aulopiformes AULOPIDAE Aulopus filamentosus (Bloch), USNM 313689, 167 mm. Plate 54 Description. LEI on lateral edge of Ebluncinate process. LE2 on anterior surface of lateral end of uncinate process at and lateral to insertion of TEb2. LE3 on cartilaginous tip of Eb3 uncinate process just dorsal to OD3 insertion. LE4 origin clusters with other LEs; insertion on dorsal bony surface of Eb4 posterolaterally. LP absent. LI1 on Pb2 dorsally at and medial to base of un- cinate process. LI2 on posterodorsal surface of Pb3 at and lateral to TPb3’ attachment and on Pb4 at and lateral to TPb4 attachment anteriorly. LI3 on dorsoanterior surface of UP5, anteriorly at and lateral to mid-point of attachment of TPb4 to Pb4. Remarks. Nelson (1967a:286) stated mistakenly that Aulopus has only two LIs. Johnson (1992) ar- gued persuasively that LI3 in Neoteleostei has shifted its insertion from Pb4 to UPS. He illustrates his in- terpretation in his fig. 3b of Aulopus. TD comprises TPb2-Pb3a, TEb2, TPb3p, TPb3’, TPb4, and TEb4. TPb2-Pb3a anteriorly on dorso- medial surface of Pb2 anterior to LI1, continuing on dorsomedial surface of anterior portion of Pb3 ante- rior to OD3 origin, continuous posteriorly with TEb2 65 by diagonal muscle strand. TEb2 on most of anterior surface of Eb2 uncinate process, ending laterally at medial edge of LE2 insertion, partially dorsal to OD3 origin and anterolateral portion of TPb3p, continuous mid-posteriorly with TPb3p. TPb3p on Pb3 mid-dor- solaterally anterior to LI2 insertion, weakly continu- ous mid-posteriorly with TPb3’. TPb3’ on Pb3 pos- terolaterally just medial to LI2 insertion, well-sepa- rated laterally, but broadly continuous mid-posteri- orly with, TPb4. TPb4 on Pb4 posterolaterally, abutting TEb4 mid-posteriorly. TEb4 attaching to Eb4 posteromedially dorsal to SO and, on left side only, ventral to dorsomedial OP fibers, posteriorly abutting SO. Remarks. The various divisions of TD are fairly well differentiated and more numerous than in any other taxon we examined. OD3 origin on Pb3 ventral to TEb2, insertion on dorsolateral surface of Eb3 uncinate process. OD4 origin on Pb3 ventral to TPb3p, insertion on dorsomedial edge of Eb4 uncinate process. LP absent. OP dorsally on dorsoposterior edge of Eb4 includ- ing uncinate process, joining raphe dorsally with OD4 insertion, ventrally joining ER at ventroposter- ior end of Eb5; not distinguishable from SO or Ad5 below ER. Ad1-—3 absent. Ad4 dorsally on dorsoposterior edge of Eb4, ven- trally broadly on Cb4 dorsal surface medial to inner angle formed by Eb4-Cb4 joint. Ad5 dorsally on distal end of Eb5 and slightly on distal end of Cb4, ventrally on Cb5 posterodistally, dorsomedially joining raphe with OP, medially not separable from OP. RD not represented externally, comprises, single, broad, thick bundle of longitudinal fibers dorsally in esophagous between outer transverse SO muscle and inner mucosal layer; divides anteriorly at about level of medial end of Eb3, inserts on medial edge of Pb4 and UPS and along medial edge of posterior half of Pb3. Remarks. Nelson (1967a:286) noted that paired RDs in Aulopus are absent, and described the inser- tion of this muscle as follows: “... fibers anteriorly attach mostly to the connective tissue of the pharyn- geal roof...” SO longitudinal fibers questionably present; his- tological verification needed (see remarks following RD under Saurida). Additional remarks. SCL present, attached anteri- orly to ventromedialmost end of Hb3, with ligament arising dorsally from mid-posterior edge and attach- ing to ventroanterior surface of Bb3. TV4 free from Cb5s. 66 SYNODONTIDAE Saurida gracilis (Quoy and Gairmard), USNM 140822, 131 mm. Plate 55 Description. LEI! on tip of Eb1l uncinate process. LE2 on Eb2 just lateral to distal end of uncinate process and at and anterior to lateralmost attachment of TEb2. LE3 on tip of Eb3 uncinate process. LE4 on Eb4 amid OD4 insertion; origin clustered with other LEs. LP absent. LI1 broadly on Pb2 anterior process just lateral to TPb2 attachment. LI2 on Pb3 posterolaterally. LI3 on posterolateral surface of UP5. Remarks. See remarks following LI3 in account of Aulopus filamentosus. TD comprises TPb2, TEb2, TEb3, and TEb4. TPb2 on Pb2 medial to LI! insertion, dorsal to M. Pb2-Eb2 origin, continuous posteriorly by diagonal muscle strap with TEb2. TEb2 on anterior bony sur- face of Eb2 uncinate process, continuous posteriorly by diagonal muscle strap with TEb3. TEb3 on dor- soposterior surface of medial end of Eb3 ventral to OD3 insertion and OD4; posteriorly, TEb3 abuts fi- bers of TEb4. TEb4 slender muscle band on mid- dorsoposterior surface of Eb4, attachment ventral to dorsal end of OP, posteriorly abuts fibers of SO. M. Pb2-Eb2 slender, longitudinal muscle, anteri- orly on Pb2 ventromedial to TPb2 attachment, pos- teriorly on anterior edge of medial end of Eb2 un- cinate process partially coincident with TEb2 attach- ment. OD3 origin broadly on Pb3 dorsomedially ventral to TPb2 and anterior to OD4 origin, extending pos- teriorly ventral to TEb2 and inserting on anterior and posterior surfaces of Eb3 uncinate process. OD4 origin on Pb3 posteromedial to posterior end of OD3 origin, insertion on anterior and posterior surfaces of dorsal crest of Eb4. OP broad, on Eb4 dorsoposterior edge medial to dorsal attachment of Ad4 and ventral to TEb4, ven- trolaterally joins—ends at—ER at level of distal end of Eb4, fuses with SO ventromedially. Ad1-—3 absent. Ad4 dorsally on lateral dorsoposterior edge of Eb4, ventrally on Cb4 dorsal surface medial to inner angle formed by Eb4-Cb4 joint. Ad5 dorsally on posterodistal end of Cb4, ventrally on posterodistal end of Cb5, ventromedially joins ra- phe with TVS, dorsally joins ER with OP. RD not represented external to SO, comprises bi- lateral pair of larger, longitudinal muscle bundles ventrally (RD) and very weakly differentiated small- BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON er pair dorsally (RD’); RD’ fibers divide at posterior end of Pb3s and each division extends anteriorly, in- serting along medial edge of most of posterior half of Pb3; RD fibers attach to UP5. Remarks. Large longitudinal muscle bundle pres- ent on each side ventrally in esophagous between transverse muscle and mucosal layers (ventral muscle bundle absent in Aulopus). There are no longitudinal muscles in the lateral walls of the esophagous. Additional remarks. SCL absent. TV4 free from Cb5s. UP4 and Eb5 absent. CHLOROPHTHALMIDAE Chlorophthalmus agassizi Bonaparte, USNM 357798, 121 mm, USNM 159379, 2 specimens, 119-125 mm. Plate 56 Description. LE1 dorsolaterally on bony edge of Eb1 uncinate process just ventral to tip. LE2 on dorsolateralmost surface of distal end of Eb2 uncinate process. LE3 dorsoanteriorly on tip of Eb3 uncinate pro- cess. LE4 on dorsal tip of Eb4 levator process. LP absent. LI1 on bony anterior edge of Pb2 uncinate process. LI2 on dorsoposterior surface of Pb3. LI3 anterior third on Pb4 lateral surface, posterior two-thirds on dorsoanterior edge of UPS at junction with Eb4. TD comprises TPb2-Pb3a, TEb2, TPb3p-Pb4, and TEb4. TPb2-Pb3a laterally on medial ends of proxi- mal Pb2 uncinate process, attaching ventromedially to Pb3 dorsoanterior surface, continuous posteriorly by diagonal muscle strand with TEb2. TEb2 very slender, on dorsomedial surface of proximal end of Eb2 uncinate process, posteriorly continuous by di- agonal muscle strand with TPb3p-Pb4. TPb3p-Pb4 on dorsomedial surfaces of Pb3 and Pb4, continuous posteriorly with TEb4. TEb4 on dorsomedial end of Eb4, continuous posteriorly with SO. OD3 origin on Pb3 just posteroventral to TEb2 and at and dorsal to posterior portion of OD4 origin, lat- erally dorsal to OD4 for much of length of OD4; insertion on Eb3 uncinate process just ventral to LE3 insertion. OD4 origin on Pb3 anteriorly ventral to TPb2- Pb3a and TEb2 and posteriorly on Pb4 anterodorsal surface ventral to OD3, laterally ventral to OD3 for much of length of OD4; insertion broadly on bony anterior surface of Eb4 levator process. OP one or two straps of muscle attaching dorsally to Eb4 dorsoposterior edge just medial to dorsal at- tachment of Ad4, and ventrally to Cb5 posterior sur- NUMBER 11 face medial to Ad5 attachment, only slightly differ- entiated medially from SO. ER absent. Ad1-—3 absent. Ad4 broadly dorsally on Eb4 dorsoposterior edge lateral to OP attachment, ventrally on Cb4 dorsal sur- face just medial to inner angle formed by Eb4-Cb4 joint. Ad5 dorsally on posterodistal ends of Cb4 and Eb5, ventrally on posterodistal surface of Cb5. RD origin tendinous, muscle undivided (unpaired), completely ventral to SO transverse muscle layer, ex- tending broadly anteriorly and becoming continuous with ventral surface muscle fibers of TPb2-Pb3a, at- taching to Pb3, Pb4, and Eb4. SO longitudinal muscle fibers questionably pre- sent; need histological verification. Additional remarks. SCL absent. TV4 free from anterior ends of Cb5s. Myctophiformes NEOSCOPELIDAE Neoscopelus macrolepidotus Johnson, USNM 358034, 127 mm; TCWC 7012.06, ca. 76 mm. Plate 57 Description. Remarks. The description is based on both speci- mens, but the illustration is based on the larger spec- imen because the smaller specimen was deeply dis- sected before it was decided to illustrate the taxon. The muscles in the larger specimen are more robust and folded than in the smaller specimen, and some elements (e.g., ER) and attachments are not readily visible without removing folds of overlying muscle, whereas in the smaller specimen the elements are more obvious. LEI very weak, on mid-dorsoposterior edge of Eb! well lateral to uncinate process. LE2 on raised dorsoposterior edge of Eb2 some- what lateral to mid-length (uncinate process absent). LE3 on anterior edge of Eb3 uncinate process pos- terior to OD3 insertion, extends anteriorly medial to LI2. Remarks. In the smaller specimen, the uncinate process of Eb3 and levator process of Eb4 are tightly bound together. In the larger specimen, they are sep- arated as illustrated, but thin, strong tendinous con- nective tissue joins the processes and muscles that the tissue passes over. LE4 on Eb4 levator process just proximal to car- tilaginous tip, surrounded anteriorly by OD4 inser- tion, extends anteriorly medial to LI2. LP absent. LI1 on Pb2 medial to uncinate process and some- what ventral toTPb2a, passes medial to LI2. 67 LI2 on Pb3 dorsolaterally just anterior to joint with medial end of Eb3. LI3 absent. Remarks. Johnson (1992) argued persuasively that the concomitant loss of LI3 and UPS are a synapo- morphy of Ctenosquamates. TD complex, comprises TPb2a, TPb2p, TEb2, TPb3, and TEb4. TPb2a dorsally on Pb2 between dorsoanteriorly projecting cartilaginous end and car- tilaginous tip of uncinate process, noticeably separate from TPb2p laterally and dorsally, but continuous posteroventrally with it. TPb2p on dorsal bony sur- face of Pb2 just medial to LI] insertion, continuous posteriorly with TEb2. TEb2 on Eb2 dorsomedially, near articulation with Pb3, continuous posteroven- trally with TPb3. TPb3 extensive longitudinally, lies dorsal to Pb3s, muscle strands extending ventrally from mid-anteroventral surface (ventral to TEb2) at- tach to medial surfaces of Pb3s, joins raphe laterally on each side with origins of OD3 and OD4, poster- oventrally continuous with TEb4. TEb4 on Eb4 be- tween OD4 insertion and OP dorsal attachment. OD3 origin on Pb3 laterally ventral to TPb2p and TEb2, joins raphe anteriorly with TPb3, continuous with OD4 origin (fibers appear separate dorsally but mesh ventrally), separating from OD4 posterolater- ally before inserting on Eb3 uncinate process anter- oventral to LE3 insertion. OD4 origin from dorsal raphe with TPb3 and dor- somedial surface of Pb3, insertion on Eb4 levator process. OP dorsally on Eb4 uncinate process medial to Ad4 attachment and ventral to OD4 insertion, inter- rupted by ER at mid-length (ER extends tendinously laterally and attaches to Cb4 distally together with dorsal end of Ad5), but continues below ER and joins raphe with anteroventral portion of Ad5 and lateral end of TV5 before attaching to Cb5 medial to raphe. Ad1-—3 absent. Ad4 dorsally on Eb4 posterolateral edge lateral to OP, ventrally on Cb4 medial to inner angle formed by Eb4-Cb4 joint. Ad5 on Cb4 distal end at attachment of ER lateral tendinous extension, joins raphe with Ad4 anteroven- trally and TV5 medially. RD large muscle cluster consisting of dorsal, in- completely separated bilateral pair of bundles with ventral, somewhat triangular, incompletely separated bundle (RD’) between them, inserting on posterior surface of Pb3 and Pb4 and posterior edge of UP4, and extending anteriorly between and attaching to medial surfaces of Pb3s. SOD ventral to TEb4. SO longitudinal muscles restricted to dorsal and ventral areas of SO. Additional remarks. SCL absent. TV4 free from 68 Cb5s. UP5 and Eb5 absent. Eb4 uncinate process ab- sent. MYCTOPHIDAE Lampanyctus macdonaldi (Goode and Bean), USNM 303167, 118 mm. Plate 58A, B Description. LE1 absent, possibly represented by tough CT in- serting on Eb! uncinate process and attaching to skull (see remarks following LE1 in Diaphus). LE2 on bony Eb2 process. LE3 on cartilaginous tip of Eb3 uncinate process. LE4 origin tendinous; insertion on cartilaginous tip of Eb4 levator process coincident with OD4 inser- tion. Originates with tendinous origins of other LEs. LP absent. LIla on Pb2 anterior to uncinate process (see re- marks under LI1p). LI1p on ventromedial surface of Pb2 uncinate pro- cess, questionably impinging on ventroposteriormost edge of LIla insertion. Remarks. LIla and LI1p are more accurately de- scribed, perhaps, as separated by the posterolateral portion of TPb2a, which lies dorsal to the insertion of LI1p, rather than by the ascending process (= our uncinate process) of Pb2, as stated by Stiassny (1996: 407). Stiassny possibly based her remark on John- son’s (1992:fig. 7B) illustration of the condition in the myctophid Diaphus mollis (but see remarks under LlIla in our description of Diaphus mollis). LI2 on posterolateralmost dorsal surface of Pb3. LI3 absent. TD comprises TPb2a, TPb2p, TEb2, TPb3, and TEb4. TPb2a on dorsal bony surface of Pb2 anterior to Pb2 uncinate process; muscle narrowly continuous posteromedianly with TPb2p. TPb2p on posterior arm of Pb2 dorsally, continuous posteriorly by di- agonal muscle strand with TEb2. TEb2 on dorsal sur- face of medialmost end of Eb2 near articulation with Pb3. TPb3, separated by large gap from TEb2, which continues onto TEb4; muscle attached to Pb3 later- ally ventral to OD3 and OD4, continuous postero- medianly with TEb4. TEb4 on mid-posterior surface of Eb4 ventral to OD4 insertion and medial to OP dorsal attachment, continuous mid-posteriorly with SOD. Remarks. A median longitudinal raphe extends from the posterior half of TPb2p to posterior margin of TEb2, and another extends from TPb3 to dorso- posterior margin of SOD. OD3 lies dorsal to mid-section of OD4; origin on Pb3 dorsomedial edge dorsal to, and coincident with, origin of OD4 mid-section; insertion dorsoanteriorly on Eb3 uncinate process ventral to LE3 insertion. OD4 origin extensive, beginning on Pb3 dorso- BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON medial edge dorsal to TPb3 and continuing posteri- orly ventral to OD3 origin, then dorsally to end of Pb3; insertion massive, on anterior surface of Eb4 levator process. OP dorsally on Eb4 posterior surface medial to Ad4 dorsal attachment, interrupted by ER ventrally, but continuing and attaching to Cb5 just anterior to combined attachment of Ad5 and TV5 to Cb5. Ad1-—3 absent. Ad4 dorsally on posterolateral surface of Eb4; ven- trally, narrowly on Cb4 just medial to inner angle formed by Eb4-Cb4 joint. Ad5 broadly on Eb4 posterolaterally and Cb5 dor- sodistally, joining raphe with TV5 on Cb5. RD on each side comprising large RD ventrally and small RD’. RD inserts completely separately from RD’, attaching to posterior end of Pb4 and UP4 and medial end of Eb4; RD’ joins longitudinal SO muscle layer at dorsoposterior end of Pb3 as layer divides bilaterally (similar to Diaphus, Plate 58C). SOD present. SO longitudinal muscle layer restricted to dorsal and ventral areas; dorsal section divides bilaterally at posterior end of Pb3s, each division extending ante- riorly between Pb3s and attaching to anteromedial surface of its respective Pb3. Additional remarks. SCL absent. TV4 free from Cb5s. UP5 and EbS absent. Eb4 uncinate process ab- sent. Jollie (1954) described and illustrated the gill-arch muscles of Lampanyctus leucopsarus (= Stenobra- chias leucopsarus Eigenmann and Eigenmann), which are very similar to those of L. macdonaldi. Diaphus mollis Taning, USNM 300877 (Gulf of Mexico), 2 specimens, 50.5—54.4 mm; USNM 274201, 54.2 mm (E. of New Zealand). Plate 58C Description. Remarks. We found no differences worth noting among the specimens of D. mollis from the two geo- graphic areas. LEI very fine, on dorsal tip of bony process just lateral to tip of Ebl uncinate process, where it is joined by insertion of medial edge of LE2. Remarks. Stiassny (1996:412) stated that reduction of LEI to a thin slip (Neoscopelidae) or ligament (Myctophidae) is a synapomorphy of myctophiform fishes. We agree that a reduced LE] may be syna- pomorphic for myctophiforms, but we find that LE1 in myctophids also may be present as a thin slip of muscle. LE2 on posterodorsal edge of Eb2 where bone widens; muscle extends anteriorly and attaches to NUMBER 11 bony process on Eb! lateral to uncinate process, thence continuing to origin on cranium. LE2’ very fine, becoming even finer and tendinous before inserting among LE2 fibers (absent in Lam- panyctus). LE3 on cartilaginous tip of Eb3 uncinate process. LE4 on cartilaginous tip of Eb4 levator process coincident with Od4 insertion. LI1 with single insertion equivalent to Llla of Lampanyctus macdonaldi. Remarks. Johnson (1992:fig. 1B) illustrated LI1 in a specimen putatively identified as Diaphus mollis (AMNH 29449). In contrast to our finding for the species, his figure shows LI1 as having anterior and posterior sections with insertions equivalent to LIla and LIlp of Lampanyctus macdonaldi (q.v.). The specimen Johnson illustrated is actually identifiable as D. dumerilii (USNM 301132). AMNH 29449, however, is identifiable as D. mollis (from off Ber- muda). M. Stiassny examined a specimen from the same AMNH lot and informed us (E-mail, 15 May 2000) that LI1 has a single insertion. Aside from D. dumerilii and L. macdonaldi, we find LI1 also has two insertions in D. lucidus (USNM 269439), D. re- gani (USNM 269381), Hygophum hygomi (USNM 206616), and Protomyctophum normani (USNM 206631). Stiassny (1996:407) interpreted the subdi- vision of LI] as a synapomorphy of the Myctophidae. The character state for LI] in D. mollis is probably autapomorphic. LI2 on posterolateralmost dorsal surface of Pb3. LI3 absent. TD comprises TPb2a, TPb2p, TEb2, TPb3, and TEb4. TPb2a on dorsal bony surface of Pb2 anterior to Pb2 uncinate process, dorsal to anteriormost edge of LIla insertion, and narrowly continuous postero- medianly with TPb2p. TPb2p on posterior arm of Pb2 dorsally, continuous posteromedially with TEb2. TEb2 on dorsal surface of medialmost end of Eb2 near articulation with Pb3. TPb3, separated by large gap from TEb2, on Pb3 ventrolaterally, continuous posteromedianly with TEb4. TEb4 on mid-posterior surface of Eb4 ventral to OD4 insertion and medial to OP, with which in joins a raphe laterally, and is continuous mid-posteriorly with SOD. Remarks. A median longitudinal raphe extends from TPb2a to posterior margin of TEb2, and another extends from TPb3 to dorsoposterior margin of SOD. TPb2a is much narrower relatively than it is in Lam- panyctus; it lacks the anterior portion without the me- dian raphe. OD3 lies dorsal to mid-section of OD4; origin on Pb3 dorsomedial edge dorsal to, and coincident with, origin of OD4 mid-section; insertion dorsoanteriorly on Eb3 uncinate process ventral to LE3 insertion. OD4 origin extensive, beginning on Pb3 dorso- medial edge, passing dorsal to TPb3 and continuing 69 posteriorly ventral to OD3 origin, then dorsally to end of Pb3; insertion massive, on anterior surface of Eb4 levator process; posteroventrally joining raphe with OP. OP a narrow strap; dorsally, narrowly on dorso- posterior Eb4 bony surface joining raphe laterally with OD4 insertion and medially with TEb4 attach- ment; ventrally on distalmost tip of Cb5, not inter- rupted by ER, which appears to be absent. Ad1-3 absent. Ad4 dorsally on posterolateral surface of Eb4, ven- trally, narrowly on Cb4 just medial to inner angle formed by Eb4-Cb4 joint. Ad5 narrowly on Eb4 posterodistally and Cb5 an- terodistally. RD comprising large RD ventrally and small RD’ dorsally on each side. RD inserts completely sepa- rately from RD’, attaching to posterior end of Pb4 and UP4; RD’ joins longitudinal SO muscle layer at dorsoposterior end of Pb3 as layer divides bilaterally. SOD present. SO longitudinal muscle fibers appear to be restrict- ed dorsally and ventrally in SO and originate at about horizontal between distal ends of Eb4s and extend anteriorly about to horizontal between distal ends of Eb3s. Additional remarks. SCL absent. TV4 free from Cb5s. See also additional remarks under Lampanyc- tus macdonaldi. UPS and Eb5 absent. Eb4 uncinate process absent. Results—Pre-Acanthomorpha Much of the raw data on the muscles is presented in Tables 1-6. The following discussion summarizes information for most of the muscles and assesses how the information bears on the pre-acanthomorph clad- ogram (Fig. 4). Muscle synapomorphies are italicized and numbered in parentheses, and summarized in Ta- ble 7. They are interpreted as such by parsimony based on how the character states are distributed among the taxa on the cladogram (for an example, see synapomorphy 9 below). We noted many cyprin- iform synapomorphies, but we include only those that occur homoplastically in other taxa (the two cyprinid genera we examined are closely related). We usually ignore autapomorphies of terminal taxa, but autapo- morphies that might be interpreted as synapomor- phies based on results of a total evidence analysis (not performed by us) are discussed. In some cases we have noted skeletal synapomorphies; these are not numbered nor indicated by italics. (1) Presence of Ads is a synapomorphy of Actin- opterygii. (2) Attachment of part of SO to a dorsal gill-arch skeletal element is a synapomorphy of Actinopteri. 70 Polypterus is the only taxon in which SO does not attach to a dorsal gill-arch element. (3) Insertion of LEI at dorsomedialmost end of Eb1 is synapomorphic for Osteoglossomorpha (nec- essarily ignoring the two osteoglossomorph taxa that lack LE1 or in which it is unrecognizable, i.e., fused indistinguishably with another muscle). Among other pre-acanthomorphs, the insertion point of LEI is quite variable, but with the exception of Elops, is always lateral to the medialmost end of Eb1. LE1 is rarely absent, and only autapomorphically. (4) Insertion of LE2 at dorsomedialmost end of Eb2 is synapomorphic for Osteoglossomorpha. Pan- todon, which is well nested among the osteoglosso- morph clades on the basis of other evidence, is ex- ceptional. Among the other pre-acanthomorphs, the insertion point of LE2 is quite variable, but with the exception of Elops, is always lateral to the medial- most end of Eb2. LE2 is rarely absent, and only au- tapomorphically, and the state is ignored in those taxa that lack it or in which it is unrecognizable, i.e., fused indistinguishably with another muscle. (5) Doubling of LE3 is a synapomorphy of Os- meroidea. A doubling of LE3 (LE3, LE3’) also oc- curs in some Clupeomorpha. LE3 is rarely absent, and only autapomorphically, except among Osteo- glossomorpha, in which three of the ten genera ex- amined lack the muscle or it is unrecognizable, 1.e., indistinguishably fused with another muscle. See also (6) for Salangidae. (6) Doubling of LE4 (LE4, LE4’) is a synapomor- phy of Osmeroidea. Too late for detailed inclusion, we examined the salangid Protosalanx chinensis (Os- beck), USNM 85840, and note that it also has LE3 and LE4 doubled. This corroborates Johnson and Pat- terson’s (1996:307) inclusion of “‘salangids” with Osmeridae. (7) LP present is a synapomorphy of Clupeoidet: see remarks following (8). (8) LP present is a synapomorphy of Characifor- mes. LP has been reported for several diverse groups of pre-acanthomorph fishes, but we reserve the name for a muscle that inserts on Eb4, with or slightly sep- arate from LE4, and originates “closely apposed to the surface of the epaxial body muscles [of clu- peoids]”’ (Greenwood and Lauder, 1981:215), or on various cranial bones in all other groups that have it. Among pre-acanthomorphs, LP is restricted other- wise to Chanos, also a member of Otocephala. Al- though absent in Denticeps, currently recognized as the sister group of all other clupeomorphs, the pres- ence of LP will probably be found to be synapo- morphic for Otocephala. LP next appears in Lampri- diformes, first clade of Acanthomorpha. Levator pos- terior has been applied to a variety of morphologi- cally similar but non-homologous muscles as the following discussion amply demonstrates. BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON Greenwood and Lauder (1981:228) stated that all of the otophysan taxa they examined, which included representatives of all four major groups (Fig. 4), have LP. We question the presence of LP in otophysans other than characiforms. The muscle Greenwood and Lauder identify as LP in cypriniforms is based on Winterbottom’s interpretation (1974:254—256), and corresponds to our LCb5 (see Winterbottom’s fig. 22a), which inserts on Cb5; that of siluriforms cor- responds to our LI4 (see Winterbottom’s fig. 21), which inserts on UP4. We are uncertain which gym- notiform muscle Greenwood and Lauder considered to be LP unless it is the muscle we identify as LE4 (inserts on Eb4), in which case they would consider LE4 as missing (see also Remarks following LI4 in description of Diplomystes). If future study should indicate that LE4 is absent in gymnotiforms, then ab- sence of LE4 would appear to be a synapomorphy of siluriforms + gymnotiforms. (9) LII appears first in Ginglymodi and is a syn- apomorphy of Neopterygii. From its first appearance, LI1 is the only levator that is present in all actinop- terygian taxa we examined. Lauder and Wainwright (1992:457) indicate that LlIs are plesiomorphic for Chondrichthyes + Actin- opterygii. We are unable to find the source on which they based their indication. LIs in Actinopterygii first appear in Neopterygii, are synapomorphic for that group, and cannot be plesiomorphic for Chondrich- thyes + Actinopterygii: chondrichthyans have no le- vators originating on the cranium. SPbI1 first appears in Chondrostei and is a syna- pomorphy of Actinopteri. LI] is attached to SPbl when LI1 first appears in Ginglymodi. Halecomorpha and Osteoglossomorpha lack SPb1, so loss of SPb1 (and hence attachment of LI1 to it) is a synapomor- phy of Halecostom1. SPb1 reappears in Elopomorpha and again LI1 is attached to it (condition present in both elopiform families and both albuliform families, but absent in all Anguillomorpha). Reappearance of SPbI is a synapomorphy of Elopomorpha, but wheth- er attachment of LI1 to SPb1 is also a synapomorphy is equivocal, because we do not know whether LI1 was attached to SPb1 in the ancestor of those forms that lack SPb1. The situation repeats. Loss of SPb1 (and attachment of LI1 to it) is a synapomorphy of Clupeocephala, but SPb1, with LI attaching to it, re- appears in, and is a synapomorphy of, the clade Pla- tytroctidae + Alepocephalidae. The apparent conclu- sion to be drawn is that when both SPb1 and LI1 are present, LI1 attaches to SPbl. (10) Division of LII (our Lila, LIIb) is a syna- pomorphy of Myctophidae. First reported by Stiassny (1996:407), our data corroborate her findings. Pres- ence of only LIla in the myctophid D. mollis appears to be autapomorphic (see also remarks under LI1 in account of D. mollis). NUMBER 11 (11) LI2 appears first in Neopterygii and is a syn- apomorphy of that clade. Aside from some special- ized osteoglossiforms, LI2 is present in all but three, distantly related, pre-acanthomorph taxa that we ex- amined. (12) Insertion of LI2 to include UP4 is a syna- pomorphy of Anguilliformes. LI2 inserts primitively on Pb3, and as a specialization may insert on other skeletal elements. (13) LIB is probably a synapomorphy of Teleostei. LI3 appears first in Teleostei together with the first appearance of Pb4, on which it inserts primitively, and is probably a synapomorphy of that clade (but if not that clade, then of Osteoglossiformes and, inde- pendently, Elopocephala). The problem in recogniz- ing LI3 unequivocally as a synapomorphy of Teleos- tei is caused by the state of the LIs in Hiodon, which is the sister group of all other Osteoglossomorpha (= Osteoglossiformes), which, in turn, is the sister group of Elopocephala (= all other Teleostei). Hiodon has only two LIs, LI1 and a second, relatively large LI, which may simply be LI2 or, conceivably, a fused LI2 and LI3. There are two possible scenarios: 1) the second LI represents only LI2, and LI3 has evolved independently in Osteoglossiformes and Elopocepha- la; 2) the second LI represents fused LI2 and LI3 or only LI2, LI3 having been lost; the presence of LI3 would, thus, represent a teleostean synapomorphy. We prefer the second scenario for reasons that follow. Osteoglossiformes, sister group of Hiodon, com- prises two clades: Notopteroidei and Osteoglossoidei. Three of the four genera of notopteroids are highly specialized with many muscle fusions or losses, and LI2 and LI3 (if the latter was present) are either lost or indistinguishably fused with other muscles. The fourth genus, Notopterus and sister-group of the other three genera, has both LI2 and LI3. The character state for LI3 at the base of Notopteroidei is “LI3 present or ?”’. Four of the five genera of Osteoglos- soidei (including all three genera in the two basal- most clades) have LI2 and LI3 (the exception has only LI2). The character state for LI3 at the base of Osteoglossoidei is ““LI3 present.’ Combining these two states, one concludes parsimoniously that the character state of LI3 for Osteoglossiformes is “LI3 present.” The state for Hiodontiformes is, at its most conservative, “LI3 absent”; therefore, the character state for Osteoglossomorpha would be “LI3 present or absent.” Given that ““LI3 present” is the plesio- morphic state for Elopocephala, combining the states for the two teleostean clades results parsimoniously in “LI3 present’ as the (synapomorphic) character state for Teleostei. (13a) Insertion of LI3 to include Pb3 is a syna- pomorphy of Osteoglossoidei. Among osteoglosso- morphs, LI3, when present, inserts variously on Pb3 and Pb4, Pb4, or Pb3 and UPS. Pb4 is plesiomorphic, Wl therefore the insertion to include Pb3 is specialized. Insertion on UP5 is autapomorphic for Pantodon. (14) Absence of LI3 is a synapomorphy of An- guilliformes. (15) Shift of the insertion of LI3 from Pb4 to UPS is a synapomorphy of Neoteleostei. See remarks fol- lowing (16). (16) Loss of LI3 and UPS is a synapomorphy of Ctenosquamata. Johnson (1992) first hypothesized these two synapomorphies (15 and 16) and our find- ings corroborate his hypotheses. TD is absent in Polypterus, possibly a result of loss or great reduction of Eb4. As such, it is not possible to decide if presence of TEb4 is a synapomorphy of Chondrostei or Actinopterygil; we assume the former (see 18). (17) Absence of attachment of TD to Eb4 is a syn- apomorphy of Albuliformes. (18) Presence of TEb4 is a synapomorphy of Chondrostei. (19) Presence of TEb4 is a synapomorphy of Clu- peomorpha. (20) Presence of TEb4 is a synapomorphy of Eu- rypterygil. (21) Attachment of TD to include Pb2 is a syna- pomorphy of Gymnotiformes + Siluriformes. (22) Attachment of TD to include Pb2 is a syna- pomorphy of Eurypterygii. (23) Presence of TPb2a and TPb2b is a synapo- morphy of Myctophiformes. (24) Attachment of TD to Pb3 first occurs in Neopterygii, and is a synapomorphy of Neopterygii. (25) Attachment of TD to Pb4 first occurs in Te- leostei (where Pb4 first appears) and is a synapo- morphy of Teleostei. (26) Attachment of TD to Eb2 is a synapomorphy of Osteoglossoidei (Table 4). (27) Attachment of TD to Eb2 is a synapomorphy of the unnamed clade: (Characiformes (Siluriformes, Gymnotiformes). Attachment of TD to Eb2 is ple- siomorphic for Characiformes and present or absent in its sister group, Siluriformes (present) + Gymno- tiformes (absent). (28) Attachment of TD to Eb2 is a synapomorphy of Neoteleostei. (29) Presence of TEb2 is a synapomorphy of Neo- teleostei. (30) Presence of TEb2 is a synapomorphy of Os- teoglossidae. (31) OD4 first appears in Neopterygii, in which it attaches to Pb3, and is a synapomorphy of that clade. After its first appearance, some component of OD4 (OD4’ or OD4 as represented in the fused OD3—4) is rarely absent. See also (32-35). (32) OD3 first appears in Teleostei and is a syn- apomorphy of that clade. An OD3 component (OD3 or OD3-4) is rarely absent. (33) Except for the osteoglossomorph Notopterus, where it occurs homoplasiously, some component of OD4 (OD4, OD4’, OD3-4) first attaches to Pb4 in Elopocephala and is a synapomorphy of that clade (Table 4). (34) Loss of attachment of an OD4 component (OD4, OD4’, or OD3-4) to Pb4 is a synapomorphy of Neognathi (Table 4). Absence of Pb4 in ateleo- podids makes it impossible to polarize the group with regard to the origin of OD4. However, the origin of OD4 in ateleopodids is well anterior on Pb3 and would appear to indicate that it is highly unlikely that Pb4 would have participated in the origin; we assume this in our analysis of the synapomorphic states for (34) and (35). (35) Attachment of an OD4 component (OD4, OD4’, or OD3-4) to Pb4 re-occurs in Stomiiformes and is a synapomorphy of that clade (Table 4). See comments in (34). (36) A vertically oriented OD4 is a synapomorphy of Albuliformes. (37) OP. first appears in Halecomorpha and is probably a synapomorphy of Halecostomi. (38) Absence of OP is a synapomorphy of Gono- rynchiformes. See comments in (40). (39) Absence of OP is a synapomorphy of Cyprin- iformes. See comments in (34). (40) Absence of OP is a synapomorphy of Platy- trochtidae + Alepocephalidae (two representatives of Alepocephaloidea). OP frequently appears to be fused to SO and/or Ad5 and its presence is often questionable. (41) ER first appears in Teleostei and is a syna- pomorphy of that clade. ER is often absent, and is never present when OP is unquestionably absent. (42a) Presence of RecD4 is probably a synapo- morphy of Chondrostei. RecDs are relatively uncom- mon among pre-acanthomorphs. RecD4 is absent in Polypterus, conceivably as a result of the loss or great reduction of Eb4. As such, it is not possible to decide with certainty if presence of RecD4 beginning with Chondrostei is a synapomorphy of Chondrostei. (42) Presence of RecD4 is a synapomorphy of Os- teoglossomorpha. (43) Presence of RecD4 is a synapomorphy of Cy- priniformes. (44) RecD2 first appears in Osteoglossiformes and is a synapomorphy of that clade. (45) Presence of RecD2 is a synapomorphy of Cy- priniformes. (46) Presence of RecD2 is a synapomorphy of An- guilliformes. Bilaterally paired Pb muscles. The muscles of this group (Table 5) have their origins on Pbs and/or UPs and their insertions on Ebs and/or Cbs. These paired Pb muscles are not homologous with TD component muscles, which, in pre-acanthomorphs, do not be- BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON come divided bilaterally by losing their median por- tions, as they may in acanthomorphs. Furthermore, the paired Pb muscles are usually aligned longitudi- nally and, within a species, often have origins on the same Pbs and UPs as the TD muscles. Paired Pb muscles, which first appear in the Osteoglossomor- pha, tend to be common only in Osteoglossomorpha and Anguillomorpha. Concerning specific bilaterally paired Pb muscles: (47) Presence of M. Pb4-Eb2 is a synapomorphy of Osteoglossoidei. (48) Presence of M. UP5-Cb4 is a synapomorphy of Albuliformes. (49) Presence M. UP5-Cb4-Eb5 is a synapomor- phy of Characiformes. (49a?) Ad1-—3 are present in only three unrelated groups of pre-acanthomorphs: Polyodontidae, Nota- canthidae, Cyprinidae. However, Edgeworth (1935: 131) reported that Ad1—4 are present in Acipenser sturio (not examined by us), but not A. ruthenus, A. fulvescens, and Scaphirhynchus. It is possible, there- fore, that Ad/—3 are synapomorphic for Chondrostei. (50) Attachment of Ad5 to Eb5 (when EDS is pre- sent either as an autogenous element or when puta- tively fused with Eb4 or Cb4) is a synapomorphy of Teleostei. This state is by far the most common among pre-acanthomorphs up to Neoteleostei. Ad5 is one of the most consistently present muscles and is always attached at its ventralmost or posteriormost end to Cb5. Among the taxa we examined it is un- equivocally absent only in Polypterus and the eel, Synaphobranchus, both of which lack Cb5. Only the attachments to other skeletal elements will be dis- cussed further. The primitive state for Ad5 in Actinopterygii is confused by the absence of the fifth arch in Polyp- terus, but it appears that the state is probably that of Ad5 attaching Cb5 to Cb4 (Table 6; Acipenser, Atractosteus, Amia). The primitive state appears au- tapomorphically above Halecomorpha in: Elopifor- mes, Anguilloida, Diplomystes, Lepidogalaxias, Eso- cidae, Ateleopus, Aulopiformes, and Neoscopelus. If attachment of Ad5 to Eb5S or to Eb4* (fused Eb5-Eb4) represents the same state (attachment to EbS), then: (51) Attachment of Ad5 to Eb4 (in the absence of EbS or Eb4*) is a synapomorphy of Stomiiformes. The state occurs otherwise, independently, only in Myctophidae, Lepidogalaxias, and in Polyodon. (52) Attachment of Ad5 to Eb4, when Eb5 is au- togenous, is a synapomorphy of Cypriniformes; it oc- curs otherwise, independently, only in Searsia. (53) Presence of RD is a synapomorphy of Neo- teleosteéi. Wiley (1976:31) observed that RDs in amiids and lepisosteids are derived from the outer circular mus- cle layer of SO, in contrast to their derivation from NUMBER 11 the inner longitudinal layer of SO in other fishes. Thus, he inferred that the RDs of amiids and lepi- sosteids are not homologous with RDs in other fishes, but can be interpreted as homologous in lepisosteids and amiids in having similar derivations. In opposi- tion to that inference, Wiley (p. 32) also described several differences between the attachments of RDs in lepisosteids and amiids, and inferred that the state of the RDs in each of these two groups is autapo- morphic, and thus their RDs are not homologous. Plotting the distribution of RD on the cladogram (Fig. 4), disregarding the differing character states of RD in lepsosteids and amiids, indicates that some kind of RD is a synapomorphy of Neopterygii (but RDs lost is a synapomorphy in the next branch, Te- leostei); thus, Wiley was justified in his first infer- ence. Lacking knowledge of the particular character- istics of RD in the hypothetical neopterygian ances- tor, there are three possibilities: RD resembled that of either extant lepisosteids or that of extant amiids or differed from both. His second inference, there- fore, may be correct, but is unwarranted. The lack of homology at a higher phylogenetic level does not necessarily preclude homology at a lower level. Another possibility is that the presence of RD may be a synapomorphy of Ginglymodi + Halecomorpha, which if corroborated by overall parsimony, will re- quire rearrangement of these two clades in relation to Teleostei. (See also summary following this sec- tion.) RD exhibits three states in pre-acanthomorphs we examined: RD completely dorsal to SO (SOD ab- sent): Amia, Atractosteus, and Lepidogalaxias (all pre-Neoteleostei); RD completely ventral to SO (SOD absent): Ateleopodiformes and Aulopiformes (both Neoteleostei); RD and SOD present: Pantodon (pre-Neoteleostei) and Stomiiformes and Myctophi- formes (Neoteleostei). Three aulopiform taxa (Gigantura chuni Brauer, USNM 221034; Omosudis lowei Giinther, USNM 206792; Parasudis truculentus (Goode and Bean), USNM 159095), in addition to those listed in Fig. 4, were examined. All agree with those listed in Fig. 4 in having RD completely ventral to SO. Although a very elongate RD was used to relate Gigantura to aulopiforms (i.e., synodontids, Rosen, 1983:440— 441), the importance of RD being ventral to SO in pre-acanthomorphs has gone unnoticed, and offers additional support for a relationship of Gigantura to aulopiforms, but not specifically to synodontids. Based on the intermusculars, Johnson and Patterson (1995:31) placed Gigantura with alepisauroids with- in aulopiforms, and Baldwin and Johnson (1996) pre- sented evidence that Gigantura and Bathysaurus form a sister group, which is the sister group of ale- pisauroids. Presence of a ventral RD in Ateleopodiformes, 73 which forms a polytomy with Stomiiformes and Eu- rypterygii might be evidence to relate Ateleopodi- formes with eurypterygian Aulopiformes. It is not possible, however, to polarize the ventral and dorsal states of RD. One character of RD, however, a single broad band of muscle that only divides at its attach- ment to the dorsal gill-arch skeleton, is unique to ateleopodids and some aulopiforms. At least Aulopus, which Baldwin and Johnson (1996:359) included in the basalmost aulopiform clade, and Chlorophthal- mus, which they included in one of the next two au- lopiform clades, share this state of RD with Ateleo- pus. Based on this evidence it is possible that Ate- leopodiformes and Aulopiformes form a sister group, possibly resolving the Neotelostean trichotomy. Such an arrangement would require addressing conflicts provided by synapomorphies (20 and 22), which sup- port an aulopiform and ctenosquamate clade. The presence of SOD basally in Acanthomorpha indicates that this state is plesiomorphic for Ctenos- quamata. An aulopiform-ateleopodiform resolution of the neoteleostean trichotomy (Fig. 4), in which the new clade is the sister group of Ctenosquamata, would indicate that presence of SOD is a synapo- morphy of Neoteleostei and its absence a synapo- morphy of Aulopiformes + Ateleopodiformes. (54) Longitudinal SO muscle layer first appears in Halecostomi and is a synapomorphy of that clade. We were unable to find a study on the composition of the SO muscle fibers in fishes and doubt that one exists. Indeed, the subject is rarely mentioned. Nev- ertheless, after discussion with several colleagues, we became aware that there is a generally erroneous con- cept that SO in fishes comprises an outer transverse muscle layer and an inner longitudinal muscle layer (e.g., Wiley, 1976:30—32). We find, however, that al- though an outer transverse layer is always present, a longitudinal layer is not. In some taxa, we were un- able to determine on gross examination if a longitu- dinal muscle layer was present. Although we usually do not describe them, there are different distribution patterns of the longitudinal muscle fibers around the esophagus (e.g., the fibers may be concentrated dor- sally and ventrally and absent laterally or they may be distributed more-or-less evenly around the esoph- agus). We suggest that a detailed histological study of the SO muscle layers is in order and would prob- ably reveal important phylogenetic information. (55) Presence of TV4 is a synapomorphy of Hal- ecostomi, and, within the pre-acanthomorph portion of that clade, TV4 is absent only in the galaxioid Retropinna. Note: attachment of TV4 to Hb4 is an autapomorphy of Halecomorpha. (55a) Attachment of TV4 to Cb4 is a synapomor- phy of Teleostei. Note: Absence of Hb4 is also a syn- apomorphy of Teleostei. (56) Attachment of TV4 to Cb5 is a synapomorphy 74 of Cypriniformes. Among pre-acanthomorphs, TV4 is primitively unattached to Cb5s. TV4 is attached to Cb5s in only three other pre-acanthomorph taxa, in each of which it is autapomorphic. (57) SCL first appears unquestionably in Albuli- formes (= Albulidae) and is a synapomorphy of that clade. Stiassny (1992:269) reported that the “‘semi- circular ligament system ... is an innovation of the acanthomorph fishes,” by which she implied the first appearance of SCL (and the way in which RecV4 and ObV3 attach to it). She further stated (p. 270) that SCL is absent in pre-acanthomorphs. Stiassny is correct in indicating that SCL is primitively absent in pre-acanthomorphs, but it occurs in various pre- acanthomorph taxa. SCL is lacking in the most ad- vanced pre-acanthomorphs, myctophiforms and most aulopiforms, but is present in the basalmost group of acanthomorphs, Lampridiformes. Only in the sense that its presence is homoplastic can it be considered to be an acanthomorph innovation. (58?) Presence of SCL is a synapomorphy of Elo- pomorpha, 1f the condition of SCL in Elopiformes (see description in additional remarks in Elopidae) can be considered to indicate its presence. The differences in character states between Elopiformes and Albulifor- mes leaves unresolved which state is plesiomorphic. (59?) If SCL is considered to be absent in Elopi- formes, the condition described in additional remarks in Elopidae is a synapomorphy of Elopiformes. (60) SCL is a synapomorphy of the Esociformes. Summary (Table 7). The dorsal gill-arch muscu- lature (and TV4 and SCL) provides support for monophyly of Actinopterygii and several of its cur- rently recognized pre-acanthomorph clades. In gen- eral, support exists mainly for clades that are already well supported in the literature. The dorsal gill-arch muscles provide conflicting ev- idence regarding the interrelationships of the Gingly- modi, Halecomorpha, and Teleostei. The monophyly of the Halecostomi (Halecomorpha + Teleostei) is supported by three synapomorphies: (37), presence of OP; (54), presence of SO longitudinal muscle layer; (55), presence of TV4. Alternatively, three synapo- morphies conflict with halecostome monophyly and, instead, support the monophyly of the Halecomorpha + Ginglymodi: presence of RD, absence of attachment of TD to Eb4, and absence of RecD4. Thus, based on our muscle evidence, each hypothesis is equally par- simonious. The first is congruent with the generally accepted hypothesis expressed in our cladogram (Fig. 4). The second is congruent with the molecular evi- dence presented in Gardiner et al. (1996). We suggest that the interrelationships of Ginglymodi and Hale- comorpha are worthy of additional study. Hilton (2003) hypothesized a different set of in- terrelationships for Osteoglossoidei than the one in our Fig. 4. His classification exchanges the position BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON of Osteoglossidae (his Osteoglossinae) with that of Arapaimidae (his Heterotinae). His classification dif- fers from ours only in aligning Pantodon as sister group of Osteoglossidae (his Osteoglossinae) instead of Arapaimidae (his Heterotinae). Given Hilton’s ar- rangement, the major change would be that synapo- morphy 13a would no longer be a synapomorphy of Osteoglossoidei, but would, instead, become a syn- apomorphy of Osteoglossidae. The other osteoglos- somorph synapomorphies would remain the same. Although not supporting monophyly of the two oto- cephalan clades, Ostariophysi and Clupeomorpha, two characters, the gongyloid cartilage (Di Dario, 2002) and LP are suggestive of close relationship of the clades, i.e., supportive of Otocephala. Among pre- acanthomorphs, the gongyloid cartilage is present only in Chanos (Ostariophysi) and most of the genera of Pristigasteroidea and Engrauloidea, (Clupeomorpha; see Additional remarks under Chanos and Cetengrau- lis). Similarly, among pre-acanthomorphs, LP is pre- sent only in characiform genera (Ostariophysi), Chan- os (and possibly Gonorynchus, in which it was re- ported present by Greenwood and Lauder, 1981:228, although we did not find it), and all Clupeomorpha, except Coilia and Denticeps. It is also possible that LCbSA in cyprinids represents a modified LP. There is evidence (not listed in Table 7) based on the state of RD and SO that Aulopiformes and Ate- leopodiformes form a clade, but other evidence may conflict. Acanthomorpha Additional material of acanthomorph taxa not mentioned in descriptive accounts (used for supple- mentary data, e.g., Table 8). Cleared and stained. ACROPOMATIDAE: Acropoma sp., USNM 287444 (2); Apogonops anomalus, USNM 287447; Doederleinia berycoides (Hilgen- dorf), USNM 290474. AMBASSIDAE: Ambassis sp., USNM 218805 (4 specimens); A. macleayi, USNM 173817 (1). ARRIPIDAE: Arripis georgianus (Valen- ciennes), USNM 267149, 287442 CARANGIDAE: Car- angoides crysos (Mitchill), USNM 167629; Seriola sp., USNM 306575; Trachinotus falcatus (Linnaeus), USNM 280104. CHAETODONTIDAE: Chaetodon trifas- ciatus Park, USNM 278739; C. melannotus Bloch and Schneider, USNM266894; Forcipiger flavissimus Jordan and McGregor, USNM 340962; Heniochus acuminatus (Linnaeus), USNM_ 147893. CompaeE: Coius sp., USNM 269799. DINOLESTIDAE: Dinolestes lewini (Griffith and Smith), USNM 59932. DINOPER- CIDAE: Dinoperca petersi (Day), USNM 269543. DREPANIDAE: Drepane africana Osorio, USNM 306264; D. longimanus (Bloch and Schneider), USNM 284472. EPHIPPIDAE: Chaetodipterus zonatus (Girard), USNM 220719, 220721; Ephippus orbis 75 NUMBER 11 Table 1.—Distribution of certain gill-arch muscles and SCL in genera of preacanthomorph fishes; homologies not implied. X = present; from Cb5; L = free either OD3 or OD4 present, or only OD3-4 present; F attached to CbS5; E longitudinal and transverse muscle layers present; T = only transverse muscle layer present; ? = presence or state questionable; * = state of LI. present or absent; A = ve TOS x >| pAd oleae lee leet fo] | | .[c,| [aaj (eaten fea) fea| ea) (oa dos|* xx >| | | [| ] x ey Ayes} ye} 4) 4jo- oo Kon 4)4 Sy) 4)4 yc: 4 ee [pe]salpe|selpe]selbel>< | Tbe] x | paoay ; | ><> aCED ; ~ je aay ‘ee [| 1a JE Ix ~ ae >|><[o-|5< >< |>e[>e[><]>e[<[ »|<)x]x x le. dO} | | | >|>[>< po pepe peepee lo-lb<|>i | > Fale a x zao |_| | 2 rT + ad Fd J ie — rot SS — Raia j a | la Ah) [ aq | ><|><|e<]>< Etedeediedees cals falta mle<| [ele (4 edie rdkadkadias »<|>< x |< ><|><|><1><| *|><]><| xd |x | | Maal FER eux x eae ea ae ii | xx va | eked | | || ra] x\x)x|x >< >< |><|><| |<] <] ><] [><|><[><[>/ |x x ea [pale | | eax <|><|><|><]>< | ><[><|<]><[><]><[><[>< ><[><|><[>< | “ “ ca gi | rasta fx *||X pe |pa[5e|5e]belbelse[se|5< ||<| [|X ]xIx ><|be]>e[><[>< Ibe] ><|>e[5<] 5<[5<]><[><]< x em [be g 2 e | eles | : SE 3 |e | (gl | Ista lel 18 | sige gs EE 5 & | (He 188 | 18 | Jasieeeleessisges Se. cs 8 es ese | eee.) see S/SSSSSS SSS 88/5 S55 SSS SS SE5 SS Salis SS esis Ss Sa SS Seas esses sss sS a8 SIEBER SSS SIS CSRS OS RISES SE SESSCSEEISCSE CESSES SSS ES SCSS MOSES EES 76 BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON Table 2.—Insertion sites or states of LI2 and LI3 and state of RD, if present, in genera of preacanthomorph fishes. X indicates insertion site or absence; ? = pres- ence or absence questionable. Data for LI2 and LI2’ combined for Zacco. 0 = RD dorsal to SO (SOD absent); 1 = RD ventral to SO (SOD absent); 2 = RD and SOD present. * = probably anomalous. Li2 LB bsent Genera Diaphus Lampanyctus Neoscopelus Chlorophthalmus Saurida Aulopus Pb4 UP3 UP4 UPS Eb2 Eb3 IEb4 %|>< 1A bsent Pb3 Pb4 UP4 UPS Eb3 b4 Ateleopus Gonostoma Maurolicus Diplophos Umbra Dallia Novumbra Esox Searsia Alepocephalus Argentina Mallotus Hypomesus Lovettia Lepidogalaxias Xx Galaxias Retropinna Coregonus Thymallus Oncorhynchus Gymnotus Diplomystes Xenocharax Brycon Zacco Opsariichthys Gonorynchus Chanos Clupea Dussumieria Chirocentrus x | | Coilia Cetengraulis Tlisha Denticeps Synaphobranchus Anguilla Conger Aldrovandia Notacanthus Pterothrissus Albula Elops Megalops Scleropages Osteoglossum Pantodon Arapaima Heterotis Mormyrus 2? 2? Petrocephalus 2 [ Y Gymnarchus K Dak || IE Notopterus x | Hiodon Amia i Atractosteus Polyodon Acipenser Polypterus ».4 NINN] S| ele ley iy| 6 ~*~ 6 | PS | P< P< | <1 >< |< x* |< mA 6 | PK 1PM | PS | PS P< | PA 12S | <1 P< S| <1 PS |S PPS PS PSPS OS S salpe]_ | |< 6 PK | PSPS | PSI PS |< IPS] [PS 1P<| P< IP |P< 15!) | <1 PS|P<) <1 P< S| PSI P<) PCI P< | PS | | PX OC << <1 PSPS) PS IPD T at Al 6 || PSP | PSPS P< | <1 OS, * PA| [PPS P< P< APS 44 PAP |PS|PS 1S) [PS |<] PIP |< IP5) PSPS) < 6 | PSPS |< J mr ><[><] ‘S) ~*~ NUMBER 11 7 Table 3—TD muscle components in genera of preacanthomorph fishes. States within components not indicated. X = present; U = unique; ** = TD absent. Eb4* treated as Eb4. See also Table 4. Genera TPb1-Pb2-Pb3-Eb1-Eb2 TPb2 |p| >< |TPb2a »|><| >< |[TPb2p TPb3-Eb4-UP3-UP4 TPb2-Pb3-Pb4-Eb4 TPb3-Pb4 TPb2-Pb3a »|><| 2 /TPb3 TPb3p-Pb4-Eb4 ITEb1-Eb2 |p| >< ITED2 TEb2a TEb2p TEb2v TEb3 || ><) ><] ><) 2 ]TED4 TPb2-Pb3-Eb4 TPb2-Pb3-Pb4 TPb3-Eb4 TPb3-Pb4-Eb4 TPb3-UP3-UP4 TPb3a TPb3-Eb2 TPb3a-Eb2 TPb3p-UP4 TPb4-Eb3 TPb4-Eb4 TUP4a TUP4p TTUPS Diaphus Lampanyctus Neoscopelus Chlorophthalmus Saurida Aulopus Ateleopus Gonostoma Maurolicus Diplophos Umbra | Dallia [ Novumbra X Esox X X Searsia x X Alepocephalus x |X Argentina XxX Mallotus IL | Hypomesus IE r Bil 1 x Lovettia Lepidogalaxias _| a Lol x Galaxias allie ima xX [ | Retropinna [ Xx Lc _ | Coregonus | ial IC xX} | x Thymallus = STs oar | { ~~ ~*~ + baci inl acl acl asl bas! bas |< ~* |< ~a\P |><|><|>< ue IL I t PAIPAIPS| [PS | PPS 6 |PS | [PSI [ * ~ P< 78 BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON Table 4.—Skeletal elements to which TD attaches (Eb4 undifferentiated from EB4*; see also Table 3) and relation of OD4 (and/or OD4' or OD3—4; see also synapomorphies (30-40) in Results section) to Pb4 in genera of preacanthomorph fishes. A = Pb4 absent; P = Pb4 present, but TD not attached to it; X = attached to; ** = TD absent; 1 = origin at least partly on Pb43; 2 = Pb4 present, but origin not on it. Ebl Eb2 Eb3 Eb4 Pbl Pb2 Pb3 Pb4 UP3 UP4 UPS OD4 Diaphus P Lampanyctus Neoscopelus Chlorophthalmus Saurida Aulopus Ateleopus Gonostoma Maurolicus Diplophos Umbra Dallia Novumbra Esox Searsia Alepocephalus Argentina Mallotus Hypomesus 4 Lovettia Lepidogalaxias Galaxias Retropinna Coregonus Thymallus Oncorhynchus Gymnotus Diplomystes 4 Xenocharax Brycon Zacco** Opsariichthys** Gonorynchus Chanos Clupea Dussumieria Chirocentrus Coilia Xx Cetengraulis llisha Denticeps DX Synaphobranchus Anguilla Conger Aldrovandia | Notacanthus xX Pterothrissus Albula Elops Megalops Scleropages Osteoglossum Pantodon x Arapaima Heterotis Mormyrus Petrocephalus Gymnarchus Notopterus Hiodon Amia Atractosteus Polyodon Acipenser Polypterus** PA PS | P< | PSPS] PS APS) PSPS] OX PRP |PS| P| PSPS | PSPS IPS! PS * ~*~ | lI D> | D> | > | > | > | bd | be | >| dd bd | | P| D> | Od | P< | P< | | O| | D> | Bd) D> | D> | D> | >< | >< | P< | P< P< || PM || PS | D> | D> | DX | DX || OM) BS <1 P< | PK | PMO P| S| S| SO S| 1PM | PS | PSO PSO > ||) DM) + Pl Ped bad ad ad bal Pao Pad be acd Pcl ad acd Pa Pal Pal al ad Pcl oa Pl a cl Fol bad a PA|PS| [PAPA IPS |S IPS IPS] TPS|PS IPS! | <) P< <>< PI S| OS ~ IL PAP || P< PAI PSIPSIPSIPS| [PSPS PX] << OS PA] P6125 | PS | PX | PS | PX | PX | PS] PSPS >< 141 241241251 >< | P< | PX | PX | PX PX P< <1 2125125 |S S| PS Pal ea DPD PD D> rol | > roll | > piel ele lel) eb >>) elm) )— |e ele lp) el >) >|) elle ||| elle ipa) |e ie} ele |b iit) |i) NUMBER 11 Table 5.—Distribution of bilaterally paired pharyngobran- chial muscles in suprageneric groups of preacanthomorph fishes (all genera within each group are included). X = present; * not illustrated. Suprageneric Groups | Aulopiformes IM. Pb2-Ebl IM. Pb2-Eb2 M. Pb3-Eb3 IM. Pb3-Eb3-Eb2 M. Pb3-Pb4-Eb2 M. Pb4-Eb2 IM. Pb4-UP5-Eb2 M. UP4-Eb2 M. UP4-Eb4 M. UP4-Eb5-Cb4 M. UPS-Cb4 M. UP5-Cb4-Eb5 Chlorophthalmus _ Saurida _ Aulopus Stomiiformes Gonostoma Maurolicus Diplophos a a | | Esociformes Umbra Dallia __Novumbra Esox Characiformes _ Xenocharax Brycon Elopomorpha Synaphobranchus — a Anguilla Conger Aldrovandia _ Notacanthus Pterothrissus Albula Elops Megalops 14-4 Osteoglossomorpha Scleropages _ Osteoglossum Pantodon _ Arapaima Heterotis Mormyrus Petrocephalus Gymnarchus Notopterus Hiodon (Bloch), USNM 257868; Platax orbicularis (Forss- kal), USNM 268668. EPIGONIDAE: Sphyraenops bair- dianus Poey, USNM 270279. HAEMULIDAE: Haemu- lon sexfasciatus Gill, USNM 292785. LOBOTIDAE: Lobotes pacificus Gilbert, USNM 82008. MoNnoDAc- TYLIDAE: Monodacytus argenteus (Linnaeus), USNM 266897. PEMPHERIDAE: Parapriacanthus ransonneti Steindachner, USNM 218867; Pempheris schwenkii Bleeker, USNM 269801. PERCICHTHYIDAE: Bostockia porosa Castelnau, USNM 218841; Gadopsis marmo- ratus Richardson, USNM 214836, 308109 (2). Po- MACANTHIDAE: Centropyge bicolor (Bloch), USNM 56995: Pomacanthodes semicirculatus (Cuvier), USNM 273043. SCATOPHAGIDAE: Scatophagus argus (Linnaeus), USNM 224393 (3), 259383; Selenetoca multifasciata (Richardson), USNM 173514, 245702 (2). SCORPIDIDAE: Microcanthus strigatus (Cuvier), USNM 267047; SIGANIDAE: Siganus vulpinus (Schle- 79 gel and Miiller), USNM 325277. TRICHIURIDAE: 771- chiurus lepturus Linnaeus, USNM 272931. Muscle preparations in alcohol. APLODACTYLIDAE: Crinodus lophodon (Giinther), USNM 227300. Car- ANGIDAE: Decapterus macrosoma Bleeker, USNM 307977; Decapterus punctatus (Cuvier), USNM 199037; Selene vomer (Linnaeus), USNM 338012. CENTRARCHIDAE: Acantharchus pomotis (Baird), USNM 237611; Ambloplites rupestris (Rafinesque), USNM 333731; Archoplites interruptus (Girard), USNM 39563; Centrarchus macropterus (Lacepéde), USNM 243775; Lepomis auritus (Linnaeus), USNM 243888; Pomoxis annularis Rafinesque, USNM 129524. CHAETODONTIDAE: Chaetodon austriacus Riippell, USNM 267044. CHIASMODONTIDAE: Chias- modon sp., USNM_ 1186139. GEMPYLIDAE: Neo- epinnula americana (Grey), USNM 366716; Pro- methichthys prometheus (Cuvier), USNM 289930. 80 Table 6.—Character states for Eb5 and skeletal elements (other than Cb5) to which Ad5 attaches in genera of preacanthomorph fishes. 0 = Eb5 absent; | = Eb5 autogenous; 2 = EbS putatively fused to Eb4 (based on configuration of distal end of Eb4); 3 = Eb5 putatively fused to Cb4; X = major attachment; x = minor attachment. Eb5 character state 2 for Osteoglossomorpha reas- sessed in “Epibranchials 5 and 4” under section ““Muscles and Skeletal Elements.” Skeletal elements Genera eet ekate | _EbS Boao Diaphus Lampanyctus Neoscopelus Chlorophthalmus Saurida Aulopus Ateleopus Gonostoma Maurolicus Diplophos Umbra Dallia Novumbra Esox Searsia Alepocephalus Argentina | Mallotus Hypomesus Lovettia Lepidogalaxias Galaxias Retropinna Coregonus Thymallus Oncorhynchus Gymnotus Diplomystes Xenocharax Brycon Zacco Opsariichthys Gonorynchus. Chanos Clupea Dussumieria Chirocentrus Coilia Cetengraulis Ilisha Denticeps Synaphobranchus Anguilla Conger Aldrovandia Notacanthus 1 Pterothrissus Albula Elops Megalops Scleropages Osteoglossum Pantodon Arapaima Heterotis Mormyrus Petrocephalus Gymnarchus Notopterus Hiodon Amia Atractosteus Polyodon Acipenser »< Polypterus Ad5 absent | >< |< APS PS |< l><|><|_ |><] >< Pol Pal eal badd bal acl bast |< ~*~ ClHlololo}He lly ye lye) el ell ele | el ele eIn RIOIN NIN) || |OlOlO} RH | |Ol/S|O)} |S! |SO|O| Oo uN ~ P| PK|P<| PSPS IPS 1PS| [P<] PS x Xx x DISISISISOININININININ|N] = |NIN/H | |W) BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON INERMIIDAE: Inermia vittata Poey, USNM 318643. Is- TIOPHORIDAE: Istiophorus sp., USNM 22999. Ky- PHOSIDAE: Kyphosus sectatrix (Linnaeus), USNM 116955. MONODACTYLIDAE: Monodactylus argenteus Linnaeus, USNM 258894 (3), 266897. NoTOGRAPTI- DAE: Notograptus guttatus Giinther, USNM 173798. POMACANTHIDAE: Centropyge bispinosus (Giinther), USNM 336476; C. vrolikii (Bleeker), USNM 210295. SCATOPHAGIDAE: Scatophagus argus (Lin- naeus), USNM 224393 SCOMBRIDAE: Gasterochisma melampus Richardson, TMH D.1982; Euthynnus al- letteratus (Rafinesque), USNM 214658; Scombero- morus cavalla (Cuvier), USNM 289928; S. commer- soni (Lacepede), USNM 297407. SCORPIDIDAE: Scor- pis sp., USNM 339348. SIGANIDAE: Siganus spinus (Linnaeus), USNM 233575. TRACHIPTERIDAE: Tra- chipterus sp., USNM 346705. Lampridiformes VELIFERIDAE Velifer hypselopterus Bleeker, NSMT-P-59516, 30°S, 168°E. 165 mm. Plate 59 Additional material. @ = Metavelifer multiradiatus (Regan), AMS I.20605013, ca. 94 mm SL; USNM 23953, cleared and stained. Remarks. AMS specimen dissection was faulty, but muscles, except for TEb2 and ODs, appear to be remarkably similar to V. hypselopterus. Description. LE1 on Eb! mid-dorsoposteriorly beginning lateral to base of uncinate process. LE2 on Eb2 mid-dorsoposteriorly. LE3 on anterior margin of cartilaginous tip of Eb3 uncinate process, just dorsal to OD3. LE4 slender tendinous insertion on tip of Eb4 le- vator process. LP absent, possibly represented by discrete liga- mentous attachments, two on Eb4 levator process, one on Eb3 uncinate process, that fuse into broad sheet of fascia dorsally, which is also continuous with sheet of fascia representing PP. LI1 on Pb2 just ventroposterior to large dorsal, cartilage-tipped process. LI2 on Pb3 dorsolaterally anterior to medial end of Eb3. TD comprises completely separate TEb2 and broad, more-or-less continuous TPb3-Pb4-Eb3-Eb4 (attachments distinct laterally), which is posteriorly continuous with SOD. TEb2 consists of two, discon- tinuous parts. Left-side TEb2 originates as broad-dor- sal and slender-ventral muscle straps (ventral strap not visible in Plate 59) attached to thick CT pad en- veloping right-side Pb2 and Pb3 cartilaginous artic- NUMBER 11 Table 7.—Synapomorphies (as numbers), based on dorsal gill- arch muscles, TV4, and SCL, supporting Actinopterygian preacan- thomorph fish groups (see Results section for explanation). Group Synapomorphy number Ctenosquamata 16 Myctophiformes 23 Myctophidae 10 Eurypterygii 20, 22 Neotelostei S283 29853 Stomiiformes 35}; Sl Esociformes 60 Neognathi 34 Platytrochtidae + Alepocephalidae 40 Osmeroidea 5,6 Characiformes + Siluriformes + Gymnotiformes 27] Gymnotiformes + Siluriformes 21 Characiformes 8, 49 Cypriniformes 39, 43, 45, 52, 56 Gonorynchiformes 38 Clupeomorpha 19 Clupeoidei 7 Elopomorpha 58? Anguilliformes 12, 14, 46 Albuliformes 17, 36, 48, 57 Elopiformes 59? Elopocephala 33 Teleostei 13, 25, 32, 41, 50, 55a Osteoglossomorpha 3,4, 42 Osteoglossiformes 44 Osteoglossoidei 13a, 26, 47 Osteoglossidae 30 Halecostomi 37, 54, 55 Neopterygii 9, 11, 24, 31, 42a?, 49a? Chondrostei 18 Actinopteri 2 Actinopterygii 1 ulating processes (pad mostly removed in Plate 59) and attaches to dorsomedial surface of left-side Eb2 anterior to LE2 insertion. Right-side TEb2 originates medially from CT on left side and passes laterally between dorsal and ventral straps of left-side TEb2 and inserts similarly on right-side Eb2. TPb3-Pb4- Eb3-Eb4 attaches to bony dorsoposterior surface of Pb3 and dorsomedial surfaces of Eb3 and Eb4 and dorsoanterior surface of Pb4. @ TEb2 damaged in dissection, but could have been continuous across gill arches, with anterior edge attaching to CT pad en- veloping Pb2 and Pb3 processes. Remarks. The specialized medial overlapping of TEb2 on each side was seen only in Lampris and Velifer, and supports close relationship of these two taxa. The condition in Metavelifer should be verified. OD3 and OD4 originate together on dorsoanterior bony surface of Pb3 and separate at about their mid- length before inserting broadly on anterior surface of Eb3 uncinate process and dorsomedial edge of Eb4 81 uncinate process. @ Both muscles consist of loose strands, but appear to be separate distal to their ori- gins. OP dorsally on posterior surface of Eb4 medial to dorsal attachment of Ad4 and ventrally on dorsodistal surface of Cb5 just posterior to posterior attachment of Ad5. Ad1-—3 absent. Ad4 attaches dorsally to posterolateral surface of Eb4 and ventrally to Cb4 dorsally just anterior to inner angle of Eb4-Cb4 joint. Ad5 attaches anteriorly to Cb4 posterodistal end and AC4 posteriorly, and posteriorly to Cb5 distal end anterior to ventral attachment of OP. SOD present. RDs adjacent. Additional remarks. SCL attached mid-dorsally to cartilaginous ventroposterior tip of Bb3. TV4 free from Cb5s. Pb4 and UP4 present. IAC absent. CT pad enveloping Pb2 and Pb3 processes gives rise to CT strap that attaches to cranium. AC4 and a much smaller AC3 present. Medial ends of Eb4 and Eb3 about equal in size. @ AC3 and AC4 absent; medial ends of Eb3 and Eb4 about equal in size. Olney et al. (1993) hypothesized that the Velifer- idae are the sister group of all other lampridiforms and that the Trachipteridae and Regalecidae are the most specialized. The Veliferidae lack ER as does the Trachipteridae (based on Trachipterus sp., USNM 346705). LAMPRIDAE Lampris guttatus (Briinnich), SIO82-70, not mea- sured (estimate >300 mm). Plate 60 Description. LE1 on anterior surface of posterior bony flange of Eb1, cartilage-tipped uncinate process absent. LE2 on anterior surface of posterior bony flange of Eb2 just anterolateral to LE2’ insertion (see re- marks following LE2’). LE2’ on dorsal edge of posterior flange of Eb2, anterior surface appressed to posterior surface of LE2, muscle fibers of LE2 and LE2’ extend dorsally at different angles. Remarks. Right-side LE2’ cleanly separable from LE2, but left-side pair did not separate cleanly. Ver- ification of distinctness in other, smaller specimens needed. LE3 sheet-like, on tip of Eb3 uncinate process an- teriorly, muscle continuous dorsoposteriorly with presumed LE4-LP complex. LE4 broad sheet, musculous distally, fascial prox- imally, inserting on Eb4 levator process, continuous ventrolaterally in fascia with presumed LP fibers; fas- 82 cla continues ventrodistally around fourth arch and attaches to Cb5 distally. LP presumably represented by small strap of fibers arising from fascia at posterolateral edge of LE4 fi- bers (information from right side only; left side dam- aged); posterior fascia presumably continuous with PP. LI1 on Pb2 dorsoanterolaterally. LI2 on Pb3 dorsoanterolaterally. TD comprises TEb2 and TEb3-Eb4. TEb2 with distinct right- and left-side muscle straps, each in- serting on respective Eb2 mid-dorsoanteriorly, ante- rior to LE2 insertion; straps originate medially from ventral surface of thick CT pad enveloping anterior cartilaginous anterior ends of Pb2s and Pb3s and dor- sally from CT attaching to Pb3; left-side strap passes through right-side strap before attaching to CT; mus- cle extends laterally and attaches on Eb2 dorsoanter- iorly anterior to LE2 insertion. TEb3-Eb4 attaches to dorsomedialmost surface of Eb3, dipping to level be- low that of attachment to dorsomedialmost surface of Eb4. Remarks. CT permeating TD obscures attach- ments; insertions should be verified on smaller spec- imen. OD3 origin on Pb2 dorsoanteriorly, more-or-less continuous with OD4 origin, divides laterally with one branch, OD3, inserting on Eb3 uncinate process, and the other branch, OD3’, inserting on Eb3 dor- soanterior surface. OD4 origin on Pb3 dorsoanteriorly; left-side in- serts on Eb4 levator process; right side inserts on both Eb3 uncinate and Eb4 levator processes. OP dorsally broadly on Eb4 posteriorly ventral to tip of uncinate process and extending laterally; ven- tral extent unclear. Ad1—3 absent. Ad4 questionably absent (unusual if true), but if present, much reduced, possibly represented by mus- cle fibers obscured from view by OP. Ad5 on Cb5 distally and Cb4 posterodistally at ventral attachment of Ad4. SOD broad. RDs adjacent. Additional remarks. SCL present, attached mid- dorsally to posteroventral cartilaginous tip of Bb3. TV4 damaged, appears to have had a median septum, apparently free from Cb5s. IAC absent. Pb4 and UP4 present. Eb4 uncinate process absent. Pb2 toothed. Tiny AC2 present on left side, absent on right side. PCI insertion restricted well medial to distal end of Cb5; muscles unusual in that right and left sides join together anteromedially. Polymixiiformes POLY MIXIIDAE Polymixia lowei Giinther, USNM 159295, 89.3 mm, USNM 202153, 124 mm. BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON Plate 61 Description. LE1 broadly on anterior surface of Eb1 uncinate process just lateral to cartilage tip. LE2 on dorsoposterior edge of Eb2 just postero- lateral to lateral end of TEb2. LE3 on tip of Eb3 uncinate process just dorsal to insertion of OD3—4 on Eb3. LE4 on Eb4 surrounding small Eb4 levator pro- cess, ventrolaterally continuous with broad PP fascia. LP slender, tendinously inserted among LE4 fibers basally. LI1 broadly on Pb2 dorsal surface just anterome- dial to base of Pb2 uncinate process. LI2 on bony Pb3 dorsoposterolaterally meeting TPb3-Eb3 insertion anteriorly. TD comprises TPb2, TEb2, and TPb3-Eb3. TPb2 with straps attaching to dorsal surface of cartilagi- nous and bony anterior end of Pb2 just anterior to LI1 insertion, lies dorsal to TEb2 anteriorly, and is complexly continuous ventrally with TEb2, and also posteriorly by diagonal muscle slip with TEb2. TEb2 broad, with crossing straps extending mid-anteriorly from attachments with TPb2 ventral surface and in- serting on Pb2 anteriorly, but muscle mainly inserting on Eb2 dorsoanteromedialmost surface, at most reaching to opposite medial edge of LE2 insertion; muscle continuous posteriorly by crossing muscle slips with TPb3-Eb3. TPb3-Eb3 inserting on Pb3 dorsoposteriorly, meeting anteromedial edge of LI2 insertion, and on Eb3 dorsomedially. OD3-4 originating broadly on Pb3 bony dorsal surface ventral to TEb2, inserting on Eb3 bony sur- face anteriorly beginning just ventral to tip of unci- nate process and on posterior bony surface of Eb4 beginning just ventral to uncinate process. OP strap of muscle on Eb4 dorsoposteriorly ven- tral to LE4 insertion and medial to Ad4 on Eb4, ven- trally on Cb4 joining raphe (ER) with Ad5. M. SO-Pb3 (indicated, but not labeled on Plate 61 as broad strap of longitudinal SO fibers lateral to each RD) extends anteriorly, passing ventral to TPb3-Eb3 and inserting on Pb3 dorsally anterior to RD inser- tion. Ad1-—3 absent. Ad4 dorsally on Eb4 posteriorly, ventrally on Cb4 medial to Eb4-Cb4 joint. Ad5 ventrally broadly on Cb5 dorsally beginning at dorsodistal tip and extending medially; dorsally joining ER with OP ventrally at attachment of both on Cb4. Tendinous tissue extends laterally from ER and is continuous with PP fascia, which attaches around Eb4-Cb4 distally (incompletely indicated in Plate 61). SOD slender, continuous anteriorly by fine, diag- onal muscle slip with TPb3-Eb3 posteriorly. NUMBER 11 RDs slender, juxtaposed. Additional remarks. SCL attached mid-dorsally to ventroposterior cartilaginous tip of Bb3. TV4 free from Cb5s. Pb4 and UP4 present. [AC absent. Pb2 toothed. Paracanthopterysgii Percopsiformes Percopsiform monophyly and intrarelationships.— Rosen (1962) implied the monophyly of the Percop- sidae + Amblyopsidae + Aphredoderidae, which subsequently became known as the Percopsiformes (Greenwood et al., 1966:396). Later, Rosen (1985: 42—45) was first to question the monophyly of the percopsiform fishes, removing the Percopsidae, but aligning aphredoderids with amblyopsids on the basis of their thoracic anus and segmented premaxilla. Pat- terson and Rosen (1989:19—20) reaffirmed Rosen’s (1982) findings, stating that they could find no char- acter synapomorphous for the Percopsiformes [of Ro- sen, 1962]. They failed to note, however, that among their many illustrations of paracanthopterygian dorsal gill-arch skeletons, the medial head of Eb4 of per- copsiforms is much broader than the combined breadth of the medial heads of any two other epi- branchials (first noted by Parenti (1993:181), who in- terpreted the character as a synapomorphy of atheri- nomorphs, percopsids, amblyopsids, and aphredod- erids). Among the other paracanthops, the medial head of Eb4 is comparatively narrow, and in only a few macrourid gadiforms, which are well removed from the percopsiforms in Patterson and Rosen’s cladogram (1989:fig. 16), is the percopsiform con- dition approximated. An additional character, not noted previously, is the presence of a levator process on the percopsiform Eb4, which is present in some or all members of the three families (e.g., the process is absent in Amblyop- sis (Patterson and Rosen, 1989:figure 13E), but pres- ent in Chologaster—both Amblyopsisdae, thus, we consider the process primitively present in percopsi- forms). The process is lacking in all other paracan- thops and present elsewhere among non-percomorph acanthomorphs only in Lampridae, Polymixia, and the zeitform Xenolepadichthys. Murray and Wilson (1999) hypothesized a Percop- siformes comprising only percopsids and aphredod- erids. They placed the amblyopsids as a deeply em- bedded clade in the Anacanthini, which they placed as the sister group of Percopsiformes. They based their classification on 47 osteological characters and included fossil taxa. They also overlooked the en- larged medial head of epibranchial 4 as a possible character, and simply disposed of the position of the anus and segmented premaxilla, which Rosen (1985) had used to unite aphredoderids and amblyopsids, by 83 stating, “The new placement of the amblyopsids re- quires viewing these characters as homoplasies [!].” Based on the three percopsiform taxa we examined (Aphredoderus sayanus, Aphredoderidae; Percopsis omiscomaycus, Percopsidae; Chologaster agassizi, Amblyopsidae) the group shares the following spe- cializations: 1. OD4’ (a branch of OD3-—4) extending posteri- orly, passing dorsal to jointly articulating Eb3 and Eb4 uncinate processes, and inserting on Eb4 levator process. We have not found this muscle in any other acanthomorph. 2. OD origin includes Pb3 (mainly) and Pb2. OD only originates on Pb3 in most other acanthomorphs; exceptions include: platycephalids, percids, caproids (Capros, but not Antigonia), elassomatids, gasteros- teids (only Apeltes), and hypoptychids. 3. M. Pb2-Eb2 present. This is an uncommon muscle with a spotty occurrence otherwise among fishes (e.g., we found it in the distantly related par- acanthop family Ophidiidae, and the percoids: Mo- rone, Moronidae, and Sillago, Sillaginidae. Endo (2002:101, as OD2) reported that it is generally ab- sent in gadiforms, but present in lotines, gaidropsar- ines, phycines, most gadines, and the morid Lotella. He (2001:fig. 34) illustrated the musculature of Gair- dropsarus and and gadiform bregmaceratid, Breg- macerops). 4. TEb2 absent. TEb2 is usually present in acan- thomorphs, but is notably absent in some other Par- acanthopterygii and in all Atherinomorpha (but pres- ent in Mugilomorpha, 1.e., Agonostomus, Mugilidae, hypothesized sister group of the atherinomorphs (Stiassny, 1993; Johnson and Patterson, 1993). 5. LE3 absent. Occurs variously among fishes, but is not common, except among Smegmamorpha and Polycentridae. 6. Medial end of Eb4 much enlarged relative to medial end of Eb3 (also common in many smegma- morphs, but not mugilomoph Mugilidae). 7. Levator process present on Eb4. This process otherwise occurs variously among acanthomorphs, but uncommonly among pre-Percomorpha. 8. An overall trapezoidal appearance, in dorsal view, of the combined transverses dorsales and ob- liqui dorsales. We have not seen such configuration in any other fishes. The combination of these specializations almost certainly indicates the monophyly of the percopsi- forms. The presence of characters 4—6 commonly among Smegmamorpha, suggests that the percopsi- forms and smegmamorphs, may be more closely re- lated than current classifications indicate, a relation- ship most recently suggested by Parenti (1993). The absence of all three of these characters in the Mu- gilidae, would seem to preclude such relationship. Evidence from the gill-arch muscles is equivocal 84 for resolving percopsiform intra-relationships. Ab- sence of an OD attachment to Eb3 is a specialization shared only by Chologaster and Percopsis. Con- versely, absence of a TPb3 attachment is a speciali- zation shared only by Chologaster and Aphredode- rus. Non-musculature characters, however, such as the advanced anus and segmented premaxilla shared by Aphredoderus and Chologaster, strongly favor a hypothesis that they are the sister group of Percopsis. APHREDODERIDAE Aphredoderus sayanus (Gilliams), USNM 238477, 71.3 mm; USNM 238466, 69.0 mm. Plate 62 Description. LEI on and medial to small, sharp prominence at base of Eb1 uncinate process. LE2 on tip of expanded bony process on posterior edge of Eb2. LE3 absent. LE4 on Eb4 levator process dorsoanteriorly, join- ing raphe medially with OD4 insertion and another laterally with LP insertion. LP insertion continuous with LE4 insertion ventro- laterally. LI] on Pb2 anteriorly, ventromedial surface join- ing raphe with TPb2 just dorsal to LI] insertion. LI2 broadly on Pb3 dorsolaterally, narrowly on Pb4 anterolaterally and adjacent dorsoanterior edge of UP4. TD comprises TPb2 (TPb2d + TPb2v), and TEb4. TPb2d dorsal to TPb2v, each sheetlike, united ante- riorly; muscles together are pad-like and notched mid-anteriorly, each with mid-longitudinal raphe, which join together mid-anteriorly. CT sheet overlies TPb2d raphe and is attached to it; TPb2d fibers form heart-shaped curve; TPb2d fibers curve broadly an- teriorly; muscles attach together on anteromedialmost surface of Pb2 uncinate process, joining raphe there with LI2, but continuing ventromedially to insert on anterolateralmost surface of Pb2 (insertion visible only in ventral view after removal of underlaying tis- sues). TEb4 triangular, apex anteriorly with fibers variously joining SO longitudinal fibers and CT in midline, posteriorly almost divisible transversely into two sections, anterior section maintains triangularity, posterior section more strap-like, two sections joining laterally and attaching along most of posterior margin of Eb4 levator process, there joining raphe with OP. M. Pb2-Eb2 on Pb2 uncinate process and dorsal surface mostly ventral to OD3—4, extending laterally and inserting on most of dorsal surface of Eb2. OD unusually broad dorsoanteriorly, comprises OD4' dorsally and OD3—4 ventrally; OD4’ and OD3—4 originate inseparably, broadly on Pb3 and Pb2 ventral to TPb2; OD4’ extends posterolaterally BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON over joined Eb3 and Eb4 uncinate processes (but is not attached to them) and inserts on anterior surface of Eb4 levator process, where it joins raphe with ven- troanterior edge of LE4; OD3-—4 splits off ventrally shortly posterior to joint origin with OD4’ and inserts on Eb3 uncinate process, with remaining portion passing ventral to joined Eb3-Eb4 uncinate processes and inserting on Eb4 anterior surface medial to in- sertion of OD4’. Remarks. It is unusual for OD to attach to Pb2. OP dorsally on Eb4 posterior surface at and medial to levator process, joining raphe there with TEb4, ventrally on Cb5 dorsolaterally, ventrolaterally join- ing raphe with Ad5. Ad1-3 absent. Ad4 on most of Eb4 posterior surface (medial half of muscle anterior to OP) and distal half of Cb4 me- dial to Eb4-Cb4 joint. Ad5 dorsally on posterolateralmost two-fifths of Cb4 and ventrally on anterodistalmost end of Cb5, joining raphe medially with OP. SOD absent. RDs separate, adjacent. Additional remarks. SCL absent. TV4 free from Cb5s. Pb4 and UP4 present. Pb3 toothed. IAC absent. PERCOPSIDAE Percopsis omiscomaycus (Walbaum), USNM 179711, 92.3 mm; USNM 334972, 72.1 mm. Plate 63 Description (based on larger specimen with remarks on smaller specimen). LEI on small bony prominence near base of Eb1 uncinate process. LE2 on expanded posterior edge of Eb2. LE3 absent. LE4 tendinously on dorsal tip of Eb4 levator pro- cess. LP on lateral surface of tendinous insertion of LE4 and distalmost end of Eb4. Smaller specimen with LP inserting only on tendinous insertion of LE4. LI1 on dorsoanteriormost surface of Pb2. LI2 on Pb3 dorsolaterally. TD comprises TPb2 and TPb3-Eb4. TPb2 roughly heart-shaped, notched anteriorly, with median raphe (from which CT sheet arises dorsally), lies dorsal to Pb3 portion of TPb3-Eb4, but not continuous with it, comprises two scarcely separable parts (more obvi- ous in smaller, non-illustrated specimen) possibly equivalent to TPb2d and TPb2v of Aphredoderus; an- terior part on dorsoanteriormost end of Pb2 and me- dial surface of uncinate process of Pb2; posterior part attaches to posterior surface of Pb2 uncinate process and forms raphe there with M. Pb2-Eb2. TPb3-Eb4 broadly triangular, blunted apex anterior, long side posterior, with irregular median raphe in anterior NUMBER 11 three-fourths, attaching by few muscle straps to Pb3 posteromedially; long side slender laterally, attaching along most of Eb4 dorsal arm, forming raphe later- ally with dorsal end of OP. Remarks. Left-side OP of large specimen compris- es two separate straps, lateral and medial. Right side has only lateral strap, with muscle area occupied by medial strap on left side forming part of SO. In small- er specimen, OP is broad, single, and occupies joint area of OP on left side of larger specimen. M. Pb2-Eb2 originates on Pb2 uncinate process at raphe with TPb2 posterior part and attaches along most of bony dorsal surface of Eb2. OD3 absent. OD unusually broad dorsoanteriorly, comprises OD4' and OD4. OD4’ originates broadly on Pb3 bony dorsal surface ventral to TPb2, passes dorsal to joined Eb3 and Eb4 uncinate processes, and inserts on Eb4 levator process just medial to LE4 insertion. OD4 originates on Pb2 dorsally ventrolateral to OD4’, joins OD4' just distal to origins, passes ven- tromedial to joined Eb3-Eb4 uncinate processes, and inserts on anterior surface of distal end of Eb4 ventral to OD4’ insertion. In smaller specimen OD4 and OD4’ are mostly fused, separating only just anterior to their insertions on Eb4. OP dorsally on posteromedial surface of Eb4, join- ing raphe there with Eb4 portion of TPb3-Eb4, ven- trally forming raphe with Ad5 before both attach dor- sodistally to Cb5. No raphe with Ad5 in smaller spec- imen. See also remarks following TD. Ad1-—3 absent. Ad4 on distal half of ventral surface of Eb4, par- tially anterior to OP, and ventrally on Cb4 distal half medial to Eb4-Cb4 joint. Ad5 on posterodistal end of Cb4 and dorsodistal end of Cb5 forming raphe ventromedially with OP just before attaching to Cb5. No raphe with OP in smaller specimen. SOD broad. RDs adjacent. Additional remarks. SCL absent. TV4 free from Cb5s. Pb4 and UP4 present. Tiny AC between right- side Pb2 and Eb! uncinate processes of larger spec- imen, possibly represents vestigial IAC, but IAC cod- ed as absent for this taxon. AMBLYOPSIDAE Chologaster agassizii Putnam, USNM 232514, 65.4 mm. Plate 64 Description. LE1 on Eb1 uncinate process about mid-dorsopos- teriorly; medial process absent. Remarks. Typically in acanthomorphs the medial 85 Eb1 process articulates with Pbl, and the uncinate process with Pb2 or IAC, which articulates with Pb2. LE2 beginning on about mid-dorsoposterior bony edge of Eb2 and extending posteriorly onto CT be- tween Eb! and Eb2 (mostly on CT). LE3 absent. LE4 on Eb4 beginning at levator process anteriorly and extending laterally along dorsoposterior edge of OD4 insertion, and just medial to LP insertion. LP on Eb4 distally, anterior and slightly lateral to lateral edge of LE4 insertion. LI1 on Pb2 anterolaterally (at anteriormost junc- tion of Pb2 and Pb3). LI2 on Pb3 dorsolaterally. TD comprising TPb2d, TPb2v, and TEb4. TPb2d a transverse strap, with median raphe ventrally at- tached to CT lining pharynx, laterally attached to Pb2 just dorsoposterior to LI] insertion and continuous and just anterior to TPb2v attachment. TPb2v a di- agonal strap on each side extending anterolaterally from raphe with posterior end of OD4’ on Pb3 on one side and attaching to Pb2 just posterior to TPb2d on opposite side, some muscle strands passing through strands of contralateral TPb2v; asymmetri- cally continuous with TEb4. TEb4, roughly triangu- lar in shape, apex anterior, commencing along com- mon line with OD4’ posteromedially, and extending posterolaterally; muscle strands, somewhat inter- laced, becoming continuous with horizontal posterior portion of muscle, and inserting on Eb4 dorsoposter- iorly; forming raphe with dorsal end of OP. Remarks. We have arbitrarily equated the two Pb2 sections of Chologaster with TPb2d and TPb2v of Aphredoderus (see also Percopsis). M. Pb2-Eb2 originating on dorsoanterior edge of Pb2 (where it articulates with Ebl medial end) and inserting along most of bony dorsal surface of Eb2. OD3 absent. OD comprises OD4’ and OD4, extremely broad dorsally and deep ventromedially, originating on Pb3 dorsomedially, continuing onto Pb2 dorsolaterally ventral to TPb2v. Muscle joins raphe posteromedially with posteromedial end of TPb2v of opposite side and divides posterolaterally into OD4' and OD4, with OD4' passing posteriorly dorsal to joined Eb3 and Eb4 uncinate processes and inserting on Eb4 levator process just ventral (anterior in view) to LE4 inser- tion and lateral to TEb4 attachment, and OD4 passing ventral to joined Eb3-Eb4 uncinate processes and in- serting on Eb4 dorsally medial to levator process. OP dorsally, broadly on posterior surface of Eb4, ventrally on dorsodistal end of Cb5 joining Ad5; me- dially continuous with SO. Ad1-—3 absent. Ad4 on most of dorsal surface of Eb4 anterior to OP attachment, and on distal half of surface of Cb4 medial to Eb4-Cb4 joint. 86 Ad5 dorsally on posterodistal fourth of Cb4, with slight tendinous extension to posterodistal end of Eb4; ventrally on distal end of Cb5, ventromedially joining raphe with OP. SOD present. RDs slightly separated. Additional remarks. SCL absent. TV4 free from Cb5s. Pb4 and UP4 present. Pb2 toothed. IAC absent. Ophidiiformes OPHIDIIDAE Dicrolene intronigra Goode and Bean, USNM 362587, 2 specimens, 226-240 mm TL. Plate 65 Description. Remarks. Only one of the two sets of gill arches was dissected to expose hidden muscle attachments, and the description of these is based mostly on that specimen. The remainder of the description is based on the undissected specimen. LE1 on anterior surface of Eb] uncinate process ventrolateral to tip. LE2 on tip of prominent bony process on posterior margin of Eb2, insertion continuous posteriorly as ligament joining Eb2 with Eb3. LE3 on Eb3 uncinate process just ventrolateral to tip, meeting insertion of OD3 portion of OD3—4. LE4 tendinously on dorsal edge of Eb4 lateral to uncinate process and immediately dorsal to Ad4 at- tachment. LP absent. LI1 on dorsomedial edge of Pb2 joining raphe with TPb2p laterally and M. Pb2-Eb2 medially. LI2 on Pb3 dorsoposteromedially lateral to M. SO- Pb3 and RD; passes medial to OD3-—4. TD complex, comprises TPb2, TPb2p, TPb3, TEb3, and, variably, TPb4 (absent in illustrated spec- imen). TPb2 very broad transversely, slender longi- tudinally, on bony edge lateral to cartilaginous edge of dorsal margin of broad Pb2 dorsal process (entire edge is cartilaginous in other specimen), continuous broadly posteriorly with TPb2p but overlapping TPb2p anterolaterally on right side (both sides non- illustrated specimen). TPb2p much shorter longitu- dinally than TPb2, passes dorsal to OD-3—4 origin, touching LI1, then extends ventrally medial to LI] and attaches on dorsomedial edge of Pb2; joins par- tial raphe with ventromedial surface of LI1; contin- uous by diagonal muscle strands with TPb3. TPb3 on dorsal surface Pb3 mid-laterally medial to lateral car- tilaginous process articulating with small AC (not il- lustrated) between process and medial end of Eb2 (posteromedial cartilage-tipped process of Pb2 is ventral to ACs and attached to them ventrally); con- tinuous posteriorly by diagonal muscle strands with BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON very broad TEb3. TEb3 dorsomedially on Eb3 bony surface beginning slightly lateral to medial cartilag- inous end, continuous posteriorly by diagonal muscle strands with narrow TPb4 (TPb4 absent in illustrated specimen, in which case continuous with SOD). TPb4 extends deeply laterally and attaches finely to Pb4 dorsomedial surface; continuous posteriorly by diagonal strands of muscle with SOD. M. Pb2-Eb2 on posterior edge of Pb2 dorsal pro- cess, just meeting lateral edge of TPb2 and the lateral edge of LI1 insertion, and on posterior surface of raised anterior edge of Eb2 just anteromedial to LE2. OD3-—4 origin on dorsoanterior surface of Pb3, di- vides posteriorly at mid-length or just before insert- ing narrowly on anterior surface of Eb3 uncinate pro- cess and medial edge of Eb4 uncinate process. On left side of illustrated specimen, an anomalous strap of OD3—4 attaches to Pb2 at the junction of LI1 and M. Pb2-Eb2. OP dorsally on Eb4 posterior surface below and medial to uncinate process, ventrally joining raphe (ER) with Ad5 dorsolaterally where both attach to medial end of CT pad enveloping posterolateral end of Cb4 (medial end of pad well separated from distal end of Cb4). M. SO-Pb3, longitudinal-fiber band originating from SO in region of Eb4 and extending anteriorly ventral to TD components and inserting on Pb3 just ventral to RD (q.v.). Ad1-—3 absent. Ad4 dorsally on dorsoposteriormost edge of Eb4 beginning at and ventral to medial end of LE4 inser- tion and ending slightly medial to distal end of Eb4; ventrally on Cb4 dorsal surface medial to Eb4-Cb4 joint. Ad5 dorsolaterally joining raphe (ER) with OP on fleshy CT pad enveloping distal end of Cb4, medially on posterior surface of lateral fourth of Cb5, joining raphe there with TV5 SOD present. RDs well separated, extending far anteriorly ven- tral to TD components and inserting on Pb3 just pos- terior to OD3—4 origin. Additional remarks. SCL absent. TV4 in two un- connected parts; anterior part with strong mid-lon- gitudinal raphe, originating on Cb4s ventral to pos- terior part and attaching mid-dorsoanteriorly to ven- tral surface of cartilaginous Bb4; ventral part contin- uous, originating on Cb4s dorsoposterior to anterior part, free from Cb5s. We have seen this type of TV4 elsewhere only in the lophiiform Chaunax (Chauna- cidae), which differs in that a fine muscle strand joins the two portions and the two halves of the anterior portion are separate at their attachment to the large cartilaginous Bb, which may be Bb4 or a complex. The dorsal tips of both Pbls are cartilaginous in One specimen and bony in the other. Pb4 and UP4 NUMBER 11 present. IAC present. Eb4 levator process absent. Pb2 toothed. Medial end of Eb3 and Eb4 about same size. There is a separate ball of cartilage between the medial end of Eb2 and the cartilage-tipped process mid-laterally on Pb3. Markle (1989:fig. 1A) illustrat- ed the dorsal gill-arch skeleton of a specimen he identified as the neobythitine Dicrolene intronigra. The illustration does not accord with the skeleton of our two specimens. For instance, he shows Eb2 ar- ticulating with Pb2 and does not indicate the presence of an accessory cartilage either between this articu- lation or one attached to Pb3. Nor does he show a cartilage-tipped process mid-laterally on Pb3. Patter- son and Rosen (1989:fig. 13g) illustrated the dorsal gill-arch skeleton of Monomitopus torvus Garman (another neobythitine), which is very similar to that of our specimens of D. nigra. Consequently, one of the characters (contact between Eb2 and Pb2) Markle used to define a gadiform-batrachoidiform relation- ship apparently is invalid or needs modification. Brotula multibarbata Temminck and Schlegel, USNM 340397, 166 mm, 214124, 133 mm. Not illustrated Description. LE1 on highest point of raised dorsoposterior bony edge of Eb1, uncinate process lacks cartilaginous tip. LE2 on apex of prominent raised dorsoposterior bony edge of Eb2. LE3 on dorsoanterior surface of Eb3 uncinate pro- cess lateral to cartilaginous tip. LE4 on bony dorsoposterior edge of Eb4 well lat- eral to uncinate process, fibers of posterior surface attaching on dorsal edge of CT covering Eb4 poste- rior surface. Remarks. Dorsolateral fibers of OP and dorsome- dial fibers of Ad4 attach to same CT ventrally. Re- leasing CT from Eb4 results in strip of muscle (= LE4-OP sling of Stiassny and Jensen, 1987) inter- rupted by band of CT. LP absent. LI1 on Pb2 dorsolaterally ventromedial to TPb2 attachment to Pb2; larger than LI2. LI2 on dorsoposterolateral surface of Pb3 just me- dial to medial head of Eb3; muscle passes amidst OD3-4 dividing it into OD3—4 and OD4 (see OD3-— 4). TD comprises TPb2, TPb2p, TEb2, and TEb3. TPb2 a bean-shaped pad, covered with thin CT, at- tachment beginning anteriorly with CT of roof of pharynx, continuing laterally and ending on anterior- most cartilaginous tip of Pb2, free edge continuing posterolaterally from this point and overlying all OD components anteriorly; muscle attached ventrally to dorsoanterior surface of Pb3 together with OD inser- 87 tions; broadly continuous posteriorly with TPb2p. TPb2p on Pb2 dorsolaterally at and medial to M. Pb2-Eb2 origin, posteriorly broadly continuous with TEb2. TEb2 on Eb2 dorsoanteriorly well medial to LE2 insertion, forming short raphe with ventral edge of M. Pb2-Eb2 insertion and posterolateralmost edge of OD2 insertion, broadly continuous posteriorly with TEb3. TEb3 on dorsal surface of Eb3 medial to base of uncinate process; broadly continuous poste- riorly with SOD. OD comprises OD2, OD3-—4, and OD3’. All OD components originate together on Pb3 dorsoanterior- ly; components separate shortly distal to origin, with OD3-—4 being variously split by penetration of LI2 on its way to its insertion and recombining as they insert on Eb3 and Eb4: on anterior and posterior sur- faces, and medial edge of Eb3 just ventral to tip of uncinate process, fusing ventrally with dorsal origin of RecD3; on anterior surface and medial edge of Eb4 just ventral to tip of uncinate process, meeting OP dorsally. OD3’ inserts on Eb3 dorsally ventral to OD3-4 insertion on Eb3, continuous there with ven- tral origin of RecD3. OD2 inserts on raised dorsoanterior edge of Eb2 immediately ventral to M. Pb2-Eb2, there joining ra- phe with RecD3 insertion; posterolateralmost edge of OD2 forming short raphe with TEb2. OP with two slightly separated attachments on Eb4 posterior surface; dorsomedial attachment on unci- nate process, posteriorly overlapping ventrolateral at- tachment on Eb4 lateral to uncinate process, fusing together before ventrally joining raphe with dorso- lateral end of Ad5 (CT extension from raphe contin- ues onto Cb4 posterior surface and posteroventral fi- bers of Ad4 join raphe). Also see remarks under LE4. M. Pb2-Eb2 tendinously on anteriormost cartilag- inous tip of Pb2, to which TPb2 and Eb! uncinate process attach; broadly, musculously on Eb2 dor- soanterior edge. RecD3 with two separate origins, dorsal origin on Eb3 uncinate process ventral to and fusing with OD3-4, ventral origin on dorsoanterior edge of Eb3, both parts fusing and inserting on Eb3 along insertion line of OD2. Ad1-—3 absent. Ad4 on Eb4 posterior surface lateral to OP, ven- trally on Cb4 dorsoposteriorly, attaching to CT ex- tending from ER. Also see remarks under LE4. Ad5 broadly on dorsodistal edge of Cb5, narrow- ing dorsally and joining raphe, which continues as CT on Cb4, with OP; attachment to Cb4 is well me- dial to distal end. SOD present. RDs present, separated. Additional remarks. SCL absent. TV4 free from Cb5s. Pb4 and UP4 present. Pb1 with cartilaginous 88 ends. Pb2 toothed. Medial end of Eb3 larger than that of Eb4. IAC absent. BY THITIDAE Calomopteryx jeb Cohen, USNM 208341, 57.6 mm. Not illustrated Description. LE1 on Eb1 mid-dorsoposteriorly, anterior (phar- yngobranchial supporting) process absent. Remarks. Anterior process considered absent (un- cinate present) because medial end of Eb] articulates with Pb2. LE2 on Pb2 posteriorly a little lateral to mid- length, just lateral to lateral end of M.Pb2-Eb2. LE3 on Eb3 just lateral to cartilage tip of uncinate process. LE4 on Eb4 dorsoposteriorly between uncinate process and distal tip of Eb4. LP absent. LI1 broadly on Pb2 dorsoanterolaterally, ventro- medially joining raphe with TPb2 laterally, larger than LI2. LI2 penetrating OD3—4, dividing it into almost two equal portions, and separating TPb3 posteriorly from TEb3, while passing ventral to TEb3 on way posteriorly to insertion on Pb3 dorsoposteriorly near medial end of Eb4. Remarks. Penetration of OD3—4 by LI2 is uncom- mon. Occurs also in related Ophidiidae and Carapi- dae, and unrelated Champodontidae and Odontobu- tidae (synapomorphic for all genera of latter). TD comprising TPb2, TPb3, and TEb3. TPb2 broad, flat, thin, attaching to Pb2 beginning at ante- rior end and extending posteriorly to medial edge of LI2, attached anteroventrally to CT of pharyngeal roof, dorsally infiltrated with filmy CT, which covers TD, continuous posteriorly by diagonal muscle strands with TPb3. TPb3 slender, extends ventrolat- erally and slightly anteriorly anterior to LI2 (as latter extends ventroanteriorly from origin) and attaches to Pb3 dorsally anterior to medial end of Eb3, contin- uous mid-dorsoposteriorly with TEb3. TEb3 broad mid-dorsally, extending narrowly laterally, passing dorsal to LI2 (as latter extends ventroanteriorly from origin) and attaches moderately broadly on Eb3 dor- somedially, broadly continuous mid-dorsoposteriorly with SOD. M. Pb2-Eb2 medially broadly on Pb2 laterally be- ginning at anterior end and extending posteriorly to LI1 insertion, laterally on Eb2 dorsally, reaching me- dial edge of LE2 insertion. OD3-—4 anteriorly, broadly on Pb3 dorsoantero- medially, almost completely divided longitudinally by passage of LI2, posteriorly on Eb3 dorsoanteriorly beginning immediately ventral to fine cartilage tip of uncinate process, there meeting LE3 insertion ante- BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON roventrally, and on medial edge and posterior surface of Eb4 uncinate process. OP dorsally, broadly on Eb4 posteriorly beginning ventral to uncinate process and extending medially, ventrally joining ER on Cb4 with Ad5 dorsally, dor- solaterally slightly overlapping Ad4 medially, which is also on Eb4 a little ventral to level of OP attach- ment. Ad1-—3 absent. Ad4 dorsally on Eb4 posteriorly, beginning me- dially just anteroventral to lateral edge of OP and extending laterally to end of Eb4, ventrally on Cb4 dorsally, beginning near lateral end and extending medially anterior to OP. Ad5 ventrally on Cb5 distally, joining ER dorsally with OP ventrally on Cb4, beginning well medial to distal end of Cb4 and continuing medially. SOD present. RDs separated by distance less than one RD di- ameter, extending far anteriorly and inserting on Pb3s dorsally. Additional remarks. SCL absent. TV4 free from Cb5s. Pb4 absent, UP4 present. Pb2 toothed. Pb1 ab- sent. [AC absent. Medial end of Eb3 larger than that of Eb4. PCI begins attachment to Cb5 well medial to distal end of Cb5, attaches by moderately long ten- don to cleithrum. CARAPIDAE Not examined. Remarks. Vandewalle et al. (1998) briefly de- scribed and illustrated the gill-arch skeleton and mus- culature of the Carapidae. The publication was avail- able to us only as a poor quality photocopy. Never- theless, several important features were apparent. Some of the specializations they report (LP absent, LI2 penetrates OD, etc.), are shared with bythitids and some ophidiids. All levators originate on inner surface of the hy- omandibula (we did not take note of the origins in our dissections of other ophidioids, which Vande- walle et al. noted would be of interest). LE1-LE4 present. LP absent. LI1 present (their LINT 2/3 or LINT3) inserting on Pb2 and Pb3 or Pb3, depending on genus. LI2 (their LINT4) inserting on “Pb4” (probably UP4, presence of cartilaginous Pb4 not mentioned), penetrates OD3, which they recognize as OD3s and OD3p. IAC absent, but IAB (with cartilaginous ends) present, which Vandewalle et al. term an “interarcual element.”’ Eb1l uncinate process apparently absent. TPb2, TEb2, and TEb3 present. OD1,2,3s,3p,4. Originating variously on Pb2, Pb2, NUMBER 11 Pb2 and 3, Pb4 [= UP4?], respectively, and inserting on Eb1, Eb2, Eb3, Eb3, Eb4 respectively. OP originating on Pb4 [= UP4?] or Eb4 depending on genus, and inserting on Cb5. RecDs [their REDOs]. RecD1 origin on Pb2, in- sertion on IAB; RecD2, origin on JAB, insertion on Eb2; RecD3 origin on Eb2, insertion on Eb3; RecD4, origin on Eb3, insertion on Eb4. Ad1-—3 apparently absent. Ad4 apparently absent, but should be verified. Ad5 on distal end of Cb5 and on Eb4 (and in one genus, also on Eb3). SOD apparently absent. RDs originate on second and third vertebrae, insert at front of Pb3s. TV4 present, not attached to Cb5s. Gadiformes Endo (2002) published a phylogenetic study of the of the Gadiformes. He discussed very little about the dorsal gill-arch musculature, but mentions (2000:82), following other authors, that TDA comprises only TEb2 in the gadiforms he examined. He (2002:fig. 34) only illustrates (partially) the dorsal gill arch musculature of Gaidropsarus (Gadropsaridae) and Bregmaceros (Bregmacerotidae). The musculature of Gaidropsarus is similar to that of Raniceps and Op- sanus (a batrachoidiform) in lacking LE3, but in- cludes M. Pb2-Eb2 (labelled as OD2), which neither of the other two genera have. Endo (2002:101) noted that M. Pb2-Eb2 “is generally absent in most gadi- forms, but present in lotines, gaidropsarines, phyci- nes, gadines (except Gadiculus, Brosme) and the morid Lotella.”’ His illustration of Bregmaceros also shows that it lacks LE3, as well as M. Pb2-Eb2 and LP. In Bregmaceros, he illustrates only the dorsoan- terior insertion of a muscle he identified as OP, which attaches to Pb3 anteromedially. We suspect that this may equate to our M. SO-Pb3. M. Pb2-Eb2 is commonly present in paracanthop- terygians. RANICIPITIDAE Raniceps raninus (Linnaeus), USNM 345222, 105 mm S; USNM 307233, 105 mm. Plate 66 Description. LE1 on Eb! dorsoanteriorly lateral to tip of unci- nate process. Remarks. Ligament extends medially from carti- laginous tip of uncinate process to dorsolateralmost cartilaginous edge of Pb2. Another ligament extends posterolaterally from cartilaginous tip of uncinate process to anterior edge of Eb2 anterior to lateralmost end of TEb2, and is here joined by a third ligament 89 extending directly posterior from the posterior edge of Eb1 posterior to LE1 insertion. LE2 on raised bony dorsal edge of Eb2 posteriorly about mid-laterally. LE3 absent. LE4 massive, on most of Eb4 bony surface dor- soposteriorly lateral to uncinate process. LP very small, easily removed or overlooked; join- ing LE4 insertion about mid-anteriorly. LI1 clasps edge of Pb3 between two anterior car- tilage-tipped processes (strut | and strut 2 of Markle, 1989:fig. 2B), dorsally ventral to OD4 origin and ventrally dorsal to Pb2. LI2 on Pb3 posterolaterally just anterior to artic- ulation with Eb4 and ventral to anteriorly extending SO muscle strap, which passes ventral to Eb4 portion of TPb3-Eb3-Eb4. TD comprises TEb2 and TPb3-Eb3-Eb4. TEb2 broad, bandana-like, attaching broadly mid-ventrally to CT of pharyngeal roof, extending laterally and at- taching to Eb2 dorsally anterior to LE2 insertion, continuing broadly posteriorly as TPb3-Eb3-Eb4. In- dividual sections of TPb3-Eb3-Eb4 relatively distinct and well separated laterally, but obscured from view by OD4, which lies dorsal to them: Pb3 section broadly dorsally on Pb3 process articulating with Eb2 medial end, separated by LI2 from Eb3 section, which attaches to dorsomedial bony surface of Eb3 and is separated laterally, but adjacent posteriorly, to Eb4 section, which attaches to ventromedial edge of Eb4 uncinate process (SO fibers attach to Eb4 dor- sally ventromedial to Eb4 section). Eb4 section con- tinuous mid-posteriorly with SOD. OD3-—4 origin on Pb3 dorsomedially ventral to TEb2; insertion mainly on dorsomedial edge of Eb4 uncinate process with slight tendinous connection to uncinate process of Eb3. OP fused laterally with Ad4 and ventromedially with SO, joined ventrolaterally to raphe (ER) with Ad5; most clearly represented by strands on bony surface of Eb4 uncinate process posteriorly and Cb5 ventrally. Ad1-—3 absent. Ad4 fused medially with OP, ventrally joining ra- phe with Ad5 dorsolaterally; most clearly represented by broad attachment to Cb4 dorsally medial to Eb4- Cb4 joint. Ad5 dorsally on Cb4 well medial to distal end; ventrally on Cb5 dorsodistally; dorsolateral margin outlined by raphe with Ad4 and OP, but free dorso- medially and medially. Remarks. It is unusual in acanthomorphs for the attachment of Ad5 to be so far removed medially from the distal end of Cb4. Condition also occurs in Opsanus (Batrachoididae), Lactarius (Lactariidae), cottoids, and Xenolepidichthys (Grammicolepidae). SOD broad. 90 RDs separated by distance less than diameter of one RD. Additional remarks. SCL absent. TV4 free from Cb5s. Pb! tiny and cartilaginous. Pb4 and UP4 ab- sent. IAC absent. Batrachoidiformes BATRACHOIDIDAE Opsanus beta (Goode and Bean), USNM 301938, 76.3 mm. Plate 67 Description. LEI on anterior surface of raised bony process on Eb1. LE2 on anterior surface of raised bony process on Eb2. LE3 absent. LE4 very broadly on dorsolateral surface of Eb4. LP absent. LI1 mainly on anterolateral edge of Pb3 with mi- nor attachment to anterior edge of Pb2 medially. LI2 on Pb3 dorsolaterally, immediately adjacent to medial end of Eb3. TD comprises TEb2 and TEb3-Eb4. TEb2 broad with mid-longitudinal raphe, which gives rise to CT sheets dorsally; attached mid-ventroanteriorly to CT of pharyngeal roof; muscle extends laterally and at- taches on Eb2 dorsally to point anterior to mid-point of LE2 insertion. TEb3-Eb4 begins on dorsoantero- medialmost edge of Eb3 and continues around to pos- teromedial edge, with ventroposterior fibers attaching onto Eb4 dorsomedial surface and posteromedial edge. OD3-—4 origin broadly on Pb3 dorsally with some ventral muscle strands extending from origin onto Eb3 dorsomedially, slightly overlapping TEb3-Eb4 on Eb3; minor insertion on medial edge of Eb3 un- cinate process, mainly on Eb4 uncinate process me- dially and posteromedially, joining raphe with OP dorsally. OP left side dorsally on Eb4 uncinate process pos- teriorly, joining raphe with OD3—4 on Eb4, right side dorsally (anomalously?) on Eb3 and Eb4 uncinate processes posteriorly, ventrally on Cb5 dorsally me- dial to distal end, ventromedially not completely sep- arable from SO. Remarks. PCI (removed in Plate 67B) attaches partly to distalmost end of Cb5, but its tendinous dor- soanterior edge is joined by OP ventrolaterally and it meshes with Ad5 ventrally just dorsal to distal end of Cb5S. Ad1-—3 absent. Ad4 dorsally on ventral edge of distal half of Eb4 medial to Eb4-Cb4 joint, its medialmost edge anter- BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON oventral to OP; ventrally dorsal edge of distal half of Cb4. Ad5 dorsally on Cb4 well medial to distal end, meeting Ad4 ventromedially, ventrally on CbS5 dis- tally, its posteromedial surface meshing with PCI just dorsal to Cb5. Remarks. It is unusual for the attachment of Ad5 to be so far removed medially from the distal end of Cb4. Condition also occurs in Raniceps (Gadidae), Lactarius (Lactariidae), cottoids, and Xenolepidi- chthys (Grammicolepidae). SOD absent. RDs slightly separated. Additional remarks. SCL attached mid-dorsally to posteroventral end of Bb3. TV4 free from Cb5s. Pb4 and UP4 absent. Pb2 toothed. IAC present, reduced. Lophiiformes CHAUNACIDAE Chaunax pictus Lowe, USNM 187752, 73.3 mm. Plate 68 Description. Remarks. The anterior ends of the Pbs of Chaunax appear to have been rotated dorsally and the gill arch- es are crowded close together. LEI slender, on dorsal edge of Eb1l uncinate pro- cess just lateral to tip. LE2 very broad, on dorsoanterior edge of Eb2, meeting M. Pb2-Eb2 insertion posteriorly. LE3 absent. LE4 stout, on most of dorsal surface of Eb4 lateral to uncinate process, insertion meeting dorsal attach- ment of Ad4 on one side and insertion fibers inter- mingling with those of Ad4 on other side in what we consider a “sling” (Stiassny and Jensen, 1987); mus- cle fibers of left side twist clockwise toward origin, those of right side twist counterclockwise. LP absent. LI1 mainly on dorsolateralmost edge of Pb3, but slight attachment to distinct spongy CT pad attaching to dorsal cartilaginous tip of Pb3 and, on right side only, a few fibers attach to dorsalmost tip of Pb2; joins raphe anteroventrally with dorsoposterior edge of M. Pb3-Eb3. LI2 on dorsoposterior surface of Pb3 lateral to base of dorsal process. TD comprises TEb2, TPb2, and TEb4; TEb2 high- ly modified, fails to attach to Eb2 (unique among acanthomorphs we examined); muscle mostly ventral to TPb2, divided mid-longitudinally by broad raphe, which gives rise to thin CT sheets attaching to skull; anteromedially muscle on dorsoposterior surface of Pb2, attaching to spongy CT pad anterolateral to mid- longitudinal raphe, laterally with short converging anterolateral and posterolateral sections joining raphe NUMBER 11 on dorsolateralmost edge of Pb2, which in turn is joined by another raphe laterally with M. Pb2-Eb2 medially. TPb2 a semicircular ribbon on each side arising anteriorly and posteriorly from mid-longitu- dinal raphe of TEb2, also forming raphe mid-laterally with TEb2. TEb4 on Eb4 posteriorly well medial to uncinate process and well lateral to medial end of Eb4. M. Pb2-Eb2 originates anteriorly on Pb2 dorsola- teralmost edge, there joining raphe with TEb2 later- ally, and inserts on Eb2 along anterior edge of broad LE2 insertion. OD3-4 originates broadly on Pb3 anteromedially, joining raphe dorsoposteriorly with ventrolateral edge of LI1; lateral fibers extend ventrally and insert on Eb3 dorsally well medial to uncinate process, re- maining fibers attenuate posteriorly insert by fine ten- don on Eb4 uncinate process just ventral to cartilage tip. OP dorsally on Eb4 posteriorly below and slightly medial to uncinate process, ventrally on Cb5 dorso- posteriorly, ventromedially just meeting TVS. Ad1-—3 absent. Ad4 dorsally broadly on Eb4 ventral to LE4 in- sertion, ventrally very broadly on Cb4 dorsal surface medial to Eb4-Cb4 joint. Ad5 absent. Remarks. The complete absence of Ad5 is uncom- mon in fishes. Even when not distinct, an indication of its fusion with OP or another muscle is usually discernible. SOD present. RDs adjacent. Additional remarks. SCL absent. TV4 in two dis- tinct parts, dorsal section a bilateral pair of muscles attaching separately to ventral surface of broad Bb cartilage plate; ventral section continuous, free from Cb5s, the two parts continuous by a fine muscle strand. A very similar TV4 is found only in Dicro- lene (Ophidiidae), q.v. Pb4 and UP4 absent. IAC ab- sent. Acanthopterygii Stephanoberyciformes MELAMPHAIDAE Poromitra capito Goode and Bean, USNM 258325, 96.1 mm. Plate 69 Additional material. @ = Scopelogadus mizolepis bispinosus (Gilbert), USNM 356388, 81.7 mm. Description. LE1 on dorsoposterior edge of Eb1 distal to un- cinate process. LE2 on raised mid-dorsoposterior edge of Eb2. 91 LE3 on Eb3 dorsally lateral to uncinate process. LE4 on posterodistal margin of Eb4 barely meet- ing LP anteriorly. LP on Eb4 at and just posterior to LE4 insertion. LI1 approximately mid-medially on dorsal surface of Pb2 anterior to uncinate process, smaller than LI2. @ Laterally on dorsal surface of Pb2 anterior to un- cinate process. LI2 on Pb3 dorsoposteriorly medial to medial end of Eb3, larger than LI1. TD comprises TPb2, TEb2, and TPb3-Eb3. TPb2 broadly on dorsal surface of anterior arm of Pb2 an- terior to uncinate process, posterormedianly dorsal to (appressed on) anterior end of TEb2, which it joins along mid-longitudinal raphe. TEb2 attaches on dor- soposterior edge of Eb2 medial to LE2 insertion. TPb2 and TEb2 are dorsal to and discontinuous from TPb3-Eb3, which is very thin, broad, sheet-like, and ventrally joins CT lining pharynx dorsally. Anteri- orly, TPb3-Eb3 attaches broadly on Pb3 medially paralleling origin of OD3-—4; posteriorly, TPb3-Eb3 expands laterally and attaches to dorsomedial end of Eb3. Change in muscle fiber direction demarcates separation of TPb3-Eb3 from SO, with which it is posteriorly broadly continuous; @ Eb3 muscle por- tion continuing posteriorly as band of crisscrossing fibers, which change abruptly posteriorly to all trans- verse SO fibers. OD3-4 originates broadly along dorsolateral edge of Pb3 and inserts dorsally on joined Eb3-Eb4 un- cinate processes. OP dorsally on Eb4 between Ad4 and SO attach- ments to Eb4, ventrolaterally attaches to dorsal edge of gill raker contained in CT dorsal to Cb5, ventro- medially forms raphe (ER) with Ad5, scarcely sep- arable medially from SO. @ Well separable medially from SO. Ad1-—3 absent. Ad4 dorsoposteriorly on bony and cartilaginous distal end of Eb4, attached by slender tendon ven- trally to dorsal surface of Cb4 slightly anterior to inner angle formed by Eb4-Cb4 joint. @ Attached musculously to Cb4. Ad5 narrowly, tendinously attached to distal end of Cb4; mid-dorsally more broadly attached to pos- teroventral surface of gill raker embedded in CT; dor- somedially joins raphe (ER) with to OP ventrome- dially; not separable medially from OP or SO; ven- trally on Cb5 dorsolaterally, ventromedially contin- uous with SO. @ Attached musculously to Cb4. SOD absent. RDs small, widely separate, inserting on Pb3 dor- soposteriorly; broad, short, thin band of longitudinal SO fibers extends anteriorly on each side and slightly overlaps respective RD insertion. @ Inserts on carti- laginous posteromedial end of Pb3 and medial end of Eb4. 92 Additional remarks. SCL absent. TV4 free from Cb5s. Slender ligament (not illustrated) attaches tip of small prominence at cartilaginous medial end of Eb2 to Pb3 anterior to uncinate process; @ ligament absent. Autogenous rod-like cartilage attached to ventroposterior cartilaginous end of Bb3; @ rod-like cartilage absent. Pb4 absent, UP4 present. Pb1 with cartilage caps dorsally and ventrally. Pb2 toothed. Eb4 levator process absent. Medial end of Eb4 larger than that of Eb3. Similar modified gill-raker noted in OP descrip- tion, otherwise noted in this study to occur only in Scomber. GIBBERICHTHYIDAE Gibberichthys pumilus Parr, USNM 214207, 79.5 mm. Plate 70 Description. LE1, weak, on Eb! near tip of uncinate process. LE2 on posterodorsal edge of Eb2 somewhat prox- imal to mid-length. LE3 on dorsoanterior surface of Eb3 uncinate pro- cess just dorsal to OD3—4 insertion. LE4 on dorsodistal end of Eb4, joining LP inser- tion posteriorly, joining narrow raphe with Ad4 dor- sally. LP slender, on Eb4 joining LE4 insertion antero- laterally. LI1 on Pb2 dorsoanteriorly ventral to anterior sec- tion of TPb2 (divides TPb2 into anterior and poste- rior sections). LI2 on Pb3 mid-dorsoposteriorly, lateral to and meeting lateral edge of TPb3 attachment. TD comprises TPb2 and TPb3. TPb2 on Pb2 an- terior process and basal area of uncinate process, di- vided into anterior and posterior sections by LI1; an- terior section partially overlies anterior end of pos- terior section; few muscle strands of anterior section attach to anterior end of right-side Pb3 and few strands of posterior section join OD3—4. TPb3 on Pb3 posterolaterally, meeting medial edge of LI2 in- sertion, posteriorly continuous with SOD. OD3-4 anteriorly broadly on Pb3 anteromedially ventral to TPb2, divides at about posterior fourth of length with anterior branch attaching to Eb3 uncinate process and posterior branch attaching to Eb4 unci- nate process. OP dorsally on Eb4 posteriorly beginning just ven- tral to tip of unciante process, ventrally joining raphe (ER) with Ad5 dorsally on Cb4. Ad1-—3 absent. Ad4 dorsally on dorsodistalmost cartilaginous end of Eb4 posteriorly, joining narrow raphe with LE4 ventrally, ventrally broadly on Cb4 dorsodistally. Ad5 dorsolaterally on Cb4, there joining raphe BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON (ER) with OP ventrally, posteroventrally on dorso- distal end of Cb5, joining raphe with TV5. SOD present. RDs separate anteriorly, united posteriorly at ori- gin, inserting on Pb3 dorsal surface posteromedially. Additional remarks. SCL present, attached dorso- medially to cartilaginous posteroventral tip of Bb3. TV4 free from Cb5s. Pb4 and UP4 present. Pb2 edentate. IAC absent. Eb4 levator process absent. PCI attaches to cleithrum by extremely long, fine ten- don. Medial end of Eb4 much larger than that of Eb3 on one side, about same size on other. STEPHANOBERYCIDAE Stephanoberyx monae Gill, USNM 304376, 104 mm. Plate 71 Description. LE] fine, on dorsal edge of Eb1 distal to uncinate process. LE2 on dorsoposterior edge of Eb2 somewhat dis- tal to mid-length of Eb2. LE3 on dorsodistal edge of Eb3, inserts jointly with OD3 insertion. LE4 on dorsodistalmost bony edge of Eb4, joined posteroventrally by raphe with dorsolateral portion of Ad4 origin, some fibers continuing across raphe. LP slender, joining LE4 dorsoanterior to insertion, wrapping partway around LE4 and becoming slightly separate just ventral to origin. LI1 on Pb2 mid-dorsolateral edge ventral to ante- rior section of TPb2. LI2 on Pb3 dorsoposteriorly somewhat lateral to medial edge of bone; anteromedial half of insertion joins posteromedial half of edge of TPb3 attachment to Pb3 on one side, but is completely posterior to TPb3 insertion on the other side; posteromedial half of insertion borders SO fibers extending dorsoanter- iorly along surface of Pb3. TD comprises TPb2, TPb3, and TEb2. TPb2 sep- arated laterally by LI1 insertion, anterior section at- taches to Pb2 anterior arm, posterior section attaches at base of Pb2 uncinate process (on right side, fibers of posterior section continue short distance anteriorly ventral to anterior section before uniting with anterior section fibers). TEb2 attaches in CT enveloping dor- somedial end of Eb2-Pb2 joint. TPb3 passes laterally dorsal to OD4 origin, then extends deeply ventrally and attaches to Pb3 dorsal surface medial surface an- terior to LI2 insertion on one side and meets antero- medial edge of LI2 insertion on the other side; ventral layer (not illustrated) of TPb3 muscle fibers passes knifelike through OD4 and attaches to Pb3 medial to attachment of lateral TPb3 attachment. OD comprises OD3, OD3'’, OD4. OD3 and OD3’ originate together on dorsoanterior surface of Pb3 be- ginning just posterior to cartilaginous anterior end, NUMBER 11 muscle forms two short divisions posteriorly; lateral division inserts with LE3 insertion on Eb3 dorsodis- tal edge; medial division ventral surface attaches to Eb4 dorsal edge, but dorsal fibers continue without interruption posteroventrally, becoming fused with posteromedial fibers of OP (we have observed a somewhat similar condition only in Hypoptychus, Hypoptychidae). OD3’ branches off ventral surface of OD3 shortly posterior to origin and inserts on dor- sal edge of Eb3 ventral to OD3. OD4 originates on dorsal surface of Pb3 just posterior to OD3-OD3’ or- igin and ventral to TPb3; on one side, muscle-fibers from ventral surface of TPb3 pass through OD4 an- teriorly before attaching to Pb3; OD4 muscle inserts on medial edge of Eb4 bony process, which, in most fishes, bears cartilaginous tip of uncinate process. OP originates on Eb4, fibers are continuous dor- sally with posteromedial division of OD3; muscle lies posterior to Ad4 ventrolaterally; inserts ventrally on Cb4 posteriorly, joining small ER with dorsal end of Ad5. Ad1-—3 absent. Ad4 origin on Eb4 dorsolaterally, joined by raphe with LE4 insertion, with some fibers continuous be- tween the muscles. Ad5 originates dorsally from raphe (ER) joining ventral end of OP on Cb4, continuous anteriorly with ventromedial OP fibers, inserting on Cb5 dorsally. SOD present. RDs originate together but separate before passing anteriorly ventral to SOD and TPb3; RD inserts on Pb3 dorsoposteromedial surface of Pb3, paralleling laterally adjacent SO longitudinal fibers, which also insert on Pb3 surface beginning at posterior end of OD4 origin. Additional remarks. SCL present, attached poster- omedianly by short ligament to posteroventral carti- laginous tip of Bb3. TV4 free from Cb5s. Pb4 re- duced. UP4 absent; tooth plate fused to Eb3 incor- rectly indicated as UP4 by Rosen (1973:472, fig. 91). Pb2 edentate. [AC absent. Eb4 levator process ab- sent. Cartilage-tipped Eb3 and Eb4 uncinate process- es absent. BARBOURISIIDAE Barbourisia rufa Parr, AMNH 29772, ca. 205 mm. Plate 72 Description. Only left half of dorsal-gill arch muscles available for study. LE1 on distal dorsal bony edge of base of Eb1 uncinate process. LE2 on dorsalmost bony edge of Eb2. LE3 on cartilaginous tip of Eb3 uncinate process. LE4 on Eb4 near dorsodistal end. LP on Eb4 at and posterolateral to base of LE4. 93 LI1 on Pb2 dorsally, anteriorly ventral to anterior portion of TPb2. LI2 on Pb3 dorsally. TD comprises three parts: TPb2, TEb2, TPb3. TPb2 is divided into two portions; anterior portion on dorsal surface of Pb2 anteriorly, becoming atten- uated posterolaterally, broadly continuous posteriorly with posterior portion, which attaches ventrolaterally on Pb2 medial to LI1 insertion and is posteriorly broadly continuous with TEb2 (see also remarks un- der SO). TEb2 attaches to Eb2 dorsally well medial to LE2 insertion and is completely separate from TPb3. TPb3 attaches dorsoposteriorly on Pb3 at joint with Eb3, and is continuously posteriorly with SOD. OD3-4 originates ventral to TPb2 and TEb2 on Pb3 and inserts broadly on joined uncinate processes of Eb3 and Eb4. OP represented by broad muscle attaching to Eb4 dorsoposteriorly, dorsolaterally partially continuous with LE4, attaching to Cb4 at lateral end of ER, and joining ER dorsally; broad Ad5 joins ER ventrally, possibly includes OP medially. Ad1-3 absent. Ad4 on Eb4 dorsoposterolaterally and Cb4 dorsal- ly medial to internal angle of Eb4-Cb4 joint. Ad5 broad, joining ER dorsally and Cb5 ventro- posteriorly well medial to distal end; medial portion of muscle probably includes OP. SOD present, continuous posteriorly with SO. RDs separate, insert on dorsomedial surface of Pb3. SO branch of transverse muscle layer arises dor- somedially from SO, curves partly around medial are of RD, and extends anteriorly, becoming continuous with ventral surface of TPb3. Two distinct strap-like muscles originate from longitudinal layer, one attach- es to Pb4 and the other to Pb3. Additional remarks. SCL present. TV4 free from Cb5s. Pb4 and UP4 present. Pb2 toothed. Eb4 levator process absent. IAC absent. RONDELETIIDAE Rondeletia loricata Abe and Hotta, USNM 206836, 74.3 mm. Plate 73 Description. LE1 on dorsolateral bony edge of Eb! uncinate process. LE2 on dorsalmost bony edge of Eb2. LE3 on tip of Eb3 uncinate process dorsal to in- sertion of Eb3 branch of OD3—4. LE4 near dorsolateral end of Eb4. LP threadlike, inserts by even finer tendon on Eb4 just distal to LE4 insertion. LI1 on anteromedial surface of Pb2 ventral to an- 94 terior half of TPb2 (splits TPb2 laterally into two parts). LI2 on posterolateral surface of Pb3, medially abutting TPb3 attachment. TD comprises TPb2 and TPb3. TPb2 overlies OD3-—4 and LI1 origins and attaches to dorsomedial surface of Pb2 anterior to Pb2 uncinate process, is divided mid-laterally by passage of LI1, and is con- tinuous posteriorly by a few muscle strands with TPb3. TPb3 attaches to dorsolateral surface of Pb3, posteriorly abutting medial edge of LI2 insertion, and is broadly continuous posteriorly with SOD. OD3-—4 originates on Pb3 dorsoanteriorly ventral to TPb2, divides well posteriorly with short branch inserting on Eb3 uncinate process and another on Eb4 uncinate process. OP dorsally on posterior surface of Eb4 uncinate process, in two sections; lateral section broader, ven- trally on Cb5 dorsally, joining ER with Ad5 dorsally; medial section slender, medially inseparable from SO, ventrally on Cb5 dorsally posterior to Ad5 me- dially. Ad1-—3 absent. Ad4 dorsally on Eb4 distal cartilaginous end, ven- trally on Cb4 dorsoposterior bony surface. Ad5 on Cb4 posteriorly well medial to distal end, joining ER with OP lateral section ventrolaterally, ventrally on distal end of Cb5 dorsodistally. SOD present. RDs separate, cross (right side dorsal to left side) just anterior to origins (tendinous), insert on Pb3 pos- teromedially. Additional remarks. SCL present. TV4 free from Cb5s. Pb4 and UP4 present. Pb2 edentate. IAC ab- sent, Eb4 levator process absent. CETOMIMIDAE Ditropichthys storeri Goode and Bean, SIO 64-24- 25, 77.7 mm. Plate 74 Description. LEI absent. Remarks. D. storeri is the only acanthomorph we examined that lacks LE1. Another specimen should be examined to verify that the absence is not an ar- tefact of a faulty dissection. LE2 on Eb2 mid-dorsally. LE3 absent. LE4 broadly, dorsally on Eb4 just dorsomedial to OD4 insertion on Eb4. LP filamentous, inserts by fine tendon on Eb4 at posterior margin of LE4 insertion. LI1 on Pb2 mid-dorsally. LI2 on Pb3 dorsal surface mid-posteriorly. TD comprises TPb2, TEb2, TEb3, and TEb4. TPb2 band-like, dorsoanteriorly on Pb2, small ven- BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON trolateral slip of fibers (not visible in illustration) at- taches to Pb3 lateral edge, muscle continuous pos- teriorly with TEb2. TEb2 string-like as it passes onto and attaches dorsally on Eb2 just medial to LE2 in- sertion, continuous posteriorly by fine, spaced, di- agonal muscle threads with TEb3. TEb3 on Eb3 dor- somedially, forming raphe anteriorly with OD3 at- tachment, continuous posteriorly with TEb4. TEb4 on Eb4 anteromedial to OD4 attachment, continuous posteriorly with SOD. OD3 and OD4 with joint origin on Pb3 dorsoan- teromedial surface; unified muscle divides at point opposite mid medial edge of LI2. OD3 inserts on Eb3 anteromedially, joining raphe with TEb3. OD4 in- serts on Eb4 dorsodistally at and anterior to LE4 in- sertion. OD4 lies mostly dorsal to OD3. OP dorsally on Eb4 posteriorly, anteriorly overlap- ping most of Ad4 and possibly fusing with Ad4 pos- teromedially; ventrally broadly on Cb5 posterodistal- ly. Ad1-—3 absent. Ad4 dorsally on Eb4 posteriorly beginning medi- ally anterior to much of OP (and possibly fusing with it) and extending laterally to end of Eb4, ventrally on Cb4 posterolaterally and CT joining distal ends of Cb4 and CbS. Ad5 apparently absent. Remarks. It is unusual for Ad5 to be absent in acanthomorphs, and is possibly the result of the close attachment of the posterolateral surface of Cb4 and the anterolateral surface of Cb5, which obviates any function Ad5 might serve. SOD slender, comprising loose threads of muscle. M. SO-Pb3 strap-like section of longitudinal mus- cle fibers, passes anteriorly from base of diverticulum of SO and ventral to SOD, TEb3, TEb4, and OD and attaches anterolaterally to Pb3 dorsal surface. RDs separate, vertically elliptical, insert on Pb3 dorsoposteromedially. Additional remarks. SCL absent. TV4 complex, free from Cb5s. Pb4 tiny, UP4 large. Pb2 edentate. IAC absent. Eb3 and Eb4 uncinate processes absent. Eb4 levator process absent. Large, vertically oriented, bulb-like diverticulum on each side of SO just pos- terior to level of SOD; SO fibers attach to surface of diverticulum, which has two small glottis-like open- ings into esophagus. Icosteiformes ICOSTEIDAE Icosteus aenigmaticus Lockington, LACM 35865-1, 199 mm; HSU 81-305, 188 mm. Plate 75 Description (based primarily on LACM specimen). LE1 shortest levator, on bony dorsal edge of Eb1 uncinate process beginning just lateral to relatively long cartilaginous medial end. NUMBER 11 LE2 on dorsoposterior edge of bony portion of Eb2 at about mid-length and well lateral to TEb2 attachment. LE3 on dorsoanterior surface of Eb3 uncinate pro- cess, meeting OD3-—4 insertion laterally. LE4 on bony dorsal edge of Eb4 well lateral to uncinate process. LP on Eb4 beginning, variably, at lateral edge of LE4 insertion, or slightly anteromedial to it, and ex- tending laterally to end of bony dorsal edge. Left- side LP anomalously doubled. LI1 on Pb2 dorsal bony surface just posterior to anteriormost cartilage tip and on anteromedial edge of Pb3 adjacent to Pb2. LI2 on Pb3 dorsoposteriorly medial to medial end of Eb3. TD flat, entire dorsal surface attached to thick muscular and CT layer that lines ventral surface of skull, comprises TPb2, TEb2, and TEb3-Eb4. TPb2 kidney-shaped, concave anteriorly, attached antero- laterally to dorsal surface of Pb2 just medial to me- dial end of IAC; anteroventral surface attached by CT (not illustrated) to dorsoanterior surface of Pb3, CT continuing and attaching to entire posterior Pb1 surface; muscle fused ventromedially and posteriorly with TEb2, which attaches to Eb2 bony surface dor- soposteriorly just lateral to medial end and well me- dial to LE2 insertion. TEb2 continuous posteriorly with TEb3-Eb4, which attaches on Eb3 dorsomedi- ally and Eb4 dorsomedial edge well medial to unci- nate process, there meeting SO fibers. TEb3-Eb4 mid-posteriorly continuous by crossing muscle strands with SOD. OD3-4 origin on Pb3 dorsally posterior to anterior cartilaginous tip, insertion on joined medial edges of Eb3 and Eb4 uncinate processes. OP bilaterally asymmetrical (undamaged side of HSU 81-305 is like right side of LACM specimen, and is here considered to be the normal state). Right- side OP dorsally on Eb4 posteriorly, beginning lateral to uncinate process (and not clearly separate medially from SO fibers on Eb4) and extending laterally to below LE4-LP insertions, ventrally joins raphe (ER) with dorsal end of Ad5; tendon continues laterally from raphe, is joined by Ad4 ventrally, and attaches to Cb4 posterodistally. Left-side OP dorsally like right side, ventrally beginning laterally on lateral end of tendon attaching to posterolateral surface of Cb4 and extending to lateral end of Cb5; muscle not dis- tinguishable from Ad5, if latter is present; Ad4 ven- trolaterally also joins lateral end of tendon. Remarks. The condition of OP on the right side resembles the condition is some stephanoberyci- forms, e.g., Poromitra, or ophidiids, e.g., Dicrolene, particularly in that OP is either interrupted centrally by ER or ends ventrally at ER, but in neither case does it attach to Cb5. For this reason, we record this 95 character as “OP wholly or partly on Cb4 and/or joining ER at level of Cb4.” Ad1-—3 absent, but fine GFM (not illustrated) on anterolateral edge of each arch. Ad4 dorsally on Eb4 posterolaterally, beginning narrowly anteromedial to OP and extending to Eb4- Cb4 joint, ventrally on Cb4 dorsal surface medial to Eb4-Cb4 joint, with some posterolateral fibers join- ing tendon at ventrolateral end of OP. Ad5 apparently, probably anomalously, absent on left side, but apparently represented by muscle strands attaching ventrally to Cb5 and dorsally join- ing raphe with ventral end of OP well medial to distal ends of Cb4 and Eb4. See remarks following OP. Remarks. Attachment to ER is somewhat similar to that of Dicrolene (Ophidiidae) and some stephan- oberycids. SOD broad. RDs short, joined at origin, slightly separated an- teriorly. Additional remarks. SCL present. TV4 free from Cb5s. Tiny AC at joint between Pb! and Eb! on one side of LACM specimen, and on undamaged side of HSU specimen. AC between Eb! and Cbl and an- other between Eb4 and Cb4 on right side of LACM specimen; AC present between Eb4 and Cb4 on only one side of HSU specimen. Pb2 edentate; Pb3 bears only one or two fine teeth. Pb4 present. UP4 present on only one side of each specimen, each represented by a single, fine tooth. Additionally, we examined USNM 49163 (ca. 200 mm), which had gill arches preserved only on one side, and in which only UP4 had teeth (one) and LACM 32682-1 (114 mm) which had no pharyngobranchial teeth on either side. Interrelationships (with comments on Amarsipidae and Stromateoidei). Aside from being assigned to the order Perciformes in its own suborder, Icosteoidei (e.g., J. S. Nelson, 1994), or the catchall suborder Percoidei (Weitzman, 1998), the only group the Icos- teidae has been allied to is the perciform suborder Stromateoidei. The basis for a stromateoid relation- ship has never been explained clearly, but appears to be a general external similarity, sometimes referred to as the “stromateoid look” (Haedrich, 1967:44; Ahlstrom et al., 1976:290). Matarese et al. (1984: 577) noted that the eggs, larvae, and early juveniles of Icosteus superficially resemble those of stroma- teoid fishes, but added that more data were needed before a precise relationship could be determined. We agree that there is a remarkable general similarity in the appearance of /costeus, particularly the young, and some stromateoids (compare Matarese et al., 1984:fig. 306 with Horn, 1984:fig. 333), but add that it remains to be shown that the similarities are ho- mologs. Haedrich (1967), who reviewed the stromateoids comprehensively (and included /costeus among his 96 non-stromateoid comparative material), did not in- clude Jcosteus among possible close relatives of stro- mateoids. Nor did he mention /costeus when he de- scribed Amarsipus (Haedrich, 1969), which he in- cluded among the stromateoids, even though it lacks the main stromateoid synapomorphy: saccular out- growths posterior to the last gill arch. Haedrich (1969) justified the inclusion of Amar- sipus within the stromateoids by indicating a group of characters that it shares with the other stroma- teoids, thus implicitly re-defining the Stromateoidei: perciform caudal skeleton, uniserial jaw teeth, ex- panded lacrimal, inflated and protruding top of the head, extensive sub-dermal canal system, bony bridge over the anterior vertical canal of the ear, which Haedrich (1971) named the pons moultoni. Johnson and Fritzsche (1989:16), discredited the pons moultoni as a stromateoid synapomorphy be- cause it is widely distributed among acanthomorphs. Of the remaining characters, /costeus clearly exhibits only the uniserial teeth and an inflated and protruding top of the head, both of which also occur variously among perciforms, and a perciform caudal skeleton, which is too nebulous a character to constitute a basis for relationships. The caudal fin of Amarsipus (Hae- drich, 1969:fig. 5) is relatively unspecialized, with autogenous: parhypural, hypurals 1—5, two epurals, and two pairs if uroneurals (the plesiomorphic actin- opterygian caudal fin differs in having three epurals). Also present are a short neural arch on PU2 and au- togenous hemal spines on PU2 and 3. The compo- sition of the fin, however, can vary considerably, de- pending on taxon, in almost every acanthomorph clade, but no percomorph has more than five hypur- als. We are uncertain of the exact structure of the cau- dal fin of Jcosteus as we only had radiographs of two specimens to assess its composition. USNM 49163 and 37327: parhypural and hypurals 1 and 2 autog- enous, hypurals 3, 4 and 5 (6 apparently not present) separate for most of their lengths, but all possibly fused proximally to urostyle; autogenous [paired] uroneural, two epurals, long neural spine on PU2. USNM 49163: procurrent + caudal-fin rays (dorsal/ ventral) 19/15, of which the central 9/9 are branched; autogenous hemal spines on PU2-—6 (additionally, dorsal fin 54, of which anteriormost is a nubbin, last ray split to base; anal fin 38, of which anteriormost is a nubbin, last ray split to base; vertebrae, including caudal fin, 21+48 or 20+49). USNM 37327: pro- current + caudal-fin rays 17/16 (damaged), autoge- nous hemal spines on PU2—4 (additionally dorsal fin 56 or 57 and anal fin 38, vertebrae 22+50). In any event, we find no basis for relating Icosteus with Amarsipus or with other stromateoids, in which the caudal fin is often more specialized than that of Amarsipus or Icosteus. We note with interest, how- BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON ever, the large number of autogenous hemal spines, 3 and 5 in the two specimens of Jcosteus we radio- graphed. It is most unusual for acanthomorphs to have more than two autogenous hemal arches, Ste- phanoberyx monae (Stephanoberycidae), a notable exception, has 3. We have noted no specializations of the gill-arch muscles that might indicate a close relationship be- tween Jcosteus and Amarsipus and/or the other stro- mateoids. Horn (1984) compiled meristics and other char- acters of stromateoid taxa, including Amarsipus, in an attempt to unravel their intra-subordinal relation- ships. Using Girellidae, Kyphosidae, and Scorpididae (last now generally included in the Kyphosidae), which he considered closely related to stromateoids, as outgroups. Horn provided a cladistic analysis of the stromateoids, for which he found only three char- acters that supported the mcnophyly of Amarsipus + stromateoids: cycloid scales (as opposed to ctenoid), lack of scales in the preopercular area (as opposed to present), and presence of 6 hypurals (as opposed to 5). While possibly valid indications of stromateoid relations to the three outgroups, the first two char- acters are not particularly innovative and occur wide- ly among percomorphs (e.g., /costeus lacks scales, except for imbedded prickles along the lateral line, and on the basis of scalation cannot be excluded from relationship with stromateoids). The hypural charac- ter is erroneous. The Girellidae and Kyphosidae (in- cluding scorpidinins) have an autogenous parhypural and five autogenous hypurals, as does Amarsipus. Horn may have been misled by Haedrich’s (1969:fig. 5) recognition of 6 hypurals in Amarsipus, one of which is actually the parhypural. Without further elongating this discussion, we, like Haedrich, find no reason to ally Jcosteus with the stromateoids. On the other hand, unlike Haedrich and more recent followers, we find no reason to consider Amarsipus closely related to the stromateoids. We, thus, consider the Amarsipidae to be incertae sedis among the percomorphs. (See also remarks following description of Amarsipus, for a possible additional character of stromateoids that is not represented in Amarsipus.) Matarese et al. (1984:577) noted that the sequence of fin formation and reduced number of pelvic-fin rays in Icosteus are “blennioid”’ characters. All re- cently published accounts of the pelvic-fin formulae of Icosteus variously state that it is 4, 5, or I,4 (blen- nioids have I,2—I,4), but we find that there is an im- bedded, greatly reduced spine closely applied to the base of the first segmented ray and that the formula is I,5 (pelvic fins are lost in the adult). We find no evidence of a relationship with the Blennioidei, or even the so-called northern blennioids, Stichaeoidei, which formerly were believed to be closely related NUMBER 11 to the true blennioids (Springer, 1993:493). The pres- ence of IAC and Pb2 and, occasionally, UP4 in Icos- teus exclude it from inclusion among the Blennioidet, which possess none of these structures. The presence of a basisphenoid in /costeus (R. Fritzsche, in litt., 4 Dec 2001, based on notes made by a former graduate student, K. Komori), two nostrils on each side of the head, and IAC, exclude it from inclusion among the Stichaeoidei, which lack a basisphenoid and IAC, and have a single nasal opening on each side of the head. Of possible bearing on the relationships of the Icosteidae, are the presence of the accessory cartilag- es mentioned in additional remarks at the end of the muscle description of /costeus. The distribution of these cartilages in acanthomorph fishes, as far as we have surveyed, is presented in Table 8. /costeus has these cartilages variably at the joints between Eb1 and Cbl and between Eb4 and Cb4 (also, uniquely in our experience, at the joint between Pb! and Eb1). We have found these cartilages on the first or fourth arches in a variety of acanthomorphs (Table 8), but on both the first and fourth arches only in Centro- pomus (Centropomidae), Decapterus (Carangidae) and Selenotoca (Scatophagidae), besides IJcosteus (Decapterus also has them on the second and third arches). For the most part, accessory cartilages are restricted to percomorphs. Fin prickles. Another character that may have bearing on the interrelationships of Jcosteus is the presence of fine prickles on the lateral surfaces of the rays of all fins (except for a reduced, imbedded pel- vic-fin spine and a vestigial anteriormost element in the dorsal and anal fins, the fins comprise only seg- mented rays). Although a variety of fishes may have fin spines that are serrated or with prickles, the pres- ence of prickles on the segmented fin rays of all or some fins is of relatively limited occurrence. Allen (2003) discusses the presence and ontogeny of the prickles in Jcosteus. Among acanthomorphs, prickles are also present in the following (1-14): 1. Early stages of the three most specialized acan- thuroid families: Luvaridae, Zanclidae, and Acan- thuridae (Johnson and Washington, 1987:504); thus, are synapomorphic for these families. 2. Adults of at least three tetraodontiform families (on dorsal, anal, and caudal fins), including the basal Triacanthodidae (Tydemania navigatoris Weber, USNM 307551: Parahollardia lineata) and the Bal- istidae (Sufflamen chrysopterus (Bloch and Schnei- der), USNM 224687, 353282; Balistes capriscus, USNM 240664) and Monacanthidae (Monacanthus ciliatus (Mitchill), USNM 353975, 353293). 3. Early juveniles (and, variously, adults) of the beryciform families Anoplogastridae, Diretmidae, Anomalopidae, and Trachichthyidae (Baldwin and Johnson, 1995; Konishi, 1999:figs. 3 & 5, provides 97 SEM photographs of the prickles in larval Anomal- ops and Hoplostethus), which together with the Mon- ocentridae have been hypothesized to form a mono- phyletic Trachichthyoidei (Moore, 1993), or together with the Monocentridae, Holocentridae, and Beryci- dae, a monophletic Beryciformes (Johnson and Pat- terson, 1993). 4. Monocentrus japonicus (Houttuyn) (Monocen- tridae; USNM 231938) has fine prickles on the seg- mented dorsal, anal, pectoral, and caudal fins (holo- centrids do not have the prickles). 5. Segmented pelvic-fin rays of an unidentified zeiform (?Macrurocyttidae, USNM 367331). 6. Capros aper (Linnaeus) (Caproidae, USNM 289207), on segmented rays of all fins. Capros is sometimes allied to the zeiforms (Tyler et al., 2003), but was excluded from that group by Johnson and Patterson (1993) and assigned to the Perciformes. (Not present in the other caproid genus, Antigonia). 7. Four most specialized of the seven lampridi- form families (at least dorsal fin; Olney, 1984:379; Olney et al., 1993:160): Radiicephalidae, Lophotidae, Regalecidae (Regalecus), Trachipteridae (Trachipte- rus, also at least caudal fin; USNM 175344; Zu, USNM 272111, also pectoral, pelvic, and caudal fins, and prickles on lateral line). 8. Stephanoberycidae, all fins (prickles also on lat- eral line like /costeus). 9. Barbourisia (Barbourisidae), USNM 215469; all fins like Jcosteus, only segmented rays (ca. 42 vert.; 21-22 dorsal-fin rays, flexible body. No IAC). 10. Chaunax (Chaunacidae, Lophiiformes), dorsal and caudal fins. 11. Antennariidae. According to Pietsch and Gro- becker (1987:30 and table 1), spinules are present on the body and fins of most taxa. We note that the spinules may occur on the interradial membranes as well as the segmented rays (USNM 73115). 12. Ogcocephalidae (Dibranchus atlanticus, USNM 158003, on base of caudal only; appears to be continuation of body armature; variably present on fins of Ogcocephalus). Not present in ceratioids. 13. Centriscidae (all segmented fin-rays of Ma- crorhamphosus); however this state is not present in the more specialized member of the family Aeoliscus (USNM 305976). 14. Priacanthidae. To sum this up, fin prickles occur in at least some members of: Lampridiformes, Paracanthopterygii (Lophiiformes), Stephanoberyciformes, Zeiformes, Caproidae (possibly closely related to zeiforms), Be- ryciformes, Acanthuroidei, Tetraodontiformes, Cen- triscidae, and Priacanthidae. The nature and disposition of the prickles on the fin rays and the lateral line exhibit differences among the groups. A study of the variation might indicate various interrelationships among the groups. In gen- 98 BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON Table 8.—Distribution of accessory cartilages (ACs) at Eb—Cb joint of arches 1—4 in acanthomorph fishes. A few taxa that lack ACs and which are described in family accounts are included (all those not included lack ACs). * denotes taxa not mentioned in descriptive accounts. P = present; — = absent; PA = present or absent. Putatively closely related groups of families are boxed. AC at joint of arch AC at joint of arch Taxa Taxa 1 2 3 4 1 2 3 4 Veliferidae Pomacanthidae Velifer hypselopterus Centropyge (three species)* - = 2 P Metavelifer multiradiatus Pomacanthodes semicirculatus - - - P Lampridae Chaetodontidae Lampris guttatus Trachipteridae Trachipterus sp.* Chaetodon austriacus* = - P Chaetodon melannotus * - - P Chaetodon triafasciatus * - 2 P - P P P Berycidae Forcipiger flavissimus * = = Beryx splendens - - - P Heniochus acuminatus * = - - Centroberyx affinis - - - P Symphysanodontidae Icosteidae Symphysanodon berryi - - - P Icosteus aenigmaticus PA - - PA Symphysanodon octoactinus - - - = Menidae Symphysanodon sp nov - 2 = 2 Mene maculata - - P P Ammodytidae Sebastidae Ammodytes dubius - - - 1» Sebastes proriger Scorpaenidae Pontinus rathbuni Neomerinhe beanorum Moronidae Morone americana - Morone mississippiensis - Belonidae Strongylura timucu Tylosurus crocodilus Scomberesocidae Cololabis saira Exocoetidae Exocoetus obtusirostris Hemiramphidae Nematistiidae Nematistius pectoralis Echeneidae Echeneis naucrates Remora remora* Rachycentridae Rachycentron canadum Coryphaenidae Coryphaena equiselis Coryphaena hippurus Carangidae Carangoides crysos * Decapterus macrosoma* Decapterus punctatus * Selar crumenophthalmus Hemiramphus far Seriola sp.* Chauliodontidae Selene vomer * Chauliodus sloani* - - - - Scomberoides tol * Centropomidae Cepolidae Centropomus undecimalis Acanthocepola limbata - - - 1p Latidae Cepola rubescens - - - P Lates niloticus Kuhliidae Ambassidae Kuhlia Ambassis sp.* - - - P Arripidae Ambassis buruensis - - - - Arripis georgianus* Tetracentrum caudovittatus 2 “ PA S Terapontidae Dinopercidae Leiopotherapon unicolor Dinoperca petersi* - - - P Terapon jarbua Epigonidae Girellidae Epigonus pandionis - - - P Girella tricuspidata Sphyraenops bairdianus * - - - J? Kyphosidae Haemulidae Haemulon (two species) Plectorhynchus pictus Pomadasys crocro Inermiidae Inermia vittata Lobotidae Cichlidae Lobotes pacificus - - - Je Paratilapia polleni - PA - - Coiidae 11 other genera & species - - = - Coius - - - P Embiotocidae Mullidae Amphisticus argenteus - - - PA Pseudupeneus maculatus - PA = - Cymatogaster aggregata - - - - Mulloides flavolineaus - P - - Embiotoca lateralis - - - IP Parupeneus maculatus - Kyphosus sectatrox * Scorpididae Microcanthus strigatus* Scorpis sp.* Monodactylidae Monodactylus argenteus * - - - P Hysterocarpus traskii = - © . NUMBER 11 Table 8.—Continued. Taxa AC at joint of arch 1 2 3 4 Percichthyidae Bostockia porosa* = = © 2 Gadopsis marmoratus * = = = = Macquaria colonorum - - - P Sillaginidae Sillago sihama - P 3 z Pempheridae Parapriacanthus Pempheris Glaucosomatidae Glaucosoma Acropomatidae Acropoma sp.* = © é PA Apogonops anomalus * - - = P Doederleinia berycoides* - - - P Synagrops bella - - - P Dinolestidae Dinolestes lewini * = - = = Sphyraenidae Sphyraena barracuda - - = = Gempylidae Neoepinnula americana * Promethichthys prometheus * Trichiuridae Trichiurus lepturus* Scombridae Euthynnus alletteratus * Scomber scombrus Scomberomorus cavalla * Scomberomorus commersoni * Gasterochisma melampus * Istiophoridae Istiophorus sp* Priacanthidae Heteropriacanthus cruentatus — - - - PA Drepanidae Drepane (two species)* Ephippidae Chaetodipterus (two species) Ephippus orbis * Platax orbicularis * Scatophagidae Selenotoca multifasciata* Scatophagus argus * Siganidae Siganus (Lo) vulpinus * Siganus (S. ) spinus* Acanthuridae Acanthurus nigrofuscus eral, prickles on the segmented fin rays are most common among deep-dwelling fishes and non-per- comorphs (acanthuroids, tetraodontiforms, and pria- canthids, not withstanding). ER, which is present in /costeus, has a limited dis- 99 AC at joint of arch 1 2 3 4 Taxa Embiotocidae con't Phanerodon atripes S = = P Phanerodon furcatus PAW LAW = Rhachochilus vacca = = = = Zalembius rosaceus - - - P Pomacentridae Amphiprion melanopus Amphiprion allardi 13 other genera & 18 species Labridae Achoerodus virids Bodianus mesothorax Bodianus rufus Cheilinus trilobatus Cheilio inermis Choerodon graphicus Choerodon cyanodus Clepticus parrae Coris julis Decodon puellaris Halichoeres hortulanus Halichoeres margaritaceus Hologymnosus doliatus Labroides dimidiatus Notolabrus celidotus Polylepion cruentum Pseudodax moluccanus Pseudolabrus miles Semicossyphus pulcher Suezichthys aylingi Symphodus roissali Tautoga onitis Tautogolabrus adspersus Odacidae Odax pullus Scaridae Leptoscarus vaigiensis Nicholsina denticulata Sparisoma aurofrenatum Notograptidae Notograptus guttatus* = 5 2 2 Aplodactylidae Crinodus lophodon* - - - - Centrogeniidae Centrogenys vaigiensis - - - P Leptobramidae Leptobrama muelleri tribution among the other acanthomorphs (Table 9), and, with the possible exceptions of its presence in Menidae (Mene) and Centriscidae (Macrorampho- sus), ER occurs only among the more basal or non- percomorph acanthomorphs: Polymixiiformes (Poly- 100 BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON Table 9.—Distribution of certain gill-arch characters in acanthomorph fishes. Dash (—) = absent; P = present; PA = present or absent; N = no; Y = yes; na = not applicable; 2' = LI] on Pb2 and IAC; 3“ = LI] on Pb3 (Pb2 absent); 3? = LI1 on Pb3 (Pb?2 not present); ? = character state unknown or questionable. Taxa Veliferidae | Velifer Pe ZIP sis ie|ielsleilsllelsielels|eltel_s|.silPslolslls|s|lelle|2l|s 2leltsileltsits|L_s_|LP/IN|P | = |< Metavelifer TWiwiMESsabeeeebesbbbeseeesabebeRSabRBeeeel. 2 NP |= Lamprididae I Lampris P|?| 2 | P|-| Pj -| -|-|-|-|-|-|-|-|-|-]-|-]-|-]-]-]-]-]-|-|-|-|P] PIP] -|-|-|-]-| - | P| 2|P] 2) - Polymixiidae a Polymixia P|P| 2 PS ees sel sie) ele p2elellete | eie|-2i[2I2/eilellsiisiisi2|lslsipeleis|_e|__= (LP INPP| IP) Aphredoderidae | Pat c Aphredoderus 2) AES a2 8243 ESS GsIsiSie sea ESS EIRiePebelsisl- sl =isi| =| SSE see eeeeSe SERS SESS SSDS slel os IPiniel-i- Ophidiidae | i | Dicrolene Pep) PSS DSSS PeSSSRISE SEs Se Seleleleleplalele eo intel Brotula P| -| 2 | P|P| -|-|P}-|-|-|-|-|-|-]-|-)-(-] -|-]-{-]--}-]-)-|-]-) Pe] -| Pl Pley2) = PINE PE Bythitidae if Calamopteryx P| -| 2 | - |P} -|-|P P|P/-|P| - | PIN} -| P| - Ranicipitidae | | | Raniceps De | PES See SeSSbRBEREBEMEESESSESESSSSSSMRSSiel =] Pini -]Pl - Batrachoididae Opsanus = | =|2,3| P| -| P| -| -|-|-|-|-)-|-|-|-|-|-/-) =] =] =| =) =f] =| =) ] l=) = |= |e [Pl =| eye = = IN| PIP Chaunacidae Chaunax FEZ Dele LRBVbebabekbbblbbabebbektbtlsebebLtsLrbeameEaneseer Melamphaidae Poromitra PI 2 | 12 |] Si) sisi 9P |] Sl) ell Sl s|/ sf el/e|[2ll Sl S|l[ell Syellellelelfsl SSS ele = Sl Sie SSeS Scopelogadus PP |e | Si] Slee lV ellellelSi [ESE] Pl sl SileilS [sl Slsi sel SS SE =] SSS DEST = Gibberichthyiidae f | Gibberichthys PPD fs el af sliPleslelelfsl/slsle| Sell self sie SllS] =] si SSNS STSISPISISTET = Stephanoberycidae im Stephanoberyx — PP 2 2 ell Sis lellellslfel(=]| Ss] sl SfelellS Vs] =e] =) else SPS) Ss PPPS See] Barbouristidae || | Barbourisia Piel 2 felelolelleslelelleleclclhalclellellell soleil ellsllollellelfalelcll alelclmlelels- Rondeletiidae | Rondeletia Pl 2s lolol] Plslelalelalelalslalelellelelolellelclelalslclelslclalplelclcir Cetomimidae Ditropichthys oP 2 || 2 Pl lie |e] ellel/sllelPfelfslfslslel ST sleeleil sls] SS] ele SPS SST Sey = Icosteidae | Icosteus PP Ash] IP SIP el Piisll se] sll sll=l/ sls] sllelfallel Self =lellsslfsl eles] =] Sl SPSS eS Oreosomatidae Allocyttus =|(P/24 PIPE le PlslSlsSs S DES S SN lTs SESS SESS PME lSrsisl es Parazenidae al Parazen =P) 3" > PPE ]|SsssessesPserlset sneer Zeniontidae Zenion - | 2|2,3| P| P| - DE eeeeleessehseeseeseeesResaeeelLeNrete Grammicolepidae | Xenolepidichthys P| P23 iP PPS PeeseeelsbeesleesebecsrrsessslasnNeese Caproidae L | iF [ sl Capros Pi -| 2 | P| -| P| -| -|-|-|-|-|-]-]-|-|-)-)-]-)-/2) 2/22] -]-]2) 2) Pe] =) =|] -] =] - 11 23] PN Pile Antigonia DPD 2 IDVSVS Vel SietslSisisieilell sielSiSlls]elfsilsllslSielelfeliele Pel s|Slelelel Ss [eIN P=] = Triacanthodidae | Parahollardia P/P| 2 | -|-| -|-| -/-|-|-|-]-|-|-/P/-|-|-]-|-|-|-[-/-|-]-|-|- |p - |-|-]-|-|-|-|) - | P/n/pl |= Menidae | IL | Mene Pipi 2 | P| -|-|-[-|-[-|-[-/-[-/-lel- (2 ESS SerPrrPi sesh EINelrer Trachichthyidae L Hoplostethus P/-| 2 | P|-| -|-| P| -|-|-|-|-|-|-|P)-|-|-] -|-|-/-|-}-|-]-|-[-]-] -]-|-|P/-|-|-| - | P/N} -| Pp] - Berycidae [ [ me le Beryx P{P/ 2 | P|-| -|P| P| -|-|-|-|-|-|-|P]-|-|-| =| -/-|-|-/-)-|-)-(-]-| -|-|-(Pl-/-|-| = | PIN -| Pe] - Centroberyx P22) PSS] =S)S[elelel sels! slelelelelelelslelelelelel =ellefPl=lela = [Pia ei = NUMBER 11 101 Table 9.—Continued. 5 ri 23 Taxa Ef a Se ao == Holocentridae Holocentrus NI -|P Sargocentron N\-/P Anomalopidae Anomalops NIPIP Photobelpharon N| PIP Centriscidae Macroramphosus NIP|P Gasterosteidae ial Gasterosteus N a Hypoptychidae Hypoptychus N = Aulichthys N = Aulorhynchidae I Aulorhynchus NIP| - Synbranchidae Synbranchus NI -|- Ophisternon Ni-l- Mastacembelidae | Mastacembelus . Elassomatidae Elassoma N 3 Mugilidae Agonostomus |= [N = Bedotiidae Bedotia N = Atherinidae Menidia N =| Odontesthes N = Aplocheilidae Rivulus Y z Cyprinodontidae Cyprinodon na P| - | P Adrianichthyidae Xenopoecilus Nip|-/P Oryzias NiP| -|P Belonidae Tylosurus P = Strongylura P 3 Scomberesocidae Cololabis Y = Hemiramphidae Hemiramphus YI P| - Exocoetidae Exocoetus Y = Acropomatidae Synagrops NIP| - Percichthyidae Macquaria NIP| - Leptobramidae Leptobrama N = Latidae Lates IN| [P| = Centropomidae Centropomus \N = Centrarchidae | Micropterus N{P| - Enneacanthus N ae Bathyclupeidae Bathyclupea N | Symphysanodontidae Symphysanodon NIP! - 102 BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON Table 9.—Continued. 3 Ps t 233 Taxa g +t Ses ao Aaa ene: ra ob a fo gk a AR ea = es Be 2 AEE S SSAQSOI SR BAB Be Seay o Ma Se 2M RAAnRAeoaeeeee een > ra} SS) Se) IS lS ie TS SE EEE EE FE [Ee lal [aay Epigonidae N ans Epigonus N a5 Moronidae Morone P|N/P| -| - Serranidae Epinephelus Plies =] = Anthias 2 22 eS See eicheleis|_sip2lelelis|_eltsisil=||silellsiieilelis|(s|_e|elleialelsiis|__2 |LP IN| 2/2 || = Lutjanidae | Pristipomoides PP) 2 | P| -|-[-(P]-|-|-|-)-)-(-1-1-[-|- Tere ieelelel-leTIeIeT- | IPE ES Pine ieie Hoplopagrus Ein ele See eseeeeseeeeslebeseeessnessL on. Haemulidae Pomadasys PP) 2 P= eee =) yyy e) Py) =e) ele ele] | Peele) = Sin Slee Haemulon Pips RESP EEE EE PEE EE EP EERE EEE ERE Er EEr Ea Nee Plectorhinchus PDP DY Pie Ee SESE SEREESEE3bESSSBER EES ESS Seles ereasarsreya = sine = fe Inermiidae | Inermia P{P| 2 [P| -(- (FEE EEE |---| e -[-[pl-|-|-| = Sinclair Apogonidae | Glossamia PPL? Dele PSS eweseEseSeS SEES ESS eiPelSlel_ = ie inipl = fe Cheilodipterus DPD | ID ets) al aP Sil ells] Sif sl] sl/eialall silslPsil Si) S| SSeS SSeS Ssleleeyevel SNPS Priacanthidae | [ eA AOA |\)2|)2\) POP sie] siS||Si/ Sissi slsllelSllelpa sis lellell ail SSi(S/S| SSS SSP (Sele =n el =] = Ostracoberycidae a ial | Ostracoberyx P|P| 2 | P|P| -|-| P| -| -|-| -| -|-|-|-|-|-|-| -|-|-|-|-|-|-|-|-|-|-| -|-|-|P)-|-|-| - | P/N/Pi-]- Cirthitidae | Parracirrhites Py De alesis Oy tate sell sey Sl ol/ololl elle} el cllallsl sll elfol sie S fe] ela a ells Cirrhitus PHP Dy Wish sal Pl slellsllelf sl SP olfelfelfsll elo sel sels] el] s] easels] Peele Pempheridae Pempheris P/P| 2 | P| -| -|-| P| -|-|-|-|-|-|-|P|-| -|-| -|-|-|-|-|-|-|-|-|-|-| -|-|-|P]-|-/- Parapriacanthus PP) 2 > SSE PESSESEEEE PEE EMS EE EIEsslslelele lelelalelsl= Glaucosomatidae | Glaucosoma PPD | wel sels lfellSlel el Sl Sllsle]eSlSSlfs]sisifel el elS]felfelfeel =e] Sa] sles Lactariidae L Le Lactarius P|P| 2 | P| -| -|-| P| -| -|-|-|-|-|-|P}-|-|-| -|-|-|-|-|-|-|-|-|-|-| -|-|-|P}-]-|- Lateolabracidae | Lateolabrax P(P| 2! | P| -| -|-| -|-|-]-|-|-|-[-]-]-]-]-]-]-[-|-)-)-|-)-]-]-]-] Pe] -]- [Piel Sciaenidae =||=]) Cynoscion PP Hal [elles ee SSS ES De SSeS SSS Se soll = [slelpallelle Polynemidae | Polydactylus DP 2 Ww iell sll lei] si] sl] el] ll ellalfelfs]fs}/s][ es] 12 lela] slf=sel} sl) sls] sls] =] >] =] sls] - Filimanus P| P| 2,3) P| -| -|-| P| - slelslsleey aa eel ee ale] Sel sel eel] cle) 7-5 Sillaginidae Sillago PP 2 Pe) Eye See Ste eee y =|) 2 S22] Eee) 2) 2) 2) PP] =e Pele Mullidae | Pseudupeneus P(P| 2 | P| -|-|-[Pl-)-(-(-(-|- yee) || -yTPiPrey ey Parupeneus P|P| 2 | P|-|-|-| P| -|-|-|-|-|-|-[Pl-[-|-[-|-|-]-[-[-[-|-|-|-|-|e]-|- [Pll Centrogentidae i ilinaal Ir | Centrogenys P| P/2,3| P| -|-|-[-|-|-|-[/-|-|-|Pl-|-|P] =| -|-[-[-)-]-[-]-[-[2 =] =] -) ely lee Ambassidae [ a Ambassis PHB IP alls lel Pi sl slfsl/sliis|[sl) oli>)| =] slfsi] ]]/Slelf sl sl ]fe]/ sls] sf olf sls] =]2i/ =|] s]] s Tetracentrum 1D P2332 el] slot Pel] sll oll all elf cll ell s][ S]] si] sl] 3 [[}2io]] =|) <]/ S|) s}}- SI] |] el] el] =] e]/ ||P] =] =] Caristiidae Caristius P/P| 2 | Pl-|-|-| P| -|-[-|-|-|-|-|P)-[-|-| =| -|-|-|-|-|-[-]-|-]-| =| -/-/P}-|-]- Bramidae ara Brama pS PES EPSSBEeESRBESEREEB EES EERE REET Te Em elelel = le iipl<|- ; 7 Toxotidae || | Toxotes P|P| 2 | P| -| -|-| P| -|-|-|-|-|-|-|-|-|-|-| P| -|-|-|-|-|-|-|-|-|-|-|-|-|Pl-|-|-| - | P/N/P|-/P Plesiopidae Assessor PPPs SE EPES SEES eel= (S| SESE SS SSE EleaDIEe fe = P|N/P| -|P Paraplesiops P2232 Sls shwelelSl sl] Sisley sll slelelslelslslelslelel sl sls sel ell sels] = [PNP] =) NUMBER 11 103 Table 9.—Continued. Percidae Perca P| P| 2,3] P| -|-|-| P| -|-|-|-)-|-|-|-]-|-|-|P]-|-|-|-|-1-)-]-|-|-]-|-|-/Pl-|-]-| - | PINiP] =] - N Percina DBESBSsisbatbbbateatbarbateasbbbakaktarese Cepolidae Acanthocepola P| P| 2,3 Cepola > P(23/> [ele lelelelelsl ssl sls] [el eel else elelelelelelelellelelelalelelel = |p Callanthiidae Callanthias allporti| P\P|| 2.| P| =| = |=) -|=|-)=)-=)-)-) 2) 2/2) =) 2) Pye) =) =) =) e112) ==) eye) eel ee le Gallanthasiausirdipl Deloss estebeLresEtbeSsstststabserneeei Grammatonotus (IP, 2 |e Wel ols] sffelle]ellells} cl] sl e]/ela| sa ]lfelf ele] =elelfelfel |] s]Slalfells Gerreidae sels Gerres P|P| 2 | P| -| -|-| -|- | =| -|-| -|P|-|-| -|-|-|-|-|-|- =||=|PA} -| =| P| -| =| =| 23) P| N| Pi} -|P Eucinostomus P| P| 2 PSS lele Sse REEeESEEseeseewn SBE ire mr Grammatidae | z | i Gramma PP 23i)P=) = (SSS SSS SSE SERS SSeS SS S22 S32 Searels PVreir Opistognathidae | j Lonchopisthus PP? ee lelelslelelelelelele Pls] - ees PS e= ep lae Opistognathus D123) DiS SS SSSSEEBRBBSESEREERSESEEERESERBEEPEEERSRknhe? Pseudochromidae j Labracinus P| P| 2,3| P| -| -|-| -|P\-|-|-|-|-|-|P)-|-|-|-|-|-]-|-|-|-|-|-|Pl-| -|-]-|P}-]-]- Pseudochromis P| P| 2,3| P| -| -|-| -|P|-|-|-|-|-|-|P/-|-]-|-|-|-|-|-|-|-|-|-|P{-| -|-|-|Pl-|-|- Leiognathidae aa Gazza P|P P Leiognathus Ppl | DP lElbaslSile SSS esSBeebesese Secutor P|P| 2 | P|-| P| -| -|-|-|-|-|-|-|-|-|-|-|-| -|-|-|-|-|-|-|-|-|-|P)P]-|-|-|-|-|-| - |P Polycentridae Afronandus PA P Monocirrhus - | P| 2,3 P P = i) ww uv a) a) a) ae) Z\Z ~~ = nN (Us) ine] iS) a} w ~\"o FAZANZA ~ ' ' tu Ww ine) < uw 1 ~ Ne WwW ime) < ine] 1 uw Polycentropsis PAI Polycentrus = Sphyraenidae Sphyraena PP? |_PiFSERERBESEBESEEEu_EESEEeEESEEERBRBERSEEEEREEERLSTRINELEITE Kurtidae Kurtus P|P|2,3| P|P| -|-|P}-|-|-|-|-|-|-|-|-|-]-]-|-|)-]-|-/-)-]-|-]-|-]- |-|-|P]-/-|-|. - | PNP) - |- Ammodytidae Ammodytes P/P| 2 | P| =| -|-| P| -|-)-|-)-|-|-|P)-l-]-] -]-)-]-)-}-}-]-}-]-]-]- | -)-)Pl-}-)-i) - | PNP} - | - Trachinidae Trachinus HRBSEBeRBBeBReBaRLabeaeaebabbbabbblwbanbelbweespNeBe Uranoscopidae Kathetostoma ReDReeeeeabbbbbaRLbeabbbbbabbebabatbrblebeaenase Xenocephalus = |P/2,3| P| P)- |=) -|-|-[-[-|-]-/-|-|-|-/-)-]-|-|Pl-[-|-)-)-]-]-) =| -]-[Pi-y-|-| = | PN} -| -|- Cheimarrichthyidae Cheimarrichthys ||P. P|'2)3 P=! =| =|) )=|=)=)=2) =| S/R =| =) ===) y=) tee] eel =} = ety} ie] = PIN] Py = | - Scorpaenidae Pontinus P|P| 2) P| -|-)-| Pl =| -)-)-|-1-]-|-|P)-1-) -]-]-]-)-1-]-]-te]-]-|- |] (Pl -|-|-| - | PIN/P] -| - Neomerinthe PP) 23) Piles 4 PESE EEE Ele Pleas) evel el slelelelelsllel elle] eellellelel. Se JPN see Sebastidae Sebastes P|P Platycephalidae Platycephalus P| P| 2,3) P| -|-}-|- |=) -)-]/-|-1-]-]-]-)Pl-| -]-]-]-1-1-]-]-|-}-} | - |] -/Pl-|-|-| - | PIN|P| - | - Inegocia = | P| 2,3 P |=) =| =| -}-]-]-|-|-]-|-|P}-|-|-]-]-]-]-]-]-]-|-]-1-[-]-]-]-|Pi-|-|-| - | PIN/P] - | - Champsodontidae Champsodon PP] 3° | P|/-|-|P Hexagrammidae Hexagrammos P| P| 2,3] P| -| -|- Anoplopomatidae Anoplopoma PP) 2,3)| Ps =I Rhamphocottidae Ramphocottus -|P| 2,3} P| -| -|- | |"O|"o 1 ' 104 BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON Table 9.—Continued. Cottidae Myoxocephalus D> PS Se ESSE SE SSSR SS Slelelelslsleleleletelslslplelslel 5 Tp P Cottus Nematistiidae Nematistius P| PP ell Sells Sllel site| si) SiS Si sifels!/ sil sislslfeilelslfallel(e] SiaSia SSeS PPT] = Carangidae Scomberoides DY wisi esl loll cl/ellsl/slel/al ela slfell syfslsllellelSlfslSl Sa] SS SS PISSTyel = iin Selar P| 2 | P|-|-|-|P|-|-|-|-|-|-|-|-|-|-|-|-|-|P)-|-|)-|-|-|)-]-|-| el] -|-|pl-|-|-| - |PInipl ele Rachycentridae Rachcentron | 2 |e fei S| see Sel sielSl eS Sle eelelSleleleelslel eS SSS sey = eines] = Coryphaenidae Coryphaena P| 2 | PP) -|P| -|-|-|-|-|-|-|-|-]-|-|-]-]-|-|-|-]-]-)-]-)-]-]P]-]-)Pl-/-)-] - | Pinte} -]- Echeneidae rT] Echeneis naucrates AAS PVE SlsisiSislseseaseskSesseSseeaR eee REESE SWS Pine) =~ Remora IZ SIDE SIS ISSelleiishelsielelsiisiellelisieslslisieleiellele|lsisleelei[sifei = |e Ni) =|) = Pomatomidae Pomatomus PP Well Sl aie al Sle] allel el aiisiisl| s]| > |/e|/Slfell Sy SSS (SSS SSeS Says TPMT = = Scombrolabracidae Scombrolabrax [P| 2 | P| -| -|-|P)-|-|-|-|-)-|-]P)-|-)-] -)-)-)-)-)-)-)-]-] 2-1-1 -]-]Pl-|-|-) = [eine] Scombridae | al ala Scomber = 2 Pel sSelelSlfslfsife Nemipteridae | | Nemipterus PMP eee seessesesleseresrrnrseeslaanese Centracanthidae Spicara P) 2 | P| -| -|-|P]-|-|-/-|-|-|-|-|-/-|-]P-|-|-|-|-|-|-|-[P]-| P| -|-|P]-|P]-|123]P|n|P] =| Pe Sparidae | L Acanthopagrus P| 2 | P| -| -|-| P| - aye ieye =| =) =) Pee] =] =| =) -) =) Pil] P|]. | P| Pl =|1 2 3) PNP EP Lagodon P| 2 | P| -| -|-| P| -|-|-|-|-|-|-|-|-|-|-]P]-|-|-|-|-|-|-|-|P]-| P| -|-|P]-|P]-|/123] PNP] - | P Sarpa P| 2 | P| -| -|-| P| -|-|-|-|-|-|-|-|-|-|-| P| -|-|-]-|-|-|-|-|-|-| P| -|-|Pl-|P]-/12 3] P(N P Lethrinidae | Lethrinus P| 2 | P| -| -|-| P| -|-|-|-|-|-|-|-|-|P}-| -|-|-|-|-|-|-|-|-|P]-| P| -|-|P|-| -|-|123/P Gymnocranius PL |? oj = js] is -|-|-|-|-|-]-| P| -|-|-]-]-|-|-|-|P]-| RP] -|-|P}-|-|-|123)P Monotaxis P| 2 | P| -| -|-| Pl -|-|-|-|-|-|-|-|-|-]-| Pl] -|-]-|-|-|-|-]-]-]-| P| -|-|P]-]-]-|123/P Girellidae Girella specimen A P| 2 | P|-|-|-|P/-|-|-|-|/-|-|-|-|-|-|-|P]-|-|-|-|-|-|-|-|-|-] -|-]-/P/-|-|-| - [Pin Girella specimen B P| 2 | P| -|-|-|Pl-|-|-|-|-|-|-|-|-|-|-|-|-|P]-|-|-|-|-|-|-|-] -|-|-|pi-]-|-| - |p Kuhliidae ; Kuhlia P| PW IP SiS sii sl slfsifellelfel/si] si[Si/sl sl2]] sf] el slfeif ell s][s][el/s]e]< |e] =)/P] else = |] P Terapontidae Leiopotherapon Pl DP PS Svsi SSS SSS SASS SSvSlSiel SSS SSS Wil sl slhaiSisial SP Terapon 22 ee eae Cichlidae ia) ~ Z'Z [ Zi Bae u~ 1 ' cal vu Zz a) 1 ll A Z Ge) Pie ~~ : : f : ~~ is Caquetaia P|P|2,3| P| -| -|-|P|P|-|-|-|-|-|-|-|-|P/-| -|-|-|-|-|-|-]-|-[Py-| = |-|-[Pi-|/-]-/1 23) |y|pl - |p Astronotus TEAS PESsRLebebbslaLsslettetetblekblsLletwvVelyrae Cichla P {P| 3? | P| -|-|-[ Ppl -[-|-[-[ [Py [=e |- [=e ete [| eye iPle y= eel ele 223] [yeep Cichlasoma Be 53 Sf 0 Copadichromis Pe ZS SS SIPe Pe SSIS] eee Sis2 sl SSS ES Sisleleelelsleietpelelel sl ails |p el = |e Crenicichla TZ EaRESSEEREEEEEEEEEEEBEEPESERREEEReS Eh." =e Cyrtocara P| P}2,3| P| -| -|-| P|P}-|-|-|-|-|-|-|-|P|-] -|-|-|-|-|-|-|-]-]Pl-| -|-]-|P]-] -|-|123] -}y|P| -|P Santanoperca P|P| 2 P= [=[PyPl- [=f EEE IPE PP / 2-23 ly [Pele Paratilapia RPS PESERPseesbercebesslstssessLelsebeLcletas eee Ptychochromoides | P|P| 2 | P| -| -|-|P|P|-|-/-|-|-|-|-|-|P/-| -|-/-|-|-|-|-|-|-|P|-| -|-|-/P|-|-|-| 23 | -|y/P P Ptychochromis P/P/2,3| P| -|-|-|P/Pl-|-(-|-|-(-|-(-(Pl--|el-[-| =) ||. |P|-|-/-EPlels|=|23)s)yele lr Pomacentridae | | Dischistodus P/P| 2 | P| -| -|-| P|P)-|-|-|-]-|P)-|]-|-|-] -|-]-]-]-]-]-|-]-|P]-| P| -| -|P)-] -|-|123)] -|YIeP P Abudefduf PP 2 [sels PIPE EES EP elSlslel SlslelelelelSele SPSS e923] = ple Chromis ped pepcelRie peep a ee Acanthochromis |p|P| 2 |P|-|-|-/PlPl-|-)-(- Pl eee SeSSe Pl P eelel -)=|0 23) lyiel =P Amblyglyphidodon | P|P| 2 | P| -| -|-|P/P| -|-|-|-|-|P|-|-|-|-|-|-|-|-|-|-|-|-|-[p|-|P]-|-|P]-|-|-/123]-ly|/p|-|p Amphiprion aI Ee er emai eae os meee a eee en Chrysiptera PP 2) PFS PP FSS SPS SBS SS SSS SSeS PS b SF PS S223 22 Sie NUMBER 11 105 Table 9.—Continued. 3 2 2 230 3 Taxa Sys RRR tres Gate OR Oe a Same eee eRe BHGR AAR OD 32ee a 8 on al ANANNANANAINMAMAMAMMMMMATtST a oh fae, oe ee GB Bog SHRRLLLLLL CLES LSELEEES AAaaREEE Sele 3 moo FPR RR E RE RR ERR ERE RE REE iS) Oss baw Dascyllus Pye 2 |e - P| P| -| - =| -|P]-|-|-|-| -|-|- - - P P| -| -| - Y|P P Lepidozygus P22 2 =e SPP Ellesse PSE SS TESS ESSER SEDI [S23 = ipl le Mecaenichthys P/P| 2 | P|-| -|-| P| P| -|-|-|-|-|P]-|-| -|-| -|-|-|-|-|-|-|-|-|P P| -|-/P|-|-|-|123] -|Y/P| -|P Microspathodon P|P} 2 P ell = (EPPS SSDS el Elle [IS EIEE| SSeS sles} esta swipe Plectroglyphidodon| P|P| 2 | P| -| -|-|P|P) -| -|-|-|-|/P|-|-|-|-|-|-|-|-|-|-|-|-|-/P]-|P)-|-]Pl-]-/-]1 23) 2S) yvipl ie Pomacentrus P|P| 2 | P| -| -|-| P{P|-|-|-|-|-|P}-|-|-|-| -|-|-)-|-|-|-|-|-|P]-| Py -|-|pl-|-|-|123] -|yipPl -|P Stegastes P/ P| 2 | P| -|-|-| P|P| -|-| -|-|-|P! -|-|-|-| -|-|-|-|-|-|-|-|-|P|-| P| -|-|P)-|-|-|123] -/yipl -[e Embiotocidae | CI if | il Amphistichus P/ P| 2,3) P| -| -|-|-|-|-|-|-|-|-|-]-|-|Pl-)- |= )-)-)2]-)-]-]-)-]-1 -]-|-/P]-)-/P]t 23] Ply/pl- Pp Embiotoca PP 2 PSE eS SS S/S eee eel eleS SESS] SSE Dee i2Ss Kael = |e Cymatogaster P|P) 2 | P| -| Pl -| -| -| =| -|-|-|-|-|-|-|-]-| -|-|-]-|-}-|-]-]-|/-)-] -|-|-|p]-|-[P/1 23] =[yv/p] =| P Hysterocarpus PIP) 2 | P| =| ==) =| -)-)-)-|-Je}-)-/-| Pl -) =) 2) (eee ie) el eyeye PFS StsaS Mpa. Phanerodon P}P] 2 | P}-| -|-| -|-|-|-|-|-]-|-|-|-|Pl-| =|2)-)-)=)-] 2/2) - (2) -] 21-1 -)Pl-]-1-]1 23] Plyipl -] Pp Rhacochilus i) 22 |S Sip Selle cls SS SSS SERRE RR ERERERMBEEDRESBEwWAS MPS. Zalembius PPI IPIERIE IEEE PEE EE EERE EP ES ET23siiiplslpP Labridae Achoerodus PPS Ze ells ]elfelfelP/SIElSslelellelel sil =slelelelelelels Pel cs lelelplclel eo 2al- plete Bodianus P| P| 3? |paAl-| -[-|P|-|-|-[-]-[- |=] Pee SE EI I=] -|P] =| = |-|-|Pl-|-|-|123] - lylpl - |p Cheilinus P/P) 3° | P| =| -|-) P| -|-|-)-|-|-]-|-)-[Pl-) =] -)-)=)-1- =] ee iPlPi ee) S a2 lyiPl/- |e Cheilio PP Sele PS) El-|-|-[P)=| eee IP) =e alee e232 lyPlaie Choerodon PP) 3" | = |=) -|-| Pl Pl -|-)-[-[-)-)-|-| Pl) (e fees E Piel- [-|-|Pl-/-|-|123]- ly] - [Pe Clepticus PPB PSE RIPeEEessreSssEsSessrrsEPsseelimsislyvirel Coris Pls 2s -|elh> | Ses SESE REE esse ERES EERE T3 2 Mpls Decodon HERMES SeSsrettsletreSeteeslslerctSeELnecteesxrrere Halichoeres hort. |P|P| 3° | P|-| -|-|P|P|-|-|-|-|-|-|-|-[Pl-|-|-[-|-[-[=/-[2|=]p)=) = |S Pl] =| -]1 23] -ly[pl -[P Halichoeres marga| P| P| 3° | = |-| -|-|P/P|-|-|-|-|-|P|-|-|P/-| -|-|-|-|-|-|-|-|-|pP]-] -|-]-|p]-|-|-]1 23] -|y[Pl-|P Hologymnosus Eee ewleeeceEesecsecsslteesersecscLceaav ner Labroides P| P) 3? | P| =| -|-|-|-|P)-|-|-|-|P|-|-|P}-| -|-|-|-|-|-|-|-]-| ppl - [Pl -|=)-)-]-]1 23] =[yiel-|P Notolabrus DPE DAS sPPESEERESERBRSEESSEESeseReeSisepeselesreip-p Polylepion Dee oll = Slee elas elelslelsee sels Dele eleblelelelaalelmpreits Pseudodax PP 2 el SEPSIS SSISESEE MESSE es SSR RE SBESEESEr 23 akMp slp Pseudolabrus DRBeSSsBeaRebebbbatbLbabessbbtbersbesrtbbwssvrsibt Semicossyphus DDL = Sells RPessesESsDer Eee sEseE Reese ReSeiZseMecir Suezichthys P/P/ 3° | = |=) =| =| P| Pl -|-/-|-)-|-|-|-|Pl-|-[-|-|-]-|-|-|-|-]P]-] -|-|-|Py-|-|-]123] - |y[pl- |p Symphodus PP) 3? | = |-|-|-|-|-|-|Pl-]-)-|-[-[-[pl-|-]-]-]-]-/2]-)- =|] =) = [py =| -|py -|-|1 23] - [yp] - | Tautoga PPS = El SE PPEEEEEEEbEREREEESSEEERE RE REREeEEtes spe Tautogolabrus PP? P=) =|=|=)=)-/ 1) Pel ESS ESP SE IrPiele 2023) = |yirle [Pe Xiphocheilus P/P| 3? | - |-|-|-[P]Pl-|-[-|-|-[-[-|-]ey-[- |=] -]- [| 7- fe Py |---|] -[- [123] - [vel - [P Odacidae Odax pullus P| 3? | P| =| -|-[Pl-|-|-|-|-|-|-|-]-|Pl-]-|-[-]-/-[-/-/Se [ele Pl] [23] [yp] - | Scaridae | | il Leptoscarus PBR PPrESeeeSEsreEsSsessseerele]PesessetZsikivireir Nicholsina PL |= | SPEiPESSSSERSREE SeeSeeee Pee RPeessetsEMPeb Sparisoma P| = ls ERPS SSEEEERESESEEEERSEBE PEE EREi23 SMple lp Pholidichthyidae ali Pholidichthys PS = le frye) PPP fel = Pll lela l tell ef l= [elf] [e123] ve) |e Luvaridae | | Luvarus 22 Sle PES EaSseeeeesseebeeeeeeeebeseeel 2Neee Ephippidae [ | | Chaetodipterus P) 2 PP EEE EEE EEEEEE EEE PIP EEE EEE, = |Pinp|-|p Zanclidae Zanclus Pi 2 | P| -[Pl-[-(--[-1-/-]-]-(-7-- Ie -PlEE EEE Pinel = Acanthuridae | | Acanthurus P| 2 | P| -|-|P| -|=-|-|-|-|-|-|-|-|-|-|-| -|-]-|-|-|-|-|-|-]-|P} -|-|-|-]-]-]-| - | P|N/P] -| - Nandidae | Nandus P| 2 | P| -|-|-|-|-[-|-|-[-]-[-|-[el-|-]-|-[-[-|-1-]-/-IE [=|] Ee P|] - [23] N[pl- |e Badidae Badis P| 2 | P|-| -|-|-]-]-|-]-|-]-| |-|-|Pl-] -]-)-]-[-]-]-]-]-]-]-] -]-]-|P]-]-]-/123) -|N/P] -| Pe Pristolepidae | Pristolepis HANS Se SVS SSS Si SS SSS SSS SS SS SSS SSeS Spy SSS) Si Sine) sys 106 Table 9.—Continued. Channidae BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON Channa asiatica C. harcourtbutleri Z|Z| |TV4 on Cb5 Anabantidae Jl Ctenopoma Sandelia Amarsipidae Amarsipus Centrolophidae Psenopsis Bathymasteridae Bathymaster Zaproridae Zaprora Stichaeidae Ulvaria Bovichtidae Bovictus Pseudaphritidae Pseudaphritis Dactylopteridae Dactyloptena Malacanthidae Caulolatilus Malacanthus Draconettidae Draconetta Callionymidae Callionymus Gobiesocidae Trachelochismus Tripterygiidae Ruanoho Lepidoblennius Blenniidae Parablennius gatto. Parablennius tasm. Scartella | |'U Scartichthys Istiblennius Dactyloscopidae Dactylagnus ae) < ce) Clinidae Gibbonsia ~~ n 1 Heterostichus las) Heteroclinus ~~ Springeratus ~~ Ophiclinus Clinus u|'U 1 ' Labrisomidae Calliclinus i Labrisomus u|'U Chaenopsidae Neoclinus Rhyacichthyidae Rhyacichthys Odontobutidae Odontobutis Micropercops Percottus ' | |'0 Xenisthmidae Xenisthmus NUMBER II Table 9.—Continued. | ILI on Pb(s) Eleotridae 107 Eleotris Ophiocara Microdesmidae Ptereleotris Nemateleotris Microdesmus Je | | | | | Gobiidae Glossogobius P|, P| 2 | 2 | Bollmannia P|P D |e Padogobius P| P| 23 | P|- Pseudapocryptes P| P| 2 [ Pp Le Gnatholepis PlPl23]| P|- Trypauchen P [P/23[ P | - Psettodidae | | Psettodes mixiidae); Paracanthopterygii: Ophidiiformes (Ophi- diidae, Bythitidae), Gadiformes (Ranicipitidae), Batrachoidiformes (Batrachoididae); Stephanoberyci- formes (Stephanoberycidae, Gibberichthyidae, Ron- deletiidae, Barbourisidae, Melamphaidae); Zeiformes (Oreosomatidae); Beryciformes (Berycidae, Anom- alopidae, Holocentridae, Trachichthyidae). Among these groups, /costeus, superficially, re- sembles the stephanoberyciform and paracanthopter- ygian families (but not Stephanoberycidae), with which it variously shares to the exclusion of the other groups (data augmented from the literature): flexible body, a high number of precaudal (20—23 in /costeus) and total vertebrae (67—72 in Jcosteus), a high num- ber of dorsal-fin rays (52—58 in /costeus), attaining a length of over a meter (ca. 2 m in /costeus). It also shares with these families and some cetomimid ste- phanoberyciforms, imbedded lateral-line scales and weakly ossified bones. It further resembles the ce- tomimids in lacking pelvic fins (at least as an adult), having a highly distensible stomach (also present in lophiiforms), and dorsal and anal fins consisting only of segmented rays and placed far posteriorly. It dif- fers, at least, from all other basal acanthomorphs in having only five, as opposed to six, hypurals, and, except for some ophidiiforms and gadiforms, in hav- ing an interarcual cartilage. The many similarities between Jcosteus and some Stephanoberyciformes, particularly Barbourisidae, mainly favors close relationship with that group; however, the presence of IAC tends to favor the Par- acanthopterygii. We assign Icosteidae to its own or- der, Icosteiformes, which we tentatively place near the Stephanoberyciformes, but, in any event, believe [123] P|N|P| that it occupies a pre-percomorph position among the acanthomorphs. Zeiformes and Possible Relatives Nelson (1994:290—291) discussed the vacillating assignments of the inter-relationships of the Capro- idae through 1993: whether they are zeiforms, per- ciforms, a sister-group of the zeoids and tetraodon- tiforms, or “in some position between the Stephan- oberyciformes and Beryciformes.” Nelson rejected Johnson and Patterson’s (1993) treatment of caproids as perciforms and retained them as the sister group of the zeoids (Nelson, 1994:253). Tyler et al. (2003), conducted a phylogenetic study of the zeiforms that also included a primitive tetrao- dontiform (Parahollardia), both genera of caproids, and seven other diverse outgroups. Relationship of zeiforms with tetraodontiforms and caproids was sup- ported in three of their four analyses (data ordered with and without most meristic characters and data unordered with most meristic characters). The rela- tionship was not supported in the fourth analysis (data unordered without most meristic characters), which they considered, “‘the most rational and best justified.” We elect to position the caproids together with the zeoids and tetraodontiforms in accordance with three of the four cladistic analyses of Tyler et al. (2003). The presence of ER in an acanthomorph is usually an indication of a pre-percomorph condition. Only Mene, the centriscid Macroramphosus, and the enig- matic /costeus (Icosteidae) among those taxa usually included in the percomorphs have it, and we present 108 evidence that Jcosteus (q.v.) is more appropriately placed near the stephanoberyciforms. We also believe that Mene, which is generally unusual morphologi- cally, is probably more appropriately placed among the pre-percomorphs, and have arbitrarily inserted it near the zeiforms. Although we believe that the cen- triscids and relatives are also probably more closely related to pre-percomorphs than to percomorphs, we place them with the other groups Johnson and Pat- terson (1993) included in their Smegmamorpha to fa- cilitate discussion of that group. Zeiformes OREOSOMATIDAE Allocyttus verrucosus (Gilchrist), USNM 329365, 82.8 mm. Plate 76 Description. Remarks. The musculature appears to be skewed atypically, but the muscles and their attachments oth- erwise appear normal. The muscle fibers are gener- ally stringy and weak, particularly the posterior mus- cles, and are infiltrated with filmy tissue that was impossible to remove without causing considerable damage to the muscles. An anomalous muscle strap on the right side attaches to Ebl mid-posteriorly, passes deeply ventrally, and attaches to Eb2 antero- medialmost edge. LE1 very broadly on Eb1 dorsoposteriorly. CT covering levators laterally indents anteriorly and cov- ers much of Eb1 posterior surface, becoming contin- uous with tendinous dorsal portion of LElventral to origin. LE2 on Eb2 dorsoposteriorly, beginning just pos- terolateral to lateral end of TEb2. LE3 absent. LE4 dorsally on bony Eb4 process posteroventral to uncinate process, muscle joined posterolaterally by LP. LP very fine, continuous with LE4 ventrolaterally slightly dorsal to combined insertion on Eb4. LI1 anterior surface attaches to Pb2 posteriorly in notch between two dorsally extending cartilage- tipped processes, continues ventrally and inserts on Pb3 dorsally ventrolateral to dorsally extending car- tilage-tipped process. LI2 on Pb3 dorsolaterally ventrolateral to medial end of Eb3. TD comprises TPb2, TEb2, TEb3, and TPb3. TPb2 very thick pad with mid-anterior U-shaped notch ex- tending more than half distance across pad and shal- low mid-posterior notch; notches joined by mid-lon- gitudinal raphe, which gives rise to CT sheets dor- sally; raphe attaches ventrally to CT of pharyngeal roof; muscle attaches anteriorly to cartilage-tipped BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON Pb3 dorsal and Pb2 dorsomedial processes, fuses ventrally with TEb2, and is continuous posteroven- trally by muscle strand with TEb3. TEb2 attaches on Eb2 dorsally just anteromedial to LE2 insertion. TEb3 attaches to posterior margin of Eb3 medial to uncinate process. A few, probably anomalous, muscle fibers branch off right-side TEb3 and insert on Eb4 dorsally. TPb3 is a very thin ribbon of muscle (not illustrated) that extends ventroanteriorly from TEb3, where TEb3 passes dorsal to dorsoposteromedial edge of Pb3. The ribbon extends deeply ventral to OD and attaches to Pb3 dorsally in a deep depression well posterior to the anterior end of Pb3. OD3, OD3’ originate inseparably on Pb3 dorsally ventral to TEb2. OD3 inserts on Eb3 uncinate pro- cess dorsoanteriorly. OD3’ branches off ventrally well lateral to origin and inserts on Eb3 dorsally an- teroventral to uncinate process. OD4 absent. OP attaching dorsally on Eb4 posteriorly begin- ning medially ventral to uncinate process and ex- tending well laterally, meeting Ad4 dorsally. A few medialmost fibers (not visible in illustration) extend ventrally to Cb5 anterior to Ad5; most fibers attach ventrally on Cb4 joining ER with Ad5, medial to distal end of Cb4, and posterior to Ad4 (ER not in view in Plate 76B). Ad1-—3 absent. Ad4 dorsally on Eb4 mostly anterior to OP later- ally, extending to near cartilaginous end of Eb4, ven- trally on Cb4 dorsally beginning short distance me- dial to Eb4-Cb5 joint and extending to joint. Ad5 dorsally on Cb4 well medial to distal end, ventrally on distal portion of Cb5 dorsally. SOD absent. RDs possibly anomalous, each divided longitudi- nally into two straps; right-side medial strap passes dorsal to left-side strap and lies appressed to right- side lateral strap; left-side medial strap passes me- dially ventral to right-side medial strap and lies ap- pressed to medial surface of right-side lateral strap. Additional remarks. SCL attached mid-dorsally to cartilaginous posterior tip of Bb3. TV4 free from Cb5s. Pb4 and UP4 absent. Pb2 toothed. IAC absent. Eb] uncinate process absent. Eb4 levator process ab- sent. PARAZENIDAE Parazen pacificus Kamohara, USNM 364277, 73.0 mm; USNM 187807, 93.9 mm. Not illustrated ZENIONTIDAE @ = Zenion hololepis (Goode and Bean), USNM 187864, 68.5 mm. Not illustrated NUMBER 11 Description. LE! on Eb! mid-dorsally (uncinate process ab- sent), attached anteriorly for entire muscular length to perpendicular section of pharyngeal roof CT, to which Pb1 also attaches for entire length. LE2 on Eb2 mid-dorsally. LE3 absent on one side, much reduced, but present on other. @ Absent on both sides. LE4 broad based, on Eb4 mid-dorsoposteriorly. LP very fine, on Eb4 at ventrolateral edge of LE4 insertion. @ Specimen damaged, unable to determine if LP was present. LI1 on Pb3 anteriorly near base of dorsoanterior- most process; about twice size of LI2. @ Almost en- tirely on Pb3, but very fine CT attachment to Pb2 present. LI2 on Pb3 dorsolaterally anterolateral to medial end of Eb3. TD comprises TPb2, TEb3, and TPb3. TPb2 a thick pad with broad, deep V-shaped notch (open an- teriorly) leading to short mid-longitudinal raphe, which attaches dorsally to CT sheet covering muscles and ventrally to pharyngeal roof CT; muscle attaches to Pb2 and Pb3 dorsalmost surfaces posteriorly with tendinous continuation to medial edge of Pb1. TPb3 a strap of muscle extending anterolaterally ventral to OD3-—4 and attaching to Pb3 along medial edge of LI2; muscle continuous mid-posteriorly with TEb3. TEb3 attaches broadly to Eb3 posterior edge, meeting OD3’ posteriorly. ® Comprises TPb2, TEb2, and TEb3. TPb2 notched anteriorly and posteriorly, fused ventrally with TEb2, which attaches on Eb2 dorsoan- teriorly ventral to LE2 insertion. OD comprises OD3—4 and OD3’, muscles origi- nate on Pb3 dorsally, OD3’ separating ventrally just lateral to origin; OD3-—4 inserts on tip of Eb3 ante- riorly, passing dorsal to tip and inserting on medial edge of Eb4; OD3’ inserts on Eb3 dorsally ventral to OD3—4. OP dorsally broadly on Eb4 posteriorly beginning near medial end and extending laterally to below lat- eral edge of LE4 insertion; ventrally, narrowly on Cb5 beginning dorsally on most proximal surface of long, rod-like cartilaginous distal end and extending short distance medially, meeting Ad5 medially and TV5 laterally. @ Possibly absent, but if present, at- taches ventrally to Cb4 (see Ad4). M. SO-Pb3 pair of SO longitudinal muscle straps, one on each side of each RD, extending well ante- riorly and inserting on medial edge of Pb3 ventral to OD3—4. @ Absent. Ad1-—3 absent. Ad4 dorsally on Eb4 posteriorly, beginning me- dially at lateral edge of OP and extending laterally to Eb4-Cb4 joint, ventrally broadly on Cb4 medial to joint with Eb4. @ Two layered; anterior layer typical; posterior layer, separates dorsally from anterior layer 109 and fuses anteroventrally with anterior layer; poste- rior layer possibly represents modified OP. Ad5 triangular, apex dorsally even though muscle attaches broadly dorsally on posterior bony surface of Cb4 beginning slightly medial to cartilaginous dis- tal end, attaches ventrally along long cartilaginous distal end of Cb5 and extends onto bony surface short distance anterior to OP. @ CbS cartilaginous distal end normal (short rounded cap), hence, AdS much less extensive ventrally. SOD absent. RDs separated by space more than 1.5X diameter of one RD. Additional remarks. SCL attached mid-dorsally to cartilaginous ventroposterior end of Bb3. TV4 free from Cb5s. Pb4 and UP4 absent.Pb2 toothed. Eb4 levator process absent. GRAMMICOLEPIDAE Xenolepidichthys dalgleishi Gilchrist, USNM 341954, 102 mm; USNM 320015, 100 mm. Plate 77 Description. LEI extremely broadly on Eb1 dorsolaterally (un- cinate process absent). LE2 on dorsally expanded posterior margin of Eb2. LE3 on Eb3 lateral to tip of uncinate process. LE4 on Eb4 levator process dorsoanteriorly. LP tendinously joining LE4 slightly dorsal to car- tilage tip of Eb4 levator process. LI1 tendinously on Pb2, tendon continuing onto Pb3. LI2 on Pb3 laterally ventral to LI1. TD comprises TPb2, TEb2, and TEb3. TPb2 a bi- lateral pair of thick, rope-like muscles, separated by shallow notch mid-anteriorly and joined along their medial edges to tough, deeply depressed CT sheet (conforms dorsally with parasphenoid keel); CT sheet is tightly applied posteroventrally to TEb2 dorsally. TPb2 attaches tightly anterolaterally to cartilaginous dorsal tip of Pb2 and weakly to adjacent cartilaginous dorsal tip of Pb3. TEb2 divided medianly by narrow raphe (obscured from view in illustration); TEb2 lat- erally on Eb2 dorsally anterior to LE2 insertion. TPb2 and TEb2 not continuous posteriorly with TEb3. TEb3 on almost entire posterior edge of Eb3 medial to uncinate process, also attached mid-ven- trally by CT to CT of pharyngeal roof. OD3, OD3’ originate together on Pb3 dorsally ventral to TEb2 and divide immediately after exiting from under TEb2, with a slender dorsal branch (OD3) attaching narrowly to Eb3 uncinate process anteriorly and a slightly larger ventral branch (OD3’) attaching on Eb3 dorsally ventral to uncinate process. OD4 absent. 110 OP on Eb4 dorsoposteriorly beginning ventral to uncinate process and extending medially almost to medial end, laterally overlapping Ad4 medially; ven- trally on Cb4 dorsally beginning well medial to distal end and extending a short distance medially, overlap- ping Ad4 medially and difficult to separate from Ad4. ER, to which OP usually attaches along with Ad5 when OP attaches to Cb4, apparently absent. Ad1-—3 absent. Remarks. As SO longitudinal fibers extend anterior to the gill arches, they spread laterally and anteriorly and form a complex mesh over the roof of the oral chamber. From this mesh, a sheet of fibers (not illus- trated) extends ventrolaterally toward each of the first three arches and either attaches directly or through connective tissue to the ventral edge of the epibran- chial, beginning about mid-laterally, and continuing laterally and ventrally and attaching extensively to the ceratobranchial dorsal surface medial to the inner angle of the Eb-Cb joint. The appearance in each case is superficially like an Ad, but acanthomorph Ads1— 3 are on the external surface of the associated Eb-Cb pair. Ad4 attaches dorsally along much of ventral edge of Eb4 beginning medially anterior to OP and ex- tending laterally almost to distal end of Eb4; ven- trally on much of dorsal edge of Cb4 anterior to OP. Ad5 reduced or absent. On one side of each spec- imen, a short muscle (almost vestigial in smaller specimen) joins Cb5 dorsodistally to Cb4 well ven- tral to its distal end. SOD absent. RDs adjacent. Additional remarks. SCL absent. TV4 free from Cb5s. Pb4 and UP4 absent. Pb2 toothed IAC absent. Ebl uncinate process absent. Eb4 levator process present. Caproiformes CAPROIDAE Capros aper (Linnaeus), USNM 289207, 2 speci- mens, 53.2—62.5 mm; USNM 327294, 64.8 mm. Plate 78 Description. LE1 on dorsolateralmost bony edge of Eb1, well lateral to long uncinate process, which is attached dorsoposteriorly by fine ligament to Pb2 anterolat- erally (no IAC). Thick, tendinous edge of thin CT sheet (not illustrated) covering LEs laterally, attaches along entire posterolateral margin of LE1. LE2 on dorsolateralmost bony edge of Eb2, ven- troanteriorly meeting Ad2 dorsally and ventromedi- ally meeting TEb2 laterally. LE3 on Eb3 uncinate process anteriorly, ventro- medial edge meeting OD3 laterally. BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON LE4 on Eb4 dorsolateralmost bony edge well lat- eral to uncinate process, joining raphe posteroven- trally with Ad4 dorsally (unclearly or only partially joining raphe with Ad4 in smaller specimen). LP absent. LI1 in depression in cartilaginous dorsalmost sur- face of Pb2 just posterior to anterior end. LI2 on Pb3 dorsoanterolaterally just ventral to overlying anteromedial edge of Eb3, insertion ante- roventrally proximate to bony posterior surface of Pb2. TD comprises TEb2 and TEb3-Eb4. TEb2 with mid-longitudinal raphe, central dorsal area with tough CT pad (not illustrated) with raised convex lat- eral margins, which attach to expanded dorsoanterior cartilaginous caps of Pb2s; pad attached mid-ven- trally to mid-longitudinal raphe; muscle laterally cov- ers much of dorsal surface of Eb2 and attaches dor- sally at ventromedial edge of LE2 and ventrally to dorsomedial edge of Ad2, variously joining raphes with each of these two muscles. TEb3-Eb4 well sep- arated posteriorly from TEb2, attaches broadly on posterolateral edge and surface of Eb3 (ventral to OD3 and OD3’), and posteromedial edge of Eb4; muscle slightly overlaps and is continuous ventrally by fine muscle strands with SOD. OD3, OD3’ origin on Pb2 posteromedially ventral to TEb2, continuing posteriorly broadly on Pb3 dor- somedially. OD3 inserting broadly on dorsoanterior edge of Eb3 uncinate process (there meeting ventro- medial edge of LE3). OD3’ separating from OD3 well ventrolateral to origin and inserting on dorsoan- terior surface of Eb3, meeting dorsomedial edge of Ad3. OD4 absent. OP on Eb4 dorsoposteriorly beginning medially near SOD and extending laterally to near uncinate process, dorsolaterally meeting TEb3-Eb4 and dor- somedial end of Ad4, ventrally on Cb5 posteriorly just medial to distal end. M. SO-Pb3 (not illustrated), band of longitudinal SO fibers on each side extends anteriorly ventral to TD and inserts on Pb3 medially. Ad1 broadly on anteroventral edge of Eb1 and dor- soanteromedial surface of Cb1. Ad2 on dorsoanterolateralmost surface of Eb2, meeting LE2 ventrally and TEb2 laterally, and on dorsoanterior surface of Cb2. Ad3 on Eb3 anterolaterally ventral to uncinate pro- cess, meeting OD3’ dorsomedially and on Cb3 dor- soanteriorly. Ad4 on Eb4 dorsoposteriorly beginning at lateral end of OP near uncinate process and extending lat- erally to end of Eb4; ventrally broadly on Cb4 lat- erally medial to Eb4-Cb4 joint, there meeting Ad5 dorsoanterolaterally. Ad5 short, dorsally on Cb4 dorsodistally and ven- NUMBER 11 trally on Cb5 dorsodistally, meeting OP ventrolater- ally. SOD present. RDs separating slightly as they pass below SOD. Additional remarks. SCL attached mid-dorsally to elongate cartilaginous ventroposterior tip of Bb3. TV4 free from Cb5s. IAC absent. Pb4 absent, UP4 present. Pb2 toothed. Eb4 levator process absent. Antigonia rubescens (Gitinther)?, USNM 365941, 76.7 mm. Not illustrated Additional material: @ = Antigonia capros Lowe USNM 163521, 78.1 mm; ® = A. combatia Berry and Rathjen, USNM 188045, 91.8 mm. Description. LE1 finely, tendinously on dorsoposterior edge of Eb1 just lateral to uncinate process. LE2 finely, tendinously on dorsoposterior edge of Eb2 much nearer lateral than medial end. LE3 finely, tendinously on dorsal tip of Eb3 un- cinate process. LE4 narrowly on dorsodistal end of Eb4, joined slightly dorsal to insertion by LP. LP very fine, joined by long slender tendon to ven- trolateral edge of LE4 dorsal to joint insertion on Eb4 dorsodistally. LI1 finely, tendinously on Pb2 posteroventrally. LI2 tendinously on Pb3 ventrolaterally just medial to cartilaginous edge articulating with medial end of Eb3. TD comprises TEb2 and TPb3-Eb3-Eb4. Very thick CT pad mid-dorsally extends anteriorly and at- taches to Pb2s dorsoanteriorly (all but obscuring their dorsal cartilaginous ends); few small ACs irregularly distributed dorsally on pad (all three species); pad attaches mid-ventroanteriorly to CT of pharyngeal roof. TEb2 continuous from one side to the other only around CT pad posteriorly; muscle otherwise attaching medially to lateral surface of CT pad and extending laterally to, or almost to, dorsodistal end of bony surface of Eb2 (well lateral to LE2 insertion). TPb3-Eb3-Eb4 transversely continuous, free from TEb2; on Pb3 dorsal surface posterolaterally ventral to OD3-OD3’, continuing onto Eb3 dorsomedially and well along posterior edges of both Eb3 and Eb4; free from SOD. @ ® TEb2 attaches to CT pad ven- trolaterally so that dorsolateral edge of pad overlaps TEb2. OD3, OD3’ origin on Pb3 dorsoposteriorly, ex- tending laterally and branching ventrally (OD3’) at about level of medial end of Eb3; OD3’ inserting on Eb3 dorsally ventrolateral to uncinate process; OD3 inserting on Eb3 uncinate process anteromedially. (Note: Eb3 and Eb4 uncinate processes closely bound 111 together. Superficially, OD3 also appears to insert on dorsomedialmost edge of Eb4 uncinate process, but processes can be readily separated and OD3 clearly inserts only on Eb3.) OD4 absent. OP dorsally on Eb4 posteriorly beginning near me- dial end and extending laterally to below uncinate process, there meeting TPb3-Eb3-Eb4 ventrolateral- ly; ventrally on Cb5 posterolaterally, just meeting Ad5 ventromedially. M. SO-Pb3, short band of longitudinal SO fibers on each side extends anteriorly ventral to TD and inserts on Pb3 medially. AD1-3 absent, reduced muscle strands (GFM 1-3) on anterolateral surfaces of respective Eb and Cb. Ad4 dorsally on Eb4 posteriorly, beginning me- dially at lateral edge of OP and extending laterally to distal end of Eb4; ventrally on Cb4 dorsally extend- ing laterally to inner angle of Eb4-Cb4 joint and an- terior to Ad5 attachment. Ad5 relatively small, dorsally on cartilaginous dis- tal end of Cb4 posteriorly, beginning at Eb4-Cb4 joint and extending ventrally to posterolateral surface of cartilaginous distal end of CbS5. SOD variously slender to moderately broad. RDs adjacent for much of length, narrowing, sep- arating, and becoming tendinous anteriorly and in- serting on Pb3 posteriorly. Additional remarks. SCL attached mid-posteriorly to greatly elongated cartilaginous ventroposterior end of Bb3. TV4 free from Cb5s. IAC reduced, ball-like suspended in ligamentous tissue (Tyler et al., 2003: table 1, indicate that IAC is absent in A. capros, but we found it in all three species of Antigonia we ex- amined). Pb4 absent. UP4 present. Pb2 toothed. Eb4 levator process absent. Tetraodontiformes TRIACANTHODIDAE Parahollardia lineata (Longley), USNM 287252, 119 mm. Plate 79 Description. LEI! narrowly, tendinously on Eb! dorsopostero- laterally. LE2 narrowly, tendinously on Eb2 dorsopostero- laterally. LE3 narrowly, on tip of Eb3 uncinate process. LE4 musculously and tendinously on Eb4 dorso- laterally. LP very fine, tendinously joining LE4 posteroven- trally just dorsal to LE4 insertion; origin also tendi- nous. Remarks. LP in tetraodontiforms (and many other fishes) is much reduced, fragile, obscured by various 112 tissues, and easily destroyed when peeling away sur- rounding tissues. Because of this, and because he was unaware of the existence of LP in fishes at that time, Winterbottom (1974a; in litt. 19 Mar 2001), in his study of tetraodontiform musculature, did not report the presence of LP in any tetraodontiform. In addi- tion to Parahollardia, we found LP in Hollardia hol- lardi Poey (USNM 289328, Triacanthodidae) and Trixiphichthys weberi (Chaudhuri) (USNM 280329, Triacanthidae). Additionally, in Balistes vetula Lin- naeus (USNM 349662, Balistidae), we found a long, slender muscle extending dorsoposteriorly from a ra- phe with LE4 somewhat dorsal to the LE4 insertion (origin of muscle not determined). This muscle may represent a modified LP, which usually joins LE4 near the LE4 insertion. Investigation of LP in tetrao- dontiforms may provide information bearing on the internal classification of the order. LI1 narrowly tendinously on dorsomedialmost tip of Pb2. LI2 by long, slender tendon on dorsal surface of cartilaginous Pb3 process that is joined laterally by medial end of Eb3. TD complex, comprising TPb2, TEb2, and TPb3- Eb3. TPb2 a pair of laterally curving, vertically raised muscles, lying dorsal to and mostly free from broad CT mid-section of TEb2; anteriorly, each TPb2 muscle joins CT of pharyngeal roof, which gives rise dorsally to thick CT shell that lies over TPb2s (and TEb2 portions underlying TPb2s) and forms cup at- taching to ventral process on skull; posteroventrally, each TPb2 joins CT, which attaches to TEb2; CT also joined by (forms tendinous mid-anterior portion of) TEb2 (mid-posteriorly, CT of TEb2 narrows consid- erably); muscle does not attach to Pb2. TEb2 mas- sive, attaching to entire bony dorsal surface of Eb2; anomalous dorsoanteromedial branch of right-side TEb?2 attaches finely, tendinously to anterior end of right-side TPb2. TPb3-Eb3 on Pb3 dorsally medial to medial end of Eb2 (Eb3 portion with mid-longi- tudinal raphe joining TPb2 and TEb2 mid-ventrally), continuing onto Eb3 dorsolaterally ventral to OD3 and meeting OD3 insertion on uncinate process, con- tinuous ventrally by diagonal strand of muscle with SOD. Remarks. Because of the large thick central CT section of TEb2, we treated Pb3 as dorsally mostly not covered by muscle. This area of the Pb3s forms a diarthrosis with a knob-like process on the ventral surface of he skull, which is very similar to that of pomacentrids, but not like that of labrids, cichlids, or embiotocids. The process has an irregular surface in Parahollardia and pomacentrids. In the other three groups, the process is divided into a bilateral pair of broad, smooth surfaces. Stiassny and Jensen (1987: 282) characterized all the “‘labroids” as having a sim- ilar knob-like process, but we disagree. BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON OD3 origin on Pb3 dorsally ventral to TEb2, in- sertion massively on anterior surface of Eb3 uncinate process. OD4 absent. OP dorsally broadly on Eb4 posteriorly beginning on or near uncinate process and extending medially almost to end of Eb4; ventrally broadly on Cb5 pos- teriorly, attachment tendinous laterally and joined on one side by Ad5 posterolaterally, but not on other. GFM 1-3 large fan of fine filamentous muscle fi- bers attach on anterior surfaces of joined distal ends of respective Eb and Cb. Ad4 broadly dorsally on Eb4 posteriorly beginning at or near uncinate process and extending laterally almost to end of Eb4; ventrally more restricted than dorsally, on Cb4 dorsally medial to Eb4-Cb4 joint. Ad5 dorsally on distal end of Cb4 posteriorly, ven- trally on Cb5 dorsodistally. SOD present, slender, entirely ventral to TPb3- Eb3. RDs slightly separated. Additional remarks. SCL free from Bb3. TV4 free from Cb5s. Small, spherical IAC present. Pb4 absent, UP4 present. Pb2 toothed. Eb4 levator process ab- sent. Winterbottom (1974a) described and illustrated the branchial arch musculature of several tetraodonti- forms. Meniformes MENIDAE Mene maculata (Bloch and Schneider), USNM 347107, 97.7 mm, USNM 102498, 87.8 mm. Plate 80 Description. LE1 on broad anterior surface of Eb1 lateral to tip of uncinate process. LE2 on mid-dorsoposterior edge of Eb2. LE3 tendinously on tip of Eb3 uncinate process, interrupted dorsally by CT and becoming musculus again before attaching to skull. LE4 fan-like, joins LP ventrally on common ten- don, which inserts on Eb4 lateral to uncinate process. LE4 continuous dorsally with CT from which various thin muscle straps continue dorsally and attach to skull. LP continuous ventrally on common tendon with LE4, dorsally joining extensive CT sheet, which dor- sally becomes musculus; dorsal extent of musculus LP not clearly separated from PP; CT sheet continues ventrally and attaches along outer curvature of 4th and 5th arches. LI1 on Pb2 tendinously just posterior to joint with IAC. LI2 on Pb3 dorsoposterolaterally. NUMBER 11 TD comprises TEb2 and TPb3-Eb3. TEb2 very short transversely, very long longitudinally, in two continuous sections: anterior section much the broad- er, with mid-longitudinal raphe giving rise to CT sheets that attach to cranium; anteroventrally muscle attaches by CT to anteriormost ends of Pb2s, mid- ventrolaterally attaches to [AC2 dorsoposteriorly, and posterolaterally attaches to dorsoanteromedial surface of Eb2 anterior to medial edge of LE2 insertion; pos- terior section attaches to posteromedial edge of Eb2. TPb3-Eb3 begins anteriorly as diagonal strap of mus- cle on each side, which attaches to Pb3 dorsally ven- tral to posterior margin of TEb2; straps cross poste- riorly just as they mesh with anterior end of broad Eb3 portion of muscle, which attaches to medial edg- es of Eb3s. Remarks. The diagonal muscle straps appear to de- rive from crossing muscle strands present in many acanthomorph TDs, and which connect the posterior end of one TD muscle with the anterior end of an- other TD muscle or SOD. The straps in Mene are so distinctive, that they possibly are not homologous with TPb3-Eb3 in other acanthomorphs. OD3, OD3’ large, columnar, originating on dor- soanteriormost end of Pb3 ventral to TEb2, dividing posteriorly as it exits from below TEb2, with ventral branch, OD3’, inserting on Eb3 dorsal surface ven- troanterior to insertion of OD3, which is on anterior surface of Eb3 uncinate process. OD4 absent. OP complex, distinctly three-parted dorsally on Eb4, less so ventrally: medial part smallest, strap- like, beginning well medial to medial end of Eb4, curving ventrolaterally around OP middle part and inserting on tendinous raphe (ER, not illustrated) at about level of Cb4; middle part originating on Eb4 just ventral to tip of uncinate process, extending ven- trally, and joining ER immediately lateral to medial OP part; lateral OP part broadest, extends laterally from middle part on Eb4 to cartilaginous distal end of Eb4 and onto AC4; sub-dorsally, medial edge of lateral OP part becomes tendinous and extends to near distal end of Cb5; tendinous edge mid-posteri- orly joins ER and continues ventrally as lateral edge of OP muscle attaching to Cb5. Ad5 joins tendinous OP edge ventroposteriorly. RecD2 slender, originating on Eb2 anteromedial edge and inserting on Eb] ventral surface just ventral to LE1, overlies CT to which gill rakers attach ven- trally. RecD3 well developed, originating on Eb3 anter- oventrolaterally and inserting on Eb2 just ventral to LE2, overlies CT to which till rakers attach ventrally. Ad1-—3 absent. Ad4 narrow, dorsally on ventrolateral surface of Eb4 fusing anteriorly with lateral OP part ventrally 113 (not as apparent in posterior view as presented in Plate 80C). Ad5 (see also OP) attaches ventrally on dorsodistal edge of Cb5, dorsally it attaches musculously and tendinously to posterodistal surfaces of Cb4 and Ad4. SOD absent. RDs proximate. Additional remarks. SCL weakly attached to ven- trally extending cartilaginous posterior tip of Bb3. TV4 free from Cb5s. Two interarcual cartilages: an- terior IAC joins Ebl uncinate process and tiny car- tilaginous process dorsally slightly posterior to an- terior bony tip of Pb2; posterior IAC (IAC2) joins cartilaginous process near posterior end of Pb2 (pro- cess is subtended by posterior extension of Pb2 tooth- plate; neither Pb2 process is visible in Plate 80C (I[AC2 unknown for any other actinopterygian taxon). AC cartilage between distal ends of Eb3 and Cb3, another between distal ends of Eb4 and Cb4. Pb4 absent, UP4 present. Pb2 toothed. Eb4 levator pro- cess absent. Leis (1994:140—142) compared characters (none including the dorsal gill arches) of Lactariidae, Men- idae, and Carangidae and suggested the possibility that the first two families are the second and first sister groups of the third. In the discussion following Gasteromorpha, below, we suggest reasons for con- sidering Menidae as being more closely related to pre-percomorphs than to percomorphs (such as Lac- tariidae and Carangidae). Beryciformes TRACHICHTHYIDAE Hoplostethus mediterraneus Cuvier, USNM 307273, 79.1 mm; USNM 361959, 98.0 mm. Plate $1 Description (based almost entirely on the smaller specimen). Remarks. LE1-LE4 becoming long tendons dor- sally and coalescing into single origin on skull. LI1- LI2 musculously attached to skull. LE1 on dorsal edge of Eb1 well lateral to tip of uncinate process. LE2 on posterodorsal edge of Eb2 at mid-length of Eb2. LE3 on cartilaginous tip of Eb3 uncinate process at and dorsal to OD3—4 insertion on Eb3. LE4 on posterolateralmost surface of Eb4. LP absent. LI1 on medial surface of Pb2 uncinate process. LI2 on dorsal surface of Pb3 mid-laterally. TD comprises TPb2’, TPb2, TEb2, and TPb3-Eb3. TPb2’ thin, roughly triangular, sheetlike, divided mid-longitudinally, closely applied to ventral surface of thin CT sheet, which attaches to ventral surface of 114 skull and envelops anterior ends of Pb2s. Mid-ven- troanteriorly, TPb2’ muscle fibers mesh with semi- circular, pad-like TPb2, which attaches to Pb2 anter- odorsally and is anteroventrally continuous with TEb2. TEb2 inserts on Eb2 dorsal surface medial to LE2 insertion. Anteroventrally, TEb2 meshes with TPb3-Eb3. TPb3-Eb3 forms raphe dorsolaterally with OD3-—4 origin (forming “‘bun-like”’ mid-dorsal section), sheet of fibers continue deep ventrolaterally and attach broadly on dorsal surface of Pb3 medial to LI2 insertion, with small muscle strap attaching to dorsomedial surface of Eb3. TPb3-Eb3 is continuous posteriorly by fine muscle strand with SOD. Remarks. TD anomalous anteriorly in larger spec- imen, with separate muscle below TPb2’ that attaches to Pb2 on one side and Pb3 on other; TPb2’ present only unilaterally. OD3-—4 massive, originates anteriorly from Pb3 anterior end (ventral to TEb2) and dorsoposteriorly at raphe with TPb3-Eb3, inserts on lateralmost dor- soanterior surface of Eb3 uncinate process and later- almost dorsoposterior surface of Eb4, forms raphe posteroventrally with dorsal end of OP. OP originates dorsally from raphe with OD3—4 on Eb4 (raphe occluded from view in Plate 81B), lat- erally continuous with Ad4, joins raphe (ER) ven- trally with Ad5. Ad1-—3 absent. Ad4 dorsally on posterolateral surface of Eb4, con- tinuous medially with OP, forms raphe posteroven- trally with Ad5 (raphe—as ER—continues medially at ventral end of OP), attaches along dorsal surface of Cb4 anteroventrally anterior to Eb4-Cb4 joint. Ad5 mid-ventrally broadly on Cb5 distally, ex- panded well ventrally beyond attachment to Cb5 with tendinous attachment to cleithrum; dorsolaterally on Cb4 distally, expands medially and joins acutely an- gled (two-sided) raphe dorsally: with Ad4 laterally and OP medially. SOD present. RDs adjacent, each inserts by short tendon on pos- terior end of Pb3 (in larger specimen tendinous in- sertion is elongate, extending well external to SO. Additional remarks. SCL absent. TV4 free from Cb5s. Pb4 and UP4 absent. Pb3 toothed. IAC absent. Eb4 levator process absent. BERYCIDAE Beryx splendens Lowe, USNM 306134, 3 specimens, 98.0-111 mm. Plate 82 Description. Remarks. LE1, 2, and 4 are tendinous (not illus- trated) along lateral edge of their insertions. LE1, may form a slender, separate muscle (section) origi- nating as a long tendon, becoming musculous ven- BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON trally and closely paralleling the remainder of LE1, and inserting as a long tendon at and lateral to the remainder of the LE1 insertion. Because of the var- iability of LE1, we do not name the separate section. LE1 origin tendinous, insertion on Eb! uncinate process dorsoanteriorly lateral to cartilaginous tip. LE2 on expanded bony edge of Eb2 dorsoanter- iorly. LE3 mostly on cartilaginous tip of Eb3 uncinate process, but few muscle fibers continue and insert on tightly abutting Eb4 uncinate process. LE4 on dorsodistalmost cartilaginous edge of Eb4. LP very slender, becoming tendinous basally and joining lateral edge of LE4 insertion. LI1 insertion begins dorsally on medialmost edge of Pb2 uncinate process just ventral to cartilaginous tip Goined there by dorsolateral edge of TPb2), and extends anteroventrally along posterior surface of un- cinate process. About same size as LI2. LI2 on Pb3 dorsolaterally near joint with Eb3 me- dialmost end. TD comprises TPb2, TEb2-Eb1, TEb2, TPb3-Eb3, and TEb4. TPb2 arises as broad strap from median raphe on dorsoanterior surface of TEb2 and attaches to anterior surface of Pb2 uncinate process just ven- tral to tip of process; joined by LI1 anterior edge along Pb2 uncinate process. TEb2-Eb1 arises as nar- row strap from median raphe that is continuous with TEb2 median raphe, curves anterolaterally, becoming dorsal to TEb2, and attenuating as short, slender ten- don that inserts in CT surrounding tips of Eb1 and Pb2 uncinate processes; continuous posteriorly by di- agonal muscle strand with TPb3-Eb3. TEb2 broad, attaches to dorsal surface of proximal bony half of Eb2, with median raphe, which gives rise dorsally to tough CT pad. TPb3-Eb3 on Pb3 bony dorsal surface medial to LI2 insertion and on Eb3 posteromedial- most surface, continuous dorsomedianly with TEb4, which is on anteromedial dorsal surface of Eb4. Remarks. Sasaki (1989:14) first reported TEb2- Eb1 in Beryx, although he did not assign a name to it, nor illustrate it. We note that there is a remarkable, probably superficial, resemblance of TEb2-Eb1 to the anterior branch of the TD plexus of Acanthurus (q.V.). OD3-4 originates broadly along most of length of bony dorsal surface of Pb3, inserts broadly on ante- rior surface of Eb3 uncinate process and medial edge of Eb4 uncinate process. OP broad dorsally, on Eb4 posterior surface medial to dorsal attachment of Ad4, dorsolaterally joining raphe with OD3-—4, and ventrally joining ER with dorsal edge of Ad5 on Cb4, with a few medial mus- cle strands continuing onto Cb5. Remarks. Raphe with OD3—4 not present on either side of one specimen. Ad1-—3 absent. NUMBER 11 Ad4 dorsally on Eb4 posterolateral surface and ventrally on Cb4 anterior to Eb4-Cb4 joint. Ad5 on Cb4 posterodistally and Cb5 dorsodistally, forming raphe (ER) along dorsal edge with OP, and continuous ventrolaterally with SO. SOD present. RDs varying from well separated, to adjacent, to overlapping, each becoming broad tendinous strap before inserting on Pb3. Additional remarks. SCL absent. TV4 free from Cb5s. Tiny AC 4 on dorsoposterior cartilaginous end of Cb4 (additional smaller AC4, dorsal to larger one on one side of each of two specimens). UP4 present. Pb4 absent (see below). Eb4 levator process absent. Medial end of Eb4 much larger than medial end of Eb3. IAC absent. Pb! bony with cartilage ends. Pb2 toothed. Rosen (1973:466, fig. 84) indicated the presence of a reduced Pb4 in B. splendens. This element was not present in any of our three specimens (nor in the one of Centroberyx). Our specimens of Beryx splen- dens were collected in the Gulf of Aden, off Somalia, whereas Rosen’s specimen was collected in the Gulf of Mexico, off Mississippi. It is, therefore, quite pos- sible that there is variation in the presence of Pb4 in Beryx and that more than one species is involved; a revision of the genus is warranted. We did not at- tempt to verify Rosen’s finding and for the purposes of our study, we have treated berycids as lacking Pb4. Centroberyx affinis (Giinther), USNM 176984, 93.5 mm. Plate 83 Description. LE1 originating as long tendon (other LEs may have tendinous component); muscle slender, on dor- solateral edge of medially projecting Eb1 uncinate process. LE2 on tip of expanded bony edge of Eb2. LE3 on anterior surface of cartilaginous tip of Eb3 uncinate process. LE4 on dorsodistalmost edge of Eb4. LP very fine, tendinously inserted at ventrodistal- most edge of LE4 insertion. LI1 on medial surface of Pb2 uncinate process. LI2 tendinously on Pb3 dorsolaterally; about same size as LI1. TD comprises TPb2, TEb2, and TPb3-Eb3. TPb2 roughly oblong with deep notch mid-anteriorly and extensive CT pad arising dorsally from broad mid- longitudinal raphe, which completely divides TPb2 (and TEb2); muscle attachment begins anteriorly on anterior tip of Pb2 and extends posteriorly to point on anterior edge of Pb2 uncinate process; muscle is 115 free anterolaterally and dorsal to anteromedial por- tion of TEb2, but is posteromedially and ventrally continuous with TEb2. TEb2 deeply notched mid- posteriorly at continuation of mid-longitudinal raphe; another, incomplete raphe mid-dorsally amidst TEb2, which extends laterally onto dorsal surface of Eb2 to point just medial to base of spade-like process. TPb3- Eb3 free from TPb2 and TEb2, attaches laterally to dorsal surface of Pb3 medial to LI2 insertion and attachment continues posteriorly, passing dorsal to anteromedial end of Eb4, onto posteromedialmost edge of Eb3. TPb3-Eb3 is dorsoposteriorly continu- ous by fine muscle strand with SOD. Remarks. The attachment of TPb3-Eb3 to Eb3 is immediately adjacent to the abutting edges of Eb3 and Eb4. We expect that other specimens might ex- hibit attachment to Eb4 as well. A well-developed TEb4 is present in the related Beryx splendens. OD3-4 massive, originates broadly along Pb3 ven- tral to TPb2 and TEb2, and inserts on anterior surface of Eb3 uncinate process and tip of Eb4 uncinate pro- cess, forming raphe posterodistally with dorsal end of OP on Eb4. OP broad dorsally, on Eb4 posterior surface medial to dorsal attachment of Ad4, dorsolaterally forming raphe with OD3-—4, and ventrally joining ER with dorsal edge of Ad5 on Cb4, with a few medial mus- cle strands continuing onto Cb5. Ad1-—3 absent. Ad4 dorsally on posterodistal end of Eb4, attached continuously on posterior edge of Eb4-Cb4 joint, and continuing medially short distance on Eb4. Ad5 dorsally on posterodistal surface of Cb4, ven- trally on distal end of Cb5. SOD very fine. RDs adjacent. Additional remarks. SCL absent. TV4 free from Cb5s. Tiny AC4 (not visible in dorsal view) on dor- soposterior edge of distal end of Cb4. UP4 present, Pb4 absent (see discussion in Additional remarks un- der Beryx). IAC absent. Medial end of Eb4 much larger than medial end of Eb3. Pb2 toothed. Eb4 levator process absent; however, it appears the narrow cartilaginous distal end of Eb4, which con- nects expanded dorsal and ventral cartilaginous bulg- es, could become ossified in larger specimens, thus isolating a levator process. HOLOCENTRIDAE Holocentrus adscencionis (Osbeck), USNM 345408, 111 mm. Plate 84 Additional material. Sargocentron diadema (Lace- pede), USNM 334012, 95.8 mm. 116 Description. Remarks. The muscles of the two taxa are essen- tially the same. LE1 broadly on anterolateral surface of expanded bony edge of Eb1. LE2 on expanded dorsoposterior margin of Eb2. LE3 on joined cartilaginous tip of Eb3 uncinate process and cartilaginous edge of Eb4 uncinate pro- cess, also joins raphe with OD3—4 on these processes (joined processes completely enveloped in muscle, not visible externally). LE4 massive, dorsolaterally on Eb4 bony surface. LP on Eb4 at and lateral to LE4 insertion; insertion continuous ventrolaterally with CT sheet attaching to fourth and fifth arches. LI1 on Pb2 posteroventral to uncinate process. LI2 on Pb3 dorsoposterolaterally, just anterior to medial end of Eb3, ventromedially joining raphe with lateral edge of TPb3. TD comprises TPb2, TEb2, TPb3, and TEb4. TPb2 on Pb2 posteromedial surface, roughly circular and pad-like, dorsally with slight ventrolateral extension on each side, notched anteriorly and giving rise to CT sheet dorsally from between notch, overlies and is broadly continuous ventrally with TEb2. TEb2 dorsal to OD3-—4 origin, attaching laterally on Eb2 anterior to LE2 insertion. TPb3 broad, not connected to TPb2 or TEb2, ventral to OD3—4 origin, on Pb3 beginning on medial edge of base of Pb3 uncinate process, and extending posteriorly along ventrome- dial edge of LI2 insertion to position medial to me- dial end of Eb3, continuous by diagonal muscle strand with TEb4. TEb4 on dorsomedial surface of Eb4, continuous posteriorly by diagonal muscle band with SOD. OD3-—4 origin on Pb3 dorsomedial edge ventral to TEb2, insertion broadly on anterior surface of Eb3 uncinate process and posterior surface of Eb4 unci- nate process, insertion joining raphe with LE3, which is on both processes (Eb3 uncinate process has well- developed rounded cartilaginous tip; Eb4 uncinate process is horizontal with narrow cartilaginous edge). OP dorsally on Eb4 posteriorly and medial to Ad4 attachment, joining raphe with OD3—4 (on Eb4) pos- teroventrolaterally, ventrally joining ER with Ad5 dorsally on Cb4. Ad1-—3 absent. Ad4 dorsally on Eb4 lateral to OP, ventrally on Cb4 at and medial to Eb4-Cb4 joint. Ad5 ventrally extensively on dorsal surface of Cb5, dorsally on Cb4 well medial to distal end, there joining raphe (ER) with OP ventrally. SOD present. RDs adjacent. Additional remarks. SCL absent. TV4 free from Cb5s. Pb4 and UP4 present. Eb4 levator process ab- sent. Pb2 toothed. BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON ANOMALOPIDAE Anomalops katoptron (Bleeker), SIO-75-466, not measured. Plate 85 Additional material. @ = Photoblepharon palpebra- tum (Boddaert), USNM 264123, 71.2 mm. Description. LE1 origin by long tendon, insertion on dorsoan- terior surface of Eb1well lateral to tip of uncinate process. @) On Eb1 immediately lateral to tip of un- cinate process. Remarks. All other levators originate musculously. Right-side uncinate process is incompletely separated from medial end of Eb1. LE2 on dorsoanterior surface of posterodorsal bony process of Eb2. LE3 on anterior surface of Eb3 uncinate process. LE4 on dorsodistalmost edge of Eb4, continuous at ventrolateralmost edge of insertion with weak mus- cle strands, representing LP, which is continuous pos- teriorly with CT sheet, which changes to muscle, rep- resenting PP; LE4 ventrolaterally joins raphe with Ad4 dorsolaterally. LP represented by weak muscle strands attaching to ventrolateralmost edge of LE4 insertion and em- bodied in CT sheet also including PP. LI1 on lateral surface of Pb2 uncinate process dor- soposteriorly, noticeably larger than LI2. @ On dorsal surface of Pb2 just ventral to tip of uncinate process, same size as LI2. LI2 on Pb3 dorsoposteriorly and slightly antero- ventral to dorsomedialmost edge of Eb3. TD comprises TPb2-Eb2 and TPb3-Eb3 (see also additional remarks end of description). TPb2-Eb2 fi- bers complex anteriorly, attaching to medial surfaces of Pb2s anteriorly, muscle fibers ““weave”’ into trans- verse muscle fibers posteriorly that insert dorsally on Eb2, slightly lateral to medial end of Eb2 and slightly medial to expanded dorsoposterior process. TPb3- Eb3 triangular, apex anteriorly between medial ends of OD3—4s, partially ventral to OD-3—4 laterally, ventrolaterally attaching to Pb3 and continuing onto dorsomedialmost surface of Eb3. @ Comprises TPb2, TEb2, and TPb3-Eb3-Eb4. TPb2 on Pb2 uncinate process, meeting LI] insertion, continuous posteriorly by diagonal muscle strap with TEb2. TEb2 on Eb2 dorsoposteriorly well medial to LE2 insertion, continuous posteriorly by diagonal muscle strap (passing ventral to OD3—4) with TPb3- Eb3-Eb4. TPb3-Eb3-Eb4 on Pb3 dorsally beginning slightly anterior to LI2 insertion and continuing along medial edge of insertion onto Eb3 dorsoposterome- dially and EB4 dorsomedially, continuous by diago- nal muscle strands with SOD. OD3-4 origin broadly on Pb3 medially ventral to NUMBER 11 TPb2-Eb?2 posteriorly, inserting on dorsomedial edge of Eb3 uncinate process and dorsoposterior edge of Eb4 uncinate process, forming partial raphe with OP dorsolaterally. @ On Pb3 medially ventral to TEb2. OP broad strap dorsally on posterior surface of Eb4 lateral to Ad4 dorsal attachment, forming partial raphe dorsomedially with OD3—4, narrowing slightly ventrally and forming raphe (ER) with Ad5 dorso- laterally, raphe continuous laterally between Ad5 and Ad4. Ad1-—3 absent. Ad4 dorsally on posterolateral surface of Eb4 lat- eral to OP dorsally, dorsoposteriorly joining raphe with LE4 ventrally, and ventroposteriorly joining ER on Cb4 with Ad5 dorsally; anterior body of Ad4 ven- trally on Cb4 dorsal surface anterior to Eb4-Cb4 joint. Ad5 ventrally broadly on Cb5 dorsodistal surface; dorsally, beginning posterodistally on Cb4, joining raphe (ER) with Ad4 ventrally and OP ventrally. SOD present, broad. RDs separate, adjacent. Additional remarks. SCL present, attached dorso- medianly to posteroventral cartilaginous tip of Bb3. TV4 free from Cb5s. CT pad of TD anteriorly almost completely envelops Pbls. Pb2 uncinate process forms small, naked articulating facet (@ facet absent). Pb4 and UP4 absent. IAC small. Eb4 levator process absent. Percomorpha Smegmamorpha Johnson and Patterson (1993) hypothesized this group to include Mugilomorpha, Atherinomorpha, Elassomatidae, Gasterosteiformes (comprising Gas- terosteidae, Aulorhynchidae, Hypoptychidae, Pegas- idae, Indostomidae, Solenostomidae, Syngnathidae, Centriscidae, Aulostomidae, Fistulariidae), Synbran- choidei, and Mastacembeloidei. The monophyly of the group, based primarily on a single synapomorphy (origin of first epineural at distal tip of a transverse parapophysis), has been controversial ever since. We do not believe the Smegmamorpha are mono- phyletic (a more formal denial is provided by Spring- er and Orrell in the Appendix), but, for the sake of convenience, we treat the members of the group that we examined together. We partition the Smegma- morpha into three groups, Gasterosteomorpha, Ath- erinomorpha, and Mugilomorpha. We believe the last two groups are closely related, but not closely related to the Gasterosteomorpha. The Atherinomorpha have been strongly supported as a monophyletic group for many years and questionably allied to the Mugilo- morpha. We believe the Gasterosteomorpha are pos- sibly polyphyletic. We recognize four groups within V7 the Gasterosteomorpha: Centrisciformes, Gasterostei- formes, Synbranchiformes, and Elassomatiformes. Johnson and Springer (1997:176) implied that there is a close relationship between Elassomatidae and Gasterosteidae. They did not follow this with a formal publication, and the matter remains unre- solved. Gasterosteomorpha We believe Centriscidae, particularly Macroram- phosus (Centriscops, Notopogon should also be ex- amined), are either a monophyletic group not closely related to other gasterosteomorphs or are the sister group of the gasterosteomorphs. Macroramphosus is one of only three taxa (others: Icosteus, Icosteidae; Mene, Menidae) that are usually classified among the percomorphs that possess ER and may have OP ven- trally attaching to Cb4, conditions otherwise known to occur among pre-acanthomorphs (commonly) and more basal acanthomorphs (1.e., polymixiids, para- canthopterygians, zeiforms, stephanoberyciforms, beryciforms). All other of Macroramphosus’s puta- tively closely related gasterosteomorphs (i.e., syng- nathoids) are highly specialized and have lost many skeletal elements; their muscles are also degenerate, but if OP is present, it attaches to CbS. If Macroramphosus is validly removed from the Gasterosteomorpha and placed in a more basal acan- thomorph position, the entire classification of the Gasterosteomorpha needs to be re-evaluated, which might result in positioning the entire Gasterosteo- morphas among the pre-percomorphs. Insofar as we can determine, the term Gasterosteo- morpha was first used by Nakabo (2002:53, 56, 512), who recognized Johnson and Patterson’s Smegma- morpha, and included under it four groups: Gaster- osteomorpha, Synbranchomorpha, Mugilomorpha, Atherinomorpha. The composition within each of these four groups mentioned only Japanese forms; therefore, it is uncertain how he would have treated Elassomatidae, although probably as a fifth group. We employ Gasterosteomorpha here to include Cen- trisciformes, Gasterosteiformes, Elassomatiformes, and Synbranchiformes. Although a reconstituted Smegmamorpha (1993) might be used for the group, we find that name marginally offensive and its orig- inal composition misleading. The ICZN does not specify that priority pertains to taxa above the family- group level. Centrisciformes CENTRISCIDAE Macroramphosus scolopax (Linnaeus), USNM 366546 and 158813, both 111 mm; USNM 366727, 110 mm, 177091, 126 mm. Plate 86 118 Description. LE1 on high, bony spike-like Eb] process dorso- medially (cartilage-tipped Eb! uncinate process ab- sent); strong, ribbon-like tendon on lateral edge of muscle, separating from muscle about halfway up from insertion and originating separately on skull; fine, thread-like tendon separates from posterior edge of ribbon-like tendon and inserts in CT around dorsal end of Pb3 laterally. LE2 tendinously on mid-dorsal edge of Eb2 and continuing tendinously posteriorly and attaching to Eb3 mid-ventroanteriorly; anteromedial edge of in- sertion joined by posterodistal edge of TEb2. LE3 absent. LE4 on Eb4 mid-dorsally lateral to uncinate pro- cess, insertion joined posteriorly by LP. LP tendinously on Eb4, joining LE4 insertion pos- teriorly. LI1 on Pb2 ventroposterior to dorsal process, and ventrolateral to origin of OD3—4 on Pb2 [sic]. LI2 on Pb3 dorsally medial to anterior portion of joint articulating with medial end of Eb3. TD comprises TEb2 and TPb3-Eb3. TEb2 rela- tively narrow, boomerang-like (anteriorly concave), with mid-longitudinal raphe continuing posteriorly halfway or completely across TPb3-Eb3; TEb2 atta- ches ventroanteriorly to complex CT pad and tissues covering dorsal ends of Pb2, Pb3, and Eb3 (tissues removed in illustration); pad continuous anteriorly and mid-ventrally with CT of pharyngeal roof (limp, but tough ligament inserts on Pb2 posteromedially ventral to cartilaginous Pb2 dorsal cap and immedi- ately medial to origin of OD3-—4 portion inserting on Pb2, and also deeply on Pb2 surrounding LI] inser- tion medially, thus straddling Pb2 portion of OD3—4 insertion; similar ligament inserts on Pb3 dorsopos- teriorly; see also Synbranchus, Synbranchidae, for similar ligaments); TEb2 attaches on Eb2 beginning at and medial to LE2 insertion and continuing dor- soanterolaterally almost to distal end of Eb2, there overlain anteriorly by GFM2; TEb2 continuous with Eb3 portion of TPb3-Eb3 anteriorly by diagonal mus- cle strands. TPb3-Eb3 broadly triangular in dorsal view, variously appearing to consist of two meshing portions that extend laterally and attach to medial edge of Eb3 uncinate process and, separately, to Eb3 dorsal surface medial to uncinate process; muscle ex- tends anterolaterally ventral to OD3—4-OD3’ and TEb2 and attaches to Pb3 dorsomedial to joint with Eb3, and is continuous posteroventrally and posteri- orly with SOD. OD3-—4, OD3’ with separate origins on Pb2 and Pb3; origin on Pb2 posteromedially well ventral to large dorsal cartilage cap, straddled laterally and me- dially by limp ligament (described in TD above): fi- bers extend posteriorly from Pb2, meshing ventrally with main body of muscle originating on Pb3 ventral BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON to TEb2; muscle divides posteriorly with ventral fi- bers (OD3’) inserting on Eb3 anterior surface ventral to tip of uncinate process and dorsal fibers (OD3—4) inserting finely on medial edges of joined Eb3 and Eb4 uncinate processes. OP dorsally on posterior surface of Eb4 beginning lateral to medial end and extending laterally a vary- ing distance, but no further than ventral to LE4 in- sertion, there variably separable or not from Ad4 dor- sally; divided at mid-level (level at Cb4) by small raphe (ER) and attaching on Cb5 medial to Ad5, var- iably appearing undifferentiated from Ad5 ventral to ER. GFM1 dorsally on and covering anterior surface of Eb] beginning a little ventral to spike-like process and extending anteroventrally most of length of Cbl1. GFM2 beginning on Eb2 dorsomedial surface and continuing laterally on dorsoanterior surface anterior to TEb2, fanning out anterolaterally from medial at- tachment to Eb2, thereby anteriorly covering TEb2 laterally and Eb2-Cb2 joint, tapering ventrally, and extending anteroventrally along most of anterior sur- face of Cb2. GFM3 beginning on Eb3 dorsomedially and con- tinuing laterally, fanning out anterolaterally, covering Eb3-Cb3 joint, tapering ventrally, and extending an- teroventrally along distal fifth of Cb3; thin ribbon of muscle fibers splits off GFM posterodistally and at- taches to Eb3-Cb3 joint anteriorly (not visible in il- lustration). Ad4 dorsally on ventral and posterior surfaces of Eb4 lateral to OP, fusing with OP anterolaterally; ventrally relatively broadly on dorsolateral surface of Cb4 medial to Eb4-Cb4 joint. Ad5 relatively deep, dorsally on posterolateralmost surfaces of Eb4 and Cb4, covering joint of these two bones, ventrally on Cb5 dorsolaterally posterior to OP. SOD broad, continuous anteriorly with TEb3. RDs moderate, adjacent or slightly separated. Additional remarks. SCL attached mid-dorsally to bony posterior end of Bb3. TV4 free from Cb5s. Pb1 cartilaginous plate, Pb4 absent, UP4 present. Pb2 toothed. [AC absent. Eb4 levator process absent. Gasterosteiformes GASTEROSTEIDAE Gasterosteus aculeatus Linnaeus, USNM 344603, 62.0 mm; USNM 58296, 64.6 mm. Plate 87 Additional material (information on these taxa in- cluded for the most part only in the data matrix, Table 12, used for the cladistic analyses). Apeltes quadracis (Mitchill), USNM 242691, 42.2 mm; Culea inconstans (Kirtland), USNM 69287 41.5 NUMBER 11 mm; Pungitius pungitius (Linnaeus), USNM 77842, 49.8 mm; Spinachia spinachia (Linnaeus), USNM 23008, 138 mm, USNM 344839, 139 mm. Description. LE1 on Eb! surrounding bony or cartilage tipped uncinate process (see remarks), joining raphe ante- roventrally with Ad1 dorsally; fine ligament extends posteriorly from posterolateral edge of raphe to an- terior edge of Eb2. See remarks following description of LE1 in Elassoma. Remarks. Primitively in acanthomorphs, the ante- rior or medial process of Eb! articulates with Pb1 and the uncinate process with IAC or Pb2 (articula- tion may be direct or ligamentous). On this basis we consider the anterior process absent in E/assoma and gasterosteiforms, and the process articulating with Pb2 to be the uncinate process. We interpret the oc- casional presence of a small cartilage tip on the ex- panded edge of Eb1 in Gasterosteus, which is com- pletely surrounded by the insertion of LE1, as a fail- ure of the edge to completely ossify, as opposed to treating it as the tip of the “true” uncinate process. We know of no case where an uncinate process is present and the anterior process of Eb! articulates with Pb2. LE2 on raised bony process on Eb2 posterolater- ally, joining raphe ventromedially with lateral end of TEb2 and another ventrolaterally with posterior por- tion of Ad2; fine ligament extends posteriorly from lateral edge of latter raphe to anterior edge of Eb3. LE3 absent. LE4 broadly on Eb4 dorsoposteriorly, joining LP anteroventrally. Remarks. Uniquely among the gasterosteids ex- amined, Spinachia apparently lacks LE4. This inter- pretation is based on the fact that LE4 in the other gasterosteids angles dorsoanteriorly towards its ori- gin, whereas LP is almost vertical, as well as joining Ad5 dorsally (see LP). LP on most of Eb4 lateral to uncinate process (may extend both lateral and medial to junction with LE4 insertion), joining raphe with Ad5 dorsally, thus forming an LP-Ad5 sling; true in all gasterosteids examined. LI1 mostly on Pb3 dorsoanteromedially dorsal to and meeting origin of OD3—4, continuing, second- arily, onto Pb2 dorsomedially. LI2 on Pb3 dorsoposteriorly opposite medial end of Eb3. TD comprises TEb2 and TPb3-Eb3-Eb4. TEb2 broad medially, narrowing considerably laterally, with mid-longitudinal raphe giving rise dorsally to CT sheets and attaching ventroanteriorly between Pb3s to CT of pharyngeal roof, extending laterally and attaching on raised bony edge of Eb2 joining raphe with medial edge of LE2 insertion (not ex- 119 tending laterally past medial edge of LE2 insertion); not continuous posteriorly with TPb3-Eb3-Eb4. TPb3-Eb3-Eb4 on Pb3 dorsolaterally just medial to LI2 insertion, continuing posteriorly and attaching narrowly on dorsomedialmost end of Eb3 (there join- ing raphe with medial end of anterior portion of Ad3), then continuing posteriorly, but well separated laterally, with attachment to Eb4 dorsally ventrome- dial to uncinate process; posteriorly continuous broadly with SOD. OD3-4 origin on Pb3 broadly dorsomedially, an- teriorly ventral to and joining LI] insertion, insertion on medial edges of Eb3 and Eb4 uncinate processes. OP dorsally beginning on Eb4 posteriorly ventral to uncinate process and extending broadly medially, ventrally on Cb5 beginning at ventrolateral edge of Ad5, medially unclearly separable from SO. Ad1 broad, consisting of a single portion (equiv- alent to posterior portion of Ad2 and Ad3) extending from raphe with LEI! ventroanteriorly across Eb1- Cb1 joint to point on Cb1 anterior bony surface just ventral to joint, ventrolateral edge attaching to fila- ments. Present and similar in all gasterosteids ex- amined. Ad2 comprising two portions, posterior portion ex- tending from raphe with LE2 ventroanteriorly across Eb2-Cb2 joint to point on Cb2 anterior bony surface just ventral to joint; anterior portion beginning on Eb2 dorsomedially and extending laterally and pass- ing anterior to most of posterior portion, with lateral edge attaching to gill-rakers and/or gill filaments. Present and similar in all gasterosteids examined. Ad3 comprising two portions, posterior portion at- taching dorsally to anterolateral edge of Eb3 uncinate process, continuing on ligament joining Eb3 and Eb4 uncinate processes and anterodistal end of Eb4; an- terior portion beginning on dorsomedial surface of Eb3, joining raphe with lateral end of TPb3-Eb3-Eb4 and extending laterally and passing anterior to most of posterior portion, with lateral edge attaching to gill-rakers and/or gill filaments. Present and similar in all gasterosteids examined. Ad4 dorsally on Eb4 broadly ventrally, ventrally broadly on Cb4 medial to Eb4-Cbé4 joint, slightly fus- ing dorsoposteriorly with Ad5 dorsoposteriorly; com- pletely obscured from view posteriorly by Ad5. Ad5 dorsally beginning on Eb4-Cb4 joint and ex- tending broadly medially on Eb4 to edge of OP, join- ing raphe with LP ventroposteriorly, ventrally on Cb5 dorsolaterally, completely overlapping Ad4 posteri- orly. SOD broad continuous anteriorly with TPb3-Eb3- Eb4. SO lateral band of fibers extends anteriorly over medial end of Eb4 and ventral to TPb3-Eb3 and in- serts on Pb3 posteriorly. RDs adjacent. 120 Additional remarks. SCL present, attached mid- dorsally to wedge of cartilage, which is attached to ventral surface of cartilaginous posterior end of Bb3; Hb3s attach to wedge posterolaterally. TV4 free from Cb5s. Pb1, Pb4 and UP4 absent. Eb3 and Eb4 un- cinate processes present. Eb4 levator process absent. Anker (1974) described and illustrated the head skeleton and muscles of Gasterosteus aculeatus. Our findings are in essential agreement with his. HY POPTYCHIDAE Johnson and Patterson (1993) removed Aulichthys from the Aulorhynchidae and included it in the Hy- poptychidae. We agree with that action. Hypoptychus dybowskii Steindachner, USNM 51494, ca. 85 mm; USNM 118009, 72.8 mm; UW 29655, 3 specimens, 65—67 mm. Plate 88 Remarks. Damage occurred in every specimen during release of the levators, and extreme difficulty was encountered in resolving the limits and attach- ments of the more posterior (non-levator) muscles. Additionally, there appeared to be complex variation. The description and Plate 88 are a composite based on all the specimens. Every muscle described or il- lustrated was seen in at least two specimens. Description. LEI! extremely fine, posterodistally on Ebl, con- tinuing just onto CT (not illustrated) between Eb1 an Eb2. LE2 extremely fine, posterodistally on Eb2, con- tinuing just onto CT (not illustrated) between Eb2 and Eb3. LE3 finest of all levators, usually lost during dis- section, on Eb3 lateral to uncinate process, continu- ing just onto CT (not illustrated) between Eb3 and Eb4; expanding at origin, and joining, or nearly so, LE4 origin. LE4 relatively large, on Eb4 dorsolaterally, joining raphe ventrolaterally with Ad5 dorsally; muscle may appear incompletely divided into lateral section, which is continuous with Ad5, and medial section, which is not; LE4 expands at origin, probably join- ing, or nearly so, LE3 origin. Remarks. The LE3 and LE4 origins are posterior to those of other levators. Usually, if not always, in percomorphs, the origins of LE3 and LE4 join the origins of the other levators, none of which ever joins the origin of LP. Our recognition of the levator on Eb4 as LE4, rather than LP is based on this fact. The alternative, that LE4 is lost and the origin of LE3 has moved posteriorly and joined that of LP, seems less parsimonious. LP absent (see remarks following LE4). BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON LI1 on Pb2 dorsolaterally just anterior to articu- lation with Eb2. LI2 on Pb3 posterodistally just anterior to medial end of Eb3. TD comprises TEb2 and TPb3. TEb2 boomerang- like, attached mid-ventroanteriorly to CT of pharyn- geal roof, extending laterally and attaching to Eb2 dorsally anterior to LE2; uniquely, ventrolaterally finely continuous with portion of OD3-—4 origin on Pb2; posteriorly meeting but not continuous with TPb3. TPb3 broad, attaching along much of Pb3 dor- sal surface, continuous posteriorly with broad SOD. OD3-—4 origin on Pb2 and Pb3 dorsally ventral to TEb2 (strands from origin on Pb2 attach on Eb2 in some specimens and join with TEb2); after passing posteriorly from under TEb2, muscle joins a branch of SO that passes posteriorly broadly over Eb4 (no raphe apparent) and continues to distal end of Cb5 medial to OP. OD3—4 extends dorsally over Eb3 and Eb4 uncinate processes (obscuring Eb4 uncinate pro- cess and, usually, Eb3 process in dorsal view), with slight insertion on Eb3 process and main insertion on Eb4 process, and is continuous with OP medial sec- tion on Eb4 (raphe absent). OP strap-like, dorsally on Eb4 posterior to unci- nate process, dorsomedialmost fibers continuous with OD3-4 posteromedially; ventrally, attaching to Cb5 dorsodistally near Ad5. Ad1 on anterodistal surface of Eb1 and dorsoan- teriormost surface of Cbl. Ad? on anterolateral bony surface of Eb2, meeting distal end of TEb2, and dorsoanteriormost surface of Cb2. Ad3 on most of bony surface of Eb3 lateral to uncinate process and on dorsoanteriormost surface of Cb3. Ad4 dorsally, broadly on Eb4 posteriorly between OP and Ad5; ventrally on Cb4, extent not deter- mined. Ad5 a small muscle strap, ventrally on Cb5 dor- soposterodistally; dorsally on Eb4 dorsoposterodis- tally, there joining raphe with LE4 ventrolaterally; small anterior portion of muscle (not visible in Plate 88) attaches to posterodistal end of Cb4. SOD very broad. RDs slender, separated by space more than twice one RD diameter. Additional remarks. SCL present, questionably free from Bb3, which has ventrally extending pos- terior end, usually indicative that SCL attaches to it. TV4 free from Cb5s. Pb4 present, Pb] and UP4 ab- sent. Pb2 toothed. IAC absent. Eb1 uncinate process present, anterior process absent (see remarks follow- ing description of LE1 under Gasterosteus for expla- nation). Eb4 uncinate process present, levator process absent. NUMBER 11 Aulichthys japonicus Brevoort, USNM 347504, 136 mm; USNM 59789, 129 mm; USNM 71104, 2 specimens, 68.5—71.8 mm. Plate 89 Description. LE1 on posterior edge of Eb1 just medial to car- tilaginous distal end. LE2 on posterior edge of Eb2 just medial to car- tilaginous distal end. LE3 on tip of Eb3 uncinate process anteriorly. LE4 short, on Eb4 dorsoanteriorly near distal end. LP short, on Eb4 at and posterolateral to LE4 in- sertion. LI1 on Pb2 dorsoanteriorly and adjacent bony sur- face of Pb3 anterolaterally. LI2 on Pb3 dorsoposteriorly anterior to medial end of Eb3. TD comprises TEb2 and TPb3. TEb2 attaches ven- tromedially to CT of pharyngeal roof and laterally to Eb2 dorsally medial to LE2 insertion, continuous mid-posteriorly with TPb3. TPb3 laterally passes ventral to OD3—4 and attaches to Pb3 broadly ante- rior to LI2 insertion; posteriorly muscle is broadly continuous with SOD; posterolaterally, anterior fibers of SO pass through and ventral to TPb3 and insert on Pb3 medially. Remarks. SO fibers originate diffusely medial to OP and extend dorsoanteriorly, passing among ante- rior SOD fibers and (difficult to see) possibly through posterior TPb3 fibers and attaching to Pb3. OD3-—4 origin on Pb2 dorsally posterior to LI1 in- sertion and on Pb3 dorsally ventral to TEb2, insertion on Eb3 uncinate process anteriorly and medial edge of Eb4 uncinate process. OP dorsally on Eb4 posteriorly, extending from near medial end to uncinate process, there apparently fusing completely with Ad4 medially; ventrally on Cb5 dorsodistally, partially fusing anteriorly with Ad5; medially fusing with SO. Ad1-—3, each beginning on anterior surface of re- spective Eb at about mid-length of Eb and extending laterally and spreading and attaching on anterior sur- face of Eb-Cb joint to include cartilaginous distal end of Cb; posterior surface of muscle takes twist at distal end and appears to pass posterior to itself. Ad4 attaches dorsally to Eb4 posteriorly lateral to uncinate process and attaches ventrally to Eb4 dor- somedially, fusing medially with dorsal half of OP. Ad5 very reduced, ventrally on Cb5 distally, pos- teroventrally fusing with OP, dorsally on Cb4 poster- odistally. SOD very broad. RDs slightly separated. Additional remarks. SCL questionably free from Bb3. TV4 free from Cb5s. Pb1, Pb4 and UP4 absent. Pb2 toothed. IAC absent. Tiny AC present on dor- 121 sodistal end of Eb1, apparently not homologous with ACI as it is completely removed from Cbl. Aulos- tomus has an elongate extension of the cartilaginous distal end of Eb1, which, if budded off, would form a similar appearing AC (Aulostomus is unique among gasterosteiforms in having IAC). Eb4 levator process absent. Eb! uncinate process present, anterior process absent (see remarks following description of LE] un- der Gasterosteus for explanation). AULORHY NCHIDAE Aulorhynchus flavidus Gill, USNM 344688, 2 spec- imens, 143-144 mm SL; USNM 167915, 89.9 mm SL. Plate 90 Description. LE1 on Eb1 dorsoposterolaterally, insertion con- tinuing posteriorly on CT (not illustrated) joining Eb1 to Eb2 anterodistally. LE2 on Eb2 dorsoposterolaterally, continuing pos- teriorly as CT (not illustrated) joining Eb3 anteriorly. LE3 absent. LE4 on Eb4 dorsolaterally. LP on Eb4 joining LE4 insertion posteriorly, join- ing raphe with Ad4 dorsally. LI1 massive, on Pb2 dorsoanteromedially and Pb3 anterolateralmost edge, which abuts Pb2. LI2 on Pb3 dorsoposteromedially anterior to me- dial end of Eb3. TD comprises TEb2 and TPb3-Eb3-Eb4. TEb2 flat, hood-like viewed dorsally; attaches finely later- ally on dorsoposterior edge of Eb2 at medial edge of LE2 insertion, attaches ventrally along mid-longitu- dinal line to CT of pharyngeal roof; continuous pos- teriorly by diagonal muscle strand with TPb3-Eb3- Eb4. TPb3-Eb3-Eb4 anterolaterally on Pb3 dorsolat- erally between joints with medial ends of Eb3 and Eb4, continuing posteriorly on medial ends of Eb3 and Eb4 dorsally, continuous mid-posteriorly by di- agonal muscle strands with broad SOD. OD3—4, OD3’ origin on almost entire dorsomedial margin of Pb3, muscle separates into OD3—4 dorsally and OD3’ ventrally on exiting posteriorly from under TEb2; OD3-4 inserts on Eb3 uncinate process an- teriorly and Eb4 uncinate process medially; OD3’ in- serts on Eb3 dorsally ventromedial to uncinate pro- cess. See also OP below. OP dorsally on Eb4 posteriorly beginning ventral to and between uncinate process and LP insertion and extending medially and becoming indistinguishable from SO (an one specimen slender group of muscle fibers from OD3-—4 extends posteroventrally medial to Eb4 uncinate process and joins OP fibers); ven- trally on Cb5 dorsodistally, apparently fusing with Ad5 medially. Ad1 begins on dorsoanterior surface of Ebl an- 122 teromedial to LE! insertion and follows anterior edge of Eb] laterally, attaching to anterior surfaces of Eb] and Cb1 where they join; short branch of muscle fi- bers separates anteriorly from remainder of muscle at distal end of Eb1 and attaches to distalmost gill raker of dorsal arch (rakers do not follow Eb1 margin, but extend directly anteriorly from distal end of Eb1). Remarks. Our description of the short branch of Ad! (and also that of Ad2 and Ad3) is problematic; Plate 90 presents another interpretation, with the short branch of fibers separating posterolaterally from the remainder of the muscle. Ad2 begins on dorsomedialmost bony surface of Eb2 ventral to TEb2 and extends laterally on anterior edge of Eb2, covering anterior surfaces of Eb2 and Cb2 where they join; short branch of muscle fibers separates anteriorly from remainder of muscle at dis- tal end of Eb2 and extends anteriorly, diminishing shortly and becoming CT that attaches to Eb1 pos- teriorly. See remarks following Ad1. Ad3 like Ad2, but involving Eb3 and Cb3. See remarks following Ad1. Ad4 dorsally on ventral surface of Eb4 posteriorly mostly lateral to OP, ventrally on Cb5 dorsal surface posteriorly medial to Eb4-Cb4 joint. Ad5 questionable small band of muscle joining posterodistal end of Cb4 and anterodistal end of CbS5, fused indistinguishably with OP ventrolaterally. SOD very broad. RDs adjacent. Additional remarks. SCL attached mid-dorsally to ventrally extending cartilaginous posterior tip of Bb3. TV4 free from Cb5s. Pb1 and Pb4 absent, UP4 pres- ent. Pb2 toothed. [AC absent. Eb1 uncinate process present, anterior process absent (see remarks follow- ing description of LE] under Gasterosteus for expla- nation). Eb4 levator process absent. Synbranchiformes The Synbranchiformes comprise the synbranchoids and mastacembeloids. Britz (1996) last supported recognition of a monophyletic group consisting of these two suborders. The interrelationships of the Synbranchiformes are unclear. SYNBRANCHIDAE Synbranchus marmoratus (Bloch), USNM 311207, 2 specimens, 298-307 mm TL. Plate 91 Additional material. @ = Ophisternon bengalensis McClelland, USNM 135205, ca. 350 mm TL. Not all data were available on this specimen; absence of comment should not be interpreted as agreement with description of S. marmoratus. BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON Description. Remarks. All of the levators extend anteriorly al- most horizontally parallel to mid-line axis of body. Their orientations in the illustrations have been mod- ified variously. @ Same. LE1 on Eb! dorsally near distal end. Uncinate pro- cess absent. @ Same. LE2 divided, with separate anterior insertion on Eb2 posterodistalmost edge and posterior insertion (LE2’) on CT between Cb1 and Eb3 or on CT but with tendon extending posteriorly from insertion to Cb3 somewhat ventral to joint with Eb3. @ Not di- vided. Remarks. In normal position LE2 and LE2' appear to be one muscle, but manipulating the muscles they easily separate basally and the separation can be con- tinued to (or almost to) their joint origin. The con- dition appears to be a derivation of the state of LE2 in the related Mastacembelus, in which LE2 is not divisible and inserts on Eb2 dorsoposterior edge and Eb3 dorsoanterior edge. LE3 absent (both sides of both specimens). @ Highly reduced LE3 present on only one side; for cladistic analysis, we consider LE3 absent. LE4 on bony dorsodistal end of Eb4, divided at origin. @ Same. LP absent. @ Same. LI1 massive, inserting in CT attached to medial end of IAB, enveloping Pb2 (which may be vestigial or absent), and attached to or enveloping medial end of Eb2 and anterior end of Pb3 (none of the insertion points are visible in illustrations). @ Same. LI2 on Pb3 dorsoposterolaterally opposite medial end of Eb3. @ Same. TD with almost complete mid-longitudinal raphe, comprises TEb2, TPb3-Eb3, and TEb4 (see addition- al remarks for discussion of absence of TPb2). TEb2 a pair of muscles, each angled anterolaterally and at- taching on respective Pb3 dorsoanteriormost end and Eb2 dorsomedially, lies dorsal to OD3—4 origin and is continuous mid-posteriorly with TPb3-Eb3. TPb3- Eb3 (see also remarks following TD description) on Pb3 dorsally along medial edge of LI2 insertion, pos- teriorly extending laterally and attaching on Eb3 dor- somedially, continuous narrowly mid-posteriorly with TEb4. TEb4 more-or-less triangular, attaching along posterior edge of Eb4 and ventral surface of ligament complex extending anteriorly to Pb3 and Eb4 (exact nature of ligaments unclear), muscle con- tinuous posteriorly with SOD. @ Comprises TEb2 (not split) and TPb3-Eb3-Eb4, shallow gap present between Pb3 and Eb3, and Eb3 and Eb4 attachments. OD3-—4 originates narrowly on joint formed by an- terolateral end of Pb3 and medial end of Eb2; inserts massively dorsally on dorsolateral surface of Eb4 (covering joint with Eb3 uncinate process and ante- rior edge of Eb4 (lacks uncinate process), which may NUMBER 11 be slightly modified as a bony process or only in- dented) and ventrally, less extensively on Eb3 unci- nate process, the tip of which may be bony or mi- nutely cartilaginous. @ Originates on Eb2 and adja- cent Pb3 and inserts similarly to S. marmoratus, ex- cept muscle does not cover Eb3 and Eb4 processes (Eb3 uncinate process with cartilaginous tip). OP dorsally on Eb4 ventrolaterally, ventrally on Cb5 dorsoposteriorly, there overlapping Ad5 ventro- posteromedially, medially mostly inseparable from SO. Ad1 (not illustrated) questionably interpreted as a fan of muscle on Eb1-Cbl joint anteriorly; mainly associated with gill filaments. Ad2 on Eb2 mid-dorsoanteriorly, extending ven- trolaterally onto anterolateral surface of Cb2, ventral edge attached to gill filaments. Ad3 on Eb3 anterolaterally, extending onto Cb3 anterolaterally, ventral edge attached to gill filaments. Adé4 dorsally on Eb4 and ventrally on Cb4, medial to Eb4-Cb4 joint. Ad5 dorsally broadly on Cb4 ventrolaterally, ven- trally on Cb5 dorsally, extending medially anterior to OP. SOD broad, continuous anteriorly with TEb4. @ Present. RDs separated by space more than one RD diam- eter. Additional remarks. SCL absent; @ Same. TV4 free from Cb5s. @ Same. Pb! and Pb4 absent. UP4 present. Eb4 levator process absent. @) Same. Small bony flange at distal bony end of Eb4 dorsally or dorsoanteriorly shields cartilage. @ Same (see Rosen and Greenwood, 1976:fig. 35, where Eb4 flange is indicated but not labeled). Pb2 is present as an edentate bony fragment in S. marmoratus. Rosen and Greenwood (1976:fig. 36) il- lustrate a moderately well-developed, but edentate, rod-like Pb2 in the dorsal gill-arch skeleton of S. marmoratus, and (figs. 28-34) and as a small, eden- tate, plate-like bone in all seven species of Ophister- non; Our specimen agrees. Rosen and Greenwood (1976:7) indicated the pres- ence of a tendon, “‘t-rab, tendon of M. retractores arcum branchialium,” (their rab = our RD), which they illustrated as inserting on UP4 posteriorly in var- ious synbranchids, but not Synbranchus. They did not discuss the tendon, which we believe is more accu- rately considered as a ligament). The ligament is in- tegral with RD ventrolaterally for the muscle’s entire length, and it remains after the muscle is digested during clearing and staining. Rosen and Greenwood also did not mention other non-muscle integrated lig- aments attaching variously to Pb3 (also attaches to cleithrum), Eb3, and Eb4 in synbranchids. On one side of the specimen of Ophisternon, the ligaments appear to unite in a complex mesh, and we are un- 123 certain if our dissections of Synbranchus did not de- stroy the integrity of the ligaments (hence, their ren- dition in Plate 91 should be verified). MASTACEMBELIDAE Mastacembelus armatus (Lacepéde), USNM 319465, 275 mm; USNM 343569, 300 mm. Plate 92 Description. LE1 on dorsoposterior edge of Eb1 and dorsoan- terior edge of Eb2, meeting Ad2 anteriorly. Cartilage- tipped Eb1 uncinate process absent. LE2 on dorsoposterior edge of Eb2 and dorsoan- terior edge of Eb3, meeting Ad3 anteriorly. LE3 absent. LE4 on Eb4 dorsally lateral to uncinate process. LP absent. LI1 massive, on joined anterior ends of Eb1, Pb2, and Pb3 dorsally, meeting anteriormost edge of OD3-—4 origin. LI2 on Pb3 dorsolaterally anterior to medial end of Eb3; muscle meets lateral edge of TPb3-Eb3. TD comprises TEb2 and TPb3-Eb3 (a few strands of TD attach to posteromedial edge of Eb4 on one side of one specimen). TEb2 strap-like with mid-lon- gitudinal raphe attaching dorsally to CT sheets (not shown) and ventrally to CT of pharyngeal roof; mus- cle attaches laterally on Eb2 dorsally, meeting pos- teromedial edge of LE1 insertion and medialmost edge of Ad2, and is continuous mid-posteriorly by diagonal muscle straps with TPb3-Eb3. TPb3-Eb3 broadly, dorsally on Pb3 at medial edge of LI2 in- sertion, continuing posteriorly and attaching on pos- terior edge of Eb3 medial to uncinate process; con- tinuous posteriorly by diagonal muscle strands with SOD. OD3-—4 origin on Pb3 dorsally ventral to TEb2, meeting posterior edge of LI1 insertion; muscle di- vides posteriorly with dorsal portion inserting on Eb4 uncinate process and ventral portion inserting on Eb3 uncinate process (including medial edges of joined processes). OP dorsally on Eb4 posteriorly in area ventral to uncinate process, laterally overlapping Ad4 medially, ventrally on Cb5 dorsolaterally meeting Ad5 ven- trally and PCI dorsally, medially not clearly separable from SO. Ad1-—3 problematically present (as GFM?). On each epibranchial dorsoanterolaterally, there is a more-or-less well-defined strip of muscle that be- comes weak and finer distally as it extends ventrally along the entire anterolateral edge of the associated ceratobranchial and attaches to the gill filaments ba- sally. Ad4 dorsally on ventroposterior surface of lateral half of Eb4, medially overlapped by OP, ventrally on 124 Cb4 dorsally medial to Eb4-Cb4 joint, there joining Ad5. Ad5 dorsally on posterolateral end of Cb4 con- tinuing short distance medially and joining Ad4 ven- trolaterally, ventrally on distal end of Cb5, there join- ing OP medially and PCI dorsoanteriorly. SOD present. SO muscle straps arise from SO in area lateral to SOD, extend anteriorly dorsal to Eb4 and attach var- iously to Pb4, Eb3, or Eb4. RDs separated by space about equal to diameter of one RD. Additional remarks. SCL absent (Bb3 with ventro- posteriorly extending cartilaginous tip). TV4 free from Cb5s. Pbl absent. Pb2 toothed. Pb4 and UP4 present. IAC absent. Eb4 levator process absent. Elassomatiformes ELASSOMATIDAE Elassoma zonatum Jordan, USNM 173343, 30.5 mm; USNM 232536, 29.9 mm. Plate 93 Description. LE1 on Eb! dorsolaterally, anterior edge of inser- tion bordering Ad1 dorsomedially. Remarks. Medial end of Eb1 articulates with Pb2 and we consider it to be the uncinate process. We consider the anterior Eb1 process, which articulates with Pb1, but never with Pb2,when present, is absent in Elassoma and the gasterosteiforms. See remarks following description of LE1 in Gasterosteus. LE2 on Eb2 dorsoposterolaterally, anterior edge of insertion bordering Ad2 dorsally, medial edge bor- dering TEb2 posterolaterally. LE3 on tip of Eb3 uncinate process. LE4 on Eb4 dorsolaterally posterior to Eb3 unci- nate process (Eb4 uncinate process absent). LP on Eb4 along posterior edge of LE4 insertion, posteriorly just dorsal to Ad4 dorsally. LI1 on anteromedial edge of Pb2 and, mainly, on adjacent anteromedial edge of Pb3 slightly anterior to OD3-—4 origin. LI2 on Pb3 dorsolaterally anteromedial to medial end of Eb3; medial edge of insertion joins lateral edge of TPb3-Eb3-Eb4. TD comprises TEb2 and TPb3-Eb3-Eb4. TEb2 very wide, mid-anterior notch leading to mid-longi- tudinal raphe that continues across anterior half of TPb3-Eb3-Eb4; attached mid-ventrally to CT of pha- ryngeal roof, giving rise dorsally to CT sheets; at- tached laterally to Eb2 dorsally at and anterior to LE2 insertion, meeting LE2 insertion anteriorly and Ad2 attachment medially; continuous posteriorly by fine diagonal muscle filament with TPb3-Eb3-Eb4. TPb3- Eb3-Eb4 attaching to Pb3 dorsolaterally at medial BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON edge of LI2, continuing onto Eb3 dorsomedialmost surface and Eb4 dorsomedially; continuous posteri- orly by diagonal muscle strand with SOD. OD3-|4 origin broadly on Pb3 dorsoanteromedially and adjacent edge of Pb2 ventral to TEb2; insertion on Eb3 uncinate process anteriorly and dorsal edge of Eb4 adjacent to Eb3 uncinate process. OP dorsally on Eb4 ventroposteromedially, ven- trally on Cb5 posterolaterally. Remarks. Although we describe Ad1 and 3 as hay- ing One or two sections, we are uncertain in those cases where only one section is present whether it is the result of variation or damage in dissection. Fur- thermore, it is possibly more accurate to describe the Ads as having one or two shapes. Ad1 variably with anterior and posterior sections; anterior section dorsally on anterolateral edge of Eb1, posterior section dorsally meeting anterior edge of LE1; sections fuse laterally and attach to anterior sur- face of distal ends of Eb1 and Cbl. Ad2 with anterior and posterior sections; anterior section dorsally extending medially and meeting an- terodistal edge of TEb2; posterior section meeting ventroanterior edge of LE2 insertion; sections fuse laterally and attach to anterior surface of distal ends of Eb2 and Cb2. Ad3 variably with anterior and posterior sections; anterior section on anterolateral edge of laterally and attach to anterolateral surface of Eb3 and Cb3; when posterior section is absent, medial attachment is to Eb3 ventroanterior to uncinate process. Ad4 dorsally on Eb4 posteriorly ventral to LP in- sertion and lateral to OP; ventrally on Cb4 dorsally medial to Eb4-Cb4 joint. Ad5 dorsally on Cb4 posterolaterally, ventrally on Cb5 dorsolaterally. SOD broad, continuous anteriorly with TPb3-Eb3- Eb4. RDs slightly separated. Additional remarks. SCL attached mid-dorsally to ventral surface of cartilaginous posterior end of Bb3. TV4 free from Cb5s. Pb1 and Pb4 absent, UP4 pres- ent. Pb2 toothed. [AC absent. Eb4 levator process absent. Mugilomorppha MUGILIDAE Agonostomus monticola (Bancroft), USNM 322336, 90.4 mm, USNM 372452, 110 mm (gill arches now cleared and stained). Plate 94 Description. LEI narrowly on cartilaginous tip of Eb1 uncinate process. LE2 on tip of dorsally expanded bony posterior NUMBER 11 edge of Eb2; ligament extends from posterior edge of insertion to anterior edge of Eb2. LE3 on cartilaginous tip of Eb3 uncinate process. LE4 narrowly on posterior edge of mid-posterior bony Eb4 process lateral to uncinate process. LP broadly on Eb4 dorsoposteriorly, beginning at LE4 insertion and extending well laterally. LI1 on Pb2 dorsoanterior process just ventral to articulation with IAC. LI2 on Pb3 dorsolaterally just anterior to medial end of Eb3. Remarks. Stiassny (1990:6—7; 1993:200—202) first proposed that the separation of the origin of LEI from those of the other LEs by the origins of LI] and LI2 is a synapomorphy of the Mugilidae and Atherinomorpha. The separation occurs in our spec- imens of Agonostomus, but it appears that it involves only LI2. TD comprises TPb2d, TPb2, TEb2, and TPb3. TPb2d, TPb2, and TEb2 originate from mid-longi- tudinal raphe which gives rise dorsally to flimsy CT sheets and is attached anteroventrally to CT of pha- ryngeal roof. TPb2d very broad, broadest anteriorly; attaches on dorsolateral half of IAC; central portion of muscle overlaps most of central portion of TEb2; laterally, beginning at medial surface of LI1 and con- tinuing posteriorly, TPb2d overlaps most of TPb2. TPb2 relatively thick, anterolaterally curving muscle on each side passing dorsal to TEb2 mid-laterally and attaching anteriorly to IAC posteromedially at joint with Pb2. TEb2 extends laterally onto dorsoanterior half of Eb2 well medial to LE2 insertion. TPb2d, TPb2, and TEb2 not connected posteriorly with TPb3, which is ventral to OD3—4; TPb3 on Pb3 dor- soposteriorly just medial to medial ends of Eb3 and Eb4. Remarks. Attachment of TPb2 (including TPb2d) primarily to IAC occurs infrequently in acantho- morphs, but most extensively, and probably homo- plastically, only in Mullidae, Moronidae, and Epi- gonidae. OD3—4 origin on Pb3 dorsomedially ventral to TEb2, insertion on dorsoanterior surface of Eb3 un- cinate process and dorsomedial edge of Eb4 uncinate process. OP dorsally on Eb4 posteriorly beginning a little medial to uncinate process and extending laterally to bony process supporting LE4 insertion, overlapping LE4 posteromedially; ventrally on bony flange ex- tending posteriorly from Cb5 posterodistally, meeting Ad5 posteriorly. Ad1 absent (small area of GFM spans Eb1-Cbl joint anteriorly). Ad2 begins medially at raphe with lateral end of TEb2, extends along anterior surface of Eb2 and spreads across Eb2-Cb2 joint. Ad3 begins slightly lateral to anteromedial end of 125 Eb3, extends laterally along anterior surface of Eb3 and spreads across Eb3-Cb3 joint. Ad4 on ventral surface of Eb4 beginning medially anterior to OP and extending laterally to near pos- terodistal end of bony edge; ventrally, attaches for equal extent along bony dorsal surface of Cb5 medial to Eb4-Cb4 joint. Ad5 dorsally beginning on cartilaginous distal end of Eb4 posteriorly and extending moderately well medially on bony surface, and on distal cartilaginous tip of Cb4 posteriorly; ventrally on distal end of Cb5 dorsoanteriorly. Remarks. It is relatively uncommon in acantho- morphs for Ad5 to insert on the bony surface of Eb4. SOD absent in smaller specimen, but present, thin and moderately broad in larger specimen. RDs proximate, insert on Pb3 posteriorly. Remarks. Harrison and Howes (1991:114) state that RD “in the majority of mugilids” they studied, which included A. monticola, inserts on Pb2. They did not mention which mugilids are exceptional or where RD inserts in them. They discussed the struc- ture and origin of RD in Agonostomus, but did not mention its insertion. They recognized that RD in- sertion on Pb2 is clearly a specialization in acantho- morphs. Harrison and Howes (1991:126) considered Ago- nostomus to be the sister group of all other mugilids, based on several characters including, among others, the slender structure of LI2 and the fact that each RD is a Single muscle, as opposed to paired in most other mugilids. The insertion of RD on Pb3 in Agonosto- mus is additional evidence of the plesiomorphic po- sition of that genus within the Mugilidae. Additional remarks. SCL free from Bb3 (posterior cartilaginous tip small, not extended posteroventral- ly). TV4 free from Cb5s. Pb4 absent, UP4 present. Pb2 toothed. Eb4 levator process absent. Both Rosen and Parenti (1981:fig. 3, AMNH 11613) and Parenti (1993:fig. 5, USNM 73742) in- dicated that Pb4 is present in Agonostomus monti- cola. Of these, the AMNH specimen is missing (B.A. Brown, e-mail, 02/06/2003), but we have seen Par- enti’s specimens, which are small, but in our opinion lack Pb4, as do our two specimens. The absence of Pb4, a specialization, albeit moderately common among acanthomorphs, is also true of atherinomorphs and might offer support for a relationship between that group and mugilids (Stiassny, 1993); however, Harrison and Howes (1991) reported the presence of Pb4 in juvenile Mugil cephalus, three species of Liza, and Chelon labrosus, but made no mention of its state in Agonostomus. We examined a few cleared- and stained juvenile specimens of specialized (have modified Ebl) mugilids (USNM 315896, unidenti- fied) and Mugil cephalus (USNM 156159), and agree 126 that they have Pb4. The absence of Pb4 in Agonos- tomus is thus a specialization. A poorly developed bony Eb4 flange is present dorsolaterally, but is easily overlooked as it only slightly overlaps the cartilaginous distal end of Eb4. It is much better developed in the larger specimen and is especially apparent when Eb4 is viewed fron- tally rather than dorsally. Atherinomorpha Atheriniformes ATHERINIDAE Menidia peninsulae (Goode and Bean), USNM 160465, 2 specimens, 88.0—95.6 mm. Plate 95A, B, D Additional material. @ = Odontesthes regia (Hum- bolt), USNM 176485, 140 mm; USNM 127863, 135 mm. BEDOTIDAE @® = Bedotia sp., USNM 301513, 48.4 mm. Plate 95C Description. LEI on Eb! at and just lateral to base of uncinate process. @ On uncinate process dorsal to base. ® On cartilage tip of uncinate process laterally, extending onto adjacent bony edge of process. LE2 on tip of expanded bony dorsal edge of Eb2, viewed laterally appears to be two longitudinally fused muscles: anterior section broader, musculously on process; posterior section tendinous ventrolater- ally, continuing posteriorly as ligament-like lateral edge of CT roofing pharynx and attaching to anterior edge of Eb3 (only CT edge shown in Plate 95.1A). @ No apparent longitudinal separation. @ Finely on tip of expanded bony dorsal edge of Eb2, no apparent longitudinal separation. LE3 slender, on Eb3 uncinate process. @) Inserts by long, slender tendon on tip of uncinate process. LE4 on tip of bony process just lateral to Eb4 un- cinate process. @ Posteroventrally joins raphe with OP dorsally and except for fine attachment antero- laterally, is free from Eb4; forms sling (sensu Stiass- ny and Jensen, 1987). LP narrowly or, usually, broadly on Eb4 bony sur- face well lateral to LE4 insertion, variably joining raphe ventrally with Ad5 or Ad5 and OP dorsally (forms sling). @ On most of bony surface of Eb4 lateral to LE4 insertion. ® Slightly separated laterally from LE4 insertion; on one side, a muscle filament continues dorsally with OP (here not considered a sling). LI1 on Pb2 just medial to attachment of IAC to BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON Pb2. ® On bony surface of Pb2 well medial to IAC attachment. LI2 on Pb3 posterolaterally and ventral to OD4 and medial end of Eb3. TD comprises TPb2a, TPb2, TPb3, and TEb4. TPb2a attaching to broad, cartilaginous anterior end of Pb2 ventrally, completely ventral to and parallel- ing anterior portion of TPb2; muscle fibers mesh dor- sally with those of TPb2 just posterior to a horizontal anterior to the separation of the two muscles. TPb2 roughly oblong, attaching to, and covering most of, broad cartilaginous anterior end of Pb2 dorsally, with irregular median raphe obscured by, and giving rise dorsally to, muscle straps changing to CT (CT not illustrated), which attaches to skull, and ventrally joining to TPb2a and ventromedian CT. CT extends posteriorly from muscle straps and covers Pb3 artic- ulating facets. TPb3 completely separated posteriorly from TPb2 and TPb2a, passing posteroventral to Pb3 dorsal articulating facets and attaching to Pb3 dor- soposteriorly medial to LI2 insertion and anterior to Pb3 articulation with Eb4 medial end, continuous by diagonal strand of muscle with TEb4. TEb4 on pos- teromedial surface of Eb4 near dorsolateral end of OP. @ ® TPb2 and TPb2a completely fused—no hor- izontal separation mid-anteriorly. OD3 absent. ® Present, see OD4. OD4 origin on Pb3 dorsomedially, beginning an- terior to flat dorsal articulating facet and ending on surface of Pb3 just ventral to facet; insertion on dor- sal surface and medial edge of Eb4 uncinate process. @ OD3-4 present, origin as in Menidia, but insertion includes anterior surfaces of Eb3 and Eb4 uncinate processes. Remarks. Eb3 and Eb4 uncinate processes are ap- pressed and bound to each other by CT, such that in Menidia and Odontesthes it appears that the anterior edge of OD4 insertion includes the medialmost edge of the Eb3 uncinate process. However, slicing through the separation of the two processes indicates that all of the insertion is on Eb4. OP dorsally broadly on posterior surface of Eb4, ventrally on posterodistal bony process of Cb5. @ OP ventrally broadly joining raphe with OP, which is mostly released from Eb4. Ad1-—3 present, each attaches along most of anter- odistal edge and surface of respective Eb and to an- terodistalmost tip of respective Cb. ® Ad1 absent. Ad4 dorsally on ventrolateral edge of Eb4, there meeting OD4 ventrally; ventrally on Cb4 dorsally immediately medial to Eb4-Cb4 joint; completely ob- scured from posterior view by Ad5 and OP. Ad5 dorsally on posterodistal surface of Eb4, ven- trally attaching to distal edge and dorsal surface of Cb5 posterodistal bony process. SOD absent. RDs proximate. NUMBER 11 Additional remarks. SCL attached mid-dorsally to posteroventral cartilaginous tip of Bb3. TV4 free from Cb5s, which are not ankylosed. Pbl and Pb4 absent, UP4 present. Pb2 toothed. Eb4 levator pro- cess absent. Cyprinodontiformes APLOCHEILIDAE Rivulus marmoratus Poey, USNM 293487, 62.7 mm. Plate 96 Description. LEI on Eb! dorsodistally (no uncinate process). LE2 on Eb2 dorsodistally. LE3 on Eb3 slightly lateral to uncinate process, insertion continues posteriorly on ligament joining Eb3 and Eb4. LE4 massive, on dorsolateral half of Eb4. LP narrowly on Eb4 at and posterior to lateral end of LE4 insertion. LI1 on Pb2 dorsoanterolateralmost surface just me- dial to IAC attachment. LI2 on Pb3 posterolaterally anterior to medial end of Eb3. TD comprises TPb2a, TPb2, TPb3, and TPb3p. TPb2a on ventroanterior surface of Pb2 and along medial margin of LI1 insertion, continuous with Pb2 along mid-longitudinal raphe, which gives rise to CT pad attaching to skull. Ventral surface of raphe atta- ches to CT sheet covering Pb3 articulating facets. TPb2 anteromedially joins longitudinal raphe and medially joins CT sheet covering articulating facets; laterally, TPb2 attaches to Pb2 dorsoanteromedial surface, dorsomedial to LI1 insertion; ventromedial- ly, TPb2 joins CT junction with OD4 dorsoanteriorly, and posteromedially joins CT junction with TPb3 an- teriorly (CT junction attaches to Pb3). Median lon- gitudinal raphe weakly represented on semicircular TPb3, which is weakly attached mid-ventrally to CT between Pb3s, but is unconnected to TPb3p. Ante- roventrally, muscle fibers of TPb3 mesh with those of OD4. TPb3p finely, tendinously attached to pos- terior edge of Pb3 slightly medial to joint with Eb4: SO fibers join tendinous attachment. OD3 absent. OD4 originates anteriorly ventral to TPb2 on dor- soanterolateral surface of Pb3 dorsal articulating fac- et (joining CT there with TPb2), and posteriorly ven- tral to TPb3 on CT covering facet, there joining with TPb3 anteriorly; inserts on broad mid-dorsal shelf on Eb4. OP dorsally very broadly on Eb4 posterior surface, ventrally on posterodistal flange of Cb5, laterally es- sentially continuous with Ad5. Ad1-—3 very broadly on respective Eb, spanning 127 respective joint with Cb anteriorly and continuing with CT supporting strip with gill rakers. Ad4 very broadly on ventrolateral surface of Eb4, and equally so on dorsal surface of Cb4, completely obscured from view posteriorly by OP and Ad5 (only visible in lateral view if obscuring CT and gill rakers are removed). Ad5 medially inseparable from OP, anterolaterally attaching to distal ends of Eb4 and Cb4, ventropos- teriorly to distal end of Cb5. SOD absent. RDs proximate. Additional remarks. SCL band-like (as opposed to more typical threadlike in other fishes), attached mid- dorsally to cartilaginous ventroposterior tip of Bb3. TV4 attached dorsally to Cb5s (continuous ventral- ly). Pb1 and Pb4 absent, UP4 present. Eb3 and Eb4 uncinate processes present. Eb4 levator process ab- sent. CYPRINODONTIDAE Cyprinodon variegatus Lacepede, USNM 107023, 3 specimens, 48.1—49.0 mm. Plate 97 Description. Remarks. Assignments of LE] and LEI’, LE2 and LE2’, and LE4 and LE4’ are arbitrary designations. LEI on Eb! just lateral to articulation with IAC, joins raphe with Ad1 along anterior margin of inser- tion. LE1’ on Eb! slightly lateral to LE1 insertion. LE2 on posterodistal end of Eb2. LE2’ laterally appressed to LE2 on posterodistal end of Eb2. Fine ligament extends from posterior edge of LE2 and LE2’ to distal end of Eb3. LE3 absent; possibly represented by LE4’; see dis- cussion in Additional remarks section. LE4 on Eb4 dorsally lateral to dorsalmost point. Remarks. LE4 and LE4’ appear to be one massive muscle before mechanical separation, but separation point was same in all three specimens. LE4’ on Eb4 at and lateral to LE4 insertion; very fine ligament connects tip of Eb3 uncinate process with LE4’ insertion; see discussion in Additional re- marks section. LP on dorsodistalmost end of Eb4. LI1 on cartilaginous dorsoanteriormost tip of Pb3; in one specimen a few muscle filaments attach to dorsoanteriormost tip of Pb2 on one side and to IAC on the other. LI2 on Pb3 dorsolaterally. TD comprises TPb2a, TPb2, and TPb3-Eb4. TPb2a on anterior surface of dorsoanteriormost Pb2 process, joined to TPb2 along mid-longitudinal ra- phe, which gives rise dorsally to CT pad covering muscles. TPb2 on posterolateral surface of Pb2 dor- 128 soanteriormost process; TPb2 and TPb2a completely, although only narrowly separate from each other on lateral surface of process. TPb3-Eb4 attachment, ten- dinous, very fine, unclear, appearing variably to at- tach at Pb3-Eb4 joint, only to Pb3, or only to Eb4. OD3 absent. OD4 originates on medial surface of Pb3 begin- ning musculously anteriorly on lateral surface of dor- soanterior process becoming CT posteriorly and at- taching to groove ventral to articulating process; in- sertion splits distally with anterior branch on Eb4 un- cinate process and posterior branch on Eb4 dorsal surface medial to LE4 insertion. OP dorsally broadly on Eb4 posterior surface, ven- trally on dorsodistal surface of Cb5; CT medial to OP covers Eb4-Pb3 joint. Ad1 with two continuous portions, dorsally on me- dial end of Eb1 at anterior edge of LE1 insertion ventrolaterally, passing over Eb1-Cb1 joint and at- taching to dorsoanterior surface of Cb1, continuous dorsally with ventral surface of crossing portion, which attaches on Eb1 medially and to gill-raker and gill-filament strip laterally. Ad2 with two continuous portions, ventral portion attaching along most of dorsal surface of Eb2 passing over Eb2-Cb2 joint and attaching to Cb2; dorsal crossing portion beginning at medialmost end of Eb2 and extending laterally and attaching to gill-raker and gill-filament strip. Ad3 with two continuous portions, ventral portion on Eb3 anteromedial surface and Eb4 ventroanterior surface, extending ventrolaterally and attaching to dorsoanterior surface of Cb3; dorsal portion attaching to most of medial arm of Eb3, extending laterally and overlapping and joining dorsal surface of ventral por- tion and attaching to gill-raker and gill-filament strip. Ad4 dorsally on ventral surface, ventrally on dor- sal surface of Cb4 medial to Eb4-Cb4 joint; obscured in posterior view. Ad5 on posterodistal end of Cb4 and anterior sur- face of Cb5 well medial to distal end. Additional remarks. SCL attached mid-dorsally to ventral surface of cartilaginous posterior end of Bb3. TV4 absent. Pb1, Pb4, and UP4 absent (or UP4 fused to Pb3, according to Parenti 1981:417, but treated by us as absent. Unless it can be shown, perhaps onto- genetically, that UP4 fuses to Pb3 in some cyprino- dontids, it is equally parsimonious to treat UP4 as absent as it is to treat it as fused to Pb3). Pb2 toothed. Eb4 levator process absent. Beloniformes ADRIANICHTHYIDAE Xenopoecilus oophorus Kottelat, USNM 340431, 2 specimens, 57.4—69.3 mm. Plate 98 BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON Additional material. @ = Oryzias latipes (Yemminck and Sclegel), 2 specimens, 29.0—29.4 mm. Description. LE1 on dorsal end of IAC, which is attached main- ly to dorsolateral end of Eb1 and less so to dorsolat- eral end of Cb1 (see also remarks following LCb2). @ IAC absent, LEI attached to lateral end of Eb] dorsally. Remarks. IAC typically joins Eb1 and Pb2 in acan- thomorphs. This is also true of IAC in atheriniforms such as Cyprinodon (Plate 97), which, similar to ad- rianichthyids, has lost the Ebl uncinate process. It seems a relatively short transition to change, for ex- ample, from the cyprinodontid state to that of the adrianichthyids (compare Rosen and Parenti, 1981: figs. 10 and 11, who recognized the states of IAC in both groups). LE2 on Eb! dorsodistally medial to LE2’, verti- cally or slightly anteriorly directed toward origin. LE2’ on Eb! dorsodistally lateral to LE2, dorso- posteriorly directed toward origin. LCb2 on dorsal end of dorsally autogenous carti- laginous end of Cb2. @ On dorsal end of Cb2. Remarks. The cartilaginous dorsoanteriormost ends of Cb2, Cb3, and Cb4 are greatly expanded in adrianichthyids and vary from continuous to autog- enous, and that of Cb4 may consist of several sepa- rate pieces of cartilage. It is also possible that ACI (see LE1 above) represents an autogenous piece of Cb1, but its position relative to the Eb1-Cb1 joint differs from those of Cb2—4. LE3 absent (see remarks following LCb5). LE4 on bony dorsoposterior surface of Eb4. LP on dorsodistal end of Eb4 and autogenous car- tilaginous end of Cb4 (not visible in Plate 98), fusing ventromedially with LCb5 laterally and with Ad5 ventrolaterally. @ On Eb4 mid-dorsally and cartilag- inous distal end of Cb3 dorsally. See remarks follow- ing LCb2. LCb5, as it extends toward Cb5, attaches antero- laterally to posterolateral surface of distal end of Eb3 (attachment released in Plate 98A, C), joins anter- oventromedialmost surface of LP on Eb4 and pos- terodistalmost surface of Cb4 just lateral to ventral attachment of Ad4, and has major portion of insertion on Cb5 dorsoanterodistalmost surface anterior to Ad5 (origin was not recorded). Remarks. Aside from these two genera, we ob- served LCb5 otherwise only in the ostariophysan cyprinids, in which the muscle inserts exclusively on Cb5. The homology of LCb5 is problematic, and the muscle possibly represents a highly modified LE3. LI1 on dorsoanterior tip of Pb2 posteriorly. LI2 broadly on Pb3 dorsoposterolaterally. TD comprises TPb2, TPb2a and TPb3. TPb2 has median longitudinal raphe and attaches to Pb2 dor- NUMBER 11 soanterior process dorsal to TPb2a. TPb2a attaches to anteroventral edges of Pb2 dorsoanterior process and is continuous posterodorsally with TPb2. TPb3 attaches to Pb3 in groove ventral to Pb3 articular surface with Eb4. OD3 absent. OD4 originates on Pb3 dorsoanteromedial surface ventral to TPb2 and inserts on bony Eb4 dorsal sur- face. OP dorsally on Eb4 posteroventral surface lateral to medial end and medial to Ad5 dorsomedially, ven- trally on Cb5 dorsodistally, only weakly separated from Ad5 laterally. M. Eb1-IAC very fine, on Eb! dorsolaterally and lateral surface of IAC ventral to LE] insertion. Ad1 absent, but long GFM present as in second arch. Ad2 broadly on Eb2 anterior surface and CT be- tween Eb2 and autogenous Cb2, and narrowly on, Cb2 dorsoanteriorly lateral to GFM dorsally. Ad3 very broadly on Eb3 anterior surface and CT between Eb2 and autogenous Cb3, and narrowly on Cb3 dorsoanteriorly lateral to GFM. Ad4 dorsally on Eb4 ventral to OD4 insertion, ventrally on Cb4 dorsodistally; muscle broad dorsal- ly, very narrow ventrally. Ad5 dorsoposteriorly on Cb4 distally and Eb4 be- ginning lateral to OP and extending to distalmost end, there joining fine raphe with ventrolateralmost edge of LP, ventrally on CbS5 distally. ® Unclear if Ad5 joins LP. SOD absent. RDs well separated. Additional remarks. SCL attached dorsomedianly to cartilaginous ventroposterior end of Bb3. TV4 free from Cb5s. A strap of SO extends anteriorly just lat- eral to RD and inserts in a deep bony pocket in Pb3 ventrolateral to LI2. Pb1, Pb4, UP4, and Eb4 levator process absent. Pb2 toothed. BELONIDAE Tylosurus crocodilus (Peron and Lesueur), USNM 128454, ca. 345 mm; USNM 137545, not mea- sured, USNM 294990, ca. 170 mm. Plate 99 Additional material. @ Strongylura timucu (Wal- baum), USNM 203575, ca. 225 mm. Description. LE1 on dorsal edge of bony process projecting me- dially from enlarged lateral end of Eb1. Remarks. Based on the configuration of the lateral end of the closely related scomberesocid, Cololabis (Plate 100), which has a cartilage-tipped uncinate process, we consider that the uncinate process of be- 129 lonids has lost the cartilage tip, hence, would not be recognized as an uncinate process. LE2 dorsally bifurcate, on posteriorly extending bony process on Eb2. @ Not bifurcate dorsally. LE3 absent (see remarks following LE4). @ On tip of all bony Eb3 uncinate process. LE4 (left side aberrant in illustrated specimen) on Eb4 more-or-less mid-dorsally continuing muscu- lously anteriorly and then inserting on CT attaching Eb4 to tip of all bony Eb3 uncinate process, insertion meeting LP insertion posteromedial edge. @ On Eb3 slightly lateral to bony uncinate process, continuing onto juxtaposed Eb4, meeting LP insertion ventro- medially. Remarks. LE4 in Tylosurus appears to have ex- tended its insertion a short distance anteriorly to in- clude the tip of bony Eb3 uncinate process. The typ- ical position of LP, at and lateral to the levator on Eb4 lends support to identification of the muscle in- serting on Eb3 as LE4. LP on Eb4 dorsodistalmost bony surface at and lateral to LE4 insertion. @ Probably present, see @ in LE4. LI1 on Pb2 dorsoposteromedially. LI2 relatively massive, on Pb3 dorsoposterolater- ally. TD complex, TDA comprising TPb2, TPb2’, TPb3, and TDP comprising TEb4. TPb2 (see remarks following this description) attaches to anterior end of Pb2 dorsoanterior process, wraps around process, and forms raphe with itself along mid-anteroventral sur- face of process; anteriorly, TPb2 extends ventrome- dially and joins a narrow raphe-like area of CT that expands posteriorly into a broad sheet that attaches to ventral surface of skull; dorsoposteriorly, TPb2 forms a flat, shallow, thin layer (barely separate from remainder of muscle), edged medially by CT, which is continuous with broad CT sheet attaching to skull; TPb2 covers most of TPb2’ and TPb3 (including dor- soanterior process of Pb3), and all three muscles join along median CT raphe-like area. TPb2’ a thin, broad band passing dorsal to TPb3 and conforming with it; laterally, TPb2’ attaches to Pb2 posterolaterally. TPb3 attaches to most of surface of Pb3 anteriorly, forming raphe posteriorly with OD4 on left side, and partial raphe on right side. TEb4 disconnected from remainder of SO, on Eb4 posteromedially, posteriorly continuous with SOD. Pb3’s dorsoposteriorly are completely ventral to muscle, no articulating facets are exposed. @ TPb2’ almost completely fused with TPb2 (TPb2’ would not be recognized as such, lacking comparison with Tylosurus); TPb3 posteriorly atta- ches by CT to dorsally exposed Pb3 articulating fac- ets, with CT continuing around facets laterally and becoming musculous OD3-—4 anteriorly. TDP com- prises TPb3-Eb3-Eb4 attaches to Pb3 a little medial 130 to medial end of LI2 continues posteriorly attaching to medial end of Eb3 and broadly along medial arm of Eb4. Remarks. TPb2 in both species is possibly a fusion of TPb2 and TPb2a, which are separate in other ath- erinomorphs. M. Pb2-Eb2 on Pb2 dorsolaterally beginning just posterior to dorsoanterior process and on Eb2 anter- omedially, laterally attaching to CT between Eb1 and Eb2. @ Muscle forks immediately anterior to its at- tachment to medial end of Eb2; arms of fork pass to either side of LI] and attach to separate areas, lateral and medial, on Pb2. OD3 absent. @ OD-3—4 origin by CT on Pb3 dor- sal articulating fact, insertion on Eb3 just medial to all bony uncinate process (very fine cartilage tip) and broadly on Eb4 dorsally medial to LE4-LP. OD4 origin dorsoanteriorly joining raphe with TPb3 posteriorly, area below raphe attaching broadly along Pb3 medially; muscle in two incompletely sep- arated parts, lateral part passes medial to bony Eb3 uncinate process and inserts on Eb4 dorsally medial to LE4-LP insertions, medial part inserts on Eb4 well medial to lateral part. Eb4 levator and uncinate pro- cesses absent (both taxa). Remarks. See similarity of insertions to those of Cyprinodon (Plate 97). OP dorsally on Eb4 posterolaterally, ventrally on Cb5 posterolaterally, fusing dorsoanteriorly with Ad4 posteroventrally. Ad] absent. Ad2 dorsally with anterior branch (GFM2?) at- taching to dorsodistal end of Eb1 and broader attach- ment to anterodistal end of Eb2, and ventrally at- taching to Cb2 dorsally. Ad3 like Ad2 but anterior branch (GFM3?) on Eb2, remainder on Eb3 and Cb3; posterior branch extends medially almost to medial end of Eb3. Ad4 attaches to Eb4 ventrally anterior to OP, pos- teroventrally joins anterior surface of OP and attaches to Cb4 dorsoanteriorly. Ad5 apparently absent. SOD present. RDs well separated. Additional remarks. SCL attached mid-dorsally by long tendon to posteroventral cartilaginous tip of Bb3. TV4 free from Cb5s. @ Dorsally attached nar- rowly to Cb5 anteriorly; ventrally free. Pb1, Pb4, and UP4 absent. Pb2 toothed. IAC absent. ACI present between Eb1l and Cblon both sides of USNM 128454 and 294990, and latter has small, questionable AC4 on both sides associated with distal end of Eb4, which is well dorsal to distal end of Cb4 (see Additional remarks in Cololabis on why this AC is not considered to be AC4); USNM 137545 lacks ACs on all arches. Strongylura timucu has AC1 on both sides, a tiny AC3 on one side, and a question- BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON able AC associated with Eb4 on one side. It also has, on both sides, a small disjunct cartilage associated with the distal (or dorsolateral) end of Cb2 and an- other with the distal end of Cb3. These two cartilages appear to be fragments of the cartilaginous distal ends of Cb2 and Cb3. The cartilages lie adjacent to the anterior bony distal ends of these two elements, each of which bears a cartilage tip posteriorly (AC3 is positioned between the cartilaginous ends of Eb3 and Cb3). Bony flange on Eb4 distally, only slightly or not at all extending over cartilaginous distal end of Eb4, hidden by LE4-LP insertions. SCOMBERESOCIDAE Cololabis saira (Brevoort), USNM 320999, 3 speci- mens, ca. 174—220 mm. Plate 100 Description. LE]! beginning on cartilaginous tip of Eb1 uncinate process and extending about half way along medially extending tendon attaching to medial end of Ebl (free from most of anterior arm of Eb1). LE2 on dorsally raised bony process on Eb2; lat- eral fibers continuous ventrally with anterior branch of Ad3. LE3 absent (see LE4). LE4 tendinously on Eb4 dorsally medial to distal end, tendinous insertion passes dorsoanteriorly and is applied closely on cartilaginous tip of Eb3 uncinate process. LP on Eb4 dorsodistally, posterolateralmost fibers ventrally variably continuous or not with OP dorsal- ly. Remarks. Although not as clearly forming an LP- OP sling (Stiassny and Jensen, 1987) as occurs in exocoetoids, the condition is, nevertheless a sling, which Stiassny and Jensen (1987:284) state is not present in scomberesocids; however, for this remark, they reference their fig. 2D, which is a belonid, which fishes, as opposed to scomberesocids, do not have a sling. LI1 on Pb2 dorsally near base of anterodorsally extending Pb2 process. LI2 on Pb3 dorsoposteriorly. TD comprises TPb2a, TPb2, TPb3a, and TPb3p. TPb2a attaches along anterolateral edge of the long dorsoanterior Pb2 process and wraps anteriorly around process, becoming continuous with TPb2 ventromedially where they join CT of pharyngeal roof. Anterolaterally, TPb2 attaches to dorsal and me- dial surfaces of long dorsoanterior process of Pb2 and posterolaterally to dorsoposterior surface of Pb2, just extending onto anteromedialmost end of Eb2; along its long ventromedial length, TPb2 joins CT of mid- line between two sides of gill arches, to which con- NUMBER 11 tralateral TPb2 also joins, and together join a sheet of CT lying dorsal to the flat, bony dorsoposterior facet on each Pb3; anteromedial extent of TPb2 com- pletely overlies TPb3a. Laterally, TPb3a attaches to much of dorsal surface of Pb3, including long dor- soanterior Pb3 process, and medially attaches (to- gether with TPb2) to CT of mid-line between two sides of gill arches; posteriorly, TPb3a continues as CT that attaches to and covers bony dorsoposterior Pb3 facet on each side. TPb3p short, joins Pb3s dor- soposteriorly, forming raphe there with SO, contin- uous posteriorly with SOD. M. Pb2-Eb2 anteriorly on ventroanterolateral sur- face of Pb2 dorsoanterior process, posteriorly on me- dial end of Eb2. OD3 absent. OD4 originating tendinously from CT covering and attaching to dorsal surface of Pb3 medial to dor- soposterior facets, passing dorsal to flat, bony facets and most of surfaces of Eb3 and Eb4 and attaching separately to Eb4 dorsoposteriorly medial to LP and anteriorly near medial end. Eb4 uncinate process ab- sent. OP dorsally on most of lateral half of Eb4 poste- riorly, overlapping Ad4 and Ad5 posteriorly and ex- cluding them completely in posterior view; postero- dorsal fibers variably continuous or not with LP pos- teroventrally; ventrally on dorsodistal end of Cb5. A splay of fibers attaches to the posterolateral surface of Cb5, passes anterior to OP ventrally and appears to join OP dorsoanteriorly in some specimens and/or Ad5 posteromedially in others. Based on the attach- ment of OP to CbS5 in belonids, these fibers appear to be OP. Ad1 absent. Ad2 with dorsoanterior branch (= GFM2?) attach- ing to lateral edge of Eb! uncinate process, continu- ing posteriorly, spanning area between first and sec- ond arches and attaching (Ad2) to ventral surface of Eb2 and dorsolateral surface of Cb2 anterior to Eb2- Cb2 joint. Ad3 like Ad2 but on Eb3 and Cb3, and anterior branch (= GFM37?) attaching to raised bony process on Eb2 to which LE2 inserts. Ad4 dorsally on dorsoposterodistalmost surface of Eb4 ventral to OP, extending medially on Eb4 ante- rior to OP, laterally joining distal ends and surfaces of Eb4 and Cb4 and fusing with dorsal end of Ad5 on Cb4. Ad5 dorsally on Cb4 posterodistally, fusing there with Ad4, extending ventrally and expanding as it attaches to posterodistal surface of Cb5. SOD present. RDs well separated. Additional remarks. SCL poorly defined; muscles crowded in region and attached to CT, which is at- tached mid-dorsally to very elongate posteroventral 131 cartilaginous tip of Bb3. TV4 ventrally free from Cb5s, but dorsally attaching narrowly, weakly to an- terior end of Cb5 (CbS5 a single bone). Pb1, Pb4, and UP4 absent. IAC absent. Two specimens have tiny ACI bi-laterally, and one of these has very fine, somewhat rod-shaped cartilage between lateral ends of Eb4 and Cb4 on both sides; the third specimen lacks ACs and the cartilage between Eb4 and Cb4. It is doubtful that the cartilage between Eb4 and Cb4 in Cololabis and belonids is homologous with AC4 of other fishes. The cartilage in the former two groups is close to the distal end of Eb4 and well separated from the distal end of Cb4. In belonids, the cartilage appears to have budded off Eb4, whereas, as far as is known, AC4 buds off the distal end of Cb4 in non-atherinomorphs. Cololabis is the only atherinomorph we examined that appears to lack an Eb4 flange. HEMIRAMPHIDAE Hemiramphus far (Forsskal), USNM 294231, 2 spec- imens, 105-111 mm. Plate 101 EXOCOETIDAE Additional material. @ = Exocoetus obtusirostris Gtinther, USNM 198465, 92.8 mm; USNM 295036, 161 mm; USNM 298987, 98.8 mm. Description. Remarks. The muscles of Exocoetus appear to be more specialized than those of Hemiramphus, al- though comparing muscles of the latter with those of the former, made it possible to identify most of them. We abbreviated descriptions of Ads, LEs, and sling of Exocoetus. LE1 tendinously on bony and cartilaginous tip of Ebl uncinate process. @ Tendinously and muscu- lously on dorsalmost tip of expanded lateral end of Eb1, continuous posteroventrally with CT joining Eb1 and Eb2 dorsodistally; uncinate process absent. LE2 fan-like or split dorsally, incompletely divis- ible into anterior and posterior sections, which fuse at insertion on bony process near distal end of Eb2. ® Similar but inserts on dorsodistal bony edge of Eb2. Remarks. Small, cartilaginously tipped uncinate process on Eb2 medial to bony process in both spec- imens (absent in Exocoetus). Among ctenosqua- mates, examined, a cartilage tipped Eb2 uncinate pro- cess is otherwise present only in some specimens of Pholidichthys (Pholidichthyidae). LE3 slender, inserts by long, slender tendon con- fluently with fine ligament joining Eb3 uncinate pro- cess with bony dorsolateral edge of Eb4 directly pos- terior (medial to Eb4 uncinate process); confluence 132 is usually on bony edge of Eb4 but may be on liga- ment mid-way between Eb3 and Eb4; musculous por- tion united with anterior surface of LE4, but easily and discretely separated. @ Musculous portion of LE3 less easily separated from LE4. LE4 massive, mostly free from Eb4, but narrowly attached medially to posterior surface of Eb4 unci- nate process at dorsolateral edge of OP; continues ventrally becoming tendinous, with OP joining ten- don dorsoanteriorly and Ad5 posteromedial surface joining tendon laterally, and together continuing ten- dinously and inserting on Cb5 dorsodistally (Ad5 at- tachment continues musculously on dorsomedial edge of Cb5); muscle variously, broadly forked ven- tral to origin. Remarks. Stiassny and Jensen (1987) first noted the LE4-OP sling of exocoetoids and its similarity to that of their labroids; however, the exocoetoid LP does not participate in the sling in the same way it does in some labroids. LP on dorsolateral surface of Eb4 and dorsopos- terior edge of Cb4, relatively distinct and separate from LE4. @ Muscle joins LE4 insertion laterally and wraps closely around LE4 posteromedially. LI1 on ventrolateral surface of elevated Pb2 an- terior process. LI2 on Pb3 dorsolaterally, posterior portion of in- sertion passing ventral to OD4 anteriorly. @ On Pb3 posterolaterally. TD complex, comprising TPb2, TPb2a, TPb2v, TPb3, and TEb4. TPb2 comprises separate muscle on each side, inserting broadly along ventral surface of cranium (muscle is considerably truncated in Plate 101); each side has broad fascia attachment on dorsal Pb2 process dorsoanteriorly and muscle generally covers TPb3, obscuring it dorsally. TPb2a completely separate from TPb2, attaches to medial edges of dor- sal Pb2 processes; muscle may represent a separation from TPb2, which wraps around Pb2 in various ath- erinomorphs (see Tylosurus, Plate 99). TPb2v super- ficially appears to be bilaterally paired muscle, but fibers fuse across ventral midline; anteriorly muscle attaches to skull, posteriorly attaches to ventroanter- ior edge of Pb2. TPb3 joins ventroanterior surfaces of Pb3 anterior processes (Pb3s are fused, but deeply cleft, ventromedially), has median longitudinal raphe. TEb4 arising on each side from lateral edge of CT sheet, passing dorsal to posterior surfaces of Pb3 apophyses, and joining opposite TEb4; muscle inserts on Eb4 uncinate process immediately posterior and essentially continuous anteriorly with OD4. SO mus- cle strap from each side joins edge of CT sheet pos- terior to TEb4 posterior edge. @ Composition same, muscles differently dis- posed; TPb3 and TEb4 are unpaired. TPb2 much less extensive, attaches dorsolaterally on Pb2 dorsoanter- ior process, extends membranously around Pb2 and BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON Pb3 anterior processes encapsulating fatty and/or glandular mass of tissue; thins and fans out posteri- orly and overlaps TPb3, extends ventromedially and very weakly attaches along midline between Pb3s and possibly dorsoposteriorly to cranium. TPb2a well developed, triangular, apex attaching to dorsomedial surface of Pb2 dorsoanterior processes, extends ven- tromedially toward midline and attaches to skull. TPb2v originates on ventrolateral surface of Pb2 dor- soanterior process and extends anterolaterally to an- teromedial end of Eb1, to which it is weakly attached by CT; posteromedially weakly attached to skull. TPb3 a narrow band of undivided muscle with mid- longitudinal raphe, situated in area between posterior ends of bases of Pb3 dorsoanterior processes and Pb3 posterodorsal articulating facets; lateral ends extend anteriorly as CT and attach to Pb3 dorsoanterior pro- cess. TEb4 interrupted at lateral edge of Pb3 dorso- posterior facet but continuous with CT across top of facets with opposite TEb4, muscle uninterrupted as it passes posterior to Pb3 posterodorsal articulating facets, attaches laterally to distal ends of Eb4s. OD3 absent. OD4 originates broadly on Pb3 dorsal surface lat- eral to anterolateral edge of Pb3 dorsoposterior facet and inserts on Eb4 uncinate process. @ Originates on Pb3 posterodorsally anterior to articulating facet and medial to LI2 insertion; inserts on Eb4 uncinate pro- cess. OP narrow muscle strap on posterolateral surface of Eb4, joining LE4 ventrally (see LE4 description, also remarks following LE4), extending ventrally and inserting on Cb5. @ Similar, but free portion dorso- laterally continuous with “‘sling.” Ad1I absent. Ad2 on entire ventral edge of Eb2 and dorsoan- terior edge of Cb2; small IAC associated with dor- soanterior edge of muscle. @ Present, IAC similar. Remarks. Rosen and Parenti (1981:fig. 12) show IAC in exocoetoids as extending between Eb1 and Eb2. The size of IAC in our specimens varied from a small sphere to a moderate-sized rod similar to that in Rosen and Parenti (1981:fig. 12A). Ad3 bipartite, anterior branch on tip of Eb2 dorsal bony process and Cb3 dorsal surface where it joins ventral portion of posterior section, which is on most of Eb3 anterior surface. @ Not bipartite (anterior por- tion absent). Ad4 on ventral edge of Eb4 and dorsoposterior edge of Cb4 medial to Eb4-Cb4 joint. @ Not clearly present; possibly fused in sling. Ad5 dorsally on Cb4 posterodistally, medial sur- face fusing with LP sling ventral tendinous end and attaching to Cb5 dorsodistally. SOD absent. RDs well separated. M. Pb3p, cone-shaped, narrowly joined to contra- NUMBER 11 lateral M. Pb3p anteroventrally at insertion on Pb3s posteroventral to Pb3 dorsal apophyses; cones are confined dorsally, but are free from, conforming con- cave bony roof formed by posterior end of skull, and are restricted ventrally by RDs and viscera; dorsally each cone has thin, traversing ribbon of muscle. Additional remarks. SCL attached mid-dorsally to cartilaginous ventroposterior tip of Bb3. TV4 inter- rupted medianly and attached to ventrolateral edge of Cb5 keel. Pb1, Pb4, and UP4 absent. Eb3 and Eb4 uncinate processes present. Eb4 levator process ab- sent. Tiny ACI present or absent, AC2 present. @ ACI present or absent, AC2 absent. Perciformes ACROPOMATIDAE Synagrops bellus (Goode and Bean), USNM 359306, 108 mm, USNM 186122, 124 mm. Plate 102 Description. LE1 with tendinous origin; insertion broadly on high Ebl uncinate process, beginning near tip and extending ventroanterolaterally almost to base. Remarks. Ligament originates on skull immediate- ly anterior to origin of LEI and inserts on Eb1 an- terior to LE] insertion; ventrolaterally, ligament is continuous with low CT sheet that attaches along al- most entire dorsolateral edge of raised anterior mar- gin of Eb1. LE2 on raised posterior margin of Eb2 at about mid-length. LE3 on tip of Eb3 uncinate process. LE4 on tip of Eb4 levator process anteriorly. LP slender, on Eb4 at and lateral to LE4 insertion. LI1 tendinously on dorsoanterior Pb2 process pos- teriorly. LI2 on Pb3 posterolaterally at joint with Eb3, me- dial edge of insertion meeting lateral edge of TPb3- Eb3. TD comprises TPb2, TEb2, and TPb3-Eb3; cov- ered dorsally by tough CT sheets, which attach along entire posterior surface of Pbls, continuing medially along anteriormost edge of Pb2s to mid-line of CT of pharyngeal roof, thence continuing posteriorly on mid-line raphe joining TPb2 and TEb2 posteriorly. TPb2 flat, thin, broadly V-shaped, with, narrow, an- terior mid-line notch deeply separating arms of V; muscle completely underlain by TEb2. TEb?2 fiat, very broad medially, narrowing considerably laterally and attaching on Eb2 dorsally anterior to LE2 inser- tion; muscle continuous posteriorly by fine muscle strands with TPb3-Eb3. TPb3-Eb3 anteriorly broadly on Pb3 dorsally ventral to OD3—4, posteriorly at- taching along medial edge of LI2 insertion and on 133 posteromedial cartilaginous end of Eb3 dorsally (passes dorsal to medialmost end of Eb4 without at- taching); muscle narrows posteriorly and is continu- ous by crossing muscle strands with SOD. OD3-—4 origin on Pb3 dorsomedially ventral to TEb2, insertion mainly on Eb3 anterior surface be- ginning just ventral to tip of uncinate process and extending medially; lesser insertion on anterior edge and surface of Eb4 just ventral to tip of uncinate process. OP dorsally on Eb4 posteriorly beginning on me- dialmost bony surface and extending laterally to or just lateral to uncinate process, laterally overlapping Ad4 dorsomedially; ventrally on Cb5 dorsoposterior- ly, beginning medially at junction of slender “horn” of Cb5 with expanded dentate portion and continuing almost to distal tip of Cb5, there joining raphe with Ad5 posteromedially. Ad1—3 absent (GFM1 well developed; GFM2 much reduced; GFM3 absent). Ad4 dorsally on Eb4 posteriorly, beginning me- dially anterior to OP and extending to posterodistal- most bony surface; ventrally on Cb4 dorsoposteriorly a short distance medial to distal end and extending laterally almost to distal end, there joining raphe with Ad5 dorsomedially. Ad5 dorsally on Cb4 posteriorly beginning ventral to medial end of Ad4 and extending laterally almost to distal end of Cb4, there joining raphe with Ad4 ventrally; ventrally on Cb5 dorsally anterior to OP, attaching for distance paralleling attachment on Cb4, joining raphe with OP ventrolaterally. SOD present. RDs moderately slender, separated by distance about equal to about half diameter of one RD. Additional remarks. SCL weakly attached mid- dorsally to elongate, posteroventrally extending car- tilaginous tip of Bb3. TV4 free from Cb5s. Pb4 and UP4 present. IAC present. Tiny AC4 present; rela- tively well developed in larger specimen. Sasaki (1989:fig. 2A) partially illustrated the dor- sal gill-arch musculature of Acropoma japonicum Giinther, which appears to differ from Synagrops bel- lus in that TPb2 is apparently broadly kidney-shaped rather than V-shaped. PERCICHTHYIDAE Macquaria colonorum (Ginther) USNM 59968, 125 mm. Plate 103 Description. Remarks. All LEs originate tendinously. LE1 on anterior surface of Eb] uncinate process ventrally. LE2 on dorsoanterior surface of Eb2 elevated bony posterior ridge. 134 LE3 on Eb3 uncinate process dorsoanteriorly with CT extending ventromedially from insertion joining OD3-4 posterolaterally on Eb4 uncinate process. LE4 on bony edge of Eb4 levator process just me- dial to cartilage tip, ventrolateral edge joining raphe with OP. LP at lateral edge of LE4 insertion, extending onto cartilage tip of Eb4 levator process. LI1 on Pb2 dorsoanteriorly; anterior edge joining raphe with TPb2 on Pb2 just medial to joint with IAC. LI2 on Pb3 dorsal surface at and lateral to TPb3- Eb3 attachment and medial to medial end of Eb3. TD comprises TPb2, TEb2, and TPb3-Eb3. TPb2 overlies TEb2, beanshaped, notched mid-anteriorly, attached to anterior cartilaginous tip of Pb2 near joint with IAC, joining raphe with ventromedial edge of LI1; muscle with mid-longitudinal raphe, anterior end of which attaches to CT of pharyngeal roof and to Pb3 anterior end dorsally; raphe giving rise dor- sally to CT mass covering muscle, ventrally joining TEb2, posteriorly continuing across middle of TEb2. TEb2 attaching laterally along medial half of Eb2 dorsal surface and lateral to LE2 insertion, continu- ous posteriorly by slender, diagonal strap of muscle with anterior end of TPb3-Eb3. TPb3-Eb3 on Pb3 medial to medial edge of LI2 insertion and dorsal surface of medial end of Eb3, continuous posteriorly by crossing muscle straps with SOD. OD3-—4 on dorsoanterior surface of Pb3, insertion on anterior surface of Eb3 uncinate process and an- terior surface and medial edge of Eb4 uncinate pro- cess. OP broadly on most of Eb4 posterior surface join- ing raphe with ventrolateral edge of LE4 insertion, ventrally, narrowly on posterodistal surface of Cb5, medially not clearly separable from SO; bilaterally asymmetrical, left side appearing to comprise two straps of muscle, right side only one. Ad1-—3 absent. Ad4 dorsolaterally on Eb4 levator process poste- rior surface lateral to OP, and medially on Eb4 ventral surface anterior to OP, ventrally on dorsoposterior surface of Cb4 medial to Eb4-Cb4 joint and anterior to Ad5. Ad5 on dorsal surface of Cb5 distally anterior to OP attachment and on Cb4 posterodistal surface pos- terior to Ad4 attachment, with dorsodistal portion of muscle continuous with tough CT (not illustrated) at- taching broadly to AC. SOD present. RDs separate, adjacent. Additional remarks. SCL attached mid-dorsally to posteroventral cartilaginous tip of Bb3. TV4 free from Cb5s. Pb4 and UP4 present. Pb2 toothed. AC4 present. BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON LEPTOBRAMIDAE Leptobrama muelleri Steindachner, WAM P.556-001, 195 mm. Plate 104 Description. LE1 shortest levator; tendinously and musculously on Eb! beginning just lateral to tip of horizontally directed uncinate process and continuing medially onto tip and on to IAC dorsolaterally. LE] and Eb1 tightly attached anteriorly to CT lining gill chamber. LE2 on tip of dorsally projecting peg-like process arising from Eb2 dorsoposteromedially; ventrolateral surface continuous with CT extending anteriorly around, and attaching to, LE] and Pbl. LE3 narrowly joining OD3-—4 just anterior to tip of Eb3 uncinate process and extending to CT cov- ering (joining) Eb3 and Eb4 uncinate processes, there joining LE4 insertion. LE4 (see also LE3) beginning on CT enveloping Eb3 and Eb4 uncinate processes and continuing short distance posteriorly to point just medial to tip of Eb4 levator process, there joining LP ventromedially. LP on Eb4 beginning at ventrolateral edge of LE4 insertion and extending to bony end of Eb4, there joining CT attaching to PP; tough fascia extends me- dially from PP, attaches to lateral edge of OP and covers OP and Ad5 posteriorly, also infiltrates SO laterally. LI1 on Pb2 dorsoanteriorly. LI2 on Pb3 posterolaterally adjacent to anterome- dialmost edge of Eb3, muscle fibers posterolaterally just failing to meet anterolateral edge of TPb3-Eb3- Eb4 muscle fibers. TD comprises TPb2, TEb2, and TPb3-Eb3-Eb4. TPb2 comprising a pair of semicircular muscle bands overlying the mid-section of TEb2; bands join raphes mid-anteriorly and mid-posteriorly and attach antero- laterally to Pbl, anteroventrally to dorsomedial edge of IAC and joint with Pb2, and also medial edge of LI1 dorsal to insertion; anterior and posterior junc- tions of muscle bands attach ventrally to CT of pha- ryngeal roof; muscle bands surround central tough CT area forming TEb2 mid-section, which is dorsal to dorsomedial surfaces of Pb3s. TEb2 consisting of anterodorsal broader section and posteroventral nar- row section joined medially by tough CT to contra- lateral muscle sections; muscle extends laterally onto Eb2 to position anterior to LE2 insertion. TPb3-Eb3- Eb4 free from TPb2-TEb2 posteriorly; beginning an- teriorly on Pb3 near ventromedial edge of LI2 inser- tion, continuing posteriorly a short distance and at- taching to posteromedialmost edge of Eb3 and to dor- somedial surface of Eb4; muscle continuous posteriorly by slender, diagonal muscle strand with SOD. OD3-—4, OD3’ originate together on Pb3 dorso- NUMBER 11 medially and branch shortly after exiting from under TEb2 into OD3—4 dorsally and OD3’ ventrally. OD3-—4 inserts on Eb3 bony surface ventral to tip of uncinate process and on posteromedial edge of bony support of Eb4 uncinate process, joining raphe with OP dorsally beginning just medial to tip of levator process and extending medially. OD3’ inserts on Eb3 dorsally ventral to uncinate process. OP dorsally on Eb4 posteriorly beginning at about mid-length of Eb4 and extending laterally to small bony process just medial to tip of levator process, joining raphe for most of its dorsal extent with OD3-— 4; laterally, OP overlaps medial half of Ad4 on Eb4; lateral edge of OP tendinous, continuous as CT sheet giving rise to PP; ventrally OP on Cb5 dorsoposter- iorly, ventrolaterally joining raphe with Ad5. Ad1-—3 absent. Ad4 dorsally on Eb4 posteriorly, beginning well medial to lateral end of OP and extending laterally almost to distal end of bone (not including cartilage), ventrally on Cb4 dorsally, extending from near inner angle of Eb4-Cb4 joint for distance equal to dorsal attachment, meeting Ad5 dorsoanteriorly on Cb4. Ad5 dorsally on AC4 medially and Cb4 beginning near posterodistal surface and extending medially a relatively short distance, meeting Ad4; ventromedi- ally joining raphe with OP ventrolaterally. SOD present. RDs slightly separated. Additional remarks. SCL present, attached mid- dorsally to tip of elongate, ventroanteriorly curving cartilaginous posterior end of Bb3. TV4 free from Cb5s. Pb4 and UP4 present. Medial end of Eb4 smaller than that of Eb3. Pb2 toothed. PCI begins at distal end of Cb5 and extends broadly medially, does join raphe with OP, but appears to join CT covering Ad5 anteriorly. LATIDAE Lates niloticus (Linnaeus), USNM 332869, 61.3 mm; USNM 332870, 123 mm, USNM 166851 (2): cleared and stained. Plate 105 CENTROPOMIDAE Additional material. @ = Centropomus undecimalis (Bloch), USNM 194201, 2:71.4—114 mm. Not illustrated Remarks. Mooi and Gill (1995:129—130) argued that the two synapomorphies Greenwood (1976) used to recognize a single family, Centropomidae, com- prising two subfamilies, Centropominae and Latinae, are invalid. They, therefore, recognized the two sub- families as families, with unresolved interrelation- ships. We find very little difference between the two families in their dorsal gill-arch musculature. 135 Description. LEI short, broadly on Eb1 uncinate process ante- riorly just ventral to tip. LE2 on raised posterior rim of Eb2 posterior to distal end of TEb2. LE3 on tip of Eb3 uncinate process anteriorly. LE4 on Eb4 dorsoposteriorly well lateral to unci- nate process, joining raphe with Ad4 dorsomedially and joined posteroventrolaterally by LP. LP on Eb4, extending medially from dorsodistal- most bony edge to, and joining, LE4 posteroventro- laterally; CT joins ventrolateral edge of LP with PP. LI1 on dorsoanteriormost surface of Pb2 immedi- ately medial to medial end of IAC and immediately lateral to anterolateralmost edge of Pb3. LI2 on Pb3 dorsoposterolaterally just medial to medial end of Eb3, meeting lateral edge of TPb3. TD comprises TPb2, TEb2, TPb3-Eb3. TPb2 flat, kidney-shaped, incurved anteriorly, divided by mid- longitudinal raphe that continues posteriorly across TEb3; raphe attaching ventroanteriorly to pharyngeal roof CT, giving rise dorsally to thin, tough CT sheets, which also attach anterolaterally to dorsoanterior sur- face on each side of TPb2; TPb2 attached anterolat- erally to Pb2 dorsoanteriormost edge and anterior edge of LI1 slightly dorsal to insertion. TEb2 almost completely underlying TPb2, with fibers of both muscles meshing on each side of mid-longitudinal raphe. TEb2 extending laterally to medial edge of raised anterior rim of Eb2, directly anterior to LE2 insertion. TPb2 and TEb2 free from TPb3-Eb3. TPb3-Eb3 attaches to Pb3 posterolaterally, meeting medial edge of LI2 insertion and attaching tendi- nously to posteromedialmost corner of Eb3. TPb3- Eb3 continuous posteriorly by diagonal muscle strands with SOD. @ TD comprises TPb2, TEb2, and TPb3. TPb2 V- shaped, arms broad, otherwise similar to Lates. OD3-4, 3’ origin on Pb3 dorsoanteromedial edge and surface, insertion on anterior surface of Eb3 un- cinate process and medial edge of Eb4 uncinate pro- cess. OD3’ present on one side in smaller specimen, but not present in larger specimen; separates ventrally from OD just ventral to origin and extends onto Eb3 dorsally to position ventral to insertion of OD3—4 on Eb3. @ OD3’ absent. OP comprises two sections, a thinner medial sec- tion and thicker lateral section; dorsally, medial sec- tion begins on Eb4 posteriorly near bony medial end and extends laterally to medial edge of lateral section, which is on Eb4 posteriorly beginning just medial to bony rise of uncinate process and extending laterally to just below LE4 insertion medially; lateral section joins raphe with OD3—4 insertion on Eb4 uncinate process. Ventrally, medial section is on Cb5 well me- dial to distal end and overlaps lateral section poster- omedially; lateral section extends laterally almost to 136 tip of Cb5, membranously overlapping Ad5 poster- oventrally. Left side of larger specimen is anomalous in that lateral section has a separate lateral branch splitting off and joining Ad5 dorsoposteriorly. @ OP lateral section meets OD3-—4 but does not join raphe with it. Ad1-3 absent. Ad4 relatively broad dorsally, be- ginning on Eb4 dorsoposteriorly ventral to LE4 in- sertion and ventrolateral to medial edge of OP thick section and continuing to Eb4-Cb4 joint; muscle at- taches on Cb4 dorsally beginning at Eb4-Cb4 joint and extending medially distance about equal to extent of dorsal attachment. Ad5 moderate, attaching dorsally on Cb4 poster- odistally near AC, and ventrally on Cb5 dorsodistally anterior to OP thick section. SOD has mid-ventral branch that separates RDs from each other. RDs adjacent. Additional remarks. SCL questionably free from cartilaginous posterior end of Bb3, which is not elon- gate, nor extends ventrally. Pb4 and UP4 present. Larger specimen has tiny AC4s and, uniquely, AC5s (latter attached to distal end of Cb5s) on both sides; smaller specimen only has AC4s. @ SCL attached mid-dorsally by short, weak ligament to ventropos- teriorly extending cartilaginous tip of Bb3. TV4 free from Cb5s. @ Larger specimen has tiny ACIs on both sides, well-developed AC4s on both sides, and tiny AC3 only on left side; smaller specimen only has AC4s, that of left side represented by two small cartilages. CENTRARCHIDAE Micropterus dolomieu Lacepede, USNM 332932, 2 specimens, 71.0—78.4 mm; USNM 332991, 78.9 mm. Plate 106 Description. LEI! on Eb! anterolateral to tip of uncinate pro- cess. LE2 on expanded dorsoposterior edge of Eb2 mid- laterally. LE3 on tip of Eb3 uncinate process anteriorly. LE4 on Eb4 just medial to minute tip of levator process. LP on Eb4 at and just anteromedial to ventrolateral edge of LE4 insertion. LI1 on dorsoanteriormost Pb2 surface, joins CT binding Pb2 and Pb3. LI2 on Pb3 dorsolaterally just anteromedial to me- dial end of Eb3. TD comprises TPb2, TEb2, and TPb3-Pb4-Eb3. TPb2 roughly heart-shaped with mid-anterior notch continuous with raphe, which widens as CT and con- tinues across TEb2; attaching anterolaterally to dor- BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON solateral surface of broad cartilaginous anterior end of Pb2 and tiny AC lying dorsal to anterior end of Pb2 (see also Additional remarks); attaching mid- ventrally to CT of pharyngeal roof; meshing poste- riorly with TEb2 anteriorly. TEb2 mostly posterior to TPb2, extending laterally and attaching to Eb2 dor- sally anterior to LE2 insertion; not continuous pos- teriorly with TPb3-Pb4-Eb3. TPb3-Pb4-Eb3 anteri- orly ventral to TEb2. Attaching to Pb3 dorsally along medial edge of LI2 insertion, continuing posteriorly and attaching to posteromedialmost edge of Eb3 and dorsomedial surface of Pb4, continuous posteriorly by fine, diagonal muscle strand with SOD. OD3-—4 origin broadly on Pb3 dorsomedially, only slightly, anteriorly ventral to TEb2; insertion mas- sively on Eb3 uncinate process anteriorly and on Eb4 dorsal edge beginning at uncinate process and ex- tending medially, with ventral fibers attaching sepa- rately to ventromedial edge of Eb4. OP dorsally broadly on Eb4 posteriorly, beginning below medial edge of LE4 and extending medially; ventrally broadly on Cb5 dorsoposteriorly; medially difficult to separate from SO. Ad1-3 absent. Ad4 dorsally on Eb4 posteriorly beginning below levator process and extending medially; ventrally on Cb4 medial to Eb4-Cb4 joint. Ad5 dorsally narrowly on Cb4 posterodistally, ventrally on Cb5 dorsally posterior to Eb4 and an- terior to OP. SOD present. RDs adjacent. Additional remarks. SCL attached mid-dorsally to tip of posteroventrally curving cartilaginous end of Bb3. TV4 free from Cb5s. Pb4 and UP4 present. Tiny AC on dorsoanteriormost tip of Pb2, not found in any other actinopterygians examined. In USNM 348876, two cleared and counterstained specimens 25.6-29.4 mm SL showed no evidence of the AC, but a third specimen, 36.8 mm SL, in the same lot, had the AC autogenous on one side and as a bud on the other. A fourth cleared and counterstained spec- imen, USNM 348877 67.6 mm, had the AC autog- enous on both sides. Enneacanthus gloriosus (Holbrook), USNM 243828, 72.8 mm; USNM 90449, 57.4 mm. Plate 107 Description. LE1 on Eb! anterolateral to tip of uncinate pro- cess. LE2 on expanded dorsoposterior edge of Eb2 mid- laterally. LE3 on tip of Eb3 uncinate process anteriorly. NUMBER 11 LE4 on Eb4 levator process anteriorly (uncinate process absent). LP beginning on Eb4 posterior to lateral edge of LE4 insertion and continuing laterally. LI1 on Pb2 dorsoanteriorly, joins CT binding Pb2 and Pb3, which lies ventral to Pb2. LI2 on Pb3 dorsally just medial to medial end of Eb3. TD comprises TEb2 and TPb3-Eb4. TEb2 very wide, interrupted mid-longitudinally by strip of CT, which gives rise dorsally to thick CT pad; attaches anteroventrally to Pb3 just posterior to insertion of LI1 on Pb3, mid-ventrally amidst SO fibers and CT of pharyngeal roof, and laterally on Eb2 dorsally lat- eral to LE2 insertion, meeting medial end of Ad2; discontinuous posteriorly with TPb3-Eb4. TPb3-Eb4 on Pb3 dorsoposterolaterally medial to medial end of Eb3, continuing posteriorly onto dorsomedial surface of Eb4; continuous posteriorly by diagonal muscle fibers with SOD. Remarks. Of all the centrarchid taxa examined (see also additional acanthomorph material section), E. gloriosus is the only one lacking TPb2. OD3-4 origin broadly on Pb3 dorsomedially ven- tral to TEb2, insertion massively on Eb3 uncinate process anteriorly and on Eb4 dorsal edge beginning at articulation with Eb3 uncinate process and extend- ing medially, with ventral fibers attaching separately to ventromedial edge of Eb4. OP dorsally on posterior surface of Eb4 medial to levator process, ventrally on Cb5 dorsolaterally, ex- tending medially anterior to SO on CbS5, ventrolateral edge of attachment finely tendinous. Ad1—3 moderately developed, on anterior surface of distal ends of respective Eb and Cb. Ad4 dorsally on Eb4 posterolaterally, beginning below levator process, ventrally narrowly on Cb4 dorsally medial to Eb4-Cb4 joint. Ad5 dorsally narrowly on Cb4 posterolaterally, ventrally narrowly on Cb5 dorsodistally. SOD slender. RDs adjacent. Additional remarks. SCL attached mid-ventrally to ventral surface of posterior tip of Bb3. TV4 com- pletely free from Cb5s or with a few dorsoanterior muscle strands attaching to anteriormost tips of Cb5s ventrally. BATHYCLUPEIDAE Bathyclupea argentea Goode and Bean, USNM 372712, 167 mm; USNM 305668, cleared and stained. Plate 108 Description. LE1 origin tendinous; insertion on Eb! anteriorly just ventrolateral to tip of uncinate process. Slender 137 ligament originates near LE! origin, expands broadly basally and inserts on Eb! anteriorly beginning an- terior to ventrolateral edge of LE1 insertion and ex- tending laterally to end of Eb1. LE2 on raised bony posterior edge at about mid- length of Eb2. LE3 on tip of Eb3 uncinate process anteriorly. LE4 beginning on tip of Eb4 levator process and extending short distance medially, joined at ventroan- terolateralmost edge by LP. LP small, short, on Eb4 at ventroanterolateralmost edge of LE4. LI1 almost entirely on dorsoposteromedial half of surface of IAC; muscle dividing longitudinally into two sections (but appear superficially as single mus- cle) with separate but adjacent tendinous insertions joining CT extending medially and joining TPb2 at- tachment to dorsalmost surface of Pb2. Remarks. The division of LI1 into two sections is apparently unusual. A non-homologous condition is duplicated in some gobioids, in which LI! complete- ly divides, with one part inserting on Pb2 and the other on Pb3. The totality of the two LI1 parts of Bathyclupea result in a muscle that is larger than LI2. LI2 from origin ventrally to level of OD3—4, mus- cle is essentially vertical and parallels LE2 closely; on reaching level of OD3—4, muscle curves sharply anteromedially and becomes almost horizontal as it passes ventral to OD3—4 to its origin on Pb3 imme- diately medial to anteromedial edge of Eb3, which lies posteroventral to Eb2 medially. TD comprises TPb2, TEb2, and TPb3-Eb3. TPb2 horizontally oblong, but with deep central anterior invagination; muscle completely dorsal to TEb2 and posteroventrally continuous with it; muscle attaches anterolaterally to Pb2 dorsoanteriorly at joint with medial end of IAC, there joining tendinous continu- ation of LI1 insertion; dorsoposterior surface of mus- cle with unilaterally extending slip, which continues tendinously with CT sheets covering dorsal surface of TD. TEb2 narrow longitudinally and relatively wide horizontally, exposed mid-dorsoanteriorly in gap exposed by TPb2 invagination; muscle with mid- longitudinal raphe attaching dorsoanteriorly to CT sheets covering TD and attaching anteriorly and mid- ventrally to CT of pharyngeal roof; muscle attaches on Eb2 beginning a little medial to LE2 insertion and continuing dorsolaterally to point anterior and a little lateral to LE2 insertion; continuous posteroventrally by fine, diagonal muscle strand with TPb3-Eb3. TPb3 originates on Pb3 beginning a little anterior to joint with medial end of Eb3 continues posteriorly along ventromedial edge of LI2 insertion and attaches to posteromedial surface of Eb3. OD3-4 origin on Pb3 dorsoanteromedially; inser- tion on medial edge and anterior bony surface just ventral to tip of uncinate process, and on medial edge 138 of Eb4 beginning just ventral to tip of uncinate pro- cess. OP dorsally on Eb4 posteriorly beginning medially about halfway to medial end of bone and extending laterally to point between uncinate process and me- dial end of LE4 insertion, failing to meet or overlap dorsomedial edge of Ad4; ventrally on Cb5 dorsally posterior to Ad5 ventrally, beginning a short distance medial to distal end of bone and extending laterally to distal end, above which OP and Ad5 join a raphe. Ad1-3 absent. Ad4 dorsally on Eb4 posteriorly, beginning me- dially ventral to levator process and extending later- ally to lateral end of bone, ventrally on Cb4 dorsally beginning laterally near Eb4-Cb4 joint and extending medially about one-fourth length of bone, joining ra- phe with Ad5 on Cb4. Ad5 dorsally on Cb4 dorsally beginning laterally at lateral end of bony portion and extending medially for distance parallel to that of Ad4, with which it forms raphe; ventrally on Cb5 dorsally anterior to OP, beginning laterally at lateral end of bony portion and extending medially a short distance; joining raphe with OP (q.v.) SOD present. RDs separated by distance equal to about half di- ameter of one RD. Additional remarks. SCL absent. TV4 free from Cb5s. Pb4 and UP4 present. Medial end of Eb3 larger than medial end of Eb4. SYMPHYSANODONTIDAE Symphysanodon berryi Anderson, USNM 370558, 128 mm; ® = USNM 204088, paratype, 85.7 mm; USNM 208500, paratype, ca. 120 mm, cleared and stained. Additional material. @ S$. octoactinus Anderson, USNM 204085, ca. 80 mm; @ S. species, USNM 371386, 86 mm. Remarks. Only the main differences exhibited by ® @ are described. Plate 109 Description. Remarks. All levators and RDs are relatively slen- der. LEI origin a fine short tendon, insertion on broad Ebl uncinate process anterolaterally just ventral to cartilage cap. LE2 on dorsalmost tip of raised posterior edge of Eb2, posterior or just posterolateral to lateral end of TEb2; lateral fibers overlap medial fibers, such that muscle can be separated artificially into two parts with separate but juxtaposed insertions. LE3 on tip of Eb3 uncinate process anteriorly. LE4 on bony dorsal edge of Eb4, slightly medial BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON to distal end, joining LP insertion medially, levator process absent. @ @ Inserts just medial to cartilage cap of levator process. LP on dorsodistalmost bony edge of Eb4, joining LE4 insertion laterally. CT extending ventrally from insertion attaches to distal cartilaginous edges of Eb4, AC4, and Cb5, and is continuous with PP; no sepa- rate Eb4 levator process in any of the three speci- mens. @ ® Inserts on levator process and joins LE4 insertion. LI1 on Pb2 dorsolaterally just ventral to cartilage cap joining IAC medially. LI2 origin a fine tendon, insertion on Pb3 dorso- posterolaterally, just medial to medial end of Eb3; medial edge of insertion joining lateral edge of TPb3- Eb3 on Pb3. TD flat, comprises TEb2, TPb2, and TPb3-Eb3. TPb2 and TEb2 with irregular mid-longitudinal raphe attaching dorsally to CT sheets (not illustrated), which are continuous anteriorly with CT of pharyn- geal roof. TPb2 dorsal to TEb2, relatively small, broadly V-shaped, arms open anteriorly, each arm arising from mid-longitudinal raphe and extending anteriorly beyond anterior margin of TEb2 and at- taching to broad, cartilaginous cap of Pb2 process that articulates with IAC medially (right-side arm de- formed). TEb2 extending laterally and attaching to Eb2 dorsally at or a little medial to LE2 insertion; posterolaterally, fine, unilateral muscle strand extends diagonally posterolaterally and becomes confluent with TPb3-Eb3 dorsally. TPb3-Eb3 on Pb3 dorsally along medial margin of LI2 insertion, extends nar- rowly posteriorly, passing dorsal to Pb4 and attaches to Eb3 posteromedially; posteriorly, diagonal muscle strap joins SOD anteriorly. @ Mid-longitudinal raphe is straight; no muscle strands join TEb2 with TPb3- Eb3. @ TPb2 transverse, not V-shaped; TPb3-Eb3- Eb4 present, with attachment broadly to Eb4 dorsally. ® TPb2 apparently absent; TEb2 broad medially, at- tenuating anteromedially with narrow extension pass- ing between dorsally naked Pb2 surfaces and attach- ing to CT of pharyngeal roof; TPb3-Eb3 present. OD3-—4 origin broadly on Pb3 dorsomedially, in- sertion on Eb3 uncinate process anteriorly ventral to cartilage tip and on medial edge of Eb4 uncinate pro- cess just ventral to cartilage tip. @ Insertion continues onto Eb4 uncinate process ventroanteriorly. ® OD3’ present. OP in two sections, dorsally on Eb4 posteriorly: medial section begins medially about mid-way be- tween medial end of Eb4 and uncinate process and extends laterally to point a little medial to uncinate process, meeting medial end of lateral section, which extends laterally to point ventral to LE4 insertion; ventrally, medial section attaches to Cb5 bony sur- face posteriorly beginning a little medial to distal end and extending medially for distance about equal to NUMBER 11 its attachment on Eb4; lateral section becomes ten- dinous ventrally, joining Ad5 posteromedially and distal end of Cb5. @ ® Not divided into two sections, present as one broad section equal to the two. Ad1-—3 absent; GFMs moderately developed, par- allel anterolateral edges of Ebs and Cbs. Ad4 broadly on Eb4 dorsoposteriorly, beginning medially ventroanterior to OP lateral section and ex- tending laterally to Eb4-Cb4 joint, ventrally on Cb4 dorsoposteriorly a short distance beginning at Eb4- Cb4 joint. Ad5 relatively small; dorsoanteriorly on AC4 pos- teriorly and Cb4 posterodistally; posteriorly on distal end of Cb5 anteriorly, posteriorly joining tendinous ventral end of OP lateral section. @ Does not appear to attach to AC4. @ AC4 absent, attaches mainly on unmodified distal end of Cb4, with tendinous contin- uation to contact between Cb4 and Eb4. © AC4 ab- sent, but distal end of Cb4 with cartilaginous poste- rior extension to which Ad5 attaches. SOD present. RDs separated by distance equal to or greater than diameter of one RD. Additional remarks. SCL attached mid-dorsally to long, ventroposteriorly extending cartilaginous pos- terior end of Bb3. TV4 free from Cb5s. Pb4 and UP4 present. AC4 present in all three specimens, but ab- sent in @ and @. We examined three additional specimens: ® = S. octoactinus Anderson, USNM 204085 (2), 69.8—81.4 mm SL, paratypes; © = S. species, USNM 371386, 86.0 mm SL. ® is similar to S$. berryi in lacking OD3’, but differs in lacking AC4 or a finger-like pro- cess extending posteriorly from the cartilaginous dis- tal end of Cb4. ® Also differs from berryi in that Eb4 has a separate levator process and the distal end of the element is a cartilaginous knob, whereas in berryi and @ there is no levator process and the car- tilaginous distal end of the element is extended dor- sally. It seems possible that the dorsal extension may be an early ontogenetic stage, and that with growth, the cartilage will ossify in its mid-portion, resulting in a separate levator process and a knob-like distal end. EPIGONIDAE Epigonus pandionis (Goode and Bean), USNM 159341, 95.2 mm, USNM 186123, 83.4 mm. Not illustrated Description. LE1 on tip of Eb1 uncinate process anterolaterally. LE2 on dorsalmost tip of raised bony posterior edge of Eb2 at about mid-length of Eb2. LE3 on tip of Eb3 uncinate process. LE4 on Eb4 dorsally just medial to tip of levator process, joining LP insertion posteriorly. 139 LP very slender, tendinously on Eb4 medial to tip of levator process, joining LE4 insertion anteriorly. LI1 on Pb2 dorsolaterally just posterior to tip of dorsoanterior process, with tendinous attachment of anteroventral edge of muscle to TPb2 where TPb2 attaches on IAC medially. LI2 on Pb3 posterolaterally, medial edge of inser- tion meeting lateral edge of TPb3-Eb3 on Pb3. TD comprises TPb2, TEb2, and TPb3-Eb3. TPb2 flat, relatively narrow, beginning slightly anterior to TEb2 mid-section and extending posteriorly dorsal to anterior half of mid-section (dorsal to all of TEb2 mid-section in smaller specimen); muscle with broad, shallow notch mid-anteriorly, notch leading into mid- longitudinal raphe extending through TPb2 and most of TEb2; ventrally raphe tightly joined to CT of pha- ryngeal roof; anterolaterally TPb2 attaches to IAC dorsally just lateral to articulation with Pb2 dorsoan- terior process, thus forming anterior half of notch with laterally extending TEb2; notch embraces me- dial edge of LI1, which has fine tendinous attachment to TPb2 near joint with IAC. TEb2 flat, relatively narrow, extending laterally and attaching on Eb2 dor- soanteriorly anterior or well lateral to LE2 insertion; TEb2 continuous mid-posteriorly by fine diagonal muscle strand with TPb3-Eb3. TPb3-Eb3 broadly on Pb3 dorsolaterally beginning posterior to joint with Eb2, and extending posteriorly along medial margin of LI2 insertion, abruptly and narrowly expanding posterior to insertion and attaching to Eb3 postero- medially; muscle continuous mid-posteriorly by broad crossing muscle strands with SOD. OD3-—4 origin on Pb3 dorsomedially, beginning a little posterior to anterior end; insertion variable: on left side, apparently abnormally, only on Eb3 broadly anteriorly beginning just ventral to tip of uncinate process; right side attachment on Eb3 similar, but in- sertion also on Eb4 beginning on medial edge just ventral to tip of uncinate process and extending a short distance anterolaterally. In smaller specimen, insertion on both sides is similar to that of right side of larger specimen (attachments to both Eb3 and Eb4). OD3’ relatively small, present only in smaller specimen; origin on Pb3 ventral to OD3—4; insertion on Eb3 dorsally ventral to, and well separated from OD3-4. OP dorsally on Eb4 posteriorly beginning about mid-way between medial end and extending laterally a little lateral to uncinate process, ventrally on Cb5 beginning moderately medial to distal end and ex- tending laterally beyond distal end onto Ad5 dorso- medially, there joining raphe with Ad5; muscle me- dially unclearly differentiated from SO laterally. Ad1-—3 absent. GFM1—3 weakly developed. Ad4 dorsally on Eb4 posteriorly beginning medi- ally below tip of levator process and extending lat- 140 erally to medial edge of cartilaginous distal end, there also attaching minimally to AC4; ventrally on Cb4 beginning medially at point ventrally opposite medial end of attachment on Eb4 (muscle ventromedially overlapped by OP) and extending laterally to medial angle of Eb4-Cb4 joint, joining raphe (not visible ex- ternally) with Ad5 anterolaterally. Ad5 very small, dorsoanteriorly on Cb4 posteri- orly a little medial to distal end, ventrally on Cb5 beginning a short distance medial to distal end and extending to distal end, dorsally joining raphe with OP ventrolaterally, and another with Ad4 ventropos- teriorly. SOD present. RDs moderate, separated by space about half di- ameter of one RD (by space almost equal to one RD in smaller specimen). Additional remarks. SCL present, but only RV4 attaches to it; appears to be fused with ventral aorta; possibly attached to elongate, ventrally curved car- tilaginous posterior end of Bb3 (removing aorta would have destroyed attachment, if present). TV4 free from Cb5s. Pb4 present. UP4 present. Pb1 most- ly bony. MORONIDAE Morone americana (Storer), USNM 61612, 99.9 mm. Plate 110 Additional material. @ = Morone mississippiensis Jordan and Eigenmann, USNM 231447, >100 mm. Description. LE1 tendinously on anterior bony surface of Eb1 uncinate process. LE2 on dorsoposterior surface of Eb2. LE3 on cartilaginous tip of Eb3 uncinate process dorsoanteriorly. LE4 on Eb4 levator process dorsodistally on one side, just proximal to dorsal tip on the other. @ Just proximal to dorsal tip on both sides. LP origin tendinous, insertion at and ventrolateral to LE4 insertion. @ Ventroanterior to LE4 insertion. LI1 on dorsoanteriormost bony surface of Pb2. LI2 on Pb3 dorsoposterolaterally, at and lateral to TPb3-Pb4-Eb3. TD comprises TPb2, TEb2, and TPb3-Pb4-Eb3. TPb2 with mid-longitudinal raphe continuing poste- riorly across TEb2, raphe attaching ventrally to CT of pharyngeal roof and dorsally giving rise to CT sheet covering TD; anteriorly, muscle attaches on most of dorsoanterior surface of IAC, posteriorly muscle is dorsal to anterior end of TEb2, to which it is fused mid-posteroventrally (no attachment to Pb2). TEb2 with dorsolaterally and ventrolaterally oriented muscle strands fusing laterally but with slightly di- BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON vided attachments on dorsal surface of Eb2, anterior attachment extending a little more laterally than pos- terior attachment, ending posteromedial to M. Pb2- Eb2 and anterior to LE2 insertion; muscle lying dor- sal to, but posteriorly connected by CT and fine di- agonal muscle strands with, anterior end of TPb3- Pb4-Eb3. TPb3-Pb4-Eb3 on Pb3 dorsoposterior bony surface at and medial to LI2 insertion, on Pb4 dor- soanteriorly, and narrowly, tendinously on Eb3 pos- teromedialmost end, continuous posteriorly by diag- onal strands of muscle with SOD. @ TPb2 on only medial third of dorsoanterior surface of IAC. M. Pb2-Eb2 on anterolateralmost bony surface of Pb2 and dorsoanterior surface of Eb2 anterior to TEb2?2. Remarks. This muscle is uncommon in acantho- morphs, but occurs at least in the ophidiids Brotula and Dicrolene, and all percopsiforms. OD3-—4 anteriorly on dorsoposterior surface of Pb3, posteriorly on anterior surface of Eb3 uncinate process and anterior surface and medial edge of Eb4 uncinate process. OP dorsally on most of Eb4 posterior surface me- dial to Ad4, ventrally, narrowly on Cb5 dorsal sur- face, joining small, strong raphe there with ventro- medial edge of Ad5; unusually well separated later- ally from Ad4; a strap of SO muscle joins OP on Cb5 but does not attach to Eb4 as it continues ante- riorly and joins SO longitudinal muscle fibers ex- tending between Pb3s. Ad1-—3 absent. Ad4 dorsally on ventral surface of Eb4 lateral to OP, ventrally on dorsoposterior surface of Cb4 medial to Eb4-Cb4 joint. Ad5 dorsally, tendinously and musculously on posterodistal surface of Cb4 and tendinously on mi- nute AC; ventrally on dorsodistal surface of Cb5. SOD present. RDs slightly separated. Additional remarks. SCL attached mid-dorsally to cartilaginous ventroposteriorly extending tip of Bb3. TV4 free from Cb5s. Tiny AC4 and AC2 present. @ AC4 relatively well developed; AC2 absent. SERRANIDAE Remarks. Imamura and Yabe (2002) hypothesized a reorganization and recomposition of the Scorpaen- iformes of previous authors. They recognized a scor- paenoid-serranoid sister group relationship within the Perciformes, but we have retained both groups sep- arately within our Perciformes pending further cor- roboration of their findings. See also remarks follow- ing Scorpaenoidei. Epinephelus merra Bloch, USNM 318074, 3 speci- mens, 68.6—92.5 mm. Plate 111 NUMBER 11 Additional material. @ = Anthias nicholsi, USNM 151904, 110 mm (very similar to E. merra). Description. LE1 with tendinous origin, insertion broadly on bony anterior surface of Eb] uncinate process. LE2 on bony dorsoposterior edge of Eb2. LE3 on dorsoanterior edge of tip of Eb3 uncinate process. LE4 on dorsoposterior edge of Eb4 medial to tip of levator process. LP slender, on Eb4 at and lateral to lateral edge of LE4 insertion. @ At and anterior to LE4 insertion. LI1 on Pb2 dorsomedially ventral to TPb2 attach- ment; slightly larger than LI2. @ On dorsal surface of Pb2 dorsoanteriorly just ventral to joint with IAC. LI2 on Pb3 dorsolaterally at anteriormedialmost end of Eb3. TD comprises TPb2, TEb2, and TPb3-Eb3. TPb2 heart-shaped, deep notch anteriorly leading to mid- longitudinal raphe, which gives rise to dorsally to CT sheets attaching to skull; on broad Pb2 dorsoanterior process and ventroposterior end of Pb1 (Pb2 and Pb1 tightly joined by CT at this point), continuous pos- teroventrolaterally with TEb2. TEb2 on Eb2 anterior to LE2 insertion, continuous posteriorly by diagonal strand of muscle with TPb3-Eb3. TPb3-Eb3 on Pb3 dorsal surface medial to LI2 insertion and anterior to Eb3 cartilaginous medial end, and on Eb3 bony sur- face dorsoposteromedially, continuous by diagonal strands of muscle with SOD. @ TPb2 does not attach to Pbl. OD3-—4 originates on Pb3 dorsomedially ventral to TEb2 and inserts on anterior surface of Eb3 uncinate process and anterior surface and medial edge of Eb4 uncinate process. OP dorsally, broadly on Eb4 posteromedially be- ginning near levator process and extending medially; ventrally on Cb5 dorsolaterally posterior and coin- cident with Ad5, medially weakly separable from SO. Ad1-—3 absent. Ad4 dorsally, broadly on Eb4 posterior surface lat- eral to OP; ventrally broadly on Cb4 dorsally medial to Eb4-Cb4 joint. Ad5 on posterodistal surface of Cb4 and dorsolat- eral surface of Cb5 anterior and coincident with OP. SOD with mid-longitudinal raphe, muscle fibers extend ventrally from raphe forming a short com- partment on each side that isolates each RD. RDs separate. Additional remarks. SCL absent. TV4 free from Cb5s. Pb4 and UP4 present. LUTJANIDAE Pristipomoides aquilonaris (Goode and Bean), USNM 158472, 101 mm. 141 Plate 112 Additional material. @ = Hoplopagrus guentherii, USNM 65544, 94.6 mm. Description. Remarks. Hoplopagrus gill-arch muscles are very similar to those of Pristipomoides; however, they are much more massive relative to the size of the spec- imen. LE1 finely, tendinously on dorsal edge of Eb1 un- cinate process just lateral to process tip. @ Tendi- nously on and lateral to tip of uncinate process. LE2 on dorsal edge of raised posterior edge of Eb2. LE3 on tip of Eb3 uncinate process anteriorly. LE4 on Eb4 levator process dorsoanteriorly. LP at and ventroanterior to LE4 insertion. LI1 on Pb2 dorsal surface ventral to articulation with IAC. LI2 on Pb3 dorsolaterally medial to anteromedial end of Eb3. TD comprises TPb2, TEb2, and TPb3-Pb4-Eb3. TPb2 is dorsal to TEb2, with median longitudinal raphe giving rise to CT sheets attaching to skull; muscle attaches anterolaterally to CT binding dor- soanteriormost cartilaginous ends of Pb2 and Pb3 and dorsomedial surface of IAC, attaches mid-ventrally to CT of pharyngeal roof, fuses mid- and ventrolat- erally with TEb2, and is posteriorly free from TPb3- Pb4-Eb3. TEb2 twisting dorsoposteriorly after pass- ing laterally from under TEb2, attaching on Eb2 dor- soanteriorly anterior to LI2 insertion. TPb3-TPb4- Eb3 on Pb3 dorsolaterally medial to medial end of Eb3, continuing posterolaterally and passing dorsal to medial end of Eb4 and attaching to posteromedial edge of Eb3; muscle fibers delaminate from ventral surface of Eb3 portion and attach to dorsal surface of Pb4; posteriorly continuous by fine muscle strands with SOD. @ Comprises TPb2, TEb2, and TEb3. TEb3 on medial end of Eb3 dorsoposteriorly. TEb2 does not twist laterally. OD3-—4 origin broadly on Pb3 dorsomedially ven- tral to TEb2, insertion mostly on anterior surface of Eb3 uncinate process, weakly on anterior surface of Eb4 uncinate process. OP dorsally on Eb4 posteriorly beginning near me- dial end and extending laterally almost to levator pro- cess; ventrally broadly on Cb5 dorsoposteriorly, pos- terior to Ad5. Ad1-—3 absent. Ad4 dorsally on Eb4 posteriorly beginning at le- vator process and extending laterally to end of bony surface; ventrally broadly on Cb4 extending laterally to Eb4-Cb4 joint. Ad5 ventrally broadly on Cb5 dorsally beginning at distal end of Cb5, joining small raphe with OP 142 ventroanterolaterally; dorsally on posterodistal end of Cb4. SOD present. RDs separated by space about equal to diameter of one RD. Additional remarks. SCL attached mid-posteriorly to very elongate cartilaginous ventroposterior tip of Bb3. TV4 free from Cb5s. Pb4 and UP4 present. HAEMULIDAE Pomadasys crocro (Cuvier), USNM 338643, 2 spec- imens, ca. 68—78.1 mm. Additional material. Haemulon scudderi Gill, USNM 181299, 70.9 mm; Plectorhinchus pictus (Thun- berg), USNM 192546, 83.0 mm. Plate 113 INERMIIDAE *@ = Inermia vittata Poey, USNM 318643, ca. 80 mm SL. Not illustrated Remarks. We found no notable differences in the muscles among the three haemulid taxa. Inermiid muscles are very similar to those of haemulids, to which family they are closely related (Johnson, 1980). Additional evidence for this relationship is that the cartilage tip of the Eb1 uncinate process is attached to the medial edge of a small, raised bony flange, rather than to a distinct, rod-like arm. A some- what similar arrangement is present in centracanthids and some sparids, e.g., Acanthopagrus. Aside from the fact that inermiid levators are less robust then those of haemulids, only the main differences are not- ed below. The inermiids and haemulids share in hay- ing the uncommon (for percomorphs) combination of absence of both SCL and SOD. Description (based on Pomadasys). LE] origin slender, tendinous; insertion broad, on Eb! beginning at joint of uncinate process with IAC and extending laterally to point almost half distance to distal end. LE2 on medial edge of raised posterior margin of Eb2. LE3 on Eb3 just anteroventral or just anterolateral to tip of uncinate process, meeting OD3—4 on Eb3. @ Inserts by long, fine tendon on tip of Eb3 uncinate process on one side, and by low, broad tendon on other. LE4 on Eb4 levator process dorsally, just posterior to uncinate process, meeting but not joining LP me- dially. @ Inserts by long, fine tendon on tip of Eb4 levator process on one side, and by low, broad tendon on other. LP broadly on Eb4 dorsally, extending medially from distal end to position just lateral to LE4 inser- BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON tion immediately anterior to tip of levator process, joining raphe with Ad4 dorsally. LI1 tendinously on Pb2 dorsoanteriorrmost tip posteriorly, extending laterally onto adjacent IAC posteromedially (not visible in illustration) up to about one-third length of IAC; tendinous attachment to Pb2 continuous with CT joining Pb2 with anter- iormost tip of Pb3 dorsally (insertion here not con- sidered to include Pb3). @ Does not insert on IAC. LI2 tendinously on Pb3 dorsolaterally immediately anterior to anteromedialmost edge of Eb3, well sep- arated from TPb3-Eb3. TD comprises TEb2 and TPb3-Pb4-Eb3. TEb2 ro- bust, with broad, irregular central CT portion at- tached mid-ventroanteriorly to CT of pharyngeal roof and giving rise mid-dorsally to CT sheets attaching to skull; muscle extends laterally and attaches on Eb2 dorsally anterior to LE2 insertion, also meeting me- dial end of GFM2 (not illustrated) as latter extends dorsally on Eb2; muscle completely disjunct from TPb3-Pb4-Eb3. TPb3-Pb4-Eb3 beginning on Pb3 dorsolaterally medial to mid-medial end of Eb3 and continuing onto posteromedialmost corner of Eb3 and dorsomedial edge of Pb4. OD3—4, OD3’ with joint origin on Pb3 dorsopos- teromedially; relatively large OD3’ splits off ven- troanteriorly shortly lateral to origin and inserts on Eb3 dorsally ventrolateral to OD3—4 portion on Eb3; OD3-4 inserts massively on Eb3 uncinate process anteriorly and on Eb4 anteriorly just ventral to tip of uncinate process. OP dorsally on Eb4 posteriorly beginning near me- dialmost bony end and extending laterally to medial edge of levator process, overlapping medial portion of Ad4 posteriorly; ventrally on Cb5 dorsoposteriorly beginning medially near TV5 and extending laterally almost to distal end, but becoming membranous ven- trolaterally and overlapping Ad5 posteriorly; ventro- medially fusing with SO. Ad1-—3 absent. GFM 1-3 (not illustrated) moderate- ly developed; GFM2 extending anteromedially, then dorsally on Eb2 to position anterior to distal end of TEDb2?2. Ad4 dorsally on Eb4 posteroventrally beginning anterior to OP near bony medial end and extending laterally to distalmost bony end; ventrally, moderate- ly broadly on Cb4 laterally, becoming fused poste- riorly with Ad5 anteromedially. Ad5 dorsally on AC4 posteriorly and Cb4 poster- odistally for short distance, there anteriorly joining Ad4 posteriorly; ventrally on Cb5 dorsoposterolater- ally, mostly anterior to OP. SOD absent. RDs adjacent. @ Separated by distance equal to diameter of one RD. Additional remarks. SCL absent; cartilaginous pos- terior end of Bb3 not elongate. TV4 free from Cb5s. NUMBER 11 Pb4 and UP4 present. Pb! with cartilaginous ends. Pb2 with teeth. AC4 present. Dorsoanterodistalmost bony margin of Eb4 and, variously, Eb2 and Eb3, with small flange in Haemulon and Plectorhinchus; flanges very reduced in Pomadasys, very well de- veloped in a cleared and stained specimen (USNM 214488, 95 mm) of Parakuhlia macropthalmus (Os6- rio). @ Flange clearly present, but so fine as to be easily overlooked. APOGONIDAE Glossamia wichmanni (Weber), USNM 344886, 2 specimens, 67.3—97.7 mm. Plate 114 (based on smaller specimen) Additional material. @ = Cheilodipterus macrodon Lacepede, USNM 276623, 82.6 mm SL. Remarks. Differs from G. wichmanni mainly in having musculature much less robust. Description. LE1 on expanded dorsoposterior edge of Eb1 lat- eral to tip of uncinate process. LE2 on expanded dorsoposterior edge of Eb2 about mid-laterally. LE3 on tip of Eb3 uncinate process anteriorly. LE4 on expanded dorsoposterior surface of Eb4 lateral to tip of uncinate process. LP on Eb4 at and lateral to lateral edge of LE4. Remarks. Narrow, lateralmost cartilaginous edge of expanded dorsoposterior surface of left-side Eb4 narrowly continuous ventrally with knob-like carti- laginous distal end of Eb4. Cartilaginous lateral end continuous ventrally with knob-like distal end of Eb4 on both sides in larger specimen and @. LI1 on dorsoanterolateralmost surface of Pb2 just medial to articulation with IAC. LI2 on Pb3 dorsoposterolaterally just anteromedial to medial end of Eb3. TD comprising TPb2, TEb2, and TPb3-Eb3, with mid-longitudinal raphe extending through all com- ponents and SOD, giving rise dorsally to CT sheet covering surface of muscles, and attaching ventrally to CT of pharyngeal roof between Pb3s. TPb2 very thin, flat, bilobed, arising dorsally from mid-longi- tudinal raphe in common with TEb2 posteriorly; an- teriorly joining raphe with TEb2 anterolaterally and at that point weakly attaching to cartilaginous ante- rior end of Pb2 just medial to LI1. TEb2 very broad, attaching on Eb2 dorsally anterior to LE2 insertion, continuous posteroventrally only by CT with TPb3- Eb3. TPb3-Eb3 on Pb3 dorsoposterolaterally begin- ning along medial edge of LI2 insertion and continu- ing posteriorly, crossing over medial end of Eb4 and attaching along posteromedial edge of Eb3; narrowly continuous mid-posteriorly with SOD. OD3-4 origin broadly on Pb3 dorsally ventral to 143 TEb2; insertion broadly on anterior surface of Eb3 uncinate process and medial edge of Eb4 uncinate process. OP dorsally broadly on posterior surface of Eb4 beginning ventral to LE4 and extending medially; ventrally on Cb5 dorsoposteriorly, joining small ra- phe ventrolaterally with Ad5. Ad1-—3 absent. Ad4 dorsally broadly on Eb4 posteroventrolater- ally, extending medially just under (anterior to) lat- eral edge of OP; ventrally broadly on Cb4 dorsally medial to Eb4-Cb4 joint, meeting Ad5 attachment on Cb4. Ad5 dorsally on Cb4 posterolaterally; ventrally on Cb5 dorsally mostly anterior to OP, joining raphe posteroventrally with ventromedial edge of OP. SOD very fine. RDs separated by space less than half diameter of one RD. Additional remarks. SCL present. TV4 free from Cb5s. Pb4 and UP4 present. Pb2 toothed. Eb4 levator process present on only one side of smaller specimen, absent on both sides of larger specimen. @ Eb4 le- vator process present on one side, absent on other. PRIACANTHIDAE Heteropriacanthus cruentatus (Lacepede), USNM 319919, 63.3 mm; @ = USNM 141752, 88.0 mm. Plate 115 Description. LEI on dorsoposterior edge of Eb! well lateral to uncinate process, but not extending onto distal quar- ter of surface. LE2 on mid-dorsoposterior edge of Eb2, insertion continuous posteriorly with ligament attaching Eb2 to mid-anterior edge of Eb3. LE3 relatively small, on joined tips of Eb3 and Eb4 uncinate processes. LE4 on dorsal edge of Eb4 levator process, inser- tion continuous ventrolaterally with ventromedial edge of CT sheet joining PP ventrally. LP finely tendinously on Eb4 at lateralmost edge of LE4 insertion. LI1 on dorsomedialmost end of Pb2. LI2 on Pb3 dorsolaterally at articulation with me- dial end of Eb3. TD comprises TEb2 and TPb3-Eb4. TEb2 broad, wing-like, depressed centrally, with lateral margin of depression on each side represented by a surface ra- phe that gives rise to thick CT sheet attaching to skull; muscle attached antero- and ventromedianly between Pb3s to CT of pharyngeal roof; fibers of anterior two-thirds of muscle curve posterolaterally, those of posterior third more-or-less transverse, fibers fuse laterally as muscle narrows and extends onto Eb2, attaching to Eb2 dorsally anterolateral to LE2 144 insertion; muscle is not connected posteriorly with TPb3-Eb4. TPb3-Eb4 on Pb3 dorsoposterolaterally, extending broadly posteriorly onto medial arm of Eb4 anteriorly, attached mid-anteroventrally to CT of pharyngeal roof. OD3-—4, OD4v origin broadly on Pb3 dorsomedi- ally ventral to TD, insertion on anterior surface of Eb3 uncinate process and medial edge of Eb4 unci- nate process, with short, separate branch (OD4v, not illustrated) attaching broadly dorsally on Eb4 ventro- medial to uncinate process. Remarks. OD4v probably autapomorphic. OP dorsally on Eb4 posteriorly, beginning on pos- terior surface of uncinate process and extending me- dially and becoming undifferentiated from SO, ven- trally on Cb5 dorsoposterolaterally, ventrolaterally joining raphe with Ad5 posteroventrally. Ad1-—3 absent. Ad4 dorsally broadly on Eb4 beginning on ventral surface of levator process and extending laterally on Eb4 posteriorly, ventrally on Cb4 dorsolaterally me- dial to Eb4-Cb4 joint. Ad5 dorsally on Eb4 posterodistally, including tiny AC4 attached to posterodistal end of Cb4 (present on right and left arches), ventrally on Cb5 dorsally an- terior to OP, joining raphe with OP ventrolaterally. @ AC absent. Remarks. Presence of AC4 is variable, and possi- bly ontogenetically correlated. In a cleared and stained specimen 58.9 mm (USNM 337685) it is present on one side but not the other. SOD absent. RDs separated by space less than one RD diameter. Additional remarks. SCL absent. TV4 free from Cb5s. Pb4 and UP4 present. IAC reduced to tiny car- tilage adjacent to tip of Ebl uncinate process, at- tached by long, slender ligament to tip of Pb2 unci- nate process. OSTRACOBERYCIDAE Ostracoberyx dorygenys Fowler, USNM 307282, 80.0 mm. Plate 116 Description. LE1 broadly on dorsoposterior edge of Eb1 begin- ning on ventrolateral edge of long uncinate process and extending laterally; origin by long tendon. LE2 very broadly on expanded posterior surface of Eb2, ventromedial edge of insertion meeting TEb2 posteriorly. LE3 on joined tips of Eb3 and Eb4 uncinate pro- cesses anteriorly, musculously on Eb3, tendinously on Eb4. LE4 broadly on Eb4 dorsolaterally, ventroposter- omedially joining raphe with Ad4 dorsomedially BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON (present on both sides and can be characterized as an LE4-Ad4 muscle sling). LP very slender, tendinously on dorsodistalmost edge of Eb4, joining ventrolateral edge of LE4 in- sertion. LI1 on dorsalmost edge of Pb2, ventromedial edge of insertion joining raphe with TPb2 anterolaterally; muscle about same size as LI2. LI2 on Pb3 dorsoposteriorly at medial end of Eb3. TD comprises TPb2, TEb2, and TPb3-Eb3. TPb2 a thick, roundish pad broadly continuous ventrally with TEb2; attaching anteriorly to CT of pharyngeal roof; attaching tendinously anterolaterally to dorsal edge of Pb2 and adjacent anterior tip of Pb3 (not illustrated); with broad mid-dorsal raphe-like area giving rise to tough, filmy CT sheets covering TD. TEb2 extending laterally onto Eb2 anterior to LE2 insertion, continuous mid-ventroposteriorly by diag- onal muscle strands with TPo3-Eb3. TPb3-Eb3 on Pb3 dorsoposterolaterally near medial edge of LI2 insertion, rising dorsally onto Eb3 beginning on pos- teromedialmost edge and extending a short distance along posteriorly, continuous posteriorly with slender SOD. OD3-—4 origin on Pb3 dorsoanterolaterally ventral to TEb2, insertion broadly on Eb3 uncinate process anteriorly and medial edge of Eb4 uncinate process, joining narrow raphe posteroventrally with OP a little medial to Eb4 uncinate process. OP dorsally broadly on Eb4 posteriorly beginning at uncinate process, extending medially, and becom- ing inseparable from SO, ventrally on Cb5 postero- laterally, joining short raphe with Ad5 ventrally. Ad1-3 absent. Ad4 on Eb4 dorsoposterolaterally and Cb4 dor- solaterally medial to Eb4-Cb4 joint, joining raphe with Ad5 dorsoanteriorly on Cb4. Ad5 dorsally beginning tendinously on Cb4 pos- terodistally and extending medially on Cb4 dorsally, joining raphe with Ad4 ventrally; ventrally, narrowly on Cb5 dorsodistally, joining raphe with OP ventral- ly. SOD moderately slender. RDs separated by space equal to about one RD diameter. Additional remarks. SCL attached mid-posteriorly to anterior surface of posteroventrally extending car- tilage tip of Bb3. TV4 free from Cb5s. Pb4 and UP4 present. IAC present. Pb2 toothed. Eb4 levator pro- cess absent. CIRRHITIDAE Paracirrhites forsteri (Schneider), USNM 278070, 91.3 mm; USNM 339098, 67.2 mm. Plate 117 NUMBER 11 Additional material. @ = Cirrhitus pinnulatus (For- ster), USNM 296746, 65.0 mm. Description. LE! broadly on Eb1 uncinate process anteriorly just lateral to cartilaginous tip; origin narrowly, ten- dinously united with posterior surface of epithelial sheet fronting gill arches dorsoanteriorly, separating them from pseudobranchs, also incorporates Pbls. Strong ligament inserts on Eb! anterior to LE! and extends to cranium. @ Ligament and epithelial sheet possibly removed without notation (early dissection). LE2 on Eb2 at about mid-length dorsoposteriorly. LE3 finely on tip of Eb3 uncinate process anteri- orly, meeting insertion of OD3—4 on Eb3. LE4 on Eb4 dorsally beginning on tip of levator process and extending medially a short distance, joined laterally or anterolaterally by LP insertion. LP relatively small, on Eb4 joining LE4 insertion laterally or anterolaterally, insertion continuous ven- trolaterally with CT sheet (not illustrated) attaching to fourth and fifth arches and incorporating PP ven- trally. LI1 broadly on Pb2 dorsally beginning just pos- teroventral to dorsal tip; medial edge of muscle, dor- sal to insertion, attaches to CT joining TPb2 anteri- orly to Pb2 and posteromedial edge of IAC. LI2 on Pb3 broadly posterolaterally opposite me- dial edge of Eb3, medial edge of insertion meeting TPb3 laterally. TD comprising TPb2, TEb2, and TPb3. TPb2, large, flat, circular, almost completely covering mid- section of TEb2, with shallow notch mid-anteriorly and mid-posteriorly connected by mid-longitudinal raphe, which gives rise dorsally to thin sheets of CT and is joined mid-ventrally by TEb2; muscle attached anterolaterally by CT to closely approximated Pbl1- Eb1 joint, [AC-Pb2 joint, and medial edge of LI1; attached mid-anteriorly and mid-ventroposteriorly to CT of pharyngeal roof. TEb2 medially ventral to TPb2, attaching along mid-longitudinal raphe ven- trally, extending laterally and attaching on Eb2 dor- sally a little anterolateral to LE2 insertion; posteriorly discontinuous from TPb3. TPb3 on Pb3 mostly ven- tral to OD3—4, attaching on Pb3 posterolaterally meeting medial edge of LI2 insertion, continuous by diagonal muscle strand with SOD. @ TPb2 on TEb2 dorsally; a flat, semicircular ribbon, open anteriorly, originating anteriorly on each side at CT attachment to closely approximated Pbl-Eb1 joint, [AC-Pb2 joint, and medial edge of LI]. OD3-—4 origin broadly on Pb3 medially ventral to TEb?2, insertion on anterior surfaces of Eb3 and Eb4 uncinate processes. OP dorsally, broadly on Eb4 beginning at medial end of bony portion and extending laterally almost to levator process, there meeting dorsomedial end of 145 Ad4; ventrally on Cb5 posterolaterally, variably join- ing or not short raphe with ventroposterior end of Ad5 and more extensive one with PCI. M. SO-Pb3 originates from SO longitudinal mus- cle layer as two broad, anteriorly extending muscle straps, lateral strap inserts on Pb3 posteriorly, medial strap (not visible in illustration) originates ventral or ventromedial to RD and inserts on Pb3 posterome- dially; RD inserts on Pb3 between M. SO-Pb3 pair. Ad1-—3 absent, but very fine muscle strand ob- scured by bases of gill-filaments on each arch an- terolaterally. Ad4 dorsally on Eb4 ventrally, beginning below levator process and extending to Eb4-Cb4 joint, ven- trally, broadly on Cb4 dorsally medial to Eb4-Cb4 joint. Ad5 dorsally, moderately broadly on posterolateral surface of Cb4, ventrally on Cb5 posterolaterally, variably joining or not OP posterolaterally in a short raphe. SOD present. RDs moderately separated or adjacent, insert by long tendon on Pb3 posteriorly between insertions of lateral and medial M. SO-Pb3 pair, except on one side of one specimen, in which RD tendon inserts into CT of pharyngeal roof posterior to Pb3. Additional remarks. SCL attached mid-dorsally to posteroventrally elongated cartilaginous end of Bb3. TV4 free from Cb5s. Pb4, UP4, and IAC present. Pbls oriented perpendicular to distal ends of Eb1s; @ Pbls oriented parallel to Eb1s. PEMPHERIDAE Pempheris schomburgkii Miller and Troschel, USNM 318588, 95.5 mm; Pempheris schwenkii Bleeker, USNM 324224, 105 mm. Plate 118 Additional material. @ = Parapriacanthus ranson- neti Steindachner, USNM 344281, 79.2 mm. GLAUCOSOMATIDAE @ = Glaucosoma magnificum, WAM P.14208-011, 93.2 mm. Not illustrated Remarks. We noted no substantive differences among the three genera. Description (based on Pempheris schomburgkii). LE1 relatively slender, on dorsal edge of Eb1 un- cinate process just lateral to cartilaginous tip; strong, fine ligament (not illustrated) extends dorsally to ven- tral surface of the skull from margin of Eb1 anterior to uncinate process; ligament is appressed to poste- rior surface of vertical CT pane lining pharyngeal area anterior to gill arches. 146 LE2 on rounded high point of dorsoposterior edge of Eb2 mid-laterally. LE3 on dorsal tip of Eb3 uncinate process. LE4 on dorsodistal edge of Eb4, joining LP ante- roventrally; levator process absent. @ LE4 and LP on levator process. ® Levator process present. LP on Eb4 at and anterior to LE4 insertion. @ ® See LE4 above. LI1 has long, slender tendinous insertion on Pb2 ventral to joint with IAC. LI2 short tendinous insertion on Pb3 dorsoposter- olateral surface at and just touching medial end of Eb3. TD comprises TPb2, TEb2, and TPb3-Eb3. TPb2, thick, pad-like, oblong, with irregular mid-longitu- dinal raphe giving rise to CT sheet covering muscle on anterior half; muscle attaches beginning anteriorly at CT of pharyngeal roof, continuing laterally over and attaching to anterior end of Pb3, dorsomedial surface of broad Pb2 uncinate process, and dorso- medial surface of IAC, remainder of muscle margin free, but muscle fusing ventromedially with TEb2; amorphous CT connects mid-posteroventral surface to mid-anterior edge of TPb3-Eb3. TEb2 extending out on dorsal surface of Eb2 to point anterolateral to LE2 insertion. TPb3-Eb3 on Pb3 posterolaterally, continuing on Eb3 posteromedial edge, continuous ventrally by crossing strands of muscle with SOD, which is ventral to TPb3-Eb3. @ TPb2 flat, attach- ment on IAC extends to about mid-length of element. @® Attachment extends laterally only to dorsolateral- most edge of Pb2 (i.e., not on IAC). OD3-—4 origin on Pb3 ventral to TPb2 and TEb2, inserting on anterior surface of Eb3 uncinate process and medial edge of Eb4 uncinate process. OP dorsally on Eb4 mid-dorsoposteriorly begin- ning at uncinate process and continuing medially about half distance to medial end of Eb4, extending ventrally posterior to dorsomedial edge of Ad4 and posteroventral surface of Ad5 and inserting on pos- terodistal surface of Cb5. Ad1-—3 absent. Ad4 dorsally on ventral surface of Eb4 mostly lat- eral to OP, and ventrally, broadly on Cb4 dorsal sur- face medial to Eb4-Cb4 joint. Ad5 ventrally on dorsal surface of distal end of Cb5 anterior to OP insertion and broadly on Cb4 ven- troposterior surface and small AC4 posteriorly. SOD slender, ventral to TPb3-Eb3; moderately well developed, just ventroposterior to TPb3-Eb3 in P. schwenkii. @ Well-developed, well posterior to TPb3-Eb3, but continuous with it by slender, diago- nal muscle strap. ® Very fine, ventral to TPb3-Eb3. RDs adjacent. Additional remarks. SCL present, attached mid- dorsally by long, loose CT to tip of elongate, ventro- posteriorly extending cartilaginous posterior end of BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON Bb3. TV4 free from Cb5s. @ SCL apparently free from short, cartilaginous posterior end of Bb3 (should be verified in additional specimens). Pb4 and UP4 present. Pb2 toothed. Johnson (1993:19), following Tominaga (1986), recommended recognizing the Glaucosomatidae as a subfamily of the Pempheridae and described addi- tional specializations shared by the two groups. Nel- son (1994:365—366), McKay (1997:5), and Mooi and Gill (2002:23), without explanation, all chose not to follow Johnson’s recommendation. McKay (1997:5) proposed that the Pempheridae and Glaucosomatidae are sister groups. To the specializations shared by pempherids and glaucosomatids previously reported in the literature, we add the ligament positioned anterior to LEI, a general similarity of the dorsal gill-arch musculature, and the shape of Eb4, which is posteriorly concave. LACTARIIDAE Lactarius lactarius (Bloch and Schneider), USNM 343873, 119 mm. Plate 119 Description. LE1 on Eb! uncinate process ventrolateral to tip; origin tendinous; long, broad-based ligament on Eb1 anterior to uncinate process (also in Glaucosoma— see additional remarks under Pempheridae). LE2 on dorsally expanded bony posterior margin of Eb2. LE3 on tip of Eb3 uncinate process medially. LE4 on Eb4 levator process beginning at ventrally positioned cartilage tip and extending medially. LP on Eb4 at and anterior to anterior margin of LE4 insertion, ventrolaterally joining CT sheet. LI1 very broad dorsally, on Pb2 dorsally, medially ventral to TPb2. LI2 on Pb3 dorsoposterolaterally just anterior to medial end of Eb3. TD comprises TPb2, TEb2, and TPb3-Eb3. TPb2 heart-shaped, attaching to Pb2 dorsoanteriorly, with median longitudinal raphe posteriorly; raphe joined ventrally by TEb2 medially. TEb2 broad, attaching along mid-longitudinal raphe ventrally between Pb3s with CT of pharyngeal roof, narrowing laterally and attaching along Eb2 dorsally anterior to LE2 inser- tion, continuous mid-posteriorly by CT (not illustrat- ed) with TPb3-Eb3 mid-anteriorly. TPb3-Eb3 on Pb3 dorsoposterolaterally medial to medial end of Eb3 and on posteromedial edge of Eb3, continuous pos- teriorly by diagonal muscle strands with SOD. OD3-—4 origin on Pb3 dorsoanteromedially ventral to TEb2; insertion on anterior surface of Eb3 unci- nate process and medial edge of Eb4. OP dorsally on Eb4 posteriorly at and medial to NUMBER 11 uncinate process; ventrally on Cb5 posterodistally, joining raphe ventrolaterally with Ad5. Ad1-—3 absent. Ad4 dorsally broadly on Eb4 posteriorly beginning ventrally at levator process and extending laterally to end of bone, ventrally broadly on Cb4 dorsally me- dial to Eb4-Cb4 joint, meeting Ad5 anterolaterally on Cb4, but diverging from Ad5 medially. Ad5 dorsally on Cb4 bony surface entirely medial to cartilaginous end; ventrally on Cb5 dorsolaterally joining raphe posterolaterally with OP. SOD slender. RDs separated by space greater than one RD di- ameter. Additional remarks. SCL attached mid-dorsally to tip of posteroventrally extending cartilaginous end of Bb3. TV4 free from Cb5s. Pb4 and UP4 present. IAC present. LATEOLABRACIDAE? Lateolabrax japonicus (Cuvier), USNM 344295, @ = 102 mm; USNM 64631, 109 mm. Plate 120 Description. LE1 on bony anterior surface of Eb! uncinate pro- cess. LE2 on dorsomedial edge of expanded bony edge of Eb2. LE3 on joined tips of Eb3 and Eb4 uncinate pro- cesses, there joining raphe with OD3-—4 insertion. LE4 on Eb4 dorsal surface lateral to uncinate pro- cess. LP on Eb4 at and lateral to LE4 insertion. LI1 on most of dorsal surface of Pb2; with few anterolateral fibers on anteromedial surface of IAC on one side and extensive attachment to medial sur- face of IAC on other. LI2 on Pb3 dorsal surface posterolaterally, begin- ning anteriorly a little anterior to articulation with medial end of Eb3 and extending posteriorly to op- posite to medial end anterior to attachment of TPb3- Pb4-Eb3 to Eb3. TD comprises TEb2 and TPb3-Pb4-Eb3. TEb2 with median longitudinal raphe, which gives rise dor- sally to CT sheets covering muscles and attaches mid-anteroventrally to CT of pharyngeal roof and to CT enveloping (attaching to) anterior ends of Pb3s; muscle attaching along Eb2 dorsally to area anterior to LE2 insertion, continuing posteriorly by diagonal ? The familial placement of Lateolabrax is problematic. It has been variously assigned to Percoidei insertae sedis (Johnson, 1984; 465) or Percichthyidae (Mochizuki, 1984:123). Springer and Raasch (1995:94, 104) assigned it to Lateolabracidae, which ap- pears to be the first published usage of a family name based on the genus. The name appeared next in Orrell et al. (2002:628). 147 strand of muscle with TPb3-Pb4-Eb3. TPb3-Pb4-Eb3 attaching laterally on dorsal surface of Pb3, dorsoan- terior edge of Pb4, and posteromedialmost surface of Eb3, continuing posteriorly by crossing strands of muscle with SOD. OD3-4 origin broadly on Pb3 dorsomedially ven- tral to TEb2, insertion on anterior surface of Eb3 un- cinate process, dorsal to OD3’, and on medial edge of Eb4 uncinate process. OD3’ branches off OD3-—4 anteroventrally just posterior to OD3-—4’s passing out posteriorly from under TEb2 and inserting on Eb3 dorsal surface ven- tral to OD3-—4 insertion on uncinate process. OP dorsally on bony posterior surface of Eb4 me- dial to uncinate process, dorsolaterally overlapping Ad4 dorsomedially, ventrally on Cb5 posterodistally, partially overlapping Ad5 ventromedially. Ad1—3 absent. Ad4 dorsally on posterior surface of Eb4 lateral to OP, ventrally on Cb4 anterior to Eb4-Cb4 joint. Ad5 dorsally on posterodistal surface of Cb4, ven- trally on dorsodistal surface of Cb5 anterior to OP. SOD present. RDs narrowly separated. Additional remarks. SCL attached mid-dorsally to ventroposterior cartilaginous tip of Bb3. TV4 free from Cb5s. Pb4 and UP4 present. Levator process present on posterior edge of right-side Eb4, not pres- ent on left side; present on both sides of @ and of cleared and stained specimen, USNM 177447. SCIAENIDAE Cynoscion nebulosus (Cuvier), USNM 176237, 76.8 mm. Plate 121 Additional material. Cynoscion arenarius Ginsburg, USNM 120047, 117 mm. Remarks. We noted no significant differences be- tween the muscles of C. arenarius and C. nebulosus, except that SOD in the former is strap-like and nor- mal. Description. LE1 short, with tendinous origin, insertion broadly on anterior surface of Eb1 uncinate process. LE2 on expanded bony dorsoposterior surface of Eb2. LE3 on Eb3 uncinate process dorsally and dorsal edge of Eb4 uncinate process just medial to cartilage tip. LE4 on Eb4 dorsally beginning lateral to uncinate process and extending to cartilaginous distal edge. LP on LE4 at and just anterior to LE4 insertion. LI1 on dorsomedialmost surface of IAC, continu- ing onto adjacent surface of Pb2, and extending to 148 CT that binds anteriormost ends of Pb2 and Pb3; much larger than LI2. LI2 on dorsolateral surface of Pb3 just anterior to anteromedial end of Eb3. TD comprises TEb2 and TPb3-Eb3. TEb2 with mid-longitudinal raphe giving rise dorsally to filmy CT covering muscle, attached anteriorly and mid- ventrally to CT of pharyngeal roof and laterally on dorsal surface of Eb2 to point anterior to LE2 inser- tion, continuing posteroventrally by fine, diagonal muscle strand with TPb3-Eb3. TPb3-Eb3 on Pb3 dor- solaterally posteromedial to LI2 insertion, and on Eb3 dorsoposteromedial surface, passing freely over medial end of Eb4 before attaching to Eb3, posteri- orly continuous by fine diagonal muscle strand with very fine SOD. OD3-—4 origin on Pb3 dorsoanteriorly ventral to TEb2, insertion on anterior surface of Eb3 uncinate process, joining raphe with anteroventral edge of LE3 insertion, and on anterior surface of Eb4 uncinate process. OP dorsally on Eb4 posterior surface extending medially beginning ventral to uncinate process, ven- trally beginning on Cb5 posterior surface distally, there joining broad raphe with PCI and extending medially; distinctly separate from Ad4 and SO. Ad1-—3 absent. Ad4 dorsally on Eb4 posterior surface extending laterally beginning from point ventral to LE4 and LP insertions, ventrally on Cb4 dorsally medial to Eb4- Cb4 joint. Ad5 on posterodistal end of Cb4 and dorsodistal end of Cb5, forming short raphe at that point with PCI and Ad4. SOD very fine, aberrant, passes dorsal to RD of left side and returns to SO between RDs. @ Strap- like, continuous from one side to the other. RDs well separated. Additional remarks. SCL absent. TV4 free from Cb5s. Pb4 and UP4 present. Pb2 toothed. Eb4 levator process absent. Sasaki (1989) described and illustrated in detail the dorsal gill-arch musculature of Ctenosciaena gracil- icirrhus (Metzelaar), with which Cynoscion agrees. Sasaki did not detail the attachments of his transver- sus dorsalis posterior, but his illustration (his fig. 37) conveys the impression that it is TPb3-Eb3, similar to that of Cynoscion. Sasaki also described and illus- trated (his figs. 42, 44) differences shown by Pogon- ius, Aplodinodus, and Leiostomus, highly specialized western Atlantic and/or freshwater inhabitants. The differences shown by these three genera mainly in- volve an interrupted TEb2; broad, musculously naked Pb3 dorsal facets; and an unusually well-developed Ad5. The insertion of LI1 to include IAC is uncommon. It also occurs in haemulids and bathyclupeids, both BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON of which also share with Cynoscion the specialized state of lacking SCL, and in bramids, kuhliids, and lateolabracids, which have SCL. POLY NEMIDAE Polydactylus oligodon (Giinther), USNM 364370, 128 mm. Plate 122 Additional material. @ = Filimanus xanthonema (Va- lenciennes), USNM 278199, 99.1 mm. Description. Remarks. LE1—4 insertions musculous and tendi- nous (tendinous portions not illustrated). LE1 on Eb! just lateral to tip of uncinate process. LE2 on tip of bony process on proximal half of posterior edge of Eb2. LE3 on anterior edge of tip of Eb3 uncinate pro- cess. LE4 on bony dorsal edge of Eb4 lateral to tip of uncinate process. LP on Eb4 beginning at lateral edge of LE4. LI1 tendinously on Pb2 dorsally posteroventral to Pb2 articulation with [AC with tendon continuing onto adjacent dorsoanterior end of Pb3. @ Muscle fibers reach lateral edge of Pb3. LI2 on Pb3 dorsolaterally just medial to medial end of Eb3. TD comprises TPb2, TEb2, and TPb3-Pb4-Eb3. TPb2 flat, roughly kidney-shaped, in two sections; triangular small anterior section fitting into mid-an- terior notch of much larger posterior section; poste- rior section with mid-longitudinal raphe continuous posteriorly through TEb2; raphe giving rise dorsally to CT sheets attaching to skull; raphe continuous mid-ventrally with CT of pharyngeal roof; muscle attaching anterolaterally to Pb2 just medial to dor- soanteriormost cartilaginous tip and joining ventrally to TEb2 lateral to raphe. TEb2 anterior fibers ex- tending laterally and passing dorsoposterior to pos- terior fibers and both portions attaching dorsally a short distance lateral to medial end of Eb2, posteri- ormost fibers continue finely tendinously dorsoanter- iorly over other TEb2 fibers and attach to IAC; mus- cle continuous posteriorly by diagonal strip of muscle with TPb3-Pb4-Eb3. TPb3-Pb4-Eb3 on Pb3 dorso- laterally, partly joining posteromedial edge of LI2 in- sertion, continuing posteriorly medial to posterome- dial end of Eb3 and attaching tendinously to it, next continuing onto Pb4 dorsolaterally medial to medial end of Eb4, then continuing posteriorly by diagonal muscle strap with SOD. @ Anterior section of TPb2 is continuous posteriorly with TEb2 ventral to pos- terior section, hence, is part of TEb2. OD3-4 robust, origin broadly on Pb3 dorsomedi- ally mostly ventral to TEb2, but partially posterior to NUMBER 11 TEb2; posteriorly on medial edge of Eb3 uncinate process and anterior surface and medial edge of Eb4 uncinate process, joining raphe with OP dorsally me- dial to tip of Eb4 uncinate process (raphe longer on right side). @ No raphe with OP. OP dorsally beginning on Pb4 posterolaterally and extending laterally onto Eb4 posteriorly to point slightly ventrolateral to uncinate process, there join- ing tendon on Eb4 with dorsomedial end of Ad4 (ten- don and associated CT sheets continue to cleithrum, with long tendinous attachment of PP to CT among sheets ventral to Eb4—only tendon illustrated); ven- trally on Cb5 ventrolaterally posterior to Ad5, joining raphe with Ad5 posterolateralmost edge; mid-medi- ally confluent with SO. @ No fibers on Pb4. Remarks. Fibers on Pb4 possibly more appropri- ately allocated to SO. Ad1-—3 absent, but frayed fan-like GFM on EbI- Cb1 joint anteriorly. Ad4 dorsally on Eb4 ventrally mostly lateral to OP, ventrally on Cb4 medial to Eb4-Cb4 joint, joining raphe ventroposteriorly with Ad5. Ad5 dorsally on Cb4 posterolaterally, anteriorly joining raphe with Ad4. SOD present. RDs adjacent. Additional remarks. SCL absent. TV4 free from Cb5s. Pb4 and UP4 present. Pb] with cartilaginous ends. Pb2 toothed. Eb4 levator process absent. SILLAGINIDAE Sillago sihama (Forsskal), USNM 347113, 139 mm. Plate 123 Description. LE1 narrowly on tip of small bony process ex- tending anteriorly from just below cartilage tip of Eb! uncinate process, muscle fanning out dorsally. LE2 narrowly on small bony process on Eb2 dor- soposteriorly, muscle expanding dorsally. LE3 on Eb3 uncinate process anteriorly, ventroan- teriorly joining OD3—4 on Eb3. LE4 on Eb4 dorsolaterally near distal end, joined ventrolaterally by LP. LP on dorsodistalmost surface of Eb4, joining LE4 insertion ventrolaterally. LI] thin, strap-like, on dorsoanterolateralmost sur- face of Pb2 and dorsomedial surface of IAC, just impinging on fine tendinous origin of M. Pb2-Eb2, ventral half and dorsal fourth musculous, separated by band of clear CT lying against LE2 medially (ar- eas of CT and adjacent LE2 about equal). LI2 on Pb3 dorsolaterally medial to articulation with Eb3, medial edge of insertion meets anterolat- eral edge of TPb3-Eb3. TD comprises TPb2, TEb2, and TPb3-Eb3. TPb2 roughly heart-shaped, notched mid-anteriorly, with 149 deep mid-longitudinal raphe, which attaches ventrally to CT of pharyngeal roof, extending posteriorly from notch and continuing across ventrally lying TEb2 broad portion, which attaches dorsally to TPb2 along raphe; raphe thence continuing across dorsal TEb2 narrow portion, becoming tendinous strand and con- tinuing across TPb3-Eb3 and anterior half of SOD; anterolaterally, TPb2 attaches to Pb2 dorsoanteriorly medial to origin of M. Pb2-Eb2. TEb2 comprising two incompletely separated parts: broad anterior part, which lies mostly ventral to TPb2, and narrow pos- terior part, which is external; anterior part attenuating laterally, passing ventral to M. Pb2-Eb2 and meshing with it as they extend almost to distal end of Eb2 bony surface, there meeting Ad2 posteromedially; posterior part narrows considerably laterally and at- taches to medial edge of bony process bearing LE2 insertion (some fibers from anterior TEb2 part also attache to this edge). TPb3-Eb3 not continuous mus- culously with TEb2, almost completely occluded from view by overlying OD3 and OD4; muscle curves anterolaterally dividing into ventral branch, which attaches to Pb3 beginning at medial edge of LI2 insertion and continuing posteriorly to bony edge bordering articulation with Pb4 and Eb4, and dorsal branch, which passes dorsal to medial end of Eb4 and attaches to posterior edge of medial half of Eb3. (See also M. Pb2-Eb2.) OD3, OD3’, OD4, essentially separate muscles. OD4 robust, originating on Pb3 dorsomedially, over- lying much of posterior half of OD3, and inserting on medial edge of uncinate process and much of pos- terior surface of Eb4, meeting LE4 and LP insertions. OD3 robust, originating on Pb3 just ventrolateral to OD4 and inserting on most of bony anterior surface of Eb3 uncinate process, meeting LE3 insertion ven- trally and slightly separated dorsally from distal end of OD3’. OD3 and OD4 fibers mingle at tightly joined Eb3 and Eb4 uncinate processes. OD3’ slen- der, originating on Pb3 ventral to OD3 and inserting a little more than halfway distally on Eb3 dorsopos- teriorly, there meeting Ad3. OP with two sections: medial section broadly dor- sally on Eb4 posteriorly beginning at medial end of bony surface and extending laterally to near LP-LE4 insertions, laterally overlapping dorsomedial half of lateral section; lateral section beginning on Eb4 pos- teroventrally (anterior to medial section), extending dorsolaterally and attaining same attachment level as medial section, and ending laterally at distal end of bony surface. Medial section broadly ventrally on en- tire posterior surface of lateral arm (horn) of Cb5; lateral section narrowly distally on horn, meeting OP distally, and both meeting Ad5 posteroventrally. OP completely occludes Ad4 in posterior view. M. Pb2-Eb2 relatively large, originating finely, tendinously on Pb2 dorsoanteriorly with and medial 150 to TPb2 insertion and extending posterolaterally onto Eb2 almost to distal end of dorsal bony surface, lying dorsal to TEb2 as it extends onto Eb2 and meshing ventrally with TEb2, meeting medial end of Ad2 pos- teriorly. Ad1 absent, weak GFM filaments on anterodistal surfaces of Eb1 and Cbl. Ad2 small, extending from dorsodistal surface of Eb2 to anterodistal surface of Cb2, meeting distal end of M. Pb2-Eb2. Ad3 extending laterally from anteromedialmost bony surface of Eb3 to anterodistal surface of Cb3, meeting anterodistal end of OD3’. Ad4 ventrally on Eb4 anterior to lateral OP sec- tion, ventrally broadly on Cb4 anterior to Eb4-Cb4 joint, posteriorly excluded from view by OP. Ad5 bulky, dorsally relatively broadly on poster- odistal surface of Cb4, there meeting distal three- fourths of Ad4 attachment posteriorly, ventrally on Cb5 dorsodistally, passing anterior to OP lateral sec- tion. SOD present. RDs adjacent; together forming cup-shaped de- pression dorsally. Additional remarks. SCL free from Bb3. TV4 free from Cb5s. Pb4 and UP4 present. Tiny AC2 present on both sides (also on both sides of C&S specimen, USNM 269802). Eb4 levator process absent. Broad bony flange overlapping cartilaginous distal end of Eb4 anteriorly. MULLIDAE Pseudupeneus maculatus (Bloch), USNM 267508, 2 specimens, 94.0-116 mm. Plate 124 Additional material. @ = Parupeneus multifasciatus (Quoy and Gaimard), USNM 267485, 2 speci- mens, 80.0—99.7 mm. Description. Remarks. We observed no noteworthy differences between any of the specimens. LE1 on dorsoanterior surface of Eb! uncinate pro- cess. LE2 on dorsal tip of expanded bony posterior edge of Eb2. LE3 tendinously on tightly bound tips of Eb3 and Eb4 uncinate processes. LE4 on dorsoposterior edge of Eb4. LP on Eb4 at and lateral to lateral edge of LE4 insertion. LI1 on Pb2 ventral to tip of dorsoanterior process. LI2 on Pb3 dorsoposteriorly just medial to medial end of Eb3. TD comprises TPb2, TEb2, and TPb3-Eb3. TPb2 consists of an anterior transverse strip of muscle at- BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON taching to each IAC dorsolaterally and continuing mid-posteriorly with TEb2 and slightly laterally with pair of slender, laterally convex ribbons of muscle lying dorsal to TEb2; each ribbon with mid-lateral raphe and fusing anteriorly and posteriorly with TEb2 lateral to mid-longitudinal raphe, which ex- tends continuously from anterior end of TD almost to posterior edge of TPb3-Eb3; mid-longitudinal ra- phe attached ventroanteromedially to CT of pharyn- geal roof and giving rise dorsally to CT sheets that cover muscle and attach to skull. TEb2 extensive, attaching on dorsal surface of Eb2 laterally anterior to LE2 insertion, posteriorly continuous (or free, var- iable in specimens) by strands of muscle with TPb3- Eb3. TPb3-Eb3 beginning anteriorly on Pb3 dorsally just medial to medial end of Eb3 and continuing pos- teriorly and abruptly expanding greatly laterally and attaching on Eb3 dorsomedially; continuous by di- agonal strands of muscle with SOD. OD3-—4, OD3’ origin on Pb3 ventral to TEb2, di- viding ventrally (OD3’) just before inserting on dor- soanterior surface of Eb3 uncinate process and me- dial edge of Eb4 uncinate process; OD3’ insertion on Eb3 dorsal surface just ventral to Eb3 portion of OD3-—4. OP dorsally broadly on Eb4 posterior surface ven- tromedial to insertions of LE4-LP, ventrally on dor- soposterior edge of Cb5. Ad1-—3 absent. Ad4 dorsally broadly on Eb4 posterior surface lat- eral to OP, ventrally on Cb4 dorsal surface medial to Eb4-Cb4 joint. Ad5 on dorsodistal surface of Cb5 and posterodis- tal end of Cb4 and Eb4. SOD present. RDs separate. Additional remarks. SCL absent. TV4 free from Cb5s. Pb1 and Pb4 absent, UP4 present. Eb4 levator process absent; however, darkly blue-stained area near posterior point of LP insertion seems to indicate that a cartilage tip may have been present early in ontogeny. AC2 (not illustrated) present on both sides of one specimen but absent on both sides of other specimen; additionally, AC2 present on both sides of two specimens of Mulloides flavolineatus (USNM 272965, cleared and stained). @ No ACs present in either specimen. Kim (2002:35—36) briefly described and diagram- matically illustrated the dorsal gill-arch muscles of Upeneus vittatus Forsskal, a member of the genus he hypothesized as the sister-group to all the other mul- lid genera. Although he describes TPb2 as including IAC among its attachments, his illustration of TPb2 does not exhibit the anterolateral extensions of the muscle that attach to IAC that are present in our spec- imens of Parupeneus and Pseudupeneus. NUMBER 11 CENTROGENIDAE Centrogenys vaigiensis (Quoy and Gaimard), USNM 183016, 84.7 mm; @ = USNM 327899, 93.5 mm. Plate 125 Description. Remarks. The musculature of Centrogenys is un- usually bulky and invested with tough CT, relative to the small size of the specimens. Determinations of extent and attachments of TD, OD, OP, and Ad4 & 5 are particularly difficult and require considerable destruction of muscles. LE1 on Eb!1 uncinate process lateral to tip. LE2 on mid-dorsoposterior edge of Eb2 almost touching TEb2 posteriorly. LE3 broadly dorsally on and lateral to tip of Eb3 uncinate process and narrowly on dorsal tip of bony Eb4 uncinate process. LE4 massive, on Eb4 broadly dorsoposteriorly lat- eral to all bony uncinate process (see remarks follow- ing OD3-—4) and anterior to levator process, joining raphe posteroventromedially with OP dorsally. @ Ra- phe more extensive in larger specimen. LP on Eb4 dorsoposterior margin medial to minute cartilage tip of levator process, joining raphe with LE4 insertion posterolaterally; finely continuous ven- trolaterally with OP in @, but ignored in coding char- acter for matrix (Table 12, Appendix). LI1 on dorsalmost edge of Pb2 and dorsoanterior surface of Pb3 posterior to anterior end (Pb2 eden- tate, much reduced and closely applied medially to lateral surface of Pb3). LI2 on Pb3 dorsoposteriorly medial to medial end of Eb3. TD comprises TPb2, TEb2, and TPb3 (see also @). TPb2 bun-like muscle pad on each side continuous with each other medially by tough, thick, broad fas- cia, which gives rise to tough CT sheets attaching to skull; muscle dorsal to medial ends of bilaterally di- vided TEb2, attaches to dorsoanterior ends of Pb2 and Pb3 and, between Pbs to CT of pharyngeal roof; muscle coverage mostly absent from dorsal Pb3 sur- face. TEb2 a muscle pair, each member becoming tendinous medially and dorsally joining ventral mid- line CT between Pbs, muscle extending onto Eb2 dorsally well lateral to LE2 insertion. TPb3 on Pb3 posterolaterally medial to medial end of Eb4, contin- uous posteriorly by diagonal muscle filament with SOD. ® Comprises TPb2, TEb2, and TPb3-Pb4. OD3-—4 massive, origin tendinous, on Pb3 antero- laterally ventral to TEb2, there joining fascia of TPb2 and TEb2; insertion extensively on Eb3 dorsoanter- iorly and on medial edge of Eb4 bony uncinate pro- cess, continuing onto posterior surface of process. Remarks. In the described smaller specimen, the Eb4 uncinate process bears an easily overlooked tiny cartilage tip (absent in the larger specimen) ventral 151 to the more dorsal bony surface that is tightly joined to the Eb3 uncinate process. The tip is visible only by severing the CT joining the two processes. It ap- pears that the cartilage tip is lost with growth, and its absence is considered typical for the taxon. OP complex, bulky, possibly representing a fusion of primitively mostly separate lateral and medial sec- tions; dorsally attached to most of broad ventral Eb4 surface anterior to Ad4, dorsoposteriorly joining ra- phe with LE4 ventroposteromedially, ventrolaterally becoming tendinous and joined laterally by Ad5 and ventrally by PCI, ventrolaterally broadly on Cb5 pos- terolaterally, joining raphe with PCI anterolaterally and TV5 anterolaterally, continuous medially with SO. @ Raphe with LE4 more extensive. Ad1-—3 absent. Ad4 broadly dorsally on ventral Eb4 surface pos- terior to OP, ventrally broadly on Cb4 paralleling Ad5 attachment medial to Eb4-Cb4 joint. Ad5 dorsally broadly on posterolateral surface of Cb4 and AC, ventrally broadly on Cb5 posteriorly, joining raphe with PCI dorsoanteriorly and OP ten- don anteromedially. SOD present. RDs adjacent. Additional remarks. SCL free from Bb3. TV4 ven- trally continuous across ventral surface of Cb5s, dor- sally interrupted, attaching to anterolateral surface of each Cb5. Pb4 and UP4 present. Pb2 edentate. [AC present. AC4 present. Cb5s are tightly bound medially along sinuous joint, one side is larger than other and overlaps the smaller side dorsoanteriorly. The bones are not fused in specimens of the sizes we examined, but it seems likely that they might be in much larger specimens (species variously reported to attain a length of 150 or 200 mm). AMBASSIDAE Tetracentrum caudovittatus (Norman), USNM 332862, 74.1 mm. Plate 126 Additional material. @ = Ambassis buruensis, USNM 305331, 55.0 mm. Description. LE! on Eb! uncinate process dorsoanteriorly just lateral to cartilage tip. @ On uncinate process just anteroventral to tip. LE2 on dorsalmost edge of raised posterior surface of Eb2. LE3 on medial edge of cartilage tip of Eb3 unci- nate process. @ On tip of process dorsally. LE4 massive; on dorsolateral surface of Eb4 reach- ing to lateralmost bony edge, there narrowly joining with anterolateralmost edge of LP; posteriorly, nar- 152 rowly, musculously continuous with OP dorsolater- ally, just medial to posteromedial edge of LP inser- tion. @ Not continuous with OP. LP little less robust than LE4; on Eb4 dorsally near distal bony end, ventroanteriorly joining LE4, meet- ing OP dorsolaterally. @ Slender, on dorsodistalmost bony edge of Eb4, ventromedially joining LE4. LI1 passing between acute, laterally open angle formed by TPb2 with TEb2, and inserting by slender tendon on Pb2 dorsally beginning just ventral to me- dial end of broad cartilaginous cap, tendon joins CT extending onto adjacent bony surface of Pb3 just pos- terior to dorsoanteriormost cartilage tip. LI2 on Pb3 dorsoposterolaterally adjacent to artic- ulation with medial end of Eb3. TD comprises TPb2, TEb2, and TPb3-Pb4-Eb3. TPb2 moderately robust, with shallow notch mid-an- teriorly leading to short, fine, mid-longitudinal sep- tum. Septum posteriorly joins broad, oblong, thick, tough CT area, representing non-muscular mid-por- tion of TEb2 covering flat, dorsal Pb3 surfaces and attaching mid-ventrally to CT of pharyngeal roof; CT sheets arise from margins of oblong CT area and cov- er TPb2 and TEb2. TPb2 entirely anterior to TEb2, but joining oblong CT area anteriorly at point just anterior to anteromedialmost musculous portion of TEb2, there forming angle that embraces LI1 tendi- nous portion dorsal to insertion. TPb2 musculously on IAC beginning about mid-dorsally and extending medially onto Pb2 anteriorly a little medial to joint with IAC (@ begins on IAC dorsomedially close to joint with Pb2); muscle is continuous across dorsal and anterior surfaces of Pb2s. CT area of TEb2 more extensive than TPb2; musculous TEb2 portion ro- bust, but narrowing considerably as it extends onto Eb2, first dorsally, then anteriorly, reaching point an- terolateral to LE2 insertion and meeting medial end of GFM2. TEb2 not continuous posteriorly with TPb3-Pb4-Eb3. TPb3-Pb4-Eb3 beginning anteriorly, tendinously on posteromedial edge of Eb3 and ad- jacent surface of Pb3 and extending posteriorly on Pb4 dorsally parallel to medial end of Eb4 (superfi- cially muscle appears to attach to medial end of Eb4). @ TPb3-Eb3 attaches to Pb3 (Pb4 absent) only on surface immediately medial to medial corner of Eb3. M. SO-Pb3 (not visible in dorsal view) arises on each side as anteriorly extending slender ribbon of SO longitudinal fibers that inserts on Pb3 just ven- troposterior to raised dorsoposterior surface. OD3-4 robust, anteriorly ventral to TEb2 on trans- verse posterior edge of raised anterior half of Pb3, posteriorly, broadly on Eb3 uncinate process anteri- orly and Eb4 uncinate process posteriorly, joining narrow raphe with OP on Eb4 medial to tip of Eb4 uncinate process. OP in two or 3 irregular and almost completely separate sections, attaching dorsally along most of BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON posterior bony surface of Eb4: medial section begin- ning at uncinate process and extending medially, middle section beginning at lateral end of medial sec- tion and extending to area ventral to LP, lateral sec- tion beginning at lateral end of medial section and extending laterally almost to end of Eb4; lateralmost section narrowly, musculously continuous with LE4 (q.v.); lateralmost section ventrally on Cb5 posteri- orly and dorsally, overlapping Ad5 posteriorly; other two sections more medially on Cb5. @ Not continu- ous with LE4, in two or three separate sections, la- teralmost section either absent or fused ventrally with Ad5, q.v. Ad1-—3 absent. GFM1 fine on anterior surface of Eb1 and Cb1 at and near joint. GFM2 better devel- oped, extending onto Eb2 anteriorly to about mid- length and meeting lateral end of TEb2; GFM3 well developed but limited to Eb3 dorsolaterally. Ad4 dorsally, narrowly on Eb4 posteroventrally near angle with Cb4; ventrally, narrowly on Cb4 dor- sally near angle with Eb4; completely overlapped posteriorly by OP and not visible in posterior or lat- eral views (distal ends of Cb4 and Cb5 unusually closely and tightly bound together). Ad5 dorsally on Cb4 posterolaterally, ventrally on Cb5 dorsolaterally, including surface of dorsodistal bony flange; muscle extends medially on Cb5 a short distance anterior to OP. @ Ad5 appears to have fused with the lateralmost part of OP, or else has been lost, such that a single muscle representing one or both, OP or Ad5, extends from the dorsolateralmost end of Cb5 to Eb4. SOD absent. RDs massive, adjacent. Additional remarks. SCL attached to ventrally ex- tending cartilaginous posterior tip of Bb3. TV4 free from Cb5s. Pb4 and UP4 present. (@ Pb4 absent, UP4 present). Eb4 levator process absent. Very tiny AC3 present unilaterally (@ no ACs present, but see discussion of ACs in section ““Abbreviations and Def- initions for Anatomical Structures” for Ambassis sp. with AC4). Bony flange of Eb4 projects laterally dor- sal to cartilaginous distal end. Diameter of medial end of Eb3 greater than that of Eb4. @ Slightly less than diameter of Eb4; however, in two cleared and stained specimens, one each as Ambassis macleayi, USNM 173817, and Ambassis sp., USNM 218805, the medial end of Eb3 is slightly larger than that of Eb4. The medial end of Eb4 is treated as smaller than that of Eb3. Anterior end of Cb5 on one side overlaps anterior end on other side. CARISTHUDAE Caristius maderensis Maul, USNM 206895, 114 mm. Plate 127 NUMBER 11 Description. LE1 on dorsal edge of Eb1 just lateral to cartilage tip of well-developed uncinate process, which is well removed laterally from direct contact with anterior end of Eb1. LE2 on Eb2 mid-dorsoposteriorly. LE3 finely, tendinously on tip of Eb3 uncinate pro- cess medially. LE4 origin tendinous; left side, tendinously on lat- eral surface of tip of Eb4 uncinate process; right side, similar to left side, but separate strip of muscle fibers extends onto bony edge of Eb4 ventrolateral to tip of uncinate process; ventrolateral edge of insertion meets ventroanterior edge of LP and together they are membranously continuous with Ad4 dorsally (see also LP). LP origin tendinous; insertion joins ventrolateral edge of LE4 insertion and extends short distance lat- erally; LP and LE4 insertions join CTI, which is joined ventrally by dorsoposterior end of medial Ad4 section, and the three muscles can be easily released as a cohesive group from Eb4. Broad CT sheet, which attaches to lateral edges of 4th and 5th arches, attaches medially to lateralmost edge of LP insertion. LI1 on bony surface of Pb2 dorsoanteriorly begin- ning just ventral to dorsalmost cartilaginous tip. LI2 on bony surface of Pb3 dorsoanteriorly just ventral to cartilaginous process articulating with me- dial end of Eb3. TD comprises TPb2, TEb2, and TPb3-Eb3. TPb2 thick, roundish pad, depressed centrally, inseparable ventrally from TEb2; with pair of longitudinal raphes that give rise to CT sheets covering muscle and at- taching to skull; CT also arising from dorsoposterior half of pad perimeter, beginning at mid-lateral raphe on each side, which is also joined ventrally by TEb2 medially (TPb2 and TEb2 inseparable medial to ra- phe); weak, tendinous strand extends anteriorly from mid-lateral raphe and attaches to Pb2 dorsolaterally; weak, filmy CT attaches muscle mid-ventrally be- tween Pb3s to CT of pharyngeal roof. TEb2 extends laterally and attaches on Eb2 dorsally well lateral to LE2 insertion; diagonal muscle strand extending from posteroventral surface is continuous with Pb3 portion of TPb3-Eb3. TPb3-Eb3 on Pb3 dorsolater- ally beginning medial to LI2 insertion and continuing posteriorly with laterally separate muscle strands at- taching to cartilaginous medial end of Eb3 dorsally and posterior bony edge of Eb3; ventral diagonal muscle strands continuous with SOD. Remarks. Vertical rotation of RDs has probably re- sulted in TPb3-Eb3 and SOD also becoming almost vertical (see also remarks following OD3-—4). OD3-—4 origin on Pb3 dorsoanteromedially ventral to TPb2 and TEb2; muscle divides posteriorly with anterior branch inserting broadly on anterior surface of Eb3 beginning on uncinate process and extending 153 laterally, posterior branch inserting equally broadly on anterior surface of Eb4; most of Eb4 insertion is hidden from view. OP dorsally on Eb4 posteriorly beginning medial to cartilage tip of uncinate process and extending me- dially with several medialmost fibers extending well anteriorly and inserting on Pb4 posteriorly; ventrally on Cb5 dorsoposterolaterally, posterior to Ad5 at- tachment, just joining raphe with Ad5 posterodistally. Ad1-—3 absent. Ad4 with lateral and medial sections. Dorsally, both sections on posterior surface of Eb4; medial sec- tion beginning lateral to OP and continuing laterally about half distance to end of Eb4, dorsoposterome- dially continuous with CT joining LE4 and LP (pos- terior Ad4 fibers separate from anterior fibers, which attach to Eb4 directly); lateral section on posterolat- eral portion of Eb4; ventrally both sections join Cb4 dorsally. Ad5 dorsally relatively narrowly on Cb4 postero- lateralmost bony surface, ventrally relatively narrow- ly on Cb5 dorsolaterally. SOD present. RDs separated by distance less than diameter of one RD; extend ventrally, vertically from origins, producing excavation of SO in region of anteriorly extending SO longitudinal fibers, then turning ante- riorly and attaching to medial edge of UP4 anteriorly, and continuing onto most of medial surface of Pb3 just dorsal to tooth plate. Additional remarks. SCL attached mid-dorsally to tip of posteroventrally extending cartilaginous tip of Bb3. TV4 free from Cb5s. Pb4 and UP4 present. IAC absent. Eb4 levator process absent. Small AC present between Eb3 and Eb4 uncinate processes on right side. Pbl bony with cartilaginous ends. Pb2 toothed. BRAMIDAE Brama brama, USNM 240541, 82.2 mm. Plate 128 Description. LEI! on raised bony dorsoposterior edge of Eb1 lateral to tip of uncinate process. LE2 on raised bony dorsoposterior edge of Eb2. LE3 absent. LE4 on cartilaginous tip of Eb4 levator process; ventroposteriorly tendinous, continuous with broad CT sheet, which includes PP, that attaches along arch- es 4 and 5 and ventral edge of Ad5. LP fine, fuses with anterolateral edge of LE4 in- sertion. LI1 on Pb2 dorsoanterolaterally, extending slightly onto IAC at joint with Pb2. 154 LI2 on Pb3 posterolaterally anterior to medial end of Eb3. TD comprises TPb2, TEb2, and TPb3-Eb4. TPb2 concave, roundish, with raised lateral edges, attach- ing ventroanteriorly to CT of pharyngeal roof, with mid-longitudinal raphe giving rise to CT sheets cov- ering muscle; raphe mid-laterally with CT extension attaching to dorsoanterior cartilaginous edge of Pb2; free ventrolaterally from, but broadly continuous ventrally with, TEb2. TEb2 attaching on Eb2 dorsally anterior to LE2 insertion. TPb3-Eb4 acutely trian- gular, with apex becoming tendinous anteriorly and joining TPb2-TEb2 mid-ventroposteriorly, mainly at- taching broadly on Eb4 bony surface dorsally medial to uncinate process with few fibers delaminating from ventral surface and inserting on Pb3 dorsopostero- medially; mid-posteroventrally continuous with SOD, which it overlaps dorsally and partially ob- scures in dorsal view. OD3-4 origin broadly on Pb3 dorsally, insertion on Eb3 uncinate process anteriorly (and, on one side, on Eb3 dorsally just ventromedial to uncinate pro- cess) and medial edge of Eb4 uncinate process. OP dorsally on Eb4 posteriorly beginning on le- vator process extending medially to uncinate process, there joining raphe with OD3—4 on one side but not other, laterally overlapping Ad4, which extends short distance medially anterior to OP; ventrally becoming tendinous and fascia-like and closely applied to pos- teromedial surface of Ad5; medially separable from SO, although fibers of both are closely adjacent. Ad1-—3 absent. Ad4 dorsally on Eb4 posteriorly beginning on le- vator process anteromedial to lateralmost edge of OP and extending to distal end, ventrally broadly on Cb4 dorsally medial to Eb4-Cb4 joint. Ad5 dorsally narrowly on posterodistalmost sur- face of Cb4 (at edge of Eb4-Cb4 joint), ventrally moderately broadly on dorsolateralmost surface of Cb5, posterior surface fused to tendinous fascia-like ventral extension of OP. SOD thin, dorsally ventral to TEb3-Eb4. RDs adjacent. Additional remarks. SCL attached weakly, if at all, mid-dorsally to elongate, posteroventrally extending cartilaginous tip of Bb3. TV4 free from Cb5s. Pb4 and UP4 present. Pb2 toothed. TOXOTIDAE Toxotes jaculatrix (Pallas), USNM 331444, 94.2 mm; 289931, 86.9 mm. Not illustrated Description. LE1 on Eb! dorsoposteriorly a little lateral to tip of uncinate process; ventral half of lateral edge of muscle tendinous. BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON LE2 on Eb2 dorsoposteriorly about one-third dis- tance from medial end of Eb2 and posterior to lateral end of TEb2; anterior muscle surface tendinous near insertion. LE3 on Eb3 beginning at medial edge of tip of uncinate process and extending medially a short dis- tance; ventral one-fourth of muscle lateral edge ten- dinous. LE4 on Eb4 medial to levator process, ventrolat- erally meeting LP, posteroventrolaterally joining nar- row raphe with OP dorsolaterally; lateral surface of ventral third of muscle edge tendinous. Remarks. The fine raphe with OP would hardly seem to qualify as a specialized character state, LE4 joins OP, but it is present on both sides of both spec- imens and as such is here considered to be a “sling” (sensus Stiassny and Jensen, 1987). LP largest levator, on Eb4 dorsally beginning at lateralmost bony edge and extending medially, meet- ing LE4 insertion medially and cupping LE4 ven- trally; LP insertion includes dorsal edge of Eb4 le- vator process, there meeting and joining raphe with musculous portion of PP, which continues ventrally as CT and all but encapsulates Ad5. LI1 narrowly, tendinously on Pb2 dorsoanterola- terally; tendon continuous with CT binding anterior ends of Pb2 and Pb3 (see also description of TPb2); about same size as LI2. LI2 on Pb3 dorsolaterally, meeting anterolateral end of TPb3-Pb4-Eb3 on Pb3. TD comprises TPb2, TEb2, and TPb3-Pb4-Eb3. TPb2 almost entirely anterior to TEb2, posteriorly overlapping a little of anterior edge of TEb2; muscle comprising an oval-shaped section on each side, which is angled a little anterolaterally from short mid-line raphe that widens and fills notch between ovals anteriorly and is continuous anteriorly with CT of pharyngeal roof; raphe continues posteriorly and becomes wide area of CT replacing central portion of TEb2 (leaving flat, dorsal Pb3 surfaces naked of muscle) and giving rise dorsally to CT sheets cov- ering TD and attaching mid-ventrally to CT of pha- ryngeal roof. Anteroventrally, TPb2 attaches to joined dorsoanterior ends of Pb2 and Pb3 and dor- somedialmost surface of IAC, and is attached to an- terior edge of ventral tendinous portion of LI1. TEb2 extends laterally and attaches on Eb2 dorsally ante- rior to LE2 insertion, muscle is dorsal to, and dis- continuous with, TPb3-Pb4-Eb3. TPb3-Pb4-Eb3 be- gins anteriorly on Pb3 dorsoposterolaterally at medial margin of L]2 insertion and continues posteriorly at- taching onto posteromedial edge of Eb3, with strands extending from muscle ventral surface and attaching to Pb4 dorsally, continuous posteriorly by fine, di- agonally crossing muscle strands with SOD anteri- orly. NUMBER 11 OD3—4 origin on Pb3 dorsomedially, insertion on Eb3 anteriorly beginning at tip of uncinate process and on medial edge of Eb4 beginning at tip of un- cinate process, joining raphe with dorsomedial edge of OP on Eb4. OP dorsally on Eb4 posteriorly beginning just ven- tral to tip of uncinate process, there meeting OD3-— 4, and extending laterally to just below tip of levator process, joining raphe dorsally with LE4, overlapping most to almost all of Ad4 dorsally; narrowly, ven- trally on Cb5 posteriorly near distal end, muscle ven- troposteriorly tendinous, joining raphe with Ad5 pos- teromedially. Ad1-—3 absent. Ad4 broadly, dorsally on Eb4 ventrally beginning just anterior to medial end of OP and extending lat- erally to posterodistalmost bony surface; broadly, ventrally on Eb4 dorsally beginning about one third from medial end of bone and extending laterally to distal end of bone at inner corner of Eb4-Cb4 joint, posterolaterally meeting Ad5 on Eb4. Ad5 dorsoanteriorly on Cb4 posterolaterally, pos- teroventrally on Cb5 dorsolaterally, anterior and pos- terior surfaces almost encapsulated by CT extending ventrally from PP. SOD present. Remarks. Johnson (1993:9) reported that Toxotes lacks SOD, but did not indicate the material on which he based his observations. He based it, however, on a set of gill arches that are from the same lot (USNM 289931) as one of the specimens on which the pres- ent description is based. If so, the muscles are dam- aged in the area where SOD would be expected and it is not possible to decide whether SOD might have been present. Johnson (1993:9) also reported SOD absent in Scorpis, Kuhlia, Kyphosus, and Pholidi- chthys, but, similarly, did not cite his material. John- son based his observations on the following dissec- tions: USNM 218922 (Scorpis), 289925 (Kuhlia), and 366512 (Kyphosus), all of which are damaged in the area where SOD might be expected to occur. Of these genera, and based on specimens dissected for the present study, we have found SOD lacking only in Pholidichthys. Our observations on these three genera are based on the following specimens: Scorpis sp. (USNM 339348) and Kyphosus (USNM 366512), both not otherwise discussed in the present study, and specimens cited in the description of Kuhlia (Kuhli- idae). RDs adjacent. Additional remarks. SCL attached mid-dorsally to tip of ventrally extending cartilaginous posterior end of Bb3. TV4 free from Cb5s. Pb! bony with cartilage ends. Pb4 and UP4 present. Very reduced Eb4 flange present. PLESIOPIDAE Assessor macneilli Whitley, USNM 269466, 47.7 mm; USNM 274581, 51.7 mm. Plate 129 Additional material. @ = Paraplesiops poweri Ogil- by, USNM 274579, 92.2 mm. ® = Acanthoclinus fuscus Jenyns, USNM 339246, not measured. Description (based on smaller specimen of Assessor with remarks based on larger specimen). LE1 on dorsoposterior rim of Eb1l beginning on uncinate process ventroanterolaterally and extending a short distance laterally. LE2 on dorsalmost edge of raised, bony dorsopos- terior rim of Eb2, well lateral to TEb2 insertion. ® Inserts posterior to distal end of TEb2. LE3 on Eb3 immediately ventroanterior to carti- lage tip of uncinate process. LE4 finely, tendinously on Eb4 dorsally a little me- dial to dorsodistalmost end of bone, there meeting LP insertion (in smaller specimen, muscle inserts above levator process, which is absent in larger spec- imen). @ @ Muscle massive, inserts broadly on Eb4 dorsally beginning laterally just anterior to levator process in ® (process absent in @), meeting OD3—4 posterodistally on Eb4; fusing with LP ventroanter- iorly (or ventromedially, depending somewhat on viewing orientation) a little dorsal to insertion. LP narrowly on dorsodistalmost end of Eb4 bony surface, meeting LE4 insertion ventroposteriorly. @ Muscle massive, ventroposteriorly joining raphe with Ad5 dorsally (thus, forming an unusual LP-Ad5 sling), fusing ventroanteriorly with LE4. ® Massive like @, but does not form LP-Ad5 sling. LI1 larger than LI2, ventrally becoming moderate- ly long, slender tendon passing ventral to TPb2 and inserting at junction between medial edge of Pb2 and adjacent impinging lateral edge of Pb3. @ @ Mus- culously and tendinously on Pb2 dorsoanteriorly pos- terior to anteriormost edge, tendinous portion con- tinuing medially as strong, slender tendon inserting on Pb3 dorsal facet mid-dorsoanteriorly. LI2 on Pb3 dorsoposterolaterally just medial to joint with Eb3; medial edge of insertion separated from lateral edge of TPb3-Eb3 in one specimen, meeting lateral edge of TPb3-Eb3 in other specimen. @ Medial edge meets lateral edge of TPb3-Eb3. TD flat, comprises TPb2, TEb2, and TPb3-Eb3. TD with mid-longitudinal raphe, which gives rise dorsally to filmy CT sheets covering TD. TPb2 an anterolaterally extending muscle on each side, in- serting on dorsoanterior surface of Pb2, almost com- pletely anterior to TEb2. TEb2 dorsoanteromedially joining TPb2 ventrally; main anterior portion contin- uous dorsally from one side to the other ventral to TPb2, lesser posterior portion interrupted medially 154 LI2 on Pb3 posterolaterally anterior to medial end of Eb3. TD comprises TPb2, TEb2, and TPb3-Eb4. TPb2 concave, roundish, with raised lateral edges, attach- ing ventroanteriorly to CT of pharyngeal roof, with mid-longitudinal raphe giving rise to CT sheets cov- ering muscle; raphe mid-laterally with CT extension attaching to dorsoanterior cartilaginous edge of Pb2; free ventrolaterally from, but broadly continuous ventrally with, TEb2. TEb2 attaching on Eb2 dorsally anterior to LE2 insertion. TPb3-Eb4 acutely trian- gular, with apex becoming tendinous anteriorly and joining TPb2-TEb2 mid-ventroposteriorly, mainly at- taching broadly on Eb4 bony surface dorsally medial to uncinate process with few fibers delaminating from ventral surface and inserting on Pb3 dorsopostero- medially; mid-posteroventrally continuous with SOD, which it overlaps dorsally and partially ob- scures in dorsal view. OD3-4 origin broadly on Pb3 dorsally, insertion on Eb3 uncinate process anteriorly (and, on one side, on Eb3 dorsally just ventromedial to uncinate pro- cess) and medial edge of Eb4 uncinate process. OP dorsally on Eb4 posteriorly beginning on le- vator process extending medially to uncinate process, there joining raphe with OD3—4 on one side but not other, laterally overlapping Ad4, which extends short distance medially anterior to OP; ventrally becoming tendinous and fascia-like and closely applied to pos- teromedial surface of Ad5; medially separable from SO, although fibers of both are closely adjacent. Ad1-—3 absent. Ad4 dorsally on Eb4 posteriorly beginning on le- vator process anteromedial to lateralmost edge of OP and extending to distal end, ventrally broadly on Cb4 dorsally medial to Eb4-Cb4 joint. Ad5 dorsally narrowly on posterodistalmost sur- face of Cb4 (at edge of Eb4-Cb4 joint), ventrally moderately broadly on dorsolateralmost surface of CbS5, posterior surface fused to tendinous fascia-like ventral extension of OP. SOD thin, dorsally ventral to TEb3-Eb4. RDs adjacent. Additional remarks. SCL attached weakly, if at all, mid-dorsally to elongate, posteroventrally extending cartilaginous tip of Bb3. TV4 free from Cb5s. Pb4 and UP4 present. Pb2 toothed. TOXOTIDAE Toxotes jaculatrix (Pallas), USNM 331444, 94.2 mm; 289931, 86.9 mm. Not illustrated Description. LE1 on Eb! dorsoposteriorly a little lateral to tip of uncinate process; ventral half of lateral edge of muscle tendinous. BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON LE2 on Eb2 dorsoposteriorly about one-third dis- tance from medial end of Eb2 and posterior to lateral end of TEb2; anterior muscle surface tendinous near insertion. LE3 on Eb3 beginning at medial edge of tip of uncinate process and extending medially a short dis- tance; ventral one-fourth of muscle lateral edge ten- dinous. LE4 on Eb4 medial to levator process, ventrolat- erally meeting LP, posteroventrolaterally joining nar- row raphe with OP dorsolaterally; lateral surface of ventral third of muscle edge tendinous. Remarks. The fine raphe with OP would hardly seem to qualify as a specialized character state, LE4 joins OP, but it is present on both sides of both spec- imens and as such is here considered to be a “‘sling”’ (sensus Stiassny and Jensen, 1987). LP largest levator, on Eb4 dorsally beginning at lateralmost bony edge and extending medially, meet- ing LE4 insertion medially and cupping LE4 ven- trally; LP insertion includes dorsal edge of Eb4 le- vator process, there meeting and joining raphe with musculous portion of PP, which continues ventrally as CT and all but encapsulates Ad5. LI1 narrowly, tendinously on Pb2 dorsoanterola- terally; tendon continuous with CT binding anterior ends of Pb2 and Pb3 (see also description of TPb2); about same size as LI2. LI2 on Pb3 dorsolaterally, meeting anterolateral end of TPb3-Pb4-Eb3 on Pb3. TD comprises TPb2, TEb2, and TPb3-Pb4-Eb3. TPb2 almost entirely anterior to TEb2, posteriorly overlapping a little of anterior edge of TEb2; muscle comprising an oval-shaped section on each side, which is angled a little anterolaterally from short mid-line raphe that widens and fills notch between ovals anteriorly and is continuous anteriorly with CT of pharyngeal roof; raphe continues posteriorly and becomes wide area of CT replacing central portion of TEb2 (leaving flat, dorsal Pb3 surfaces naked of muscle) and giving rise dorsally to CT sheets cov- ering TD and attaching mid-ventrally to CT of pha- ryngeal roof. Anteroventrally, TPb2 attaches to joined dorsoanterior ends of Pb2 and Pb3 and dor- somedialmost surface of IAC, and is attached to an- terior edge of ventral tendinous portion of LI1. TEb2 extends laterally and attaches on Eb2 dorsally ante- rior to LE2 insertion, muscle is dorsal to, and dis- continuous with, TPb3-Pb4-Eb3. TPb3-Pb4-Eb3 be- gins anteriorly on Pb3 dorsoposterolaterally at medial margin of LI2 insertion and continues posteriorly at- taching onto posteromedial edge of Eb3, with strands extending from muscle ventral surface and attaching to Pb4 dorsally, continuous posteriorly by fine, di- agonally crossing muscle strands with SOD anteri- orly. NUMBER 11 OD3-—4 origin on Pb3 dorsomedially, insertion on Eb3 anteriorly beginning at tip of uncinate process and on medial edge of Eb4 beginning at tip of un- cinate process, joining raphe with dorsomedial edge of OP on Eb4. OP dorsally on Eb4 posteriorly beginning just ven- tral to tip of uncinate process, there meeting OD3-— 4, and extending laterally to just below tip of levator process, joining raphe dorsally with LE4, overlapping most to almost all of Ad4 dorsally; narrowly, ven- trally on Cb5 posteriorly near distal end, muscle ven- troposteriorly tendinous, joining raphe with Ad5 pos- teromedially. Ad1-—3 absent. Ad4 broadly, dorsally on Eb4 ventrally beginning just anterior to medial end of OP and extending lat- erally to posterodistalmost bony surface; broadly, ventrally on Eb4 dorsally beginning about one third from medial end of bone and extending laterally to distal end of bone at inner corner of Eb4-Cb4 joint, posterolaterally meeting Ad5 on Eb4. Ad5 dorsoanteriorly on Cb4 posterolaterally, pos- teroventrally on Cb5 dorsolaterally, anterior and pos- terior surfaces almost encapsulated by CT extending ventrally from PP. SOD present. Remarks. Johnson (1993:9) reported that Toxotes lacks SOD, but did not indicate the material on which he based his observations. He based it, however, on a set of gill arches that are from the same lot (USNM 289931) as one of the specimens on which the pres- ent description is based. If so, the muscles are dam- aged in the area where SOD would be expected and it is not possible to decide whether SOD might have been present. Johnson (1993:9) also reported SOD absent in Scorpis, Kuhlia, Kyphosus, and Pholidi- chthys, but, similarly, did not cite his material. John- son based his observations on the following dissec- tions: USNM 218922 (Scorpis), 289925 (Kuhlia), and 366512 (Kyphosus), all of which are damaged in the area where SOD might be expected to occur. Of these genera, and based on specimens dissected for the present study, we have found SOD lacking only in Pholidichthys. Our observations on these three genera are based on the following specimens: Scorpis sp. (USNM 339348) and Kyphosus (USNM 366512), both not otherwise discussed in the present study, and specimens cited in the description of Kuhlia (Kuhli- idae). RDs adjacent. Additional remarks. SCL attached mid-dorsally to tip of ventrally extending cartilaginous posterior end of Bb3. TV4 free from Cb5s. Pb! bony with cartilage ends. Pb4 and UP4 present. Very reduced Eb4 flange present. PLESIOPIDAE Assessor macneilli Whitley, USNM 269466, 47.7 mm; USNM 274581, 51.7 mm. Plate 129 Additional material. @ = Paraplesiops poweri Ogil- by, USNM 274579, 92.2 mm. ® = Acanthoclinus fuscus Jenyns, USNM 339246, not measured. Description (based on smaller specimen of Assessor with remarks based on larger specimen). LEI on dorsoposterior rim of Ebl beginning on uncinate process ventroanterolaterally and extending a short distance laterally. LE2 on dorsalmost edge of raised, bony dorsopos- terior rim of Eb2, well lateral to TEb2 insertion. @ Inserts posterior to distal end of TEb2. LE3 on Eb3 immediately ventroanterior to carti- lage tip of uncinate process. LE4 finely, tendinously on Eb4 dorsally a little me- dial to dorsodistalmost end of bone, there meeting LP insertion (in smaller specimen, muscle inserts above levator process, which is absent in larger spec- imen). @) @ Muscle massive, inserts broadly on Eb4 dorsally beginning laterally just anterior to levator process in ® (process absent in @), meeting OD3—4 posterodistally on Eb4; fusing with LP ventroanter- iorly (or ventromedially, depending somewhat on viewing orientation) a little dorsal to insertion. LP narrowly on dorsodistalmost end of Eb4 bony surface, meeting LE4 insertion ventroposteriorly. @ Muscle massive, ventroposteriorly joining raphe with Ad5 dorsally (thus, forming an unusual LP-Ad5 sling), fusing ventroanteriorly with LE4. @ Massive like @, but does not form LP-Ad5 sling. LI1 larger than LI2, ventrally becoming moderate- ly long, slender tendon passing ventral to TPb2 and inserting at junction between medial edge of Pb2 and adjacent impinging lateral edge of Pb3. @ @® Mus- culously and tendinously on Pb2 dorsoanteriorly pos- terior to anteriormost edge, tendinous portion con- tinuing medially as strong, slender tendon inserting on Pb3 dorsal facet mid-dorsoanteriorly. LI2 on Pb3 dorsoposterolaterally just medial to joint with Eb3; medial edge of insertion separated from lateral edge of TPb3-Eb3 in one specimen, meeting lateral edge of TPb3-Eb3 in other specimen. @ Medial edge meets lateral edge of TPb3-Eb3. TD flat, comprises TPb2, TEb2, and TPb3-Eb3. TD with mid-longitudinal raphe, which gives rise dorsally to filmy CT sheets covering TD. TPb2 an anterolaterally extending muscle on each side, in- serting on dorsoanterior surface of Pb2, almost com- pletely anterior to TEb2. TEb2 dorsoanteromedially joining TPb2 ventrally; main anterior portion contin- uous dorsally from one side to the other ventral to TPb2, lesser posterior portion interrupted medially 156 and joining lateral margin of small, roundish CT pad overlying naked, flat Pb3 facet posteriorly; muscle extends laterally, attaching to Eb2 dorsally medial to LE2 insertion; muscle slip (absent in larger speci- men) arises from TEb2 mid-anteriorly on right side and extends medially as fine tendon, which inserts on Pb2 near LI1 insertion. TPb3-Eb3 mid-anteromedi- ally continuous by CT with TEb2, attaches to Pb3 dorsoposteriorly along LI2 insertion and has fine CT attachment to anteromedialmost edge of Eb3; muscle continuous posteriorly by fine, diagonal muscle strands with SOD. Remarks. The flat Pb3 dorsal facet forms the dor- somedialmost surface of Pb3. In the larger specimen of Assessor, TEb2 is continuous only by CT across the facets, the anterior one-third of each facet is ven- tral to TPb2; the posterior two-thirds are naked mus- culously. In the smaller specimen, only a small area of the facet is naked. @ Mid-dorsal CT area giving rise dorsally to tough CT sheets; most of Pb3 dorsal surface not covered by muscle; muscle fibers of anterior portion of TPb2 more-or-less transversely oriented, those of postero- lateral portion thinner, more longitudinally oriented, two portions meet at fine CT notch laterally near point where ventroanterior edge of LI] impinges on TPb2. TEb2 extends laterally to point anterior to LE2 insertion. @ TPb2 and TEb2 musculously continuous across Pb3s; TEb2 like Assessor. OD3-4 origin on Pb3 along and ventral to lateral edge of flat, dorsal Pb3 facet (ventral to roundish CT pad), insertion on Eb3 uncinate process dorsoanter- iorly and Eb4 uncinate process medially. @ ® Muscle massive. OP in two parts, medial part dorsally on Eb4 pos- teriorly beginning just ventral to uncinate process and extending short distance laterally, posteriorly over- lapping medial edge of lateral OP part, which extends laterally on Eb4 to below medial edge of LE4-LP insertions; ventrally two parts join raphe with ventro- posterior surface of Ad5 dorsal to attachment to Cb5, and attach to dorsal surface of small, posterodistally projecting bony Cb5 process that extends laterally past medial margin of cartilage tip. OP of larger spec- imen comprises single part occupying same area as two parts of smaller specimen. @ Lateral portion much thicker than medial portion, overlying medial portion posteriorly. ® Comprises single part. Ad1-3 absent; fine GFMs present. @ GFMs mod- erately well developed, that on arch 2 extending dor- sally on Eb2 and just meeting distal end of TEb2. Ad4 relatively small, dorsally on Eb4 ventropos- teriorly lateral to lateral edge of OP, ventrally, nar- rowly on Cb4 medial to Eb4-Cb4 joint. Ad5 dorsally moderately narrowly on Cb4 pos- terolaterally; ventrally very narrowly on Cb5 dorso- BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON distally. @ Very well developed, joining raphe dor- sally with LP. @ Like Assessor. SOD relatively fine. RDs narrowly separate. Additional remarks. SCL present. TV4 free from Cb5s. Pb4 absent, UP4 present. Pbl mostly bony. Pb2 toothed. IAC present. Although, TV4 is free from Cb5s in Paraplesiops, it consists of two layers, dorsal and ventral. The ven- tral layer is musculously continuous and unmodified as it passes across the Cb5s, whereas, the dorsal layer changes to thick CT as it crosses the flattened ven- troanterior tips of the Cb5s. Bony Eb4 flange present. Medial end of Eb4 smaller than that of Eb3, also true of Trachinops (USNM 269557), Plesiops (USNM 264268), and Acanthoclinus (USNM 200546), all cleared and stained specimens, and @, but medial end of Eb4 is larger than medial end of Eb3 in Paraple- SIODPS. Mooi (1993:291) hypothesized the phylogeny of the Plesiopidae. He recognized Trachinops as com- prising the basalmost clade of the family, Assessor as comprising the next clade, and Paraplesiops and Acanthoclinus, separately, as members of the next, and final two, clades. We, therefore, consider that the state of the medial end of Eb4 in Paraplesiops is apomorphic for that genus, as is the position of the lateral end of TEb2 reaching laterally anterior to LE2 insertion, which is usually unspecialized for perco- morphs. On the other hand, the musculously uninter- rupted condition of TPb2 and TEb2 of Acanthocli- nus, which generally appears to be plesiomorphic in perciforms, is apomorphic for Acanthoclinus. PERCIDAE Perca flavescens (Mitchill), USNM 193135, 91.7 mm. Plate 130 Additional material. @ = Percina caprodes (Rafin- esque), USNM 230758, 2:101—-103 mm. @ = Per- cina sciera (Swain), USNM 161594, 91.5 mm. Description. LE1 on Eb! uncinate process just lateral to carti- lage tip. ® @® Uncinate process appears to have moved medially and fused with cartilaginous medial end of Eb1; anterior arm of Eb1 represented by slight cartilaginous bulge of ventromedial end in @, which does not join Pb2; bulge absent in @; muscle inserts on raised bony process on Eb1 mid-dorsoposteriorly. LE2 on bony dorsoposterior surface of Eb2. LE3 on cartilaginous tip of Eb3 uncinate process. LE4 dorsoposteriorly on Eb4 just lateral to unci- nate process. LP on Eb4 at and lateral to LE4 insertion. NUMBER 11 LI1 mainly on Pb2 dorsoanteriorly, but few muscle strands appear to continue into CT covering Pb2 and anterior end of closely joined Pb3. @ ® Mainly on Pb2, but some muscle strands clearly continue onto Pb3. LI2 on Pb3 dorsoposteriorly. TD comprises TPb2, TEb2, and TPb3-Pb4-Eb3. TPb2 thin, flat, laterally curving ribbon of muscle on each side arising from TEb2 at anterior end of mid- longitudinal raphe and continuing to posterior end of raphe; not attached to Pb2. TEb2 with mid-anterior and mid-posterior notches joined by mid-longitudinal raphe giving rise dorsally to sheets of CT and con- necting ventrally to CT of pharyngeal roof; antero- lateral and posterolateral converging fibers fusing be- fore attaching on most of dorsal surface of Eb2. TPb3-Pb4-Eb3 on Pb3 dorsally just medial to LI2 insertion, extending onto dorsomedial end of Eb3 and, weakly, onto dorsal surface of Pb4, posteriorly continuous by diagonal slip of muscle with SOD. @ TPb2 thin, flat, laterally curving ribbon of muscle on only one side of each of two specimens. @ @ TEb2 scarcely notched posteriorly; TPb3-Pb4-Eb3 on Pb3 dorsoposteriorly, Pb4 dorsally, and Eb3 dorsoposter- omedialmost end. OD3-4 origin broadly on Pb2 posterior to LI] in- sertion and continuing broadly medially onto Pb3 ventral to TEb2, inserts on anterior surface of Eb3 uncinate process and anterior surface and medial edge of Eb4 uncinate process. OP dorsally on all of bony posterior surface of Eb4 medial to LE4 insertion, ventrally on Cb5 bony sur- face beginning medial to distal end, overlapping pos- teriorly much of Ad5 posterior surface and joining Ad5 attachment to Cb5. M. Pb2-Eb2 absent. @ On one side of one speci- men, small strip of muscle extends from lateral edge of Pb2 to anterior edge of Eb2 anteroventral to TEb2. ® On both sides, relatively broad strap of muscle originates on lateral edge of Pb2 ventral to OD3—4 and attaches to anterior edge of Eb2 ventral to TEb2. Ad1—3 absent. @ ®@ Ad1 on dorsoanterior surfaces of, and covering joint between, Ebl and Cb1; Ad2 beginning medially at raphe with lateral end of TEb2, extending laterally and spreading anterolaterally and attaching to anterior surfaces of Eb2-Cb2 at joint; Ad3 broad dorsomedially, on dorsolateral surface of Eb3, spreading anterolaterally and attaching to ante- rior surfaces of Eb3-Cb3 joint. Remarks. GFMs (not illustrated) present in Perca in positions similar to Ad1—3, which attach to GFMs in @ and ®, but relatively weakly developed, hence, difference in interpretation. Ad4 broadly dorsally on posterior surface of Eb4 lateral to OP; ventrally, less broadly on Cb4 dorsal surface anterior to Eb4-Cb4 joint, joining Ad5 at- tachment on Eb4. 157 Ad5 on Cb4 posterolateral surface and Cb5 dor- soposterior surface beginning lateral to OP attach- ment and extending medially anterior to OP attach- ment. SOD present. RDs slightly separated. Additional remarks. SCL attached mid-dorsally to tip of cartilaginous ventroposterior end of Bb3. TV4 free from Cb5s. Pb4 and UP4 present. IAC well de- veloped. LE4 levator process absent. @) @ Pb1 absent. @ JAC a tiny sphere on both sides in larger specimen, absent in smaller specimen. ® IAC absent. CEPOLIDAE Cepola rubescens Linnaeus, USNM 285452, ca. 260 mm. Plate 131 Additional material. @ = Acanthocepola limbata (Valenciennes), NSMT P. 64733, 375 mm. Description. Remarks. All levators and RDs relatively slender. LEI origin by short, fine tendon; insertion on dor- soanterior surface of Eb1l just ventral to cartilage tipped process lateral to uncinate process, which ar- ticulates with IAC; dorsoanterior surface of LE1, and entire anterior surface of Pbl, attaches to posterior surface of CT arising from anterior edge of Eb1. LE2 on dorsoposteriormost edge of Eb2 posterior to lateral end of TEb2. LE3 on tip of Eb3 uncinate process. LE4 on bony surface of Eb4 immediately medial to cartilage tip of levator process, posteriorly joining LP insertion. @ On levator process together with LP. LP beginning on tip of levator process and ex- tending medially, joining LE4 insertion posteriorly. @ On levator process with LE4. LI1 divides into anterior and posterior branches as it passes over Pb2 articulation with IAC, anterior branch inserts on dorsalmost surface of Pb2, posterior branch, inserts on Pb2 dorsally well ventral to ante- rior branch and continues onto adjacent lateral edge of Pb3 anterior process. @ IAC absent on both sides; insertion divided on only one side; insertion on other side passes posterior to Eb2-Pb2 joint and inserts like posterior branch of other side. LI2 on Pb3 dorsoposterolaterally immediately me- dial to medial end of Eb3 anteriorly. TD comprises TEb2, M. TEb2-Pb2, and TPb3- Eb3. TEb2 with mid-longitudinal raphe giving rise to CT pad dorsally and attaching anteriorly to CT of pharyngeal roof and laterally on Eb2 dorsally anterior to LE2 insertion; muscle is joined ventromedially by M. TEb2-Pb2, and is continuous mid-posteroventral- ly by fine, diagonal muscle strap with TPb3-Eb3 dor- 158 soanteriorly (® TEb2 not continuous with TPb3- Eb3). M. TEb2-Pb2 broad, strap-like, originating ventrally from TEb2 posteriorly and extending ante- riorly along mid-longitudinal raphe for about half length of raphe (@ muscle originates along mid-pos- terior half of raphe); muscle extends ventroanterola- terally dorsal to OD3—4 origin and inserts on Pb2 dorsomedially just posterior to base of dorsoanterior process. TPb3-Eb3 on Pb3 dorsally beginning ante- rior to joint with medial end of Eb3 and extending posteriorly and attaching to posteromedial corner of Eb3 on both sides, but on one side a fine, probably anomalous, muscle strand extends to opposite side and attaches to anteromedial edge of Eb4. (@ No at- tachment to Eb4); muscle continuous posteriorly with SOD. @ TPb3-Eb3 joins raphe posterolaterally on each side with SO muscle strap passing dorsoanter- iorly immediately lateral to OP on Eb4. Remarks. M. TEb2-Pb2 is not present in any other acanthomorph we examined. It is difficult to equate the muscle with TPb2, which arises dorsal and/or an- terior to TEb2. If present in Owstonia and/or Sphen- anthias, M. TEb2-Pb2 will either represent another synapomorphy of the Cepolidae or it will define a monophyletic group within the family. OD3-—4 origin of Pb3 dorsally ventral to M.TEb2- Pb2, beginning a little posterior to anteriormost tip of Pb3; insertion on Eb3 anteriorly just ventral to tip of uncinate process and on bony medial edge of Eb4 uncinate process. OP strap-like; dorsally on Eb4 posteriorly begin- ning medially ventral to tip of uncinate process and extending laterally about half distance between un- cinate and levator processes (@ extends to or almost to levator process); ventrally on Cb5 beginning lat- erally on distal cartilaginous end medial to Ad5, and extending medially a short distance and meeting TVS. Ad1-—3 absent. GFM1—3 moderately developed. Ad4 on Eb4 dorsoposteriorly beginning medially just medial to lateral edge of OP and extending lat- erally to distalmost bony edge, ventrally on Cb4 dor- sally beginning at Eb4-Cb4 joint and extending me- dially a short distance. Ad5 short, ventrally on Cb5 posterodistally, dor- sally on Cb4 posterodistally and AC4 posteriorly. @ does not join AC4. SOD slender. RDs separated by distance greater than diameter of one RD. @ Separated by about one-half RD diameter. Additional remarks. SCL attached mid-dorsally to ventrally extending posterior cartilaginous tip of Bb3. TV4 free from Cb5s. Pb4 and UP4 present. ® SCL attached mid-posteriorly to ventroanteriormost sur- face of ventrally extending posterior cartilaginous tip of Bb3. IAC small, ball-like; AC4 relatively small. @ IAC absent; AC4 relatively large. BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON The presence of two uncinate processes on Eb1 is unique among the Actinopterygii (the tip of the la- teralmost process is attached by CT to Eb2 anteri- orly). Gill and Mooi (1993:331 and fig. 2) noted that two rod-like uncinate processes on Eb] support monophyly of the Cepolidae and Owstoniidae, which they recognized, as do we, as a single family, Ce- polidae. Nelson (1994:378), however, recognized both families in a superfamily Cepoloidea. CALLANTHIIDAE Callanthias allporti Giinther, USNM 350940, 129 mm. Plate 132 Additional material. Callanthias australis Ogilby, AMS I.18079—002, 87.9 mm. @ = Grammatonotus laysanus Gilbert, BPBM 22757, 126 mm. Description. Remarks. We noted no substantive differences in the musculature between the two species of Callan- thias, or in the skeletal structure in both genera. Dif- ferences reported are for Grammatonotus. LEI! on dorsal edge of Eb1l immediately lateral to broad uncinate process, continuing very slightly onto CT joining Eb! and Eb2. LE2 on posterodorsal edge of Eb2 a little lateral to mid-length and continuing on CT between Eb2 and Eb3; muscle is cleanly divisible into two sec- tions, one on bone and one on CT. LE3 on tip of Eb3 uncinate process. LE4 on tip of Eb4 levator process. LP on Eb4 levator process at and lateral to LE4 insertion. LI1 on dorsal bony edge of Pb2 uncinate process just posterior to cartilaginous tip. LI2 on Pb3 dorsolaterally at and anterior to medial end of Eb3. TD comprises TEb2, and TPb3-Pb4-Eb3. Thick CT pad attaches anterolaterally to anteriormost end of Pb2 and mid-anteroventrally to anterior end of mid-longitudinal TEb2 raphe, which joins CT of pha- ryngeal roof. TEb2 attaches ventrally along mid-lon- gitudinal raphe to CT of pharyngeal roof; muscle ex- tends laterally attaching along most of Eb2 dorsal surface; TEb2 not continuous with TPb3-Pb4-Eb3. TPb3-Pb4-Eb3 on Pb3 dorsal bony surface medial to medial end of Eb3, continuing posteriorly onto pos- teromedialmost edge of Eb3 and adjacent dorsome- dial edge of Pb4, continuous posteriorly by diagonal muscle straps with SOD. @ Attachment to medial end of Eb3 absent, hence TPb3-Pb4. OD3-—4, OD3’ origin on dorsoanterior surface of Pb3 ventral to TEb2, splitting shortly into dorsal (OD3) and ventral (OD4; not illustrated) sections. NUMBER 11 Dorsal section completely overlies ventral section, hiding it in dorsal view; dorsal section splits laterally with ventroanterior portion (OD3’) inserting on Eb3 dorsal surface well ventral to uncinate process and dorsal portion (OD3) inserting on anterior surface of uncinate process. Ventral section inserts on Eb4 an- terior surface medial to levator process. @) OD3’ ab- sent; OD3-—4 attachment on Eb4 only to tiny section of dorsal edge of Eb4 ventromedial to levator process (Eb4 section essentially absent). Remarks. The Eb3 uncinate process is tightly joined to Eb4, which has lost the uncinate process. That the Eb4 process posterior to this joining is not a displaced uncinate process is evidenced by the in- sertions of LE4 and LP on the process. These two muscles rarely, if ever, insert on the Eb4 uncinate process. OP broadly on posterior surface of Eb4 medial to base of levator process and broadly on posterodistal surface of Cb5 posterior to Ad5S. Ad1-—3 moderately developed, but gill-filaments at- tach to these muscles and they could be interpreted simply as well-developed cord-like GFMs, present on Ebs and extending only to dorsoanteriormost surface of respective Cbs; however, AdsIl—3 well developed in @). Ad4 broadly on posterior surface of Eb4 beginning at levator process and extending laterally, ventrally on posterodorsal surface of Cb4 medial to Eb4-Cb4 joint. Ad5 small, on posterodistal surface of Cb5, partly anterior to OP ventral attachment, and on distalmost cartilaginous tip of Cb4 posteriorly. SOD present. RDs well separated. Additional remarks. SCL attached mid-dorsally to tip of ventrally extending posterior cartilaginous end of Bb3. TV4 free from Cb5s. Pb! cartilaginous. Pb4 and UP4 present IAC present. Eb4 uncinate process absent. @ SCL questionably absent (should be veri- fied in another specimen); Pbl cartilaginous, very slender, embedded in tough CT; one side in two seg- ments. GERREIDAE Gerres cinereus Eigenmann, USNM 331928, 90.4 mm; USNM 337861, 2: 60.9—95.7 mm. Plate 133 Additional material. @ = Eucinostomus sp., USNM 342106, 92.7 mm. Description. LEI tendinously and musculously on anterior sur- face of Eb1 uncinate process just ventral to cartilag- inous tip. 159 LE2 narrowly tendinously and musculously on tip of bony process on Eb2. LE3 tendinously and musculously on dorsomedial bony edge of Eb3 uncinate process. LE4 tendinously on mid-dorsoposterior surface of Eb4, fusing with medial end of LP insertion. @ In- sertion not fused with that of LP. LP massive, on much of dorsoposterior edge of Eb4, fusing with LE4 insertion, meeting, but not join- ing OP dorsally. @ Smaller than LE4, insertion en- tirely lateral to that of LE4. LI1 tendinously and musculously on bony poste- rior surface of Pb2 dorsoanterior process ventral to cartilaginous tip, with tendon continuing ventrally to posterior surface of dorsoanteriormost Pb3 process. LI2 slender tendinously on Pb3 posterolaterally ventral to medial end of Eb3, which overlies Pb3. TD comprises TEb2 and TPb3-Eb3-Eb4. TEb2 pair of muscles, each member joined to lateral edge of thick CT pad covering Pb3 dorsal articulating fac- ets and attaching on medial half of Eb2 dorsal surface lateral to LE2 insertion. TPb3-Eb3-Eb4 on dorsopos- terolateral surface of Pb3, posteromedialmost edge of Eb3, and dorsomedialmost surface of Eb4. OD3-—4, OD3’. OD3—4 origin on lateral edge of Pb3 dorsal articulating facet, insertion on medial edge and anterior surface of Eb3 uncinate process and medial edge and posterior surface of Eb4 unci- nate process; just posterior to origin, OD3’ originates with OD3-—4, diverges well laterally and inserts on dorsal surface of Eb3 ventral to uncinate process, forming raphe with posteromedial end of Ad3; OD3’ absent in 60.9 mm specimen and @. OP massive, on much of posterior surface of Eb4 and extending ventrally and attaching by tendons to posterior surface of Cb5 somewhat medial to distal end; Ad5 fibers joining OP tendons anteriorly; slen- der, separate OP strap attaches dorsally narrowly to Eb4 medial to massive OP attachment and to Cb5 medial to Ad5 attachment. Ad1 relatively small, restricted, spans anterior sur- face of Eb1-Cb1 joint. Ad2 begins on Eb2 dorsoanteriorly adjacent to lat- eral end of TEb2 and expands laterally attaching over anterior surface of Eb2-Cb2 joint. Ad3 begins medially at raphe with lateral end of OD3’ and expands laterally attaching over anterior surface of Eb3-Cb3 joint. In smallest and largest specimens and @ occupies most of dorsal and an- terolateral surfaces of Eb3 (area of attachment equals that of OD3’ + Ad3 of 90.4 mm specimen). Ad4 on ventral surface of Eb4 and dorsal surface of Cb4 medial to Eb4-Cb4 joint; completely excluded from view posteriorly. Ad5 dorsally tendinously on distal end of Cb4, ten- don continues onto distal end of Eb4; muscle fusing 160 (larger specimen) or not anteriorly with OP, ventrally on Cb5 distal end. SOD present. RDs adjacent. Additional remarks. SCL attached mid-dorsally to ventroposteriorly extending cartilaginous tip of Bb3. TV free from Cb5s. Rosen and Patterson (1990:13) reported that Pb4 is absent in gerreids. Although it is reduced in size, we find that it and UP4 are present. IAC present. Eb4 levator process absent. Well-devel- oped bony flange (occluded from view by LP in Plate 133) at end of Eb4, completely overlies cartilaginous distal end of Eb4. Medial end of Eb4 varies in di- ameter from about equal to a little larger than that of Eb3, but is smaller than that of Eb3 in @ (two cleared and stained specimens of Eucinostomus, USNM 397354 and 306708, also examined for this charac- ter). GRAMMATIDAE Gramma loreto Poey, USNM 369710, 53.3 mm; 199487, 54.4 mm; 319222, 50.9 mm; 360710, 52 mm. Plate 134 Description. LE1 on dorsoposterior edge of Eb] lateral to un- cinate process. LE2 on dorsoposterior surface of Eb2 at about mid-length. LE3 on Eb3 uncinate process just ventral to car- tilaginous tip. LE4 on bony edge of Eb4 levator process just me- dial to cartilaginous tip. LP on bony edge and cartilaginous tip of Eb4 le- vator process, joining LE4 insertion posteriorly. LI1 broad; finely tendinously on anterolateralmost edge of Pb3 adjacent to medial edge of Pb2. LI2 on Pb3 dorsoposteriorly along medial edge of attachment of TPb3-Pb4 and near medial end of Eb3. TD comprises TEb2 and TPb3-Pb4. TEb2 a pair of muscles, each separated from (and continuous with) counterpart by CT band equal to width of area separating Pb3s; each element of pair comprising an- terior and posterior portions, anterior portion attaches to dorsal surface of Eb2 anterior to LE2 insertion, posterior portion attaches to posterior surface and edge of Eb2 medial to LE2 insertion; free from TPb3-Pb4. TPb3-Pb4 on Pb3 dorsal surface medial to LI2 insertion and on anterior edge of Pb4, contin- uous posteriorly by diagonal strands of muscle with SOD. OD3-4 origin on dorsal surface of Pb3 ventral to TEb2, insertion on dorsoanterior surface of Eb3 un- cinate process and medial edge of Eb4 uncinate pro- cess. OP dorsally on Eb4 posterior surface beginning BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON below LP and extending medially for variable dis- tance (unclear separation from SO on one side), ven- trally on Cb5 dorsodistally posterior to Ad5. Ad1-—3 absent. Ad4 dorsally on Eb4 beginning below LP insertion and continuing laterally to near distal end of Eb4, ventrally on Cb4 dorsally medial to Eb4-Cb4 joint. Ad5 on dorsoposterior end of Cb5 anterior to OP and on posterior surface of Eb4-Cb4 joint. SOD present. RDs separate. Additional remarks. SCL free from Bb3 (cartilag- inous posterior end not elongate). TV4 ventrally con- tinuous across Cb5s, but dorsally attaching to Cb5s (necessary to bisect muscle ventrally to expose at- tachment). Pb4 present. UP4 present. IAC present. OPISTOGNATHIDAE Lonchopisthus higmani Mead, USNM 186058, 73.1 mm; USNM 292182, 85.1 mm. Plate 135 Additional material. @ = Opistognathus darwiniensis Macleay, USNM 174042, 60.5 mm. Description. LE1 on anteroventral surface of Eb] uncinate pro- cess. LE2 on dorsoposterior surface of Eb2. LE3 on tip of Eb3 uncinate process anteriorly. LE4 on Eb4 adjacent to tip of levator process an- teriorly. @ Broadly on Eb4 dorsoposteriorly (levator process absent). LP at and posterior to LE4 insertion. LI1 on tendinous covering of anterior end of Pb2 dorsolaterally, covering continues medially over ad- jacent anterior end of Pb3. Remarks. Articulation of Pb1l with anteromedial tip of Eb1 joins tiny CT pad that covers joined an- terior ends of Pb2 and Pb3. Pb2 edentate. @ Condi- tion of preparation prevented determining if tiny CT pad was present. LI2 on Pb3 dorsoposterolaterally, lateral to attach- ment of TPb3-Eb3-Eb4 and just medial to joint with medial end of Eb3. TD comprises TEb2 and TPb3-Eb3-Eb4. TEb2 a pair of muscles connected by tough CT across flat dorsal Pb3 surfaces, CT continuous anteriorly be- tween Pb3s with CT of pharyngeal roof; muscle ex- tending laterally on Eb2 dorsal surface anterior to LE2 insertion. TPb3-Eb3-Eb4 on Pb3 dorsoposter- iorly well medial to LI2 insertion on right side and at medial edge of insertion on left side, with tendi- nous attachments (not illustrated) to medialmost ends of Eb3 and Eb4, continuous mid-posteriorly by cross- ing muscle strands with fine SOD. @ Thick CT pad over Pb3 dorsal surfaces; pad continuous with CT NUMBER 11 joining TEb2 muscular portions, partially separable from more ventral CT covering and attaching directly to Pb3s. OD3-—4 origin ventral to TEb2 on posterolateral edge of Pb3 flat dorsal surface, insertion broadly on anterior surface of Eb3 uncinate process and medial edge of Eb4 uncinate process, cartilage tip of which is extremely small in smaller specimen and absent in larger specimen. @ On anterior surface of Eb3 un- cinate process and dorsal edge of Eb4, which is at- tached to Eb3 uncinate process (no Eb4 uncinate pro- cess). OP dorsally on posterior surface of Eb4 beginning at levator process and continuing well medial to un- cinate process, ventrally broadly on Cb5 beginning laterally a little medial to distal end. @ Dorsally, broadly on posterior surface of lateral half of Eb4, posteriorly overlapping and almost completely ex- cluding Ad4 from posterior view. Ad1-—3 absent. Ad4 dorsally on Eb4 posteriorly beginning at le- vator process and extending laterally to just medial to distal end, ventrally on Cb4 dorsally medial to Eb4-Cb4 joint, fusing there with anterior surface of Ads. Ad5 dorsally on posterodistal end of Cb4, ventrally on Cb5 dorsodistally. SOD present. RDs slightly separated. Additional remarks. SCL present. TV4 two lay- ered, ventral layer free from Cb5s; dorsal layer di- vided mid-longitudinally, each portion attached to re- spective Cb5 lateral surface. IAC present @ Each dorsal portion attaches to ventral edge of respective Cb5. Pb4 and UP4 present. PSEUDOCHROMIDAE Labracinus cyclophthalmus (Miller and Troschel), USNM 290900, 2 specimens, 78.3—82.4 mm; USNM 345601, 90.7 mm SL. Plate 136 Additional material. ®@ = Pseudochromis porphyreus Lubbock and Goldman, USNM 290595. ® = Pseudochromis persicus Murray, USNM 147902. Description. Remarks. The muscles of both species of Pseu- dochromis are essentially the same as those of La- bracinus. LE1 on Eb! posteriorly just lateral to base of un- cinate process. LE2 on raised posterior margin of Eb2. LE3 on and lateral to Eb3 uncinate process ante- riorly. LE4 massive, on Eb4 dorsolateral surface. LP narrowly on Eb4 dorsal surface at and posterior 161 to LE4 insertion, completely covering minute tip of levator process. @ Levator process very reduced on one side, absent on other. ® Levator process like La- bracinus. LI1 tendinously on fascia covering anterior end of Pb2 (which overlaps Pb3) and continuing onto dor- solateralmost edge of Pb3. LI2 on Pb3 dorsoposterolaterally just anterior to medial end of Eb3. TD comprises TPb2a, TEb2, and TPb3-Eb3 (TPb3-Eb3-Eb4 in largest specimen). TPb2a on an- terior surfaces of Pb2s, with mid-vertical raphe con- tinuing posteriorly and joining thick CT pad covering most of dorsal surfaces of Pb3s, which are only partly roofed by muscle. TEb2 a pair of muscles joined me- dially by broad area of CT and joining mid-ventral surface of CT pad overlying Pb3 articulating facets; muscle extending laterally and attaching to Eb2 dor- soanteriorly unusually well lateral to LE2 insertion; TEb2 not continuous posteriorly with TPb3-Eb3. TPb3-Eb3 on Pb3 dorsoposteriorly ventral and lateral to articulating facet and posteromedial end of Eb3 (and finely, tendinously on Eb4 dorsoanteromedially in largest specimen); continuous posteroventrally by diagonal muscle strand with SOD. CPb originates posteriorly from longitudinal SO layer, extends anteriorly between Pbs, with continu- ous major branch extending anterolaterally around Pb2s and attaching to each UP4 anterolaterally; mi- nor branch extends posteromedially from continuous major branch, passes medial to Pb3 and attaches to UP4 medially. OD3-—4 origin on Pb3 dorsoposteriorly ventral to TEb2, insertion broadly on anterior surface of Eb3 uncinate process and medial edge of Eb4 uncinate process. OP dorsally on most of Eb4 posterior surface, ven- trally broadly on Cb5 dorsoposteriorly. Ad1 anterodistally on Eb! and Cbl. Ad2 medially joins raphe with distal end of TEb2 on anterior edge of Eb2 and extends laterally attach- ing on anterior surfaces of Eb2 and Cb2 at Eb2-Cb2 joint. Ad3 medially on Eb3 dorsally ventral to lateral portion of OD3—4, laterally attaching to anterior sur- faces of Eb3 and Cb3 at Eb3-Cb3 joint. Ad4 broadly on Eb4 ventrally almost entirely an- terior to OP, ventrally broadly on Cb4 medial to Eb4- Cb4 joint. Ad5 dorsally on Eb4-Cb4 joint posteriorly, con- tinuing, ventrally on Cb5 broadly anterodistally an- terior to OP. SOD slender. RDs adjacent. Additional remarks. SCL present. TV4 dorsal sec- tion attaches to Cb5s anteriorly, ventral section con- tinuous across anteroventral ends of Cb5s. Pb4 and 162 UP4 present. Pb2 toothed. Eb4 levator process with minute cartilage tip. CbS5 bears a tiny bit of cartilage at or near its distal end in Labracinus, but is usually completely ossified in Pseudochromis, although there is a “‘spot” of car- tilage on one Cb5 laterally well anterior to its distal end in ®. Labracinus has a tiny IAC, which is absent in @ and ®. Labracinus has a tiny, cartilaginous Pb1. @ Pbl has relatively long cartilaginous dorsal and ventral ends separated by small central ossified area. @ Pbl state unknown. LEIOGNATHIDAE (See additional remarks following description for discussion of valid family name for this taxon.) Gazza sp., USNM 268914, 2 specimens, 68.6—69.2 mm. Plate 137 Additional material. @ = Leiognathus equula Forss- kal, USNM 349512, 58.6 mm. ® = Secutor insi- diator (Bloch), USNM 329585, 70.9 mm. Description. LEI! broadly on Ebldorsolaterally and on anterior and medial surfaces of large, dorsally expanded tri- angular process at lateral end of Eb! (lateral surface of process concave, receives gill filament bases; pro- cess apparently not an uncinate process as similar process is serially present on Eb2 and gill filaments are not known to attach to Eb1l uncinate process; car- tilaginously tipped uncinate process absent); muscle with lateral tendinous portion extending to origin. @ Like Gazza, but lateral edge of Ebl not expanded nearly as much. ® Narrowly, tendinously on EbI| posterolaterally medial to scarcely expanded lateral end of Ebl1. LE2 on medial edge of large, dorsally expanded triangular process at lateral end of Eb2; ventrolater- ally joining raphe with posterolateral edge of TEb2; gill filaments attach to concave lateral edge of tri- angular process. @ @® Lateral end of Eb2 not ex- panded dorsally nearly as much as in Gazza. LE3 musculously and, variably, tendinously on bony portion of Eb3 uncinate process anteriorly (pro- cess has minute cartilage tip posteromedially, visible only after separation from Eb4 uncinate process). LE4 + LP fuse, well dorsal to Eb4, along tendi- nous raphe and form arms of a Y; raphe continues along anterior edge of musculous shank of Y, which is presumably formed by lateral section of OP, and attaches between uncinate process and lateral end of Eb4 posteriorly; another tendinous raphe extends along lateral section of OP posteriorly and attaches to hook-like cartilaginous distal end of Cb4; latter tendinous raphe on Cb4 expands into CT sheet that continues dorsoposteriorly as PP (not illustrated); BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON dorsoposteriorly, ventromedial portion of medial OP section attaches to Cb5 posterolaterally. @ Tendinous continuation from fused LE4 + LP continues along anterior edge of lateral OP section and attaches to posterodistalmost bony edge of Eb4, OP continues ventrally and meets Ad5 on Cb5; tendinous raphe absent along lateral section of OP posteriorly. ® LE4 + LP fuse along strong tendon that divides before attaching between uncinate process and distal end of Eb4, tendon divides ventral to attachment, with lat- eral section of OP attaching to it. LI1 on Pb2 anterolaterally well ventral to cartilage- tipped dorsally extending process of Pb2. LI2 tendinously on Pb3 just medial to joint with posteromedial end of Eb3. TD comprises TEb2 and TEb3-Eb4. TEb2 com- prises a muscle on each side joined medially by ten- dinous CT to ventrolateral edge of a thick, cone- shaped CT pad (attenuated end ventral) that attaches medially to a conforming dorsal shelf of Pb3 and sits freely in a conforming depression of dorsoanterior surface of OD3-—3'; pads on either side joined by CT dorsally across dorsal edge of conforming Pb3 shelf; TEb2 attaches on Eb2 dorsally, joining LE2 ven- troanteriorly and extending well lateral to it; muscle free from TEb3-Eb4. TEb3-Eb4 on Eb3 posterior edge medial to base of uncinate process, continuing posteriorly and passing dorsal to medial end of Eb4 but attaching ventrally to posteromedial edge of Eb4 and, on left side only of one specimen, joining dor- somedial edge of OP medial section; muscle ven- troanteriorly continuous by diagonal muscle strand (obscured from view in illustration) with SOD. @ ® TD comprises TEb2 and TEb3; TEb3 attaches on me- dial edge of Eb3 uncinate process ventral to OD3 insertion. OD3, OD3’ massive, joint origin on lateral surface of conforming Pb3 dorsal shelf (see description of TEb2), dividing laterally with small OD3 section in- serting on Eb3 uncinate process dorsoanteriorly, and large OD3’ section inserting separately on remainder of Eb3 uncinate process anteriorly immediately ven- tral to OD3, insertion continues ventrally covering all of bony dorsolateral surface of Eb3 (superficially ap- pears as if OD3 and OD3’ are entirely fused; sepa- ration most distinct in @). OD4 absent. OP presumably in two sections, for lateral section see LE4 above; medial section dorsally, broadly on Eb4 posterolaterally, ventrally more broadly on Cb5 posteriorly, joining posterior raphe of OP lateral sec- tion with Ad5 posteriorly; lateralmost portion of me- dial OP section joining tendinous raphe with lateral OP section on posterodistal end of Cb4 where raphe expands into CT sheet which continues to PP (not illustrated). @ Posterior raphe of lateral OP section NUMBER 11 absent; two OP sections join on Cb5; CT of PP at- taches to distal end of Cb5. M. SO-Pb2 strap of SO longitudinal fibers extend- ing anteriorly and inserting on Pb2 posteroventrally, passes posteriorly along Pb3 medially and then ven- tral to TEb3-Eb4 and, on left side only, divides, with one portion inserting on Eb4 at and medial to OP medial section and the other becoming continuous with SO and SOD anteriorly; on right side, Eb4 in- sertion is absent. Remarks. This muscle does not insert on Eb4 in Leiognathus (only one side is adequate for determi- nation) or on one side in Secutor, but does insert on Eb4 on the other side. Ad1-3 absent. Ad4 relatively small, scarcely if at all visible ex- ternally, on Eb4 ventrally and Cb4 dorsally medial to Eb4-Cb4 joint. Ad5 small, on dorsodistal end of Cb5 and cartilag- inous finger-like process at distal end of Cb4, joining short raphe there with ventromedial end of OP lateral section and ventrolateral end of OP medial section. SOD slender. RDs well separate anteriorly, swelling posteriorly and becoming proximate, each consisting of strap- like lateral section, RD’, and much larger medial sec- tion, RD. RD’ anteroventrally continuous (infiltrat- ed?) with CT of SO and inserting on UP4 and Pb4 posteriorly; RD with minor insertion on posterome- dial edge of UP4 and major insertion on Pb3 poste- riorly. Additional remarks. SCL weakly attached mid-an- teriorly to posteroventrally extending cartilaginous end of Bb3 (SCL easily separated from Bb3). TV4 mostly continuous ventrally, but dorsoanterior fibers attach to anterolateral surfaces of Cb5s (hence similar to condition in to many other fishes that have an LE4- OP sling; e.g., labrids, cichlids, centrogenyids). Pb4 and UP4 present. [AC absent. Eb4 with small, bony flange dorsally at distal end. Dorsal end of Pb1 bony; . @ Dorsal end cartilaginous. Eb1 uncinate process absent. Eb3 uncinate process with minute cartilagi- nous tip. @ Cartilage tip of uncinate process normal. @® Tip of process bony on both sides. Eb4 uncinate process present. ® Uncinate process on one side with minute cartilage tip, bony tip on other side. Eb4 le- vator process absent in all. Correct family name. Bleeker (1859:xxii1) coined the family name Equuloidei, in which he included fishes currently recognized in the families Leiognath- idae and Menidae. Equuloidei was emended to Equulidae by Gill (1893:134), which Gill listed as a synonym of ““Liognathidae Gill, 1892.” We were un- able to find an 1892 or earlier publication of Gill’s that refers to Liognathidae. Equuloidei Bleeker was based on the genus Equula Cuvier, which is a junior synonym of Leiognathus 163 Lacepéde; however, Equulidae has date seniority over Liognathidae Gill (first emended as Leiognathidae by Jordan and Evermann, 1902:338). Hence, Equulidae has seniority and, ostensibly, should replace Leiog- nathidae, but as far as we can tell has not been used as a senior synonym since before 1899. Leiognathi- dae has been used as a valid senior synonym for the family for many years and we elect to continue to use it and recommend that Equulidae be suppressed. POLY CENTRIDAE Remarks. Until recently (Britz, 1997), Polycentrus, Polycentropsis, Monocirrhus and Afronandus were included with Nandus in the family Nandidae. Britz (1997) demonstrated that a Nandidae that included these four genera was polyphyletic and removed them. He hypothesized the monophyly of the first three genera based on egg morphology, but only “tentatively” included Afronandus with them based on a single shared character: presence of adhesive filaments on the vegetal egg pole. Britz noted that this character also had been reported for some Pseu- dochromidae and Cichlidae, which he indicated were not closely related to the Nandidae, but he did not assign either the first three or all four genera to a separate family. Berra (2001:427), based on Britz (1997), apparently was first to define the Polycentri- dae as comprising only these four genera. Kullander and Britz (2002:301) similarly recognized a Polycen- tridae, but again only “tentatively” included Afron- andus, based on Britz’s (1997) character, which they erroneously described as “‘presence of attachment fil- aments around the micropylar area of the egg.” We find at least three synapomorphies, in addition to the vegetal-pole filaments, that support monophyly of the Polycentridae: loss of LE3, loss of sensory canals on most lateral-line scales (including all of those on posterior half of body), and presence of slender ligament attaching dorsal margin of Eb! an- terior to LEI to ventral surface of skull at or near origin of LE1. Notwithstanding its unilateral vestigial condition in Monocirrhus and Afronandus, loss of LE3 is a relatively uncommon character state for percomorphs (various scorpaenoids, Psettodidae, many smegma- morphs, Bramidae, Uranoscopidae, dactyolscopid blennioids, callionymids, gobiesocids). Absence of lateral-line scale tubes, or restriction of the tubes to a limited area on the anteriormost portion of the body, is also uncommon in perciform fishes, partic- ularly those inhabiting freshwaters, although Britz (pers. comm.) informs us that the condition is also true of some deeply nested taxa within the Anaban- toidei. Among freshwater percomorphs, the reduction or absence of tubed lateral-line scales is generally restricted to mugilids, various atherinomorphs and 164 gasterosteomorphs, elassomatids, percichthyids, nan- nopercids, and specialized members of other groups (e.g., gobiids). The distribution of the Ebl ligament in perciforms is less well known, as we may have overlooked or removed it in some taxa early in our study (see Introduction, methods section); however, it is not present in anabantoids, ambassids, and Tox- otes. We have noted the ligament at least in pem- pherids, cirrhitids, bathyclupeids, lacteriids, acro- pomatids. Another character, presence of a hook-like post- premaxillary process (reduced in Afronandus), is also uncommon among acanthomorphs, and may repre- sent another polycentrid synapomorphy (see Liem, 1970:figs. 13-16). A similar post-premaxillary pro- cess is also present in Nandus (Liem, 1970:fig. 12), but is reduced in the closely related Badis, possibly a reflection of the small size attained by this genus. Liem (1970), who examined the osteology and gill-arch musculature of all four polycentrid genera, included those genera in the Nandidae, along with Nandus, which he also examined (Liem, 1970:9, 11). Liem (1970:56) reported that all five genera had only three external levators, but did not indicate which levator was missing. He suggested that the posteri- ormost levator might represent a fusion of the third and fourth. Liem was correct in asserting that Poly- centrus, Polycentropsis, Monocirrhus and Afronan- dus have only three LEs (he could easily have missed a vestigial LE3, if present, which we found unilat- erally in Afronandus and Polycentropsis), but he was incorrect in implying the presence of only three LEs in Nandus, in which LE3 and LE4 are both well de- veloped. Commonly in acanthomorphs, including Nandus, LE3 and LE4 appear to be fused along their entire length, except for their easily overlooked sep- arate insertions on the surfaces of Eb3 and Eb4. LE3 and LE4 are easily separated and there is no inter- mingling of their fibers. Other specializations shared by the polycentrids are the absence of SOD and Pb4 and the presence of more than three anal-fin spines and only one epural. All these characters have a broad and varied distri- bution. In summary, the specializations shared by the four polycentrid genera, strongly imply their monophyly. Polycentropsis abbreviata USNM 302514, 34.2 mm. Not illustrated Boulenger, Additional material. @ = Polycentrus schomburgkii Miiller and Troschel, USNM 226071, 43.0 mm. Description. Remarks. The arrangements of the muscles, except for TD, are very similar in the two taxa. BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON LEI! on Eb! posteriorly noticeably lateral to car- tilage tip of uncinate process. Slender ligament orig- inates on skull with LEI and inserts on Eb1 anterior margin anterior to LEI insertion. LE2 on Eb2 about mid-dorsoposteriorly. LE3 present only on left side; thread-like, less than 10% width of LE4, inserting on tip of Eb3 uncinate process. @) Absent on both sides. LE4 broadly on Eb4 dorsoposteriorly beginning ventral to OD3—4 insertion on posterior surface of Eb4 just ventral to tip of uncinate process, meeting LP insertion posteriorly; posteroventral edge of in- sertion meets OP dorsolaterally and Ad4 dorsoanter- iorly; Eb4 levator process absent. CT sheet extends from near LE4 and LP insertions to distal end of Eb4 and becomes PP dorsally. @ Like Polycentropsis but beginning lateral to OD3—4 insertion on posterior surface of uncinate process and extending laterally to tip of levator process, meeting LP insertion posteri- orly; CT sheet begins at levator process. LP on Eb4 at and anterior to LE4 insertion. LI1 on lateral edge of dorsoanteriormost tip of Pb2, almost appearing to continue onto adjacent me- dialmost edge of IAC, and on dorsoanteromedialmost edge of adjacent Pb3. @ Like Polycentropsis, but in- sertion more extensive on Pb2, and, on one side, a few lateralmost muscle fibers continuing onto medi- almost surface of IAC. LI2 on Pb3 dorsoposteriorly immediately medial to medial end of Eb3. TD comprises TEb2 and TPb3-Eb3, and TEb4. TEb2 musculously continuous only narrowly poste- riorly, with mid-anterior notch continuing posteriorly as mid-longitudinal raphe, which gives rise dorsally to CT sheets covering muscles and is continuous an- teriorly with CT of pharyngeal roof; ventroanteriorly, muscle attaches loosely to Pb3; laterally muscle at- taches on Eb2 dorsally anterior to LE2 insertion; pos- teriorly muscle is free from TPb3-Eb3. TPb3-Eb3 rel- atively narrow, begins anteriorly on Pb3 just anterior to medial end of Eb3 and continues posteriorly on posteromedialmost edge of Eb3; muscle is free from TEb4. TEb4 broader than TPb3-Eb3, attaches along Eb4 posteriorly between medial end and uncinate process. @ TD comprises TPb2, TEb2, TPb3-Eb3, and TEb4. TPb2 a semicircular ribbon of muscle on each side, arising ventroanteromedially from TEb2 dorsal- ly and finely attached by CT to dorsoanteriormost surface of Pb3 (not attached to Pb2); muscle curves posterolaterally and joins medial band of CT cover- ing Pb3s posteriorly and joining the bi-lateral muscle straps of TEb2. CT joining muscle sections of TEb2 notched deeply mid-anteriorly; anteriorly joining CT of pharyngeal roof and dorsally giving rise to CT sheets; muscle extends laterally from CT covering Pb3s and attaches on Eb2 dorsally well anterolateral NUMBER 11 to LE2 insertion; free from TPb3-Eb3. TPb3-Eb3 ventral to OD3—4, beginning anteriorly a little ante- rior to LI2 insertion, continuing posteriorly along me- dial edge of LI2 insertion, and attaching finely to posteromedialmost corner of Eb3; posteriorly contin- uous by slender, diagonal muscle strand with TEb4. TEb4 slender, on Eb4 a little lateral to cartilaginous medial end, meeting OP dorsomedially. OD3-—4 on Pb3 dorsomedially ventral to TEb2, in- sertion on Eb3 anteriorly ventral to tip of uncinate process and on Eb4 posteriorly ventral to tip of un- cinate process. OP strap-like; dorsally on Eb4 posteriorly, begin- ning medially ventral to TEb4 attachment, continuing laterally and meeting LE4 insertion posteriorly, and posteriorly overlapping Ad4 dorsomedially; ventrally on Cb5 dorsoposteriorly, anteriorly meeting TV5 posterolaterally, laterally posteriorly overlapping Ad5 ventromedially. @ Dorsally on Eb4 posteroventrally beginning medially ventral to OD3—4, continuing lat- erally almost to levator process (lateral to posteroven- tral edge of LE4 insertion), posteriorly overlapping Ad4 dorsomedially; ventrally like Polycentropsis, but joining well-developed raphe with Ad5 posterome- dially. Ad1-3 absent. Ad4 dorsally on Eb4 beginning ventrally anterior to lateral edge of OP and extending laterally almost to Eb4-Cb4 joint; ventrally on Cb4 dorsally just me- dial to Eb4-Cb4 joint. Ad5 strap-like, dorsoanteriorly on Cb4 beginning just medial to distal cartilage tip, posteroventrally on Cb5 beginning just medial to distal cartilage tip. @ Dorsoanteriorly meets Ad4 ventroposterolaterally on Cb4; begins posteroventrally on distal tip of Cb5 and joins raphe with OP ventrolaterally. SOD absent. RDs separated by space about one-half diameter of one RD. ® Space less than one-half diameter of one RD. Additional remarks. SCL attached mid-dorsally to ventral surface of cartilaginous posterior tip of Bb3. TV4 free from Cb5s. Pb2 toothed Pb4 absent; UP4 present. Afronandus sheljuzhkoi (Meinken), USNM 372183, 65.9 mm. Plate 138 Description. LE1 on Eb! dorsoposteriorly, well lateral to car- tilaginous tip of uncinate process; slender ligament originates with LE1 and inserts on Eb] anteriorly an- terior to LE] insertion. LE2 on Eb2 about mid-dorsoposteriorly, posterior to posterolateralmost edge of TEb2. 165 LE3 absent on right side; on left side, a very fine filament inserting on tip of Eb3 uncinate process. LE4 on Eb4 posteriorly medial to lateral end, an- teromedially meeting OD3-—4 and laterally meeting LP insertion medially. LP on Eb4 dorsally beginning at lateral edge of LE4 insertion and extending laterally; CT sheet ex- tends from lateral edge of insertion and becomes PP dorsoposteriorly. LI1 on Pb2 dorsally just posterior to dorsoanter- iormost cartilage tip, inertion becoming tendinous as it joins adjacent anterior end of Pb3; muscle fibers abut medial IAC surface, but no fibers basally insert on IAC. LI2 narrowly on Pb3 dorsolaterally, medial edge meeting TPb3-Eb3-Eb4 lateral edge. TD comprises TEb2 and TPb3-Eb3-Eb4. TEb2 broad medial to much narrower extension onto Eb2 dorsally, there attaching anterior to LE2 insertion; broadly, mid-anteriorly consisting of thick CT pad, which attaches anterolaterally to dorsoanteriormost ends of Pb2s and is mid-anteriorly continuous with CT of pharyngeal roof; muscle fibers continuous nar- rowly and only posteriorly, but interrupted there by very short, fine, mid-longitudinal raphe; muscle not continuous posteriorly with TPb3-Eb3-Eb4. TPb3- Eb3-Eb4 anteriorly ventral to OD3—4s, attaching to Pb3 at medial edge of LI2 insertion, continuing pos- teriorly and attaching to posteromedialmost corner of Eb3, abruptly narrowing at that point and then ex- panding considerably and attaching to Eb4 postero- medially, there just reaching dorsomedialmost edge of OP. OD3-—4 origin on Pb3 dorsomedially ventral to TEb?2, insertion on Eb3 anteriorly immediately ven- tral to tip of uncinate process and on Eb4 posteriorly on and ventral to tip of uncinate process; meets LE4 insertion ventroanteriorly. OP dorsally on Eb4 posteriorly beginning medially near lateral end of TPb3-Eb3-Eb4 and extending lat- erally to below LE4, overlapping medial half of Ad4 posteriorly; ventrally on Cb5 dorsally beginning near lateral end and extending medially and meeting pos- terolateral edge of TV5; anteriorly overlapping Ad5 posteromedially. Ad1-—3 absent. GFMs1-—3 (not illustrated) moder- ately well developed. Ad4 dorsally on Eb4 posteroventrally beginning medially anterior to OP and extending laterally al- most to distal end of bony surface; ventrally on Cb4 dorsally beginning medially anterior to OP (which is on Cb5) and extending laterally along Ad5 attach- ment dorsally to near lateral end of bony surface of Cb4. Ad5 dorsally on Cb4 posteriorly beginning well medially and extending laterally along ventral edge of Ad4 almost to lateral end of bony surface of Cb4; 166 ventrally on Cb5 posteriorly beginning near distal end of bony surface and extending medially; over- lapped posteriorly by OP. SOD absent. RDs almost adjacent. Additional remarks. SCL present. TV4 free from Cb5s. IAC articulates with cartilaginous medial end of Eb1 uncinate process. Pb4 absent. UP4 present. Eb4 levator process absent on one side, greatly re- duced on other (treated as absent). Monocirrhus polyacanthus Heckel, USNM 103840, 78.1 mm; USNM 269969, cleared and _ stained, 56.1 mm. Not illustrated Description. LEI on Eb! well lateral to tip of uncinate process; strong, ribbon-like ligament originates with LE] and inserts on Eb] anteriorly anterior to LEI. LE2 on raised process on Eb2 posteriorly at about mid-length of element. LE3 absent. LE4 broadly on Eb4 dorsoposteriorly beginning lateral to uncinate process and extending laterally much of distance to distal end, laterally meeting slen- der LP insertion posteriorly. LP slender, at and anterior to LE4 insertion later- ally. LI1 on Pb2 dorsally just posterior to anteriormost tip, insertion continuing on adjacent anterolateral- most margin of Pb3 and impinging on medialmost surface of IAC, but no fibers basally inserting on it. LI2 on Pb3 laterally medial to medial end of Eb3. TD comprises TEb2 and TPb3-Eb3-Eb4. TEb2 musculously continuous from attachment on Eb2 dorsoanteriorly anterior to LE2 insertion on one side across Pbs to attachment to Eb2 on other side; area over Pbs attached mid-ventroanteriorly to pharyngeal roof CT and dorsally to CT sheets, with strongest attachment on each side to muscle surface just medial to extension of muscle onto Eb2; strong CT continues anteriorly, attaching to ventroanterior edge of LI, and tightly enveloping anterior ends of Pb2 and Pb3 and medial surface of Pb1-Eb1 joint; muscle not con- tinuous with TPb3-Eb3-Eb4. TPb3-Eb3-Eb4 origin ventral to OD3—4, beginning on Pb3 along medial edge of LI2 insertion, continuing onto posteromedi- almost corner of Eb3, followed posteriorly by con- siderable lateral extension, which attaches on Eb4 mid-posteriorly. OD3—4, OD3’ anteriorly on Pb3 dorsomedially ventral to TEb2; OD3-—4 posteriorly on bony surfaces of Eb3 uncinate process anteriorly and Eb4 uncinate process posteriorly, almost covering tip of latter pro- cess; OD3’ branches off OD3—4 ventroanteriorly at BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON about mid-length and attaches to Eb3 dorsally ventral to OD3—4 on Eb3. OP dorsally on Eb4 posteriorly beginning medially at about mid-length of Eb4 and extending laterally to below mid-point of LE4 insertion, there joining Ad4 dorsomedialmost edge; ventrally on Cb5 dorsally be- ginning at posterolateral edge of TV5 and extending laterally a short distance and joining Ad5 postero- medially; not clearly differentiated from SO medial- ly. Ad1-—3 absent. Ad4 on Eb4 dorsally beginning below about mid- point of LE4 insertion, where Ad4 and OP join, and extending laterally almost to lateral end of bony sur- face of Eb4; ventrally on Cb4 dorsally beginning me- dially at point ventral to its medial attachment on Eb4 and extending laterally almost to posterodistalmost end of bony surface of Cb4, meeting Ad5 dorsoan- terolaterally. Ad5 strap-like, dorsoanteriorly on Cb4 dorsopos- teriorly well medial to lateral end of Cb4, there meet- ing Ad4 and extending medially a short distance; posteroventrally on Cb5 dorsally beginning near dis- tal end and extending a short distance medially, meet- ing OP ventroanteromedially. Remarks. It is unusual in a perciform for Ad5 to attach well medial to the distal end of Cb4. SOD absent. RDs separated by distance about diameter of one RD. Additional remarks. SCL attached mid-dorsally to tip of posteroventrally extending cartilaginous end of Bb3. TV4 free from Cb5s. Pb4 absent; UP4 present. IAC present. Eb4 levator process absent. Medial end of Eb4 larger than that of Eb3 in 78.1 mm specimen, about equal in 56.1 mm specimen. SPHYRAENIDAE Remarks. Johnson (1986) included Sphyraenidae in the Scombroidei, but Orrell et al. (2003 and in preparation) excluded it based on a molecular study. Sphyraena barracuda (Edwards), USNM 331677, 139 mm. Plate 139 Description. LE1 on anterolateral surface of Eb1 uncinate pro- cess. LE2 on raised bony posterior edge of Eb2. LE3 long tendinous origin, insertion on tip of Eb3 uncinate process. LE4 long tendinous origin, insertion on dorsal sur- face of Eb4 just lateral to base of uncinate process. LP medial half of insertion at and anterior to in- sertion of LE4, posterior half is lateral to LE4. NUMBER 11 LI1 larger than LI2, broadly on Pb2 dorsoanterior surface beginning at and ventral to TPb2. LI2 on posterolateral edge of Pb3 anterior to me- dial end of Eb3. TD comprises TPb2, TEb2, and TPb3-Eb3. TPb2 dorsal to TEb2, V-shaped, arms anterior, joined pos- teriorly along mid-longitudinal raphe, which attaches ventrally to CT of pharyngeal roof, each arm attaches to respective Pb2 dorsoanteriormost surface; muscle fuses ventrally with TEb2 on either side of raphe. TEb2 broad, short laterally, attaching to dorsal sur- face of Eb2 well medial to LE2 insertion, posteriorly continuous by diagonal strap of muscle with TPb3- Eb3. TEb3-Pb3 on Pb3 dorsoposteriorly medial to LI2 insertion and on posteromedial edge of Eb3, free from SOD. OD3-4 originating on Pb3 anteromedially ventral to TEb2 and inserting on medial edges of Eb3 and Eb4 uncinate processes. OP relatively slender, dorsally on posterior surface of Eb4 at and medial to uncinate process, ventrally on posterodistal end of Cb5, joining raphe with Ad5 ventrolaterally. Ad1-—3 absent. Ad4 relatively broad, on Eb4 posteriorly beginning ventral to LE4 insertion and continuing laterally to distal end of bony portion of Eb4, broadly ventrally on Cb4 dorsal surface medial to Eb4-Cb4 joint and medial to broad Ad5 attachment to Cb4. Ad5 dorsally broadly on posterolateral bony sur- face of Cb4 just medial to distal end, fusing medially with SO; ventrally broadly on CbS5 laterally, fusing anteriorly with TV5. SOD present. RDs separated by space greater than width of one RD. RD extends anteriorly between two longitudinal SO muscle straps: one originating from SO dorsally in region ventral to SOD and extending anteriorly medial to RD and inserting on Pb3 dorsally at and medial to OD3—4 origin, and the other originating in area anterior to OP and extending anteriorly lateral to RD, becoming dorsal to RD anteriorly and insert- ing on most of medial edge of Pb3 ventral to OD34. Additional remarks. SCL absent (note Bb3 has posteroyentrally extending cartilaginous tip which at- taches to the ventral aorta where this vessel divides bilaterally and extends anteriorly). TV4 free from Cb5s. Pb4 absent, UP4 present. Pb2 toothed. IAC present. Eb4 levator process absent. Medial end of Eb4 larger than that of Eb3. PCI originates by long tendon on cleithrum and attaches on Cb5 beginning well medial to distal end of Cb5 and continuing me- dially. KURTIDAE Kurtus sp., USNM 345060, 92.4 mm. Not illustrated 167 Description. LE1 on dorsolateral edge of Eb1 uncinate process beginning just lateral to cartilage tip. LE2 on dorsal edge of expanded posterior margin of Eb2. LE3 on tip of Eb3 uncinate process anteriorly. LE4 on Eb4 dorsoposterolateral to tip of uncinate process, ventroposteriorly meeting LP insertion ven- troanteriorly. LP on Eb4 beginning at and extending posterior to LE4 insertion. LI1 inserting on Pb2 dorsoanteriormost process posteriorly, continuing posteroventrally and attaching to anteriormost edge of Pb3; joining posterior edge of TPb2 attachment to Pb2; smaller than LI2. LI2 on Pb3 posterolaterally immediately adjacent to medial end of Eb3. TD comprises TPb2, TEb2, and TEb3. TPb2 roughly V-shaped, arms broad, flat, open anteriorly, completely dorsal to TEb2, with raphe coursing through posterior apex and across TEb2 (thus com- pletely dividing both muscles) and attaching ventrally to dorsomedial edge of each Pb3; muscle attached mid-anteriorly to CT of pharyngeal roof, attached an- terolaterally to dorsoanterior tip of Pb2, there joining LI1 just dorsal to its insertion. TEb2 extending lat- erally and attaching to Eb2 dorsally anterior to LE2 insertion; free posteriorly from TEb3 except for fine posterior CT continuations from dorsomedial edges of Pb3s to mid-longitudinal raphe of TEb3. TEb3 posteroventral to level of TEb2, attaching broadly to posteromedial margin of Eb3; raphe continuing pos- teriorly across SOD. OD3-—4 origin on Pb3 dorsomedially ventral to TEb2; inserting on Eb3 broadly anteriorly ventral to tip of uncinate process and on Eb4 somewhat less broadly ventral to tip of uncinate process. OP a relatively narrow strap, dorsally on Eb4 pos- teriorly beginning medial to uncinate process and ex- tending medially, ventrally on Cb5 dorsally posterior to Ad5, beginning laterally near distal end of Cb5 and extending medially. Ad1-3 absent. Ad4 dorsally broadly on Eb4 posteriorly beginning at lateral edge of OP and extending laterally to end of bone, ventrally on Cb4 dorsoposteriorly beginning a little medial to lateral end and anterior to Ad5 at- tachment and extending medially. Ad5 dorsally on Cb4 posteriorly beginning near distal end and extending medially, ventrally on Cb5 dorsoposteriorly and extending medially anterior to OP. SOD very fine, medially with pair of fine ventral extensions from mid-dorsal raphe, each encircling one RD. RDs separated by distance less than half diameter of one RD. 168 Additional remarks. SCL interrupted medially by attachment to autogenous ball of cartilage attached to ventral tip of posteroventrally extending cartilaginous posterior end of Bb3 (autogenous ball also present in cleared and stained specimen of Kurtus indicus, USNM 305690). TV4 free from Cb5s. Pb4 absent. UP4 present. IAC present. Eb4 levator process ab- sent. Pb1 bony with cartilage tips and, ventrolaterally, uniquely, has toothplate with fine teeth adjoining similar toothplate on medial end of Eb1 (consistent with scale-eating behavior). Pb2 toothed. Medial end of Eb4 smaller than that of Eb3. The next four families were included with nine others in a suborder Trachinoidei by Nelson (1994: 395 et seq.), who indicated that the composition of the suborder was probably polyphyletic. AMMODYTIDAE Ammodytes dubius Reinhardt, USNM 302478, 180 mm. Plate 140 Description. LE1 on broad tip of Eb1 uncinate process antero- laterally. LE2 on tip of raised mid-posterodorsal edge of Eb2. LE3 on tip of Eb3 uncinate process anteriorly. LE4 together with LP (laterally) tendinously on dorsal bony edge of Eb4 lateral to uncinate process. LP together with LE4 (medially) tendinously on bony distal dorsal edge of Eb4 at and lateral to LE4 insertion. LI1 on bony dorsolateral edge of Pb2 just ventral to articulation with IAC. LI2 on Pb3 dorsoposteriorly at and lateral to me- dial edge of Pb3 portion of TPb3-Eb3. TD comprises TPb2, TEb2, and TPb3-Eb3. TPb2 small, thin, V-shaped, dorsal to TEb2 anteriorly, each arm attaching to cartilaginous tip of Pb2 anterior end, junction of arms divided by longitudinal raphe, which continues across TEb2, attached ventrally to CT of pharyngeal roof. TEb2 transversely broad, longitu- dinally narrow, attaching mid-ventrally to CT of pha- ryngeal roof, attaching laterally to Eb2 dorsally an- terior to LE2 insertion; free and well separated from TPb3-Eb3. TPb3-Eb3 on Pb3 dorsoposteriorly at and medial to LI2 insertion, continuing posteriorly and attaching finely to medial tip of Eb3, continuous mid- posteriorly by crossing strands of muscle with SOD. OD3-4 origin broadly on Pb3 dorsoposterolateral edge, anteriorly ventral to TEb2, but mostly posterior to TEb2; insertion on Eb3 anteriorly ventral to tip of uncinate process and on medial edge of Eb4 uncinate process. OP dorsally posteriorly on Eb4 beginning on un- BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON cinate process and extending laterally almost to be- ginning of cartilaginous distal end, ventrally on Cb5 dorsolaterally, joining CT ventrolaterally with pos- terior end of Ad5, medially continuous with SO. Remarks. This description was completed much earlier than that of Symphysanodon berryi (Symphy- sanodontidae, q.v.), in which the dorsal gill-arch mus- culature is otherwise remarkably similar. Re-exami- nation of our A. dubius specimen indicates the possi- bility that OP is divided into two parts, similar to S. berryi, but this should be verified in another specimen. Ad1-—3 absent. Ad4 broadly on Eb4 dorsoposteriorly lateral to OP, ventrally on Cb4 beginning at inner angle of Eb4- Cb4 joint and continuing a short distance medially. Ad5 posteriorly on distal end of Cb5 and anteriorly on Cb4 posterolaterally and AC4 posteroventrally, joining tendinous raphe with OP ventromedially. SOD present, slender. RDs separated by distance greater than diameter of one RD. Additional remarks. SCL attached mid-dorsally to ventroposteriorly extending cartilaginous posterior end of Bb3. TV4 free from Cb5s. Pb4 and UP4 pres- ent. Pb2 toothed. IAC present. AC4 present (not yet budded off in 111 mm, cleared and stained specimen, USNM 302247). Medial end of Eb3 larger than that of Eb4. Kayser (1962) described and illustrated the skele- ton and musculature of A. tobianus Linnaeus. The dorsal gill-arch muscles of A. tobianus differ most notably from those of A. dubius in that TPb2 is ab- sent, the origin of OD3—4 is completely ventral to TEb2, LI1 and LI2 are much more massive, and AC is apparently absent or has not budded off (thus Ad5 attaches anteriorly only to Cb4). Kayser illustrated the GFMs, which we did not. TRACHINIDAE Trachinus draco Linnaeus, USNM 349536, 2 speci- mens, 80.6—84.3 mm. Plate 141 Description. LE1 on Eb! anteriorly lateral to tip of uncinate process. LE2 on raised dorsoposterior bony edge of Eb2. LE3 on tip of Eb3 uncinate process anteriorly. LE4 on Eb4 dorsoposteriorly between uncinate and levator processes. LP on Eb4 posterior to LE4. LI1 on most of Pb2 bony surface dorsomedially with muscle fibers separating dorsoposteriorly and in- serting on Pb3 bony surface slightly posteromedial to anterior end of Pb3. LI2 on Pb3 dorsally medial to medial end of Eb3. TD comprises TPb2, TEb2, and TPb3-Eb3. TPb2 NUMBER I1 flat, broad, bi-lateral pair of anterolaterally extending muscles originating posteriorly from mid-longitudinal raphe or somewhat heart-shaped ribbon of muscle di- vided mid-longitudinally; either type dorsally on mid- section of TEb2 and attaching anterolaterally to car- tilaginous dorsoanteriormost tip of Pb2; anterior fibers may coalesce medially with those of TEb2; CT sheets attaching to skull arise from mid-longitudinal raphe. TEb2 broad, extending laterally onto Eb2 dorsally an- terior to LE2 insertion, not continuous posteriorly with TPb3-Eb3. TPb3-Eb3 on Pb3 dorsolaterally beginning anteriorly medial to LI2 insertion and continuing pos- teriorly and attaching to posteromedialmost edge of Eb3 (see Remarks following), continuous posteriorly by diagonal muscle strand with SOD. Remarks. On right side only of larger specimen (illustrated), TPb3-Eb3 continues onto Eb4 dorso- medially. We consider this continuation to be anom- alous. OD3-4 anteriorly on Pb3 dorsomedially ventral to TEb2, posteriorly on anterior surface of Eb3 uncinate process and medial edge of Eb4 uncinate process. OP dorsally on Eb4 posteriorly, extending from uncinate process about half distance to medial end of Eb4, ventrally on Cb5 dorsoposterolaterally. Ad1-—3 absent. Ad4 broadly dorsally on ventral surface of Eb4 lateral to levator process, ventrally equally broadly on Cb4 dorsally medial to Eb4-Cb4 joint. Ad5 dorsally on Cb4 posterolaterally (reaching distal end), joining raphe with Ad4 ventral attach- ment; ventrally anterior to OP on Cb5 dorsally, reaching distal end. SOD present. RDs separated by space slightly less than diameter of one RD. Additional remarks. SCL present. TV4 free from Cb5s. Pb4 and UP4 present. [AC present. Eb4 levator process present. URANOSCOPIDAE Kathetostoma cubana Barbour, USNM 187953, 81 mm. Additional material. @ Xenocephalus egregius (Jor- dan and Thompson), USNM 186212, 89.1 mm. Plate 142 Description. LE1 on anterior surface of broad bony flange lat- eral to tip of Eb] anterior process (uncinate process and Pb1 absent). @ Uncinate process and tiny Pbl present. LE2 short, bulky, on dorsoanterior surface of Eb2 mid-laterally, medial edge of insertion joining raphe with TEb2 lateralmost edge. LE3 absent. 169 LE4 long, bulky, on Eb4 dorsoposterolaterally. LP very fine, at lateralmost edge of LE4 insertion; easily overlooked and lost during dissection. LI1 on dorsoanterolateralmost surface of Pb3 (Pb2 absent); just dorsoanterior to insertion, attached to thick pharyngeal roof CT enveloping Eb! medial end and Pb3 dorsoanterior end (dorsoanterior surface of OD3-4 origin also attached to CT medial to LI1; CT thins considerably medially). @ On dorsoanterior- most surface of Pb3 and medial end of small, eden- tate Pb2; dorsomedialmost edge of insertion attached to CT to which mid-anteromedial edge of TEb2 and anteromedialmost edge of OD3—4 origin also attach; Eb1 medial end not enveloped by same CT; pharyn- geal CT not thickened. LI2 on Pb3 dorsoposteriorly opposite medial ends of Eb3 and Eb4. TD comprises TEb2 and TPb3-Eb3. TEb2 broad, flat, covering entire Pb area dorsally, with mid-lon- gitudinal raphe giving rise dorsally to CT sheets at- taching to skull, attaching mid-ventrally to CT of pharyngeal roof between Pbs; muscle attaches on Eb2 dorsomedially, joining raphe with medial edge of LE2 insertion, and overlies, but is not continuous with, anterior end of TPb3-Eb3. TPb3-Eb3 on Pb3 dorsally medial to LI2 insertion and ventral to OD3— 4, continuing on to Eb3 broadly dorsomedially. @ Comprises TEb2, TEb3, and TPb3-UP4. TEb2 broad- ly continuous posteriorly with TEb3, which attaches on Eb3 dorsally medial to base of uncinate process; TPb3-UP4 completely ventral and weakly attached dorsally to TEb3; muscle attaches on Pb3 posterior edge and adjacent dorsomedialmost edge of UP4, and is continuous posteriorly by diagonal muscle strand with SOD. OD3-—4 origin on Pb3 dorsoanteriormost surface ventral to TEb2, medial to LI1 insertion, and dorsal to LI2 insertion and TPb3-Eb3 anteriorly; dorsoan- teriormost surface of muscle strongly attached to thick CT of pharyngeal roof; insertion on medial edg- es of Eb3 and Eb4 uncinate processes. @) Origin on Pb3 dorsomedially ventral to TEb2; anteromedial- most edge attaches to CT also joined by LI1 and TEb2 (see LI1 above); muscle is dorsal to LI2 inser- tion and TPb3-UP4 anterolaterally. OP in two sections: lateral section dorsally on Eb4 posteriorly beginning below LE4 insertion medially and extending medially to below uncinate process, medial section dorsally, narrowly on Eb4 uncinate process, laterally folded over medial section posteri- orly (fibers of sections continuous in crotch of fold); both sections attaching together on Cb5 medially and join raphe with PCI dorsoanteriorly (PCI passes be- tween two sections to insert on Cb5); medial section incompletely separated medially from SO. @ Medial section joins tendinous anterior extension of PCI ven- troanteriorly and lateral section joins dorsoanteriorly 170 (also, Ad5 posteriorly joins lateral surface tendon); cartilaginous and adjacent bony area of Cb5 distal end not included in attachments. Ad1-—3 absent. Ad4 dorsally broadly on Eb4 ventrally below LE4 insertion, ventrally on dorsolateral half of Eb4 medial to Eb4-Cb4 joint; medially, Ad4 attachment on Cb4 is separate from Ad5 attachment on Cb4, but grad- ually the two muscles meet and distally are insepa- rable. @ See Ad5. Ad5 dorsally broadly on Cb4 posterolaterally (see also Ad4), ventrally on Cb5 distally. @ Musculous portion on Cb4 well posteromedial to distal end, but joins CT attaching along posterolateral edge of Cb4; entire musculous portion joins Ad4 ventrally on Cb4; posterior surface joins lateral surface of tendinous an- terior end of PCI. SOD absent. @ Present. RDs exceptionally large, slightly separated. @ Moderately large, separated by space equal to about half one RD diameter. Additional remarks. SCL absent. TV4 free from Cb5s. Pb1, Pb2, and Pb4 absent, UP4 present. Medial end of Eb4 larger than medial end of Eb3. Eb] un- cinate process absent (see also Rosen and Patterson, 1990, fig. 36b). Eb4 levator process absent. @ Pbl small, partly bony. Pb2 small, rod-like, edentate. Pb4 absent, UP4 present. IAC absent. Me- dial end of Eb4 not noticeably larger than that of Eb3. Eb! uncinate and anterior processes present. Eb4 levator process present, completely covered by LE4 insertion. Pietsch (1989) hypothesized Gnathagnus Gill Gu- nior synonym of Xenocephalus Kaup) and Pleuros- copus Barnard as the plesiomorphic sister group of a clade including Kathetostoma Gunther. CHEIMARRICHTHYIDAE In a phylogenetic study, Imamura and Matsuura (2003) excluded Cheimarrichthys as a close relative of the Pinguipedidae, in which some authors have included it. They were, however, unable to propose a sister-group relationship for Cheimarrichthys and recognized it, as have other authors, as the sole mem- ber of the Cheimarrichthyidae. Cheimarrichthys fosteri Haast, USNM 362725, USNM 362725, 71.6 mm. Not illustrated (superfically resembles Callanthias, Plate 132) Description. LEI on Eb! uncinate process anteriorly ventral to cartilage tip. LE2 broadly on Eb2 dorsoposteriorly, medial edge meeting posterolateral end of TEb2. BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON LE3 broadly on Eb3 anteriorly, beginning at tip of uncinate process and extending laterally. LE4 largest levator, on Eb4 dorsolaterally, reach- ing distal end of bony surface, posterolaterally join- ing LP insertion. LP on Ebé4 at and lateral to LE4 insertion. LI1 on Pb2 dorsomedially and adjacent anterior end of Pb3, there meeting OD3-—4 anteromedially. LI2 on Pb3 posterolaterally, medial edge of pos- terior half of insertion meeting TPb3-Eb3. TD comprises TEb2 and TPb3-Eb3. TEb2 band- like with mid-longitudinal raphe, which attaches ven- trally to pharyngeal roof CT and anterior edge of TPb3-Eb3, muscle extends laterally to point anterior to LE2 insertion. TPb3-Eb3 band-like attaching on Pb3 posterolaterally beginning anterior to LI2 inser- tion, continuing along medial edge of LI2 insertion, and attaching to posteromedialmost corner of Eb3; continuous posteriorly by diagonal muscle strap with SOD. OD3-—4 broadly, anteriorly on Pb3 dorsomedially, meeting LI1 posteriorly; posteriorly on medial edges of Eb3 and Eb4 uncinate processes. OP thick, strap-like; dorsally on Eb4 uncinate pro- cess posteriorly, there meeting OD3—4, and extend- ing laterally to below medial end of LE4 insertion, which it also meets, and, at that point, overlaps dor- somedial half of Ad4 posteriorly; ventrally on Cb5 posteriorly beginning laterally at posteroventral end of Ad5 and extending medially for distance about equal to extent on Eb4; medially clearly differenti- ated from SO on one side, unclearly on the other. Ad1-—3 absent. Ad4 dorsally on Eb4 posteriorly, beginning me- dially anteroventral to OP and extending laterally to distalmost end of bony surface; ventrally on Cb4 dor- sally beginning medially ventral to a perpendicular from dorsomedial origin on Eb4 and extending lat- erally to distalmost end of bony surface, meeting Ad5 anteriorly for most of latter’s length. Ad5 short, anteriorly on Cb4 posterodistal edge, meeting Ad4, posteriorly on Cb5 dorsally, beginning medially a little anterior to OP ventrolaterally and ending near distalmost end of bony surface. SOD present. RDs separated by space equal to about one RD diameter. Additional remarks. SCL present, apparently free from ventrally curving cartilaginous posterior end of Bb3. TV4 free from Cb5s. Pb4 and UP4 present. Pb1 short, cartilaginous, oriented medially (horizontally). Pb2 toothed. Eb4 levator process absent. IAC pres- ent. Scorpaenoidei Imamura and Yabe (2002) hypothesized a reorga- nization and recomposition of the Scorpaeniformes NUMBER I1 of previous authors. They recognized a scorpaenoid- serranoid sister group, on the one hand, and a zoar- coid-cottoid sister group on the other. They did not imply that the two sister groups are closely related, but assigned both to the Perciformes. Furthermore, they excluded the Champsodontidae and Normani- chthyidae as possible members of either of the two sister groups. We retain the scorpaeniform composition of earlier studies in our arrangement of the taxa that follow, but treat the group as a suborder. We do so only for convenience and absent of intent to imply judgment on Imamura and Yabe’s study. In fact, see additional remarks following the description of Hexagrammos stelleri, Hexagrammidae. SCORPAENIDAE Pontinus rathbuni Goode and Bean, USNM 190360, 90.8 mm. Plate 143 Additional material. @ = Neomerinthe beanorum (Evermann and Marsh), USNM 187913, 80.6 mm. SEBASTIDAE @ = Sebastes proriger (Jordan and Gilbert), USNM 1066601, 106 mm. Not illustrated Description. Remarks. Ishida (1994) hypothesized the phylog- eny of the scorpaenoid fishes, in which he described various aspects of the dorsal gill-arch musculature. He hypothesized that the Sebastidae are the sister group of all other scorpaenoids. We noted few tren- chant differences in the muscles of the three genera of scorpaenoids we examined, although some carti- laginous elements of the gill arches exhibit differ- ences. We illustrate Pontinus because our preparation was much better than those of the other two genera. LE1 origin tendinous, insertion on base of Eb! un- cinate process anteroventrally. @ Origin not ob- served. @ Origin musculous. @ ® Insertion on Eb1 uncinate process anterolaterally. LE2 on Eb2 mid-dorsoposteriorly. LE3 on tip of Eb3 uncinate process anteriorly. LE4 on medial bony edge of Eb4 levator process. LP at and lateral to LE4 insertion, just extending onto tip of levator process. LI1 broadly on Pb2 dorsoanterior process begin- ning just posterior to dorsal tip, attaching to CT join- ing medial edge of process to adjacent surface of Pb3 dorsoanterior process (medial edge of Pb2 process, where LI1 inserts, just overlaps lateral edge of Pb3 process); ventral surface of muscle fans out posteri- orly toward insertion such that in dorsal view it ap- pears that there is a separate dorsoanterior insertion 171 overlying a ventroposterior insertion along medial edge of Pb2 process. @ Insertion more restricted, but muscle inserts on both Pb2 and Pb3. ® Insertion does not fan out ventrally, insertion area relatively about half that of Pontinus. LI2 on Pb3 dorsoposterolaterally at lateral edge of TPb3 and medial to medial end of Eb3. TD comprises TPb2, TEb2, TPb3-Eb3-Eb4. TPb2 centrally oblong with mid-anterior and mid-posterior notches joined by mid-longitudinal raphe, which gives rise dorsally to CT sheets attaching to skull; attached anterolaterally to dorsoanteriormost surface of Pb2 (with weak CT strands attaching to dorsoan- teriormost tip of Pb3); attached anteriorly to CT of pharyngeal roof between dorsoanterior ends of Pb2s; muscle lying dorsal to all but small posterior portion of TEb2, ventrally joining with TEb2 along mid-lon- gitudinal raphe and fusing with TEb2 posteromedi- ally on left side (less so on right side). TEb2 mostly ventral to posterior half of TPb2 and joining TPb2 along mid-longitudinal raphe, muscle attaching on Eb2 dorsally well medial to LE2 insertion; well sep- arated (possibly anomalous) from TPb3 posteriorly. TPb3-Eb3-Eb4 broadly dorsolaterally on Pb3, joining medial edge of LI2 insertion, fine, tendinous attach- ment to medial end of Eb3 on one side and poster- omedialmost bony edge of Eb3 on other, broadly on Eb4 dorsoanterior surface well medial to uncinate process, continuous mid-posteriorly with slender SOD. @ ® TPb2 unnotched posteriorly, completely dorsal to central part of TEb2; TEb2 continuous pos- teriorly by fine strands of muscle with TPb3-Eb3- Eb4; fine, tendinous attachment to medial end of Eb3 in both taxa. OD3-—4 origin on dorsoanterior surface of Pb3 ven- tral to TEb2, insertion on dorsoanterior surface of Eb3 uncinate process and medial edge of Eb4 unci- nate process. OP dorsally beginning on Eb4 uncinate process posteriorly and extending laterally to posteroventral surface of levator process, there meeting Ad4; ven- trally broadly on Cb5 dorsoposteriorly, overlapping Ad5 ventroposteriorly; indistinguishable from SO medially. Ad1-—3 absent. Ad4 continuous sheet beginning dorsally on ven- tral surface of Eb4 levator process and continuing laterally to Eb4-Cb4 joint, from there attaching dor- sally along lateral third of Cb4, meeting Ad5 narrow- ly dorsally on Cb4 ventral to Eb4-Cb4 joint. Ad5 dorsally narrowly meeting Ad4 on Cb4 pos- terior surface medial to distal end, ventrally on Cb5 dorsodistally anterior to OP attachment. SOD slender. RDs separated by space equal to about half di- ameter of one RD. @ Separated by space equal to 172 about diameter of one RD. ® Separated by space more than twice diameter of one RD. Additional remarks. SCL present. TV4 free from Cb5s. Pb4 and UP4 present. IAC attached by loose ligament to Pb2. @ IAC attached directly to Pb2. ® IAC attached directly to Pb2; AC4 present. PLATYCEPHALIDAE Platycephalus endrachtensis Quoy and Gaimard, USNM 173884, 113 mm. Plate 144 Additional material. @ = Inegocia japonica (Tile- sius), USNM 99761, 67.3 mm. Description. LE1 on Eb! dorsally anteroventral to uncinate pro- cess. LE2 on dorsally expanded posterior margin of Eb2 anteriorly. LE3 on Eb3 uncinate process anteriorly. @ Absent. LE4 on Eb4 levator process dorsally just medial to cartilaginous edge. LP insertion fused anteroventrally with posterov- entral edge of LE4 insertion. LI1 on Pb2 dorsomedially, continuing medially along anterolateralmost edge of Pb3 adjacent to Pb2; little, if any, larger than LI2. LI2 on Pb3 dorsoposteriorly, medial edge of at- tachment meeting lateral edge of TPb3-Eb4. TD comprises TEb2 and TPb3-Eb4. TEb2 broad with mid-longitudinal raphe giving rise dorsally to filmy CT sheets covering muscles; anterolateral and posterolateral muscle fibers coming together and at- taching to Eb2 dorsally anterior to LE2 insertion; muscle continuous posteriorly by fine diagonal mus- cle strand with TPb3-Eb4. TPb3-Eb4 on Pb3 dorsally meeting medial edge of LI2 insertion; muscle fibers extend dorsomedially joining ventral surface of Eb4 portion of muscle, which extends well laterally and inserts on Eb4 dorsally between medial end and sur- face ventral to uncinate process; on one side, an ap- parently anomalous muscle strand of Eb4 portion in- serts on dorsomedialmost bony surface of Eb3; mus- cle continuous posteriorly by fine diagonal muscle strand with SOD. @ Comprises TEb2 and TPb3-Eb3; TEb3 on Eb3 mid-posteriorly. OD3-4 origin finely on Pb2 anteriorly, continuing broadly on entire dorsomedial edge of Pb3; insertion broadly on anteromedial edge of Eb3 uncinate pro- cess and posterior surface of Eb4 uncinate process. @ On Pb2 and Pb3 on one side, but only on Pb3 on other. OP dorsally on Eb4, extending medially from ven- tral surface of levator process to posterior surface of uncinate process, ventrally on Cb5 dorsoposteriorly, medially continuous with SO. BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON Ad1-—3 absent. Ad4 in two sections; anterior section broadly dor- sally on Eb4 ventrally lateral to OP, ventrally broadly on Cb4 dorsally anterior to posterior section; poste- rior section about half as broad as anterior section, fibers angled ventrolaterally (versus almost vertical for anterior section), dorsally on ventral surface of Eb4 levator process, ventrally on Cb4 dorsally pos- terior to anterior section, joining raphe with Ad5 on Cb4. @ Consists of anterior section only. Ad5 dorsally on posterolateral bony surface of Cb4 well medial to distal end, joining raphe with Ad4; ventrally on bony surface of Cb5 dorsolaterally, join- ing raphe posteriorly with OP ventroanteriorly. SOD present. RDs separated by distance greater than diameter of one RD. Additional remarks. SCL attached mid-dorsally by CT to tip of posterior cartilaginous end of Bb3. TV4 free from Cb5s. Pb4 and UP4 present. Pbl mostly bony. IAC present (see discussion below). Medial end of Eb3 larger than that of Eb4. Pb2 toothed. @ Pb4, if present, greatly reduced. IAC absent. Platycephalus appears to be more plesiomorphic than Inegocia in having IAC, LE3, and Pb4, and more specialized, perhaps, in having TD attaching to Eb4. Only Platycephalus is included in the cladistic analysis. Imamura (1996) published a phylogenetic analysis of the Platycephalidae and related taxa. He described the dorsal gill-arch musculature (1996:172-173) of platycephalids in general, provided a generalized il- lustration (1996:fig. 43) of this musculature in the platycephalid Sorsogona tuberculata Cuvier, and dis- cussed variation he found in other platycephalid taxa. In so far as they are comparable, our findings agree with his. Imamura (1996:132—133) also illustrated and re- ported briefly on the gill-arch muscles of the Bem- bridae and Hoplichthyiidae, outgroup families most closely related to the Platycephalidae. He illustrates (but does not label) the most plesiomorphic of these, Bembridae, as having or lacking TPb2 (depending on genus), which muscle he reports is lacking in Hopli- chthyiidae and Platycephalidae. CHAMPSODONTIDAE Champsodon atridorsalis Ochiai and Nakamura, USNM 297752, ca. 94 mm. Additional material. USNM 297752, ca. 82 mm; C. vorax Giinther USNM 122578, 126 mm, partial in situ dissections to determine presence of LP, q.v. Not illustrated Description. Remarks. Mooi and Johnson (1997) present a de- tailed morphological description of Champsodon vor- NUMBER 11 ax, which is representative of the genus. We agree with most of their description of the dorsal gill-arch muscles, but expand on it and note a few differences. LE1 broadly on Eb! dorsally well lateral to tip of uncinate process, beginning near point of anteriorly deflected (normally medially directed) arm of Eb1 (see Mooi and Johnson, 1997:fig. 8b for dorsal gill- arch skeletal structure). LE2 on apex of prominent, raised Eb2 triangular process. LE3 slender, on tips of joined Eb3 and Eb4 unci- nate process anteriorly. LE4 largest levator, on Eb4 dorsoposteriorly. LP very fine, fragile, origin near dorsoanteriormost edge of PP, insertion on Eb4 co-incident with ventro- lateralmost edge of LE4 insertion. Remarks. Mooi and Johnson (1997:152) reported LP absent, which VGS verified (ibid., p. 174), based on what now appears to have been a defective dis- section. LP is often fragile and destroyed in acantho- morphs unless special care is taken to ascertain its presence early during dissection while the levators are still attached at both their origins and insertions. LI1 on Pb3 dorsally beginning just posterior to an- teriormost tip. LI2 on Eb3 posteriorly a little lateral to medial end, penetrates OD3—4 on way to insertion. Remarks. Champsodon is the only taxon we en- countered in which LI2 inserts exclusively on an Eb. In percomorphs, LI2 usually inserts on Pb3 at, or occasionally ventral, to articulation of Pb3 with Eb3, and it is possible that we have overlooked the fact that some LI2 filaments might be associated with the distal end of Eb3. Penetration of OD3—4 by LI2 oc- curs uncommonly in acanthomorphs (e.g., Brotula, Ophidiidae, and related Calomopteryx, Bythitidae; synapomorphic for gobioid family Odontobutidae). TD comprises a modified TPb2 and TEb2 (togeth- er with continuous mid-longitudinal raphe), and TEb4 (no raphe). TPb2 in three parts: anterior part a broad, transversely continuous strap attaching antero- laterally to tip of Eb1 uncinate process and adjacent IAC dorsomedially, also weaker attachments to ad- jacent dorsal edges of anteriormost tips of Pb2 and Pb3; posterior two parts (one on each side) each con- sisting of slender muscle slip arising from surface of TEb2 lateral to mid-longitudinal raphe and curving anterolaterally before finely, tendinously joining short raphe extending posteriorly from attachment of anterior TPb2 part to Eb1; anteroventral edge of LI1 also finely joined to same raphe, which marks shal- low constriction between anterior TPb2 part and TEb2; TPb2 otherwise broadly continuous posteri- orly with TEb2. TEb2 a broad, transversely contin- uous muscle strap narrowing laterally and twisting as it passes between LI1 and LI2 and attaches on TEb2 dorsally anterior to LE2 insertion; TEb2 posteriorly 173 is free from and dorsal to anterior edge of TEb4. TEb4 attaches broadly on posterior edge of Eb4. Remarks. TPb2 is unusual but readily derivable from a common acanthomorph TPb2 muscle state in which TPb2 consists of a depressed, roundish pad dorsal to and continuous partially or completely ven- trally with TEb2. Reduction of much of the pad, leav- ing only its lateral edges and anterior portion would result in the condition found in Champsodon. In some acanthomorphs (e.g., Psenopsis, Plate 177) only the anterolaterally curving portion of TPb2 is present. The attachment of TPb2 to the tip of Ebl uncinate process appears to be concomitant with a reduction in size and ventral displacement of IAC, which, unusual for an acanthomorph, articulates with the bony surface of the Eb1 uncinate process ventral to its cartilage tip. While not common among acan- thomorphs, TPb2 may attach well out on the dorsal surface of IAC (e.g., Pseudupeneus, Plate 124). OD3-—4 originates very broadly on Pb3 dorsome- dially and inserts on Eb3 anteriorly ventral to tip of uncinate process and Eb4 posteriorly ventral to tip of uncinate process; muscle penetrated by LI2 (see re- marks following LI2). OP dorsally on Eb4 posteroventrally (slightly ven- tral to level of SO attachment on Eb4), beginning medially ventral to LE4 insertion and extending lat- erally and meeting Ad4 dorsomedially; curving strongly ventromedially and attaching on Cb5 nar- rowly posteriorly medial to posterior end of Cb5; mid-medially not clearly differentiated from SO. Remarks. Mooi and Johnson (1997:fig. 12a) do not differentiate OP and Ad4 from SO. Ad1-—3 absent. Ad4 dorsally on Eb4 ventrally beginning medially at lateral edge of OP and extending laterally almost to distalmost end of bony Eb4 surface, attaching ven- trally on Cb4 dorsally, meeting entire edge of Ad5 attachment on Cb4. Ad5 relatively long, dorsoanteriorly on Cb4 begin- ning medial to distal end and continuing medially on Cb4 for about half length of Ad5, joining raphe with Ad4 attachment on Cb4; posteroventrally on Cb5 dorsally beginning at distal end and extending me- dially a short distance, meeting OP insertion. SOD absent. RDs separated by about diameter of one RD. Additional remarks. SCL present. TV4 free from Cb5s. Pb4 and UP4 present. Pb2 toothed. Eb4 levator process absent. HEXAGRAMMIDAE Hexagrammos stelleri Telesius, USNM 130279, 2 specimens, 119-123 mm. Plate 145 174 Description. LEI on raised posterior surface of Eb! dorsally lat- eral to tip of uncinate process. LE2 on Eb2 mid-dorsoposteriorly, meeting distal end of posterior branch of TEb2 anteriorly and, on right side only, meeting distal end of anterior branch posteriorly (fails anteriorly to meet TEb2 posteriorly on left side). LE3 on Eb3 just anteroventral to tip uncinate pro- cess. LE4 on Eb4 dorsoposteriorly lateral to uncinate process, ventrolateral margin of insertion joined by IL} e2 LP on Eb4 dorsolaterally, joining LE4 insertion ventrolaterally and extending laterally to dorsodistal- most bony end of Eb4. LI1 extends ventrally between overlapping anterior ends of Pb2 (anteroventral) and Pb3 (posterodorsal), inserting almost entirely on Pb3 ventroanteriorly, in- cluding dorsoanteromedial edge of anterior end of Pb3 dorsal to dorsoanteriormost origin of OD3—4 on Pb3; a few fibers insert on Pb2 dorsolateral surface. LI2 on Pb3 dorsoposteriorly just medial to medial end of Eb3; posterior half of medial edge of insertion meets lateral edge of TPb3-Eb3 attachment on Pb3. TD comprises TEb2, TPb3-Eb3, and, questionably, TPb2 (see remarks following TD description). TEb2 medially broad, notched mid-posteriorly, with irreg- ular mid-longitudinal raphe attaching dorsally to filmy CT sheets and ventrally to CT of pharyngeal roof; medial portion lies dorsal to, and extends well anterior to, all Pbs except Pb1l; dorsolateral margin on each side of central portion forms fine, flat muscle ribbon, which unites anteromedially and posterome- dially with remainder of muscle; muscle narrows and extends laterally, dividing and attaching on Eb2 dor- sally anterior and posterior to LE2 insertion (see also LE2); few fine muscle strands join heart-shaped por- tion posteroventrally to mid-anterior edge of TPb3- Eb3 (strands not visible in dorsal view). TPb3-Eb3 on Pb3 dorsoposteriorly joining posterior half of LI2 insertion medially and continuing posteriorly onto Eb3 dorsomedially; posteromedially continuous by diagonal muscle strands with SOD. Remarks. Shinohara (1994:50) illustrated the dor- sal gill-arch musculature of the hexagrammid Pleu- rogrammus azonus Jordan and Metz. The illustration appears generalized, and he reported no variation in the muscles either among hexagrammids or between hexagrammids and his comparative material. His re- port that TDA includes attachment to Eb4 is not true of our specimens of H. stelleri (a species included in his material), nor is his report that LI1 inserts [only] on Pb2. The extent to which unexplored variation probably exists in his comparative material should be obvious from the variation in muscles often reported in the present study. BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON He faintly indicates a fine circular perimeter of muscle fibers on a central differentiated plate-like muscle section between the right and left “‘arms”’ of what is clearly TEb2, but does not label or describe it. Imamura and Yabe (2002:fig. 14A) illustrate the dorsal gill-arch musculature of Hexagrammos lago- cephalus (Pallas), which, insofar as can be told, ap- pears mostly similar to ours of H. stelleri. It differs in having TPb2 well-developed and unambiguous and in indicating that the attachment of TEb2 to Eb2 is only anterior to LE2 insertion, which may be an oversight. Based on Imamura and Yabe’s illustration, we rec- ognize the presence of TPb2 in our specimen of H. stelleri. OD3-—4 originating broadly on posteroventral shelf of Pb3 and inserting on joined Eb3 and Eb4 uncinate processes medially; dorsoanteriorly, origin meets LI1 insertion on Pb3. OP dorsally on Eb4 posteriorly beginning medially on uncinate process, meeting OD3—4 there in a raphe, and continuing laterally to point ventral to LE4 in- sertion; ventrally, broadly on Cb5 dorsally, meeting and narrowly joining ventrolateral end of Ad5 on dorsodistalmost end of Cb5. M. SO-Pb3 absent. Ad1-—3 absent; fine GFM on anterolateral surfaces of Eb1 and Cbl, and anterolateral edges of Eb2 and Cb2, and Eb3 and Cb3. Ad4 dorsally on Eb4 posteroventrally, beginning just medial to lateral edge of OP and extending lat- erally to Eb4-Cb4 joint, ventrally on Cb4 broadly an- terior to Ad5. Ad5 dorsally on Cb4 beginning posterodistally and extending a short distance medially, ventrally, nar- rowly on Cb5 dorsally, mostly anterior to OP. SOD present. RDs adjacent or narrowly separated, inserting on Pb3 posteriorly. Additional remarks. SCL present. TV4 free from Cb5s. IAC reduced to one or two small cartilages contained in ligament attaching Eb! uncinate process to Pb2. Pb4 and UP4 present. Pb! cartilaginous. In a C&S specimen of H. superciliosus (Pallas), USNM 290478 (ca. 135 mm SL), Pb! is cartilaginous on one side and ossified, with cartilaginous dorsal and ven- tral ends, on the other. Eb4 levator process absent. The attachment of TEb2 to Eb2 both anterior and posterior to LE2 insertion was otherwise encountered only in Imamura and Yabe’s (2002:fig. 14B) illustra- tion of Pholis nemulosa. Another similarity of H. stelleri to Pholis is that both have cartilaginous Pb1s. The ligament attaching Eb1 uncinate process to Pb2 in Pholis appears similar to that of H. stelleri, except that the latter has a couple of small pieces of cartilage representing [AC included in the ligament. NUMBER 11 We believe that the TEb2 attachment, and probably the interarcual ligament and cartilaginous Pb1, sup- ports Imamura and Yabe’s (2002) hypothesis of a sis- ter-group relationship between their cottoid and zoar- coid lines. ANOPLOPOMATIDAE Anoplopoma fimbria (Pallas), USNM 269910, 149 mm. Not illustrated Description. LE1 on Eb! dorsoposteriorly a little medial to mid- length. LE2 on Eb2 dorsoposteriorly a little lateral to mid- length. LE3 on Eb3 just ventral to tip of uncinate process. LE4 on Eb4 dorsally somewhat medial to distal end, meeting LP insertion anteriorly. LP on Eb4 dorsally at bony posterior edge at me- dial end of thin cartilaginous extension of distal car- tilaginous tip of Eb4, meeting LE4 insertion poste- riorly. LI1 on Pb2 broadly anterodorsally and extending posteriorly on anterolateralmost edge of Pb3. LI2 on Pb3 dorsoposterolaterally medial to medial end of Eb3. TD comprises TPb2, TEb2, and TPb3-Eb3. TPb2 not attached to Pb2, comprises two flat, semi-circular bands dorsal to TEb2 mid-section, which floors space between bands; muscle confluent mid-laterally with TEb2; anteriorly and posteriorly joining mid-longi- tudinal raphe which gives rise dorsally to CT sheets covering muscle and is attached mid-ventrally to CT of pharyngeal roof. TEb2 extends laterally and atta- ches on Eb2 anteriorly a little lateral to LE2 insertion, muscle not continuous posteriorly with TPb3-Eb3. TPb3-Eb3 on Pb3 dorsolaterally ventral to OD3—4, beginning along medial margin of LI2 insertion and continuing posteriorly and joining medialmost edge of Eb3, broadly continuous posteriorly with SOD. OD3-—4 origin broadly on Pb3 dorsoanteriorly, posteriorly on Eb3 anteriorly beginning just ventral to tip of uncinate process, meeting LE3 insertion, and on Eb4 posteriorly beginning just ventral to tip of uncinate process. OP stringy, infiltrated with amorphous CT, begin- ning dorsally on Eb4 ventral to LP insertion and ex- tending medially to attachment of OD3—4, broadly overlapping Ad4 medially; ventrally on Cb5 begin- ning dorsodistally posterior to Ad5 and extending medially a short distance. Ad1 muscle flat, beginning narrowly on anterior surface of lateral quarter of Eb1, fanning out laterally and covering Eb1-C1 joint; not as well-developed relatively as Ad1 of other fishes, but better developed 175 than most GFM1Is; it and Ad2 and Ad3, possibly should be treated as GFMs. Ad2 similar to Ad1, but muscle extends medially on Eb2 and meets distal end of TEb2. Ad3 similar to Ad1l, but muscle extends medially to medial end of bony surface. Ad4 dorsally broadly on Eb4 posteriorly, begin- ning medially near about medial margin of OP and extending laterally to axilla formed by posteriorly ex- tending (but medial in orientation) cartilaginous pro- cess on distal end of Cb5 and main portion of carti- laginous distal end; ventrally much more narrowly on Cb4 medial to angle formed by Cb4 and Eb4. Ad5 relatively short, ventrally on Cb5 dorsodistal- ly anterior to OP, dorsally on Cb4 distally ventral to posteriorly extending cartilaginous process (see Ad4). SOD broad. RDs robust, barely separated from each other. Additional remarks. SCL present. TV4 free from Cb5s. Pb4 and UP4 present. Pb1 absent (anterior pro- cess absent). Pb2 toothed. [AC absent. Medial end of Eb4 smaller than that of Eb3. Eb4 levator process absent. PCI begins a little medial to distal end of Cb5 and extends well medially. There is a remarkable similarity of the gill-arch muscles of Anoplopoma and the zoarcoid Bathymas- ter. RHAMPHOCOTTIDAE Rhamphocottus richardsonii Giinther, USNM 49141, 57 mm. Plate 146 Description. LE! on Eb! lateral to broad uncinate process. LE2 on Eb2 mid-dorsoposteriorly. LE3 absent. (See additional remarks.) LE4 on Eb4 dorsoposterolaterally medial to distal end, meeting LP insertion medially. LP at and lateral to LE4 insertion. LI1 partially on Pb2 dorsoanteriorly, but mostly on dorsoanteriormost surface of Pb3, joining raphe me- dially with OD3-—4 origin present). LI2 on Pb3 dorsolaterally just medial to medial end of Eb3. TD comprises TPb2, TEb2, and TPb3-Eb3. TPb2 a bilateral pair of laterally convex muscle straps, not attached to any skeletal element (muscle usually at- taches to Pb2 in acanthomorphs), arising broadly, seamlessly anteriorly and posteriorly from TEb2. TEb2 broadly oblong mid-dorsally with mid-longi- tudinal raphe giving rise dorsally to filmy CT; muscle attaching beginning mid-anteroventrally and continu- ing posteriorly, to CT of pharyngeal roof; attaching laterally on Eb2 dorsoanteriorly anterior to LE2 in- sertion; overlapping anterior end of, and posteroven- 176 trally continuous by fine muscle strands with, TPb3- Eb3. TPb3-Eb3 on Pb3 dorsoposterolaterally medial to medial end of Eb3 and ventral to attachment on dorsomedial end of Eb3, broadly continuous poste- riorly with SOD. OD comprises OD3—4 and OD3’. OD3-—4 origin broadly on Pb3 dorsomedially, branching ventrolat- erally almost immediately into short OD3’ (not illus- trated), which inserts on medial end of Eb3 dorsally, and OD3-—4, which inserts on Eb3 uncinate process anteriorly and Eb4 uncinate process medially. Remarks. OD3’ usually attaches to Eb3 dorsally ventral to OD3—4 or OD3 attachment to anterior sur- face of Eb3 uncinate process. OP dorsally on Eb4 posteriorly beginning on pos- terior bony surface of uncinate process and extending laterally to below mid-insertion of LE4, ventrally on Cb5 posterolaterally, joining raphe ventrolaterally with Ad5, inseparable medially from SO. Ad1-3 absent. Ad4 dorsally on Eb4 posteriorly beginning below medial end of insertion of LE4 and extending later- ally to Eb4-Cb4 joint, ventrally on Cb4 dorsally me- dial to Eb4-Cb4 joint. Ad5 dorsally on Cb4 well medial to distal end, ventrally on Cb5 dorsodistally. SOD very broad. RDs adjacent. Additional remarks. SCL attached mid-dorsally to cartilaginous posterior end of Bb3. TV4 free from Cb5s. Pb4 and UP4 absent. IAC absent. Eb4 levator process absent. Yabe (1985) described the dorsal gill-arch mus- culature of cottoids and various related taxa. He hy- pothesized Rhamphocottus as the sister group of all other cottoids. Although his gill-arch muscle descrip- tions are generalized, we agree with him that there is limited variation in these muscles. Yabe reported, however, that the absence of LE3 is a synapomorphy of cottoids, although in some taxa a few strands of LE4 attach to Eb3. We found LE3, varying from well developed to moderately reduced in all three species of Myoxocephalus we examined, two of which were also included in Yabe’s material (see Myoxocephalus, Cottidae). The presence of LE3 in Myoxocephalus is either autapomorphic for the genus among cottoids, or more widely distributed among cottoids than Yabe noted. Yabe and Uyeno (1996) repeated Yabe’s (1985) cottoid synapomorphies in excluding Normanichthys crockeri (Normanichthyidae) from among the cot- toids. COTTIDAE Myoxocephalus niger (Bean), USNM 70823, 2 spec- imens, 83.8 mm (second specimen, 89.2 mm, ex- BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON amined only for presence of LE3, LE4, LP); M. jaok (Cuvier), USNM 127057, ca. 130 mm, and M. stelleri Tilesius, USNM 54232, 94.7 mm (both examined only for presence of LE3, LE4, LP). @ = Cottus carolinae (Gill), USNM 163051, not measured. Not illustrated Description. LE1 dorsoposterolaterally on Eb1 uncinate pro- cess. @) On Eb! lateral to uncinate process. LE2 dorsoposteriorly on Eb2 at about mid-length. @ Near distal end of Eb2 dorsoposteriorly. LE3 well developed, dorsoposteriorly on Eb3 slightly lateral to tip of uncinate process. @ Absent. (See also additional remarks under Ramphocottus (Ramphocottidae.) LE4 broadly on posterodorsal surface of Eb4 lat- eral to uncinate process. LP on posterodorsal surface of Eb4 at and poste- rior to LE4 insertion. LI1 has split insertion: on dorsoanterolateralmost edge of Pb3 (along origin of OD3-—4 anterolaterally) and on ventroanterior surface of Pb3 sandwiched be- tween Pb3 and dorsoanterior surface of Pb2. @ Sim- ilar, but Pb2 is absent. LI2 on Pb3 dorsal surface ventral to OD3—4 and anterior to medial end of Eb3. TD comprises TEb2 and TEb3. TEb2 attached an- teroventrally to CT dorsal to LI1 and along ventral midline to CT of pharyngeal roof; mid-longitudinal raphe dorsally giving rise to filmy CT sheets; attach- ing laterally on Eb2 dorsally anterior to medial end of LE2 insertion; faint indication dorsally on one side of laterally curving muscle fibers overlying main por- tion of muscle laterally (possible vestigial remnant of TPb2); continuous by fine strand of muscle posteri- orly with TEb3. TEb3 very narrow, finely, tendi- nously attaching to Eb3 about one-fourth length lat- eral to medial end; posteriorly continuous with SOD. @ Mid-longitudinal raphe is continuous on anterior portion of TEb3; TEb2 extends laterally to opposite LE3 insertion, is broadly continuous with, and over- laps slightly, anterior end of TEb3; no indication of TPb2 fibers; TEb3 attaches musculously to the me- dial end of Eb3. OD3-—4 origin on dorsoanterior surface of Pb3, in- sertion on medial edges of Eb3 and Eb4 uncinate processes. OD3’ origin on Pb3 dorsoanteriorly ventral to OD3-4, insertion on Eb3 anteromedial surface. @ Muscle not distinct, represented as muscle fibers originating on Pb3 immediately medial to OD3—4 in- sertion and inserting on medial end of Eb3 at and ventral to TEb3 attachment. (See remarks following OD3’ in Rhamphocottus). OP broadly dorsally on posterior surface of Eb4 NUMBER 11 uncinate process and extending laterally to below mid-insertion point of LE4, ventrally on Cb5 dorso- laterally, joining raphe there with Ad5; only slightly separated dorsolaterally from Ad4. @ Distinctly sep- arated from Ad4. Ad1-3 absent. Ad4 dorsally on Eb4 posterolaterally, ventrally joining raphe with Ad5 on Cb4 dorsolaterally. @ Ventrally broadly on Cb4 dorsally medial to Eb4-Cb4 joint. Ad5 dorsally on Cb4 posterior surface beginning about one-fourth length from lateral end and extend ing medially to point a little past mid-length of Cb4, ventrally on Cb5 dorsodistally. @ Dorsally on Cb4 a little medial to distal end, ventrally on distal end of Cb5 dorsolaterally. SOD broad. RDs adjacent. @ RDs separated by space equal to about half width of one RD. Additional notes. SCL attached mid-dorsally to ventrally extending cartilaginous posterior end of Bb3. TV4 free from Cb5s. Pb4 and UP4 absent. NORMANICHTHYIDAE Normanichthys crockeri Clark. Not illustrated Remarks. Specimens unavailable. See Yabe and Uyeno (1996) for anatomical description and illustra- tions. Information below extracted from their study and modified to be consistent with present study. Fig- ures mentioned are theirs. Although they excluded Nomanichthys as having close cottoid relationships, Yabe and Uyeno were unable to propose a reasonable sister group for it. They assigned it as incertae sedis to Scorpaeniformes on the sole basis that it has a suborbital stay. LE1, LE2, LE3, LE4, LP present. LI1 on anterior margin of Pb2. LI2 on dorsolateral margin of Pb3. TEb2 present [TPb2 absent, based on their fig. 7c]. An additional transversus of undescribed attachments present posteriorly. OD3-—4 originating on Pb3 dorsally, inserting broadly on Eb4, some fibers inserting on “‘dorsal pro- cess” of Eb3. OP not described, but present in fig. 7c. Ad1—4 present (fig. 8). Ad5 joins Cb5 and Eb4. SOD not described, but present in fig. 7c. RDs separate, inserting on dorsomedial margin of Pb3 (fig. 3c). Slender, muscular SO branch [modified CPb?] ex- tending anteriorly lateral to Pb4 [= UP4?] and Pb3 and attaching to posterolateral corner of Pb2. [See their fig. 3b]. This muscle differs from M. SO-Pb2 in that the latter passes medial to UP4 and Pb3. 177 SCL present (see their fig. 7b, but not labeled). TV4 apparently undivided. Pb] and IAC absent. Pb2 toothed. Pb4? UP4 present. Uncinate and levator pro- cesses not described. Their fig. 3c illustrates an all bony Eb3 uncinate process and no indication that Eb4 has either an uncinate or levator process. Carangoidei Although Freihofer (1978) first noted a synapo- morphy uniting the four families of carangoid fishes, Smith-Vaniz (1984) first formally defended their monophyly. NEMATISTIIDAE Nematistius pectoralis Gill, USNM 82203, 153 mm. Plate 147 Description. LEI! on Eb! uncinate process just lateral to carti- laginous tip. LE2 on raised dorsoposterior edge of Eb2 about one-third length Eb2 laterally. LE3 on medial edge of cartilaginous tip of Eb3 uncinate process. LE4 broadly on Eb4 posterior surface lateral to uncinate process. LP on Eb4 posterior surface beginning a little an- terior to medial end of LE4 insertion, and extending a little lateral to LE4 insertion. Right-side LP only (not illustrated) with slender strap of LP muscle pass- ing across posterior surface of LE4 and inserting on Eb4 at mid-posterior point of LE4 insertion. LI1 insertion ventral to TPb2, on dorsoanterior- most bony and cartilaginous surface of anterior end of Pb2. LI2 on Pb3 dorsolaterally, extending posteriorly to margin of medial end of Eb3. TD comprises TPb2, TEb2, and TPb3-Eb3. TPb2 kidney-shaped pad, deeply notched mid-anteriorly with posterior end of notch continuing as laterally offset raphe (one side only; raphe gives rise dorsally to CT covering muscles); attached anterolaterally to cartilaginous anteriormost end of Pb2 at junction with IAC, free edge of muscle from this point later- ally giving rise to CT sheet covering muscles; ven- trally, broadly confluent with TEb2. TEb2 ventral to TPb2 and dorsal to OD3-4 origin, extending out about one-third distance along dorsal surface of Eb2 to point anterior to LE2. TPb3-Eb3 on Pb3 dorsal surface along medial edge of LI2, and, passing dorsal to medial end of Eb4, inserts on dorsoposterior sur- face of medial end of Eb3; continuous dorsoposter- iorly by slender diagonal muscle strap with slender SOD. OD3-—4 anteriorly on dorsomedial surface of Pb3, posteriorly on anterior surface of Eb3 uncinate pro- 178 cess and anterior surface and medial edge of Eb4 uncinate process. OP dorsally, broadly on Eb4 posterior surface just ventral to LE4 and LP insertions, ventrally, broadly on Cb5 bony posterior surface beginning medial to cartilaginous posterior tip, posteriorly overlapping Ad5 attachment to Cb5. On right-side only, OP ven- trally becomes CT, which attaches to Ad5 posterior surface dorsal to Cb5 (not illustrated). Ad1-—3 absent. Ad4 dorsally on ventral surface of distal one-third of Eb4, partly overlapped posteriorly by lateral edge of OP, ventrally on Cb4 medial to Eb4-Cb4 joint. Ad5 on distal fourth of dorsal surface of Cb5 an- terior to OP and dorsoposterior surface of Cb4 just ventral to AC. SOD present. RDs well separated. Additional remarks. SCL attached mid-dorsally to posteroventral tip of Bb3. TV4 free from Cb5s. Pb2 toothed. AC dorsoposteriorly on cartilaginous distal tip of Cb4. Pb4 and UP4 present. Eb4 levator process absent. CARANGIDAE Selar crumenophthalmus (Bloch), USNM (not measured). 189251 Plate 148 Additional material. @ = Scomberoides tol (Cuvier), USNM 76607, 136 mm. Description. LE! on base of Eb! uncinate process anteriorly. @ On uncinate process dorsoanteriorly. LE2 on tip of expanded bony dorsoposterior mar- gin of Eb2. ® On prominent bony process arising from posterior margin of Eb2. LE3 finely, tendinously on tip of Eb3 uncinate pro- cess anteriorly. @ Musculously on Eb3 uncinate pro- cess anteriorly, there meeting OD3—4. LE4 on Eb4 dorsally just lateral to uncinate pro- cess. @) On Eb4 dorsally between uncinate and le- vator processes. LP on Eb4 beginning at and anterior to LE4 in- sertion and extending ventrolaterally, completely covering broad levator process, with posterior fibers continuous with Ad4 dorsoposteriorly. @ Levator process relatively small, muscle not continuous with Ad4. LI1 on Pb2 broadly dorsally, beginning anteriorly near articulation with IAC. LI2 on Pb3 dorsolaterally opposite medial end of Eb3. TD comprises TPb2, TEb2, and TPb3-Pb4-Eb3- Eb4. TPb2 dorsal to TEb2, pad-like with shallow lat- eral folds, anterior half of muscle transversely con- BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON tinuous, with mid-longitudinal raphe, which expands broadly replacing the muscle medially with an area of thick CT; CT sheets arise from lateral edge of raphe and from thick CT and attach to skull; muscle attaches anterolaterally to Pb2 dorsoanterior process adjacent to IAC and mid-ventrally along raphe with CT of pharyngeal roof. TEb2 joins TPb2 ventrally along mid-longitudinal raphe, and attaches strongly anteromedially to Pb3 dorsoanterolaterally posterior to LI1 insertion, and laterally on Eb2 dorsoanterola- teral to LE2 insertion (strong attachment of TEb2 to Pb3 is unusual in acanthomorphs). TPb2 and TEb2 not joined musculously with TPb3-Pb4-Eb3-Eb4. TPb3-Pb4-Eb3-Eb4 broad, with mid-longitudinal ra- phe on anterior half; muscle attaches to Pb3 dorso- posteriorly medial to medial end of Eb3, continues posteriorly and attaches along posteromedial edge of Eb3, dorsal surface of Pb4, and dorsomedial surface of Eb4. @ TEb2 destroyed befere determining if it was attached to Pb3. TD posteriorly appears to com- prise only TPb3-Eb3-Eb4; what appears to be in same position as Pb4 in other carangids, appears to be an elongate cartilaginous tip of Pb3 to which UP4 is attached dorsally. OD3—4, OD3’. OD3-4 origin on Pb3 dorsoposter- omedially, below TEb2 posteriorly; insertion on an- terior surfaces of Eb3 and Eb4 uncinate processes; OD3’ (not illustrated) splits off ventrally from OD3— 4 about half-way between origin and uncinate pro- cesses and inserts on Eb3 dorsally ventral to OD3—4. OP dorsally on Eb4 uncinate process posteriorly, ventrally broadly tendinously on Cb5 posterolateral- ly, joining Ad5 posteroventrally. Ad1-—3 absent, but ropy GFMs present on antero- lateral surfaces of each Eb-Cb arch. Ad4 dorsally on Eb4 posterolaterally, mostly lat- eral to OP, continuous with LP; ventrally broadly on Cb4 dorsoposteriorly medial to Eb4-Cb4 joint. Ad5 dorsally on medial surface of AC4 and pos- terodistal end of Cb4, ventrally on Cb5 dorsally, join- ing tendinous ventral end of OP. @ Dorsally joining raphe with OP. SOD slender. RDs separated by space less than one RD diameter. Additional remarks. SCL attached mid-dorsally to ventroposteriorly extending cartilaginous tip of Bb3. TV4 free from Cb5s. Pb4 and UP4 present. Pb2 toothed. AC4 present. A moderately well-developed Eb4 flange is pres- ent. Flange is absent in @, moderately well developed in Carangoides crysos, very well developed in Tra- chinotus falcatus, and weakly developed and scarcely noticeable in Selene vomer and Decapterus macro- soma. RACHYCENTRIDAE Rachycentron canadum (Linnaeus), USNM 341455, 113 mm. NUMBER 11 Plate 149 Description. LE1 broadly on Eb1 uncinate process anteriorly ventral to cartilage tip. LE2 on dorsally expanded bony posterior edge of Eb2. LE3 on tip of Eb3 uncinate process anteriorly. LE4 narrowly tendinously on Eb4 just medial to tip of levator process. LP broadly on dorsal surface of Eb4 anterior to tip of levator process and anteroventral to LE4 insertion. LI1 on Pb2 dorsoanteromedially. LI2 on Pb3 dorsally medial to broad medial end of Eb3. TD comprises TEb2 and TPb3-Pb4-Eb3 (on left side only, a flat, thin, laterally curving, semicircular strand of muscle dorsal to TEb2, arising anteriorly, represents vestigial TPb2). TEb2 very broad central- ly, ventrally continuous along mid-longitudinal raphe with CT of pharyngeal roof, narrowing laterally and joining medial edge of LE2 insertion, continuing lat- erally and attaching to Eb2 dorsally anterior to LE2 insertion, posteriorly continuous by diagonal strand of muscle with TPb3-Pb4-Eb3. TPb3-Pb4-Eb3 dor- solaterally attaching to dorsoposteromedial surface of Eb3, ventrolaterally attaching to dorsoposterolateral surface of Pb3 and dorsal surface of Pb4, continuous by diagonal strand of muscle with SOD. OD3-—4, OD3’ originates broadly on Pb3 dorsally ventral to TEb2, branches ventrally just lateral to or- igin; dorsal branch (OD3—4) inserts on bony anterior surfaces just ventral to cartilage tips of Eb3 and Eb4 uncinate processes. OD3’ branches off OD3—4 ven- trally shortly after emerging posteriorly from below TEb2, and inserts on Eb3 dorsally ventroanterior to OD3-4 insertion on Eb3. OP dorsally on Eb4 posteroventrally beginning near medial end and extending laterally to point me- dial to levator process, ventrally on Cb5 dorsolater- ally, joining raphe with ventroposterior end of Ad5, medially continuous with SO. Ad1-—3 absent. Ad4 broadly on Eb4 ventrally beginning medially below levator process and extending laterally to bony end of Eb4, ventrally broadly on Cb4 medial to Eb4- Cb4 joint, joining Ad5 attachment on Cb4. Ad5 dorsally on Cb4 posterolaterally and adjacent AC4, ventrally on Cb5 dorsolaterally anterior to OP. SOD present. RDs separated by space less than diameter of one RD. Additional remarks. SCL questionably free from Bb3. TV4 free from Cb5s. Pb4 and UP4 present. IAC present. Pb2 toothed. 179 CORY PHAENIDAE Coryphaena equiselis Linnaeus, USNM 158126, 118 mm. Additional material. @ = Coryphaena hippurus Lin- naeus, USNM 340988, 105 mm; damaged prepa- ration. Plate 150 Description. LEI originating tendinously and joining tendinous origin and anterior margin of LE2; inserting very broadly on Eb1 dorsoposteriorly and bony dorsoan- terior surface of long uncinate process. LE2 on expanded posterior edge of Eb2; anterior edge of muscle tendinous, attaching to tip of Ebl uncinate process as muscle extends anterodorsally; tendinous edge joining tendinous dorsal extension of LE] ventral to origin. @ Not possible to tell if LE2 attached to Eb1l uncinate process. Remark. Ligament (Plate 150C, not labelled) at- taches tip of Eb] uncinate process posteriorly to an- terior edge of Eb2 near anterolateral end of TEb2. LE3 tendinously on tip of Eb3 uncinate process. LE4 on Eb4 just lateral to uncinate process. LP on Eb4 beginning at ventrolateral edge of LE4 insertion, continuing laterally to anterior edge of notched cartilaginous distal end of Eb4, crossing notch, and ending on AC4 medial to distal end of Eb4 and dorsal to medial end of Cb4. See also OP. LI1 on Pb2 dorsoanteriorly, beginning on anterior process that joins medial end of IAC. LI2 on Pb3 dorsolaterally. TD comprises TPb2, TEb2, TEb3 (or possibly TEb3-Eb4, see remarks). TPb2 circular, pad-like, centrally thin, thickened laterally, with mid-longitu- dinal raphe, completely covering and apparently completely continuous with central portion of TEb2 and almost completely continuous laterally with TEb2, tenuously attached by CT only to dorsal sur- face of Pb2 anteriormost tip (attachment easily bro- ken, not illustrated); circular area overlain with thin, very tightly applied CT sheet (could not be removed without damaging muscle). TEb2 inserting on most of dorsoanterior surface of Eb2, narrowly, weakly continuous mid-posteriorly with mid-anterior end of TEb3. TEb3 triangular, apex anteriorly continuous with TEb2, inserts along Eb3 dorsal surface ventral to OD3—4 and posterior to OD3’, posteriorly contin- uous by diagonal muscle strap with SOD. @ TPb2 deeply undercut laterally and separated from TEb2; TEb2 well separated and unconnected with TEb3; TEb3 strap-like, not triangular. Remarks. Very weak CT fibers questionably arise from mid-lateral dorsal surface of TEb3 and narrowly attach to medial edge of cartilaginous tip of Eb4 un- cinate process. The connection was destroyed on both 180 sides inadvertently and was so tenuous that we de- cided against identifying TEb3 as TEb3-Eb4. @ Con- dition precluded determination if Eb4 connection was present. There is no “substantive” attachment of TD to Pb3. Weak, filmy CT attaches the ventral surface of the combined TPb2-medial TEb2 to a tough CT sheet covering the dorsal surface of Pb3 (OD3-—4 originates on dorsolateral edge of sheet). M. Pb3-Eb2 (not illustrated) small, possibly anom- alous muscle ventral to TEb2, present on only one side; originating on Pb3 dorsoanteriorly, joining ra- phe there with OD-3—4, and inserting on dorsome- dialmost bony surface of Eb2. @ Not present. OD3-—4, OD3’ on Pb3 dorsolaterally ventral to TEb2, branches ventrally just lateral to origin; dorsal branch (OD3-—4) inserts on bony dorsoanterior sur- face of Eb3 uncinate process and bony dorsomedial edge of Eb4 uncinate process; ventral branch (OD3’) inserts along most of Eb3 dorsal surface ventral to OD3-4 insertion. OP on left side broadly on posteroventral surface of Eb4 beginning ventral to uncinate process and ex- tending medially, dorsolaterally joining LP along fine line of CT, muscle narrows as it extends ventrally and inserts on dorsodistal end of Cb5 together with Ad5 insertion. Right side OP, possibly abnormal (illustrated reversed in rear view), differs in having two separated portions dorsally, junction with LP along line of CT more extensive, fusing dorsoposteriorly with Ad4 dor- sal attachment to Eb4. @ Both sides with single OP similar to that of left side of C. equiselis. Ad1-—3 absent. Ad4 on Eb4 ventrolaterally, ventrally on Cb4 an- terior to Eb4-Cb4 joint, posterolaterally joining raphe with Ad5 near insertion on Cb4 (not visible in PI. 150). Ad5 ventrally broadly on dorsolateral surface of Cb5 and dorsally on posterodistal surfaces of Cb4 and AC4. SOD present. RDs separate. Additional remarks. SCL absent. TV4 free from Cb5s. IAC broadest laterally. Pb4 and UP4 present. Pb2 toothed. Eb4 levator process absent. AC4 present attaching to posterodistal end of Eb4 medially and dorsoposterodistal end of Cb4 (autogenous on one side and completely fused on other in @). Small ACs present on first and second arches, possibly derived from segmentation of distal ends of Eb] and Eb2, rather than from distal ends of Cbs. @ No ACs on first and second arches. Johnson (1984:497) hypothesized that Coryphaen- idae, Rachycentridae, and Echeneidae form a mono- phyletic group with Echeneidae as the sister group to the other two families. The results of a cladistic anal- ysis reported by O’ Toole (2002:617) corroborated the BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON monophyly of the three families, but indicated Cor- yphaenidae as the sister group of the other two fam- ilies. ECHENEIDAE Echeneis naucrates, USNM 202201, 2: 150, 175 mm SL; USNM 206662, 145 mm. Additional material. @ = Remora remora (Linnaeus), USNM 181890, 82.7 mm. Not illustrated Description. Remarks. @ Differs from Echeneis primarily in composition of TD and relationship of LI2 to OD3+4. LEI broadly on bony anterior surface of Eb] un- cinate process. LE2 broadly, dorsally, on anterior surface of ex- panded posterior margin of Eb2 at about mid-length. LE3 variable: on tip of Eb3 uncinate process an- teriorly or on joined tips of Eb3 and Eb4 uncinate processes anteriorly. @ Same variation (each side dif- ferent). LE4 and/or LP. See following remarks. Remarks. Whether the relatively well-developed muscle inserting broadly on the flat bony surface of Eb4 well lateral to the uncinate process represents only LP or also includes LE4 is unclear (the muscle was destroyed during dissection of Remora). The muscle originates well posterior to the origins of the other levators, which originate near each other or to- gether. This probably indicates that the muscle in question represents LP and that LE4 is absent, which is rare among acanthomorph fishes (also probably true of Pholidichthys and Spinachia). The gill-arches of echeneids are greatly appressed against the ventral skull surface and the origin of LE4 (if the muscle includes that levator) may have been separated pos- teriorly from those of the other levators with which it is normally associated. Additionally, the muscle on each side of each of the three specimens appears to incorporate two incompletely separated parts. We conclude parsimoniously that LP is present, and that LE4 is questionably absent. @ Information unavail- able. LI1 on dorsoanterior surfaces of Pb2 and adjacent Pb3; insertion is ventral to tough CT extending an- teriorly from, and forming broad mid-section of TEb2. LI2 largest levator, penetrating OD3—4 on way from origin to insertion on Pb3 broadly posterolat- erally. @ Does not penetrate OD3-—4, but passes ven- tral to it, as is usual in most acanthomorphs. TD comprises TEb2, TEb3, and TUP4. TEb2 a broad muscle on each side joined by broad median area of tough CT covering large, musculously naked dorsal Pb3 facets; muscle forms anterior two-thirds NUMBER 11 of lateral CT margin and attaches laterally in two separate areas on Eb2: on medial edge of bony pro- cess supporting LE2 and dorsally anterior to LE2 in- sertion. TEb3 a slender muscle on each side extend- ing along posterior one-third of median area of tough CT covering Pb3 dorsal facets; muscle attaches on Eb3 dorsally medial to bony support of uncinate pro- cess. TEb3 continuous posteriorly by slender, diag- onal muscle strand with broad, transversely muscu- lously uninterrupted TUP4, which attaches to UP4 dorsally posterior to Pb4; posterolateral corner of muscle meets SO dorsolaterally and is broadly con- tinuous posteriorly with SOD. @ Comprises TEb2 and TUP4-Eb4. TEb2 relatively slender, forms about one-fourth to one-third of lateral CT margin. TUP4- Eb4 transversely musculously continuous, attaching narrowly anterolaterally on Eb4 dorsomedially and on medial edge of UP4. OD3-4 robust, origin broadly on lateral surface of musculously naked Pb3 dorsal facet, insertion on bony anterior surface and medial edge of Eb3 unci- nate process and bony medial edge of Eb4 uncinate process; muscle penetrated by LI2, separating most of portion inserting on Eb3 from most of portion in- serting on Eb4; posterolaterally, muscle forms raphe with OP dorsally. ® Muscle relatively flat, not pen- etrated by LI2, inserting on medial bony edges of Eb3 and Eb4 uncinate processes. OP with two separate sections; dorsally medial section on posterior bony surface of Eb4 uncinate process, lateral section on posterior surface of levator process; two sections meet in raphe ventrally and at- tach on Cb5 posteromedially, posteromedial to me- dial end of Ad5; dorsally, OP sections overlap much of Ad4 posteriorly. Ad1-—3 absent. Ad4 dorsally on Eb4 beginning posteroventrally and posterorlaterally; medially mostly anterior to OP, extending laterally to end of Eb4 bony surface; ven- trally on Cb4 anterior to both Ad5 and OP. Ad5 dorsally on AC4 and distal end of Cb4 pos- teriorly, ventrally on Cb5 beginning dorsodistally and extending medially anterior to OP. SOD present. RDs moderately slender, separated by distance greater than one RD diameter. Additional remarks. SCL present, but highly mod- ified; apparently forming circle, the anterior portion of which is a sheet of tough ligamentous tissue. TV4 free from Cb5s. Pb4 and UP4 present. AC4 present. IAC absent. Scombroidei Johnson (1986) included the Sphyraenidae in the Scombroidei; however, Orrell et al. (2003:45 and in preparation) concluded based on a molecular study 181 involving outgroups and representatives of most pu- tative scombroid genera that “there is no support for a close relationship between barracudas (Sphyraeni- dae) and the Scombroidei.”’ POMATOMIDAE Pomatomus saltatrix (Linnaeus), USNM 289926, 3 specimens, 113-126 mm. Plate 151 Description. LE1 on Eb! uncinate process dorsoanteriorly ven- tral to cartilage tip. LE2 on raised bony dorsoposterior edge of Eb2. Remarks. In a large articulated gill-arch skeleton of Pomatomus (USNM 016528, specimen size un- known; gill arches 200 mm from basihyal to 5th Cb), a distinct, bony, prong-like process arises from Eb2 dorsoposteriorly. LE3 on tip of Eb3 uncinate process dorsoanterior- ly. LE4 on dorsal edge of Eb4 levator process lateral to uncinate process, posteroventromedially joining raphe with OP dorsally. Remarks. The cartilaginous portion of the Eb4 le- vator process varies in and among the three speci- mens: small, round cartilage, two separated small car- tilages, or a single linear cartilage, all on the posterior edge of Eb4. The cartilaginous portion(s) is well lat- eral to the uncinate process and medial to the distal cartilaginous end of Eb4. These minute levator pro- cesses persist in large specimens, as indicated by remnants of their presence in a large skeleton (see remarks following LE2). LP at and anterior to LE4 insertion. LI1 on Pb2 dorsoanteriorly beginning posterior to IAC attachment to Pb2. LI2 on Pb3 dorsolaterally well anterior to articu- lation with medial end of Eb3 and at and lateral to attachment of TPb3-Eb3 to Pb3. TD comprises TPb2, TEb2, and TPb3-Eb3. TPb2 squarish, dorsal to TEb2, with mid-longitudinal ra- phe, continuing posteriorly across TEb2; muscle at- tached anteroventrolaterally to dorsal end of Pb2 and adjacent dorsomedialmost surface of IAC, fusing ventroposterormedially with TEb2. TEb2 attaching laterally on dorsal surface of Eb2 at point anterior to ventromedial edge of LE2 insertion, posteriorly free and well separated from TPb3-Eb3. TPb3-Eb3 on Pb3 dorsoposteriorly beginning at and medial to LI2 and continuing onto dorsomedialmost surface of Eb3, continuous posteriorly by crossing muscle straps with SOD. M. Pb3-Eb2 (not illustrated, easily overlooked.) small, hidden muscle originating on Pb3 at and an- terior to LI2 insertion and inserting on Eb2 ventral to LE2 insertion. 182 OD3-—4, OD3’ origin ventral to TPb2 on dorsoan- terior surface of Pb3, insertion on Eb3 uncinate pro- cess dorsoanteriorly and on medial edge and dor- soanterior surface of Eb4 medial to uncinate process. OD3' splits off ventromedial surface of OD3—4 shortly posterior to origin and inserts on Eb3 dorsally anteroventral to uncinate process. OP dorsally on medial half of posterior surface of Eb4, joining raphe with LP posteroventromedially, ventrally on Cb5 dorsolaterally posterior to Ad5, joining raphe with Ad5 posterolaterally. Ad1-—3 absent. Ad4, dorsally on posterolateral surface of Eb4, overlapped posteromedially by OP, ventrally on Cb4 medial to Eb4-Cb4 joint. Ad5 ventrally on Cb5 dorsally mostly anterior to OP, dorsally on posterodistalmost surface of Cb4. SOD present. RDs adjacent or separated by space less than one- fourth diameter of one RD. Additional remarks. SCL attached mid-posteriorly to ventroposterior cartilaginous distal end of Bb3. TV4 free from Cb5s. Pb4 and UP4 present. Pb! with dorsal and ventral cartilage ends. Pb2 toothed. PCI attaches to Cb5 beginning well medial to distal end and extends medially. SCOMBROLABRACIDAE Scombrolabrax heterolepis Roule, USNM 187651, 2 specimens, 98.5—104 mm. Plate 152 Description. Remarks. LE1 origin tendinous, all other levators originate musculously. LE1 on dorsoposterior edge of Eb1 just lateral to tip of uncinate process. LE2 on mid-dorsoposterior edge of Eb2. LE3 on dorsomedial edge of tip of Eb3 uncinate process. LE4 on dorsal edge of Eb4 between tips of unci- nate and levator processes. LP at and lateral to LE4 insertion, ventrolateral edge continuous with CT sheet attaching along edges of fourth and fifth arches and containing PP, which impinges on LP insertion. LI1 on bony surface of Pb2 just ventral to joint with IAC. LI2 on bony Pb3 dorsal surface lateral to TPb3- Eb3 attachment and anterior to medialmost end of Eb3. TD comprises TPb2, TEb2, and TPb3-Eb3. TPb2 completely overlying TEb2 medially, almost circular, notched anteriorly with mid-longitudinal raphe, which attaches dorsally to CT sheets; muscle attached mid-anteriorly to CT of pharyngeal roof, attachment continuing dorsolaterally to (and over) dorsoanterior BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON end of Pb2, at which point flat laterally convex rib- bon of muscle arises (on each side) and continues to posterior end of mid-longitudinal raphe; ventral sur- face fused with TEb2, but muscle partially replaced by CT on mid-portion of right side (TEb2 visible through CT); TPb2 strongly attached to Pb2, overlies anterior end of Pb3 to which it is loosely attached. TEb2 extending laterally on dorsoanterior surface of Eb2 anterior to LE2 insertion, well separated from, but connected mid-posteriorly by CT (not illustrated) to TPb3-Eb3. TPb3-Eb3 anteriorly broadly on Pb3 dorsal surface medial to LI2 insertion, posteriorly narrowly on posteromedial surface of Eb3, continu- ous posteriorly by transverse muscle strands with SOD. OD3-4 originating on dorsoanteromedial edge of Pb3 ventral to TEb2 and inserting on joined anterior surface of Eb3 uncinate process and medial edge of Eb4 uncinate process. OP dorsally broadly on posterior surface of medial arm of Eb4 beginning just lateral to levator process and extending medially to end of bony surface, mid- medially inseparable from SO, ventrally broadly on dorsoposterior edge of Cb5 mostly posterior to Ad5. Ad1-—3 absent. Ad4 broadly dorsally on Eb4 dorsoposterior sur- face beginning below levator process and extending laterally to bony distal end, ventrally narrowly on Cb4 medial to Eb4-Cb4 joint. Ad5 broadly on Cb5 dorsolaterally, mostly antero- medial to OP, very narrowly on posterodistalmost end of Cb4. SOD present. RDs slightly separated. Additional remarks. SCL absent. TV4 free from Cb5s. Pb4 and UP4 present. Pb2 toothed. PCI atta- ches on Cb5 beginning medial to distal end and con- tinues medially; origin is from cleithrum by long ten- don. SCOMBRIDAE Scomber scombrus Linnaeus, USNM 203841, 143 mm. Plate 153 Description. LE1 very short, on tip of Ebl uncinate process, insertion tendinous. LE2 inserts by long tendon on dorsomedialmost edge of long bony Eb2 process. LE3 inserts by long tendon on dorsomedialmost edge of cartilaginous tip of Eb3 uncinate process, muscle becomes tendinous dorsally and then mus- culous again before attaching to skull. LE4 inserts by long tendon on posterodorsal edge of Eb4 lateral to uncinate process. LP absent. NUMBER 11 LI1 tendinously on Pb2 dorsally slightly anterior to joint with medial end of Eb2, joining raphe anter- omedially with anterolateral side of TPb2, where an- teromedial edge of TEb2 meets TPb2; about same size as LI2. LI2 on Pb3 dorsoposterolaterally. TD comprises TPb2, TEb2, TPb3-Eb3. TPb2 thin, oval, attached dorsomid-longitudinally to thin, tough CT sheet, mid-anteroventrally to CT of pharyngeal roof, and anteroventrolaterally to broad cartilaginous end of Pb2 just medial to articulation with IAC, con- tinuous ventrally with TEb2 (only narrow, arcing, free lateral edge distinguishes TPb2 from TEb2). TEb2 on Eb2 dorsally medial to LE2 insertion on long bony process and just anterior to anterior at- tachment of GFM2. TPb3-Eb3 broadly on Pb3 bony surface beginning anteriorly at about mid-length of bone and well medial to lateral margin and LI2 in- sertion, continuous posteriorly by thin diagonal mus- cle strands with Eb3 portion, which is on postero- medial surface of Eb3. M. Pb2-Eb1 (not visible in dorsal view) on Pb2 anterolaterally ventral to joint with IAC, and on Eb1 posterior surface ventral to uncinate process. Remarks. M. Pb2-Eb1 occurs otherwise only in pre-acanthomorphs. OD3-—4, OD3’ origin on most of dorsoanterome- dial surface of Pb3 ventral to TEb2, dividing poster- oventrally, with short ventral branch (OD3’) inserting on dorsomedial surface of Eb3 and combined dorsal insertion on medial surface of Eb3 uncinate process ventral to cartilage tip and bony dorsomedial edge of Eb4 uncinate process. OP dorsally on Eb4 posteriorly beginning medially at point between medial end and uncinate process and extending laterally well past uncinate process, to be- low LE4 insertion, ventrally on Cb5 joining tendon (raphe) with posterior edge of Ad5. Ad1-—3 absent. GFM1 and GFM2 present, super- ficially appear to be RecD2 and RecD3, but are as- sociated with gill filaments. Not interpreted as Ad1 and Ad3, because they do not extend onto associated Cbs. GFM2 on anterior edge of Eb2 anterior to prom- inent bony process supporting LE2, narrowing to point at attachment to ventrolateral edge of bony pro- cess supporting cartilage tip of Ebluncinate process, lateral edge of muscle associated with gill filaments. GEM3 on dorsolateral edge of prominent Eb2 pro- cess and anterior edge of Eb3 mid-laterally, lateral edge of muscle associated with gill filaments. Ad4 on ventral surface of Eb4 dorsolaterally and dorsodistal surface of Cb4 anterior to Eb4-Cb4 joint, ventroposteriorly fusing with Ad5 on Cb4. Ad5 moderately broadly on dorsodistal margin of Cb5 and narrowly on AC4 and posterodistalmost end of Cb4, fusing at about mid-anterior surface with Ad4 and ventrally joining tendinous edge of OP. SOD present. RDs separated by distance greater than twice di- ameter of one RD, each with small separate branch. Additional remarks. SCL present attached mid- dorsally to cartilaginous ventroposterior tip of Bb3. TV4 free from Cb5s. Pb4 absent, UP4 present. Eb4 levator process absent. AC4 attached to posterodistal end of Eb4 and dorsodistal end of Cb4 (also present in Rastralliger kanagurta (Cuvier), USNM 192526; not present in Scomberomorus cavalla (Cuvier), USNM 289928, which has a similarly positioned AC between Eb3 and Cb3). PCI attaches on Cb5 well medial to distal end and continues medially. Two enlarged, modified gill-raker patches are pres- ent on each side of SO (seen in posterior view, PI. 153B); internally these patches support filamentous teeth. Similarly positioned gill-raker patch noted in present study only in Melamphaidae. Sparoidei Orrell et al. (2002), based on a molecular phylo- genetic study, provided evidence that Lethrinidae are the sister group of the Sparidae, within a weakly sup- ported monophyletic group comprising Sparidae, Centracanthidae, Lethrinidae, and Nemipteridae. Car- penter and Johnson (2002), in a morphological phy- logenetic study (not involving muscles), however, hy- pothesized Nemipteridae (Lethrinidae (polytomus Sparidae-Centracanthidae)). In another molecular phylogenetic study, Orrell and Carpenter (2004) found that Sparidae are monophyletic only with in- clusion of Centracanthidae, which was not monophy- letic. In the same study, Sparoidei was not monophy- letic with inclusion of either Nemipteridae or Leth- rinidae, nor did the latter two families form a mono- phyletic group. NEMIPTERIDAE Nemipterus furcosus (Valenciennes), USNM 349457, 2 specimens, 93.5—104 mm. Plate 154 Description. LE1 on tip of bony uncinate process at mid-length of Eb1. Cartilage tipped Eb1 uncinate process absent. LE2 on raised posterior edge of Eb2; insertion pos- terorventrally continuous with ligament attaching Eb2 to Eb3 anteriorly. Remarks. Imamura (2000:214) described LE2 in nemipterids as inserting on both Eb2 and Eb3, a con- dition he found limited otherwise among perco- morphs to malacanthids. We believe the state in ma- lacanthids is different from that in nemipterids; see remarks following description of LE2 in Malacanthi- dae. The nemipterid condition is more common than Imamura recognized, although we did not always note it in our descriptions. 184 LE3 on tip of Eb3 uncinate process anteriorly. LE4 on Eb4 levator process anteriorly. LP on Eb4 beginning at LE4 insertion posterolat- erally and extending laterally. LI1 on dorsoanteriormost surface of Pb2 beginning dorsally just ventral to cartilage tip. LI2 on Pb3 dorsolaterally just anterior to medial end of Eb3. TD comprises TPb2, TEb2, and TPb4-Eb3. TPb2 a pair of muscles, each laterally convex, joined to each other posteromedially and to TEb2 ventropos- teromedially; TPb2 and TEb2 attach to dorsalmost tip of Pb2 posteriorly. TEb2 attaches mid-anteroventral- ly to CT of pharyngeal roof, with mid-longitudinal raphe, which gives rise dorsally to CT sheets cover- ing muscles; TEb2 attaches laterally on Eb2 surface lateral to LE2 insertion; muscle not continuous pos- teriorly with TPb4-Eb3. TPb4-Eb3 anteriorly ventral to TEb2, continuous posteriorly with longitudinal SO muscle that extends anteriorly between Pb3s; TPb4- Eb3 dorsally slender, winglike, attaching to Eb3 pos- terior edge medial to uncinate process, ventromedi- ally attaching to Pb4 dorsally, continuous posteriorly with SOD. Remarks. Attachment of TEb2 to Pb2 is uncom- mon; also present, homoplastically, in Amarsipus (Amarsipidae). CPb (not illustrated) moderately well-developed, extending along lateral surfaces of Pb2, Pb3 and UP4 and attaching to lateral ends of Pb2 and UP4, with weak anterior branch extending medially between Pb2 and Pb3 and posterior branch between Pb3 and UP4, anterior branch joining even weaker muscle at- taching to Pb2 posteromedially and, together with posterior branch, fading into sparse SO muscle fibers of pharyngeal roof. In the smaller specimen, CPb does not extend anteriorly past its attachment to Pb2; in the larger specimen, muscle continues anteriorly from the attachment and begins to attenuate greatly at about the mid-anterior Pb2 margin, then fades into SO fibers medially. OD3, OD3’, OD4 all well developed, originating massively and essentially together from a dorsome- dially raised bony flange on Pb3 and the Pb3 surface ventrolateral to it, and separating almost immediately into OD3 and OD4 dorsally and longitudinally, and OD3’ ventrally from the other two. OD3 inserts on Eb3 uncinate process anteriorly, OD3’ on Eb3 dor- sally ventral to uncinate process, and OD4 on Eb4 levator process (uncinate process absent). OP dorsally on Eb4 posteriorly medial to levator process and ventrally on Cb5 posterolaterally, joining small raphe at its ventrolateralmost edge with Ad5 ventroposteriorly. Ad1-3 absent. Ad4 dorsally on Eb4 posteriorly beginning anterior to OP and extending laterally and attaching to Eb4 BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON ventrally near joint with Cb4, ventrally broadly on Cb4 dorsally medial to Eb4-Cb4 joint. Ad5 dorsally, moderately broadly on Cb4 postero- laterally, ventrally less broadly on Cb5 dorsolateral to OP. SOD broad. RDs separated by space less than half one RD di- ameter. Additional remarks. SCL very fine, easily broken, attached mid-dorsally to cartilaginous ventroposterior tip of Bb3. TV4 free from Cb5s. Pb4 and UP4 pres- ent. IAC absent. Pb2 toothed. Eb4 uncinate process absent, as inferred from in- sertion of LE4 on levator process (in acanthomorphs, LE4 inserts on the uncinate only in a few basal taxa, e.g., Lampris, Velifer. Also, indicated by apparent re- duction in size of the uncinate process and juxtapo- sition of it near the levator process in the closely related Lethrinus (Lethrinidae, q.v.). Carpenter and Johnson (2003:120) report that the absence of the Eb4 uncinate process is apomorphic for the Nemip- teridae among the sparoids. Cartilaginous, meniscus-like pad (not illustrated) present, tightly and closely attached and conforming with dorsal surface of ventral cartilaginous end of Pb1. Pad is easily overlooked, but appears to act as cushion between skull and Pb1, which curves around skull and attaches dorsally to it. Carpenter and Johnson (2002) hypothesized the monophyly of the Sparoidea, in which they included nemipterids as the sister group to lethrinids, and these together as the sister group of sparids and centracan- thids. LETHRINIDAE Lethrinus obsoletus (Forsskal), USNM 309317, 2 specimens, 84.3—103 mm. Plate 155 Additional material. © Lethrinus harak (Forsskal), USNM 259390, 85.5 mm. @ = Gymnocranius gri- seus (Temminck and Schlegel), 350957, 115 mm. @® = Monotaxis grandoculis (Valenciennes), USNM 264135, not measured. Description. LE1 broad based, on Eb1 just lateral to tip of un- cinate process; origin long, tendinous. LE2 on dorsally expanded posterior margin of Eb2. LE3 finely tendinously on tip of Eb3 uncinate pro- cess anteriorly. LE4 on tips of Eb4 levator and uncinate processes (see Additional remarks). LP finely, tendinously on Eb4 lateral to and not coincident with LE4 insertion (see Additional re- NUMBER 11 marks). @) @ LP insertion joins LE4 insertion later- ally. LI1 on Pb2 dorsolaterally beginning ventral to TPb2 attachment to Pb2 and extending ventrally. LI2 on Pb3 dorsoanterolaterally just posterior to Pb2 posterolaterally. TD comprises TPb2, TEb2, and TPb3-Eb4. TPb2 thick, laterally convex muscle pair lying dorsal to TEb2 and joined posteriorly by CT, which extends anteriorly as mid-longitudinal raphe of TEb2, which gives rise dorsally to CT sheets covering muscles and attaching to skull; anteriorly each member of TPb2 pair attaches to Pb2 dorsoanteriorly; posteromedially, along extension of raphe, and just anterior to its pos- terior end, each TPb2 member is joined ventrally by TEb2, which is transversely continuous anteriorly. Anterolaterally on each side, TEb2 joins TPb2 at at- tachment to Pb2; laterally TEb2 attaches on Eb2 dor- sally lateral to LE2 insertion; lateral end of TEb2 divides into anterior and posterior branches, with me- dial end of fine Ad2 inserting into divide (not appar- ent on Plate 155A); TEb2 attached mid-ventroanter- iorly between Pb3s to CT of pharyngeal roof; muscle not continuous posteriorly with TPb3-Eb4. TPb3-Eb4 ventral to origin of OD3—4 and OD3’, begins on Pb3 dorsolaterally just medial to posteromedialmost edge of Eb3 and is dorsomedially continuous with Eb4 portion of muscle anteroventrally; Eb4 portion is mostly separate from Pb3 section, and inserts on dor- somedial bony surface of Eb4; Pb3 portion is ventro- medially continuous with SOD, which is completely obscured in dorsal view of gill arches. ®@ TD includes TPb3-Eb3-Eb4 instead of TPb3-Eb4; only few strands of muscle attach to posteromedial surface of Eb3. ® TD includes TPb3-Pb4-Eb3 instead of TPb3- Eb4; only few strands of muscle attach to dorsome- dial surface of Pb4. CPb very prominent externally, especially anteri- orly, where it is transversely continuous; completely encircles Pb2 and Pb3 and attaches to posterolateral and posteromedial corners of UP4; medial fibers pass into SO longitudinal fibers. @ Muscle extends only from mid-lateral surface of UP4 on one side anteri- orly around to mid-lateral surface of UP4 on other side. OD3-—4, OD3’ origin broadly on Pb3 dorsomedi- ally; OD3—4 insertion on entire anterior surface of Eb3 uncinate process and entire medial edge of Eb4 leading to levator and uncinate processes; OD3’ split- ting off from OD3—4 almost immediately distal to origin and inserting on Eb3 dorsal surface ventral to Eb4 insertion of OD3—4. OP dorsally on most of bony posterior surface of Eb4 medial to levator process, ventrally broadly on CbS posteriorly, ventrally joining raphe with PCI (see Additional remarks). Ad1-3 relatively weak, on anterior surfaces of rel- 185 evant Eb and Cb; Ad2 extending medially onto dorsal surface of Eb2 between split lateral end of TEb2. Ad4 dorsally broadly on Eb4 posteriorly beginning medially on levator process and extending almost to lateral end of bone; ventrally broadly on Cb5 poste- riorly. Ad5 dorsally relatively narrowly on Cb4 postero- laterally and ventrally on Cb5 dorsolaterally. SOD present, ventral to TPb3-Eb4, not visible in dorsal view in illustrated specimen, but posterior to TPb3-Eb4 and visible in dorsal view in other speci- men and taxa. RDs adjacent. Additional remarks. SCL absent (but Bb3 has a very elongate posteroventrally extending cartilagi- nous end, normally present when SCL is present). TV4 free from Cb5s. IAC present. Pb! bony with cartilage tips. Pb2 toothed. Pb4 and UP4 present. Displacement of LP insertion from joining or meeting LE4 insertion appears to be unique to Leth- rinus among acanthomorphs. In lethrinids, the uncinate processes on Eb3 and Eb4 are not bound together as they are in most acan- thomorphs. The bony support of the Eb4 uncinate process is indistinguishable from that of the levator process and its cartilaginous tip, when present, is greatly reduced (vestigial) and barely separated from that of the levator process. In the nemipterids, the uncinate process has been entirely lost. The associ- ation of LE4 with the tip of Eb4 uncinate process is not the same condition as occurs in Caristius (Car- istiidae) in which the uncinate process is well devel- oped and in its usual position (joining Eb3 uncinate process) and there is no Eb4 levator process. PCI inserts broadly on Cb5 reaching the cartilage tip of the element in Gymnocranius and Monotaxis, failing to extend even near the tip in both species of Lethrinus. OP joins raphe with PCI. CENTRACANTHIDAE Spicara smaris (Linnaeus), USNM 269800, 2 speci- mens, 86.0—90.6 mm. Plate 156 SPARIDAE @ = Acanthopagrus bifasciatus (Forsskal), USNM 191682, 71.8 mm. Additional material. ® = Lagodon rhomboides (Lin- naeus), USNM 143843, not measured. © = Sarpa salpa (Linnaeus), USNM 343618, 73.6 mm. Not illustrated Remarks. Carpenter and Johnson (2002) hypothe- sized that centracanthids and sparids form a mono- phyletic group “with placement of centracanthids un- resolved with respect to sparid genera [p. 114]” 186 among the Sparoidea, which also includes nemipter- ids and lethrinids. Except for CPb, muscles for all three sparid taxa we include were recorded only as present or absent, and all are present as in Spicara, although the finer details of the descriptions for Spicara may not apply to the sparid taxa. Description. LEI on dorsoanterior surface of bony Eb] uncinate process. Remarks. The medial edge of the tip of the unci- nate process is minutely cartilaginous and easily overlooked. LE2 narrowly tendinously on raised posterior edge of Eb2 anteriorly. LE3 finely tendinously on tip of Eb3 uncinate pro- cess anteriorly. LE4 finely tendinously on tip of Eb4 levator pro- cess anteriorly. LP on dorsal surface of Eb4 lateral to levator pro- cess. LI1 on Pb2 dorsally at base of cartilage-tipped dor- sal process that articulates with IAC. LI2 on Pb3 dorsolaterally medial to medial end of Eb3. TD comprises TPb2, TEb2, and TPb3-Pb4-Eb3. TPb2 pair of flat, laterally curving muscles dorsal to TEb2; each member of pair attaching anteriorly to Pb2 cartilage-tipped anterior process; pair joined pos- teromedially by raphe, which continues anteriorly along center of TEb2; dorsally, raphe gives rise to CT sheets covering muscles and attaching to skull; TPb2s join TEb2 at raphe, which is attached ventrally to CT of pharyngeal roof between Pb3s; TPb2 and TEb2 not continuous posteriorly with TPb3-Pb4-Eb3. TEb2 broad, extending laterally and attaching to Eb2 dorsally anterolateral to LE2 insertion, meeting me- dial end of Ad2, failing to cover dorsoposterior Pb3 surfaces in illustrated specimen (probably unusual), but does so in other specimen and three sparid genera examined. CPb. TPb3-Pb4-Eb3 on Pb3 dorsoposter- olaterally at medial end of Eb3, continuing posteri- orly on small Pb4 dorsally, and extending well lat- erally and attaching finely on posterior bony edge of Eb3 medial to uncinate process and posterior to OD3'; TPb3-Pb4-Eb3 continuous posteroventrally by crossing strands of muscle with SOD. CPb (not illustrated) relatively weak muscle strands beginning on posterolateral corner of UP4, continuing anteriorly along Pb3 laterally and attach- ing to posterolateral end of autogenous tooth plate (Pb2’; see also remarks) joined to Pb2 posteriorly and absent from Pb2’ laterally, but beginning again on Pb2’ anterolaterally and continuing anteriorly around Pb2 and across mid-line of arches to opposite side; muscle strands from posterior end of Pb2’ also pass BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON medially between Pb2 and Pb3, with short branch attaching to Pb2 posteromedially and longer branch continuing posteriorly along medial side of Pb3 and attaching to posteromedial side of UP4, with strands also passing laterally between Pb3 and UP4 and join- ing strands passing anterolaterally to UP4 and Pb3. Muscle absent on Pb2 laterally, Pb2’ mid-laterally, and UP4 posteriorly. Remarks. CPb present in @ and ®, in which strands may be continuous anteriorly along Pb2’ lat- erally, but absent in @. Pb2’, which articulates closely with the posterior end of Pb2, has been treated as an Eb2 tooth plate (e.g., Carpenter and Johnson, 2002:120, character no. 34; see also discussion by Johnson (1992:19, item 4)) and is arbitrarily treated here as a part of Pb2. OD3—4, OD3’ origin broadly on Pb3 dorsally, an- teriorly ventral to TEb2; muscle dividing ventroan- terolaterally well lateral to origin, with separate ven- tral insertion (OD3’) on Eb3 dorsally ventroanterior to uncinate process, meeting medial end of Ad3, and (OD3—4) on Eb3 uncinate process anteriorly and Eb4 uncinate and levator processes medially. OP dorsally on Eb4 posteriorly medial to levator process, ventrally on Cb5 dorsoposteriorly, laterally joining raphe with Ad5 medially. M. Pb3-Cb5 diagonal strap of muscle attaching to Pb3 posteriorly and extending posteriorly medial to medial end of Eb4, then ventrally and attaching ten- dinously to posterodistal surface of Cb5. Remarks. Muscle appears restricted to centracan- thids and sparids and provides additional evidence of the close relationship of these two groups. Ad1-3, short, each on anterodistal surfaces of rel- evant Eb and associated Cb. Ad4 dorsally on Eb4 posteriorly beginning on le- vator process and extending laterally, ventrally on Cb4 dorsally medial to Eb4-Cb4 joint. Ad5 anteriorly on posterodistal surface of Cb4 and posteriorly on Cb5 dorsally, joining raphe dorsome- dially with OP ventrolaterally. SOD present. RDs separated by space about one-half diameter one RD. Additional remarks. SCL present (also @). @ ® Attached mid-dorsally to tip of ventroposteriorly curving cartilaginous end of Bb3. TV4 free from Cb5s. Pb4 and UP4 present. Carpenter and Johnson (2002) described the pe- culiar relationship of Pb1 to Eb! in centracanthids. The cartilaginous ventral end of Pb1 tightly joins and conforms with the dorsomedial surface of the anterior arm of Ebl, whereas the two elements are loosely articulated in the sparids, lethrinids and nemipterids. Additionally, we found that Pb1 of centracanthids ap- pears to have pivoted laterally from its articulation with Eb! and that it is the ventrolateral surface of NUMBER 11 the cartilaginous ventral end that is joining Eb1. Also, there is a thin wafer of fibrocartilage (not il- lustrated in Plate 156) that is tightly joined to the dorsal surface (or dorsomedial surface if Pbl were upright) of the cartilaginous ventral end of Pb1. The wafer rests against the ventral surface of the cranium and probably serves as a cushion. Girelloidei Girelloidei here coined as a subordinal taxon to facilitate designating a group of families (Girellidae, Scorpididae, Microcanthidae, Kyphosidae, Kuhliidae, Terapontidae, Arripidae, Oplegnathidae, but exclud- ing those belonging to the Stromateoidei) that John- son and Fritsche (1989) hypothesized formed a monophyletic group. The group, including stroma- teoids, was based primarily on their possessing Freih- ofer’s (1963) pattern 10 of the ramus lateralis acces- sorius (RLA). Johnson and Fritsche did not examine Amarsipus, sole member of the Amarsipidae, which has been included as a stromateoid since its original description (Haedrich, 1969). Amarsipus lacks the striking complex specialization that characterizes all other stromateoids (see both Amarsipus and Icosteus, Icosteidae, for discussions of the inter-relationships of Amarsipus), and if it is a stromateoid, is most probably the sister group of all other stromateoids. We examined Amarsipus for the presence of pattern 10 of RLA, and found it absent, but are uncertain which of the other patterns it possesses. Amarsipus’s relationship with other stromateoids is, therefore, an Open question. We only examined the musculature of three of the families of Girelloidei, and noted no information bearing on its monophyly. GIRELLIDAE Girella simplicidens Osburn and Nichols, USNM 167579, 95.6 mm, USNM 321278, 78.0 mm. Not illustrated Description. LE1 on Eb! posteriorly at about mid-length and well lateral to tip of horizontally directed uncinate process; tendon runs along lateral surface of muscle. LE2 on Eb2 posteriorly at about mid-length; slen- der tendon runs along ventral half of lateral edge of muscle. LE3 on tip of Eb3 uncinate process medially. LE4 on Eb4 posteriorly projecting levator process dorsally. LP on Eb4 dorsally a little lateral to LE4 insertion. Remarks. Among percomorphs, only Girella, Ra- chycentron (Rachycentridae), Spicara (Centracanthi- dae), lethrinids, and several atherinomorphs have LP inserting completely separate from LE4. In other per- 187 comorphs, the two muscles almost always insert to- gether and the insertions are often fused. LI1 on Pb2 dorsally just posterior to anteriormost tip; about same size as LI2. LI2 finely, tendinously on Pb3 dorsally immedi- ately medial to articulation with medial end of Eb3, near, if not bordering lateral edge of TPb3-Pb4-Eb3 on Pb3. TD comprises TPb2, TEb2, and TPb3-Pb4-Eb3 (TPb3-Pb4-Eb3-Eb4 in smaller specimen). TPb2 a thick, laterally curving, semicircular muscle on each side dorsal to TEb2 and ventral to thick CT pad; muscle arises posteriorly from posterolateral edge of moderately broad CT area, which narrows anteriorly into mid-longitudinal raphe as it extends across TEb2, gives rise to thick CT pad dorsally, and con- tinues anteriorly with CT of pharyngeal roof; anter- omedially muscle fades into TEb2; muscle impinges on Pb2 dorsoanteriorly, but is at most weakly, if at all attached to Pb2. TEb2 extends laterally onto Eb2 dorsally to point anterior to LE2 insertion, meeting medial end of GFM2. TPb3-Pb4-Eb3 beginning on Pb3 dorsolaterally a little anterior to articulation with Eb3, extending posteriorly and attaching to posterior corner of medial end of Eb3 and, not visible exter- nally, ventrally attaching on dorsal surface Pb4 (and on dorsal surface of medial end of Eb4 in smaller specimen); muscle continuous posteroventrally by fine muscle strands (not visible externally) with SOD. OD3-4 anteriorly on Pb3 dorsomedially ventral to TEb2, posteriorly on Eb3 broadly anteriorly begin- ning just ventral to tip of uncinate process and on Eb4 anteriorly beginning ventrolateral to uncinate process (muscle divides as it extends posterolateral from medial edge of Eb3 uncinate process). OP dorsally on Eb4 ventrally, beginning a little lateral to medial end of Eb4 and extending laterally and curving posteriorly as it follows posteriorly pro- jecting bony Eb4 shelf, but ending on shelf well an- terior (medial) to its posterior cartilage tip (levator process); ventrally on Cb5 posteriorly, beginning me- dial to distal end and continuing medially a short distance past attachment of Ad5; lateral edge of mus- cle is tendinous, becoming fascia-like on attaching to Cb5 and joining dorsomedial edge of Ad5; medially muscle is indistinguishable from SO. When gill arch- es are viewed posteriorly, only OP dorsal and ventral portions are visible, as Ad4 occludes mid-portion from view. Ad1-—3 GFMs1-3. Ad4 very broad, dorsally on Eb4 ventrally, begin- ning medially anterior to dorsolateral end of OP, ex- tending laterally posterolaterally around posteriorly projecting Eb4 shelf (supporting levator process), then laterally to end of bony Eb4 surface; ventrally relatively narrowly on Cb4 dorsally medial to Ad5. absent; moderately well-developed 188 Ad4, other than SO, is main muscle visible in pos- terior view of gill arches. Ad5 relatively small, anteriorly on posterodistal surface of Cb4 and Eb4 at joint and lateral to OP laterally, joining CT with OP ventrolaterally; poste- riorly on Cb5 posterodistally. SOD present. RDs adjacent. Additional remarks. SCL weakly attached mid- dorsally to ventrally projecting cartilaginous posterior end of Bb3. TV4 free from Cb5s. Pb! present, mostly bony. Pb2 toothed. Pb4 and UP4 present. IAC pres- ent. Medial end of Eb4 larger than medial end of Eb3. Eb4 flange absent. KUHLIIDAE Kuhlia mugil (Forster), USNM 114998, 2 specimens, 87.0—94.8 mm. Plate 157 Description. LEI on bony surface of Eb1l uncinate process be- ginning just lateral to joint with IAC. LE2 on dorsalmost edge of raised dorsoposterior margin of Eb2. LE3 on Eb3 extending laterally from medial edge of cartilage tip of uncinate process. LE4 on Eb4 bony surface just medial to cartilage tip of levator process. LP on Eb4 bony surface beginning just anterior to cartilage tip of levator process and extending to and joining LE4 insertion anteriorly. LI1 insertion mainly on dorsoposteriormost edge of Pb2 anterior process, with CT attachments to IAC dorsoposteromedialmost surface (adjacent to Pb2 in- sertion) and anterolateralmost edge of TPb2. LI2 on Pb3 dorsoposteriorly immediately medial to anteromedialmost edge of Eb3. TD comprises TPb2, TEb2, and TPb3-Pb4-Eb3. TPb2 divided; medially concave cord-like muscle on each side dorsal to TEb2; attaches anteromedially to dorsoanteriormost end of Pb2 with membranous con- tinuations onto IAC medially and adjacent LI1 inser- tion; anteromedially joining irregular mid-longitudi- nal raphe with TEb2 anteriorly, posteromedially fad- ing into TEb2; CT sheets arising from irregular mid- longitudinal raphe attach also on surface of TPb2. TEb2 flat medially, with irregular mid-longitudinal raphe, which is continuous ventroanteriorly with CT of pharyngeal roof; muscle thickening laterally and attaching broadly dorsally on Eb2 anteroventral to LE2; muscle discontinuous with TPb3-Pb4-Eb3. TPb3-Pb4-Eb3 attaching on Pb3 dorsolaterally ven- tral to OD3—4 and just medial to mid-medial edge of Eb3, continuing posteriorly on Eb3 posteromedial edge and ventrally on Pb4 dorsally (attachment on BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON Pb4 obscured in dorsal view), continuous by diagonal muscle strand with SOD. OD3-—4 origin broadly on Pb3 dorsoposteromedi- ally ventral to TEb2, insertion broadly on Eb3 dor- soanteriorly beginning just ventral to tip of uncinate process and narrowly on medial edge of Eb4 just ventral to tip of uncinate process or on anterior sur- face just ventral to tip of uncinate process. OP dorsally on Eb4 posteriorly beginning at me- dial end of bony surface and extending laterally to below, or slightly lateral to, uncinate process, ven- trally joining tough, clear CT (not illustrated) sur- rounding Ad5, and attaching to Cb5 ventromedially continuous with SO. Ad1-—3 absent (GFMs moderately developed). Ad4 dorsally on Eb4 posteriorly beginning slightly anteromedial to lateral edge of OP and extending lat- erally to below levator process, there becoming sharply less robust and extending somewhat ventrally to end of bony surface; ventrally, narrowly on Cb4 dorsally medial to Eb4-Cb4 joint. Ad5 dorsally on Cb4 posterodistally and AC4 ven- trally; ventrally on Cb5 posterolaterally. Muscle en- cased in tough, clear CT (removed in Plate 157), which is joined by OP ventrally. SOD present in both specimens. Johnson (1993:9) reported that SOD is absent in Kuhlia (see also remarks following SOD in Toxotes (Toxotidae) description). RDs adjacent or very slightly separated. Additional remarks. SCL attached mid-dorsally to ventroposteriorly extending cartilage tip of Bb3. TV4 free from Cb5s. Pb4 and UP4 present. Pb2 toothed. AC4 present (also in cleared and stained specimen of K. sandvicensis (Steindachner), USNM 289471. TERAPONTIDAE Leiopotherapon unicolor (Giinther), USNM 173654, 96.3 mm; USNM 173858, 120 mm, cleared and stained gill arches. Plate 158 Additional material. @ = Terapon jarbua (Forsskal), USNM 173657, 81.1 mm. Description. LE! slender, on tip of Eb1l uncinate process; mus- cle and Pb1 attached anteriorly to pharyngeal roof CT. @ Not especially slender. LE2 finely, tendinously on Eb2 mid-dorsoposter- iorly. @ Insertion musculous, not fine. LE3 on Eb3 uncinate process dorsoanteriorly just ventral to cartilage tip. LE4 on Eb4 posteriorly projecting levator process; ventroposterolateral and ventroposteromedial fibers continuous with Ad4. @ LE4 and Ad4 not continu- ous. NUMBER 11 LP on Eb4 beginning at ventro-anterolateral edge of LE4 insertion and continuing laterally a short dis- tance; posterolateralmost fibers continuous with Ad4 dorsally. ® Fibers not continuous with Ad4. LI1 on Pb2 dorsally just posterior to anteriormost tip. @ On Pb2 similarly, but, with additional attach- ment to posterior surface of Pb2-IAC joint. LI2 on Pb3 dorsolaterally just medial to medial end of articulation with Eb3. TD comprises TPb2, TEb2 and TPb3-Eb3. TPb2 a thick, irregularly round, concave pad dorsal to TEb2, with mid-posterior notch extending anteriorly as raphe, from which tough, filmy CT sheets arise and also attach to pad dorsolaterally; muscle attaches anterolaterally to Pb2 and fuses ventrally with mid- medial area of TEb2. TEb2 attaches by CT mid-ven- trally to CT of pharyngeal roof, extends on Eb2 dor- sally to medial surface of tiny, low, diagonal bony strut anterior to LE2 (medial end of GFM2 attaches to lateral surface of strut). TPb2 and TEb2 not con- tinuous with TPb3-Eb3. TPb3-Eb3 on Pb3 postero- laterally beginning just anterior to LI2, continuing posteriorly medial to LI2 insertion and extending onto Eb3 posteromedially; muscle continuous poste- riorly by diagonal muscle strand with SOD. ® Strut weakly developed, would have been overlooked without knowledge of occurrence in Leiopothera- pon—development possibly related to size of speci- men. OD3-—4, OD3’ originate together on Pb3 medially ventral to TEb2 and TPb3-Eb3, extend laterally with short OD3’ branch separating anteroventrally and 1n- serting on Eb3 dorsally ventral to uncinate process, there meeting medial end of GFM3. Major portion of muscle continues laterally with anterior portion in- serting broadly on Eb3 uncinate process anteriorly and posterior portion inserting on Eb4 uncinate pro- cess posteriorly, with fibers passing between uncinate processes and inserting on anterior surface of Eb4 just ventral to tip of uncinate process. @ OD3’ absent. OP dorsally on Eb4 posteriorly beginning medially near end of bony surface and extending laterally to just medial to uncinate process; ventrally on Cb5 be- ginning laterally as broad CT raphe with mid-poste- rior surface of Ad5 and continuing medially about same extent as muscle occupies on Eb4. GFM1-3 each begin as fragile sparse muscle fan on anterodistal surfaces of respective Eb and Cb. GFM1 continues dorsoanteromedially a short dis- tance on Ebl. GFM2 continues dorsomedially be- coming dorsal and more compact and inserting on lateral surface of tiny, bony diagonal strut anterior to LE2 insertion (TEb2 inserts on medial surface of strut). GFM3 follows path similar to GFM2, ending near OD3’. @ OD3’ absent. Ad4 dorsally begins on Eb4 dorsoposteriorly lat- eral to LP insertion, extends medially to uncinate 189 process, with few muscle strands continuous with LP posteroventrally and others joining raphe with LE4 posteromedially; ventrally, muscle attaches on Cb4 dorsally beginning just medial to medial end and ex- tends medially about a quarter length of Cb4. @ Mus- cle completely separated from LE4 and LP. Ad5 on Cb4 and dorsoposterodistally and ventrally on Cb5 beginning dorsodistally and extending me- dially slightly less than distance occupied by Ad4; posterior surface joins broad CT raphe with OP ven- trally. SOD present. RDs adjacent. Additional remarks. SCL attached mid-dorsally to posteroventrally extending cartilaginous tip of Bb3. TV4 free from Cb5s. Pb4 and UP4 present. AC4 pre- sent on both sides, but present on only one side and in two pieces in cleared and stained specimen. Nei- ther AC4 nor any indication of a posterior cartilagi- nous extension of distal end of Cb5 present in @. Labroidei CICHLIDAE Caquetaia kraussii (Steindachner), USNM 258004, 62.8 mm. Plate 159 Additional material (lengths not recorded). Astrono- tus ocellatus (Agassiz), USNM 329642; Cichla ocellaris (Bloch and Schneider), USNM 226019; Cichlasoma bimaculatum (Linnaeus), USNM 181457; Copadichromis jacksoni (Iles), USNM 261845; Crenicichla alta Eigenmann, USNM 226024; Cyrtocara moorii Boulenger, USNM 280311; Santanoperca leucosticta (Miiller and Troschel), USNM 289647; Paratilapia polleni, USNM 344609; Ptychochromoides katria Reinthal and Stiassny, USNM 344607; Ptychochromis oli- gacanthus (Bleeker), USNM 344605. Remarks. Muscles of all the taxa are generally sim- ilar. The nature and relationships of LE4, LP, and OP, however, are particularly complex (see discussion in Additional remarks), and variable among the taxa. The description of these three muscles is based al- most entirely on Caquetaia. Only a few variations pertaining to the other muscles are mentioned in the description. Description. LE] on and lateral to Eb1 uncinate process. LE2 on Eb2 mid-dorsoposteriorly. LE3 on Eb3 uncinate process ventral to tip and at OD3-4 attachment to Eb3. LE4 essentially free, inserting on Eb4 only at point posterolaterally where LE4 joins with LP insertion medially; continuous ventrally with central portion of 190, OP (separation shown by a raphe), which inserts on Cb5. LP on Ebé4 laterally, insertion fusing with LE4 me- dially and partly joining raphe with putative lateral part of OP dorsally. LI1 on Pb2 dorsoanteriorly and Pb3 ventroanter- iorly (sandwiched between Pb2 and Pb3). LI2 dorsolaterally on Pb3 lateral to OD3—4 origin and medial to medial end of Eb3. TD comprises TPb2, TPb2a, TEb2 and TPb3-Eb4. TPb2 is deeply notched mid-anteriorly and almost completely divided mid-longitudinally; the division is an expansion of the mid-longitudinal raphe fre- quently present in the acanthomorph TD. The divi- sion also completely divides TEb2, but not TPb3- Eb4. TPb2 attaches to Pb2 dorsoanterolateral process (process not visible externally) and to tiny IAC; pos- teriorly, TPb2 joins CT sheet that covers Pb3 artic- ulating surfaces. TPb2a has mid-anterior raphe and attaches to anterior Pb2 surfaces ventral to TPb2; CT extends posteriorly from raphe and passes between Pb2s and Pb3s; separation of TPb2 from TPb2a is indistinct (but separation may be distinct in other cichlids). TEb2 a pair of muscles, joined medially by CT sheet covering Pb3 articulating facets, and at- taching laterally on Eb2 to position anterior to LE2 insertion. TPb2 and TEb2 well separated, not contin- uous posteriorly with TPb3-Eb4. TPb3-Eb4 attaches on Pb3 posterolaterally slightly ventral to attachment to posteromedial end of Eb4 and continues slightly anteromedially on Pb3. Posteriormost component in some other cichlids attaches only to Pb3 (see also Table 9; and discussion in Additional remarks). CPb comprises a pair of well-developed sub-epi- thelial muscles (Anker, 1978:261), each originating posterolaterally on UP4, muscle divides anteriorly with branch attaching to Pb2 laterally and branch passing medially and attaching to Pb3 anteromedi- ally. OD3-—4 origin on Pb3 articulating facet anterolat- erally, insertion on medial edge of Eb3 uncinate pro- cess and on medial edge of, and enveloping Eb4 un- cinate process, joins raphe posteroventrally with a OP medial portion. OP comprises four parts (apparently not all men- tioned or illustrated in the literature; see discussion in Additional remarks): OP1, OP2, OP3, and OP4. The first three parts attach ventrally near the distal end of Cb5 and the fourth part attaches ventrally, broadly on Cb5 dorsally medial to the distal end. OP1 (appears to be the same as Aerts’s (1982:233) pars lateralis) is dorsally on Eb4 posteriorly just medial to the membranous dorsal attachment of Ad5; dorso- posteromedially it joins a small raphe with LP ven- troposteriorly and is continuous ventromedially with OP2 and ventrolaterally with Ad5. OP2 (the same as Aerts’s (1982:233) pars centralis) is continuous with BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON LE4 ventrally, their junction indicated by a raphe. OP3 is on Eb4 dorsally, extending broadly medially from the uncinate process and joining a raphe with OD3-—4 (dorsal attachment appears to be similar to Aerts’s (1982:233) pars medialis). OP4 is on most of the posteroventral edge of Eb4 (ventral attachment appears to be similar to that of Aerts’s (1982:233) pars medialis). Ad1 small, on Eb! anterolaterally and Cb1 anter- odistally. Ad2 and Ad3 well developed, on anterolateral half of respective Eb and anterodistally on respective Cb. Ad4 (not visible in illustration) dorsally, broadly on Eb4 ventral surface, fusing posteriorly with OP4 dorsoanteriorly, ventrally, narrowly on Cb4 dorsal surface medial to Eb4-Cb4 joint. Ad5 dorsally, tendinously primarily on posterolat- eral end of Eb4, secondarily on posterodistal end of Cb4 dorsally; ventrally on posterodistal end of CbS. SOD absent, but strap-like branch of SO arises lat- erally on each side and attaches to Pb3 mid-poster- oventrally, giving impression of an interrupted SOD. RDs adjacent. Additional remarks. SCL free from Bb3 (SCL ab- sent in Cichla and Crenicichla). TV4 with continuous ventral portion across Cb5 and split dorsal portion attaching laterally to each Cb5 anteriorly (ventral portion absent in Cichla; also reported absent in var- ious African cichlids by Stiassny, 1992:265—267). Pb4 absent, UP4 present. IAC present, small. Anker (1978) termed our TPb2 as M. craniophar- yngobranchialis 2 and our TPb2a as M. transversus pharyngobranchialis 2. Both muscles appear to be derived from TPb2, and the second, which is not al- ways separate from the first in cichlids, should not carry the main part of the name. TPb2a in acantho- morphs is a specialized part of TPb2 that is restricted to “labroid” (sensus Stiassny and Jensen, 1987), pseudochromid, atherinomorph, and pholidichthyid fishes, and is questionably synapomorphic for these fishes as group. Anker (1978:256—257) described the musculature of the cichlid Haplochromis elegans Trewavas. He termed the posteriormost TD element the transversus epibranchialis 4, and described it as attaching to Pb3, Pb4 [actually UP4, Pb4 is absent], and Eb4. Stiassny (1981:96), who examined the dorsal gill-arch mus- culature in several cichlids (including some genera we also examined), followed Anker in recognizing a TEb4. She appears to have noted no variation in the attachment of this muscle, which she described as originating “from the caudal eminence formed at the junction of Pb3 and UP4.” In cichlids there is a tight association of UP4 dorsally with Pb3 at the joint of Pb3 with the medial end of Eb4, and the three ele- ments are bound by CT. The posteriormost TD mus- cle is attached to this complex dorsally, and as such NUMBER 11 is removed from UP4. The posteriormost TD muscle never attaches to the bony portion of UP4, but may attach to Eb4 and Pb3 or only to Pb3 (Table 9). Remarks. Aerts (1982) studied the development of OP and LE4 in Haplochromis elegans Trewavas. He reported that during ontogeny OP forms in three parts (lateral, central, and medial) and that LE4, which is initially separate, combines with the dorsal fibers of the OP medial part. Claeys and Aerts (1984) dis- cussed further the ontogeny of LP, LE4, and OP in H. elegans (which they placed in Astatotilapia). They found that during ontogeny a few fibers of LP attach dorsolaterally to Eb4, but most become attached to the aponeurotic system of Eb4 dorsolaterally directly opposite the insertion area of the lateral bundle of OP and later join end to end with them to form a compound muscle. Thus, for H. elegans, and many or most other cichlids, the “‘sling’’ comprises LE4, OP2, and LP, although we find that the contribution of LP to the sling is usually considerably more lim- ited than that of LE4. We further note that OP1 may fuse almost completely with Ad5 (e.g., Ptychochrom- is). POMACENTRIDAE Dischistodus fasciatus (Cuvier), USNM 328190, three specimens, 57.1—68.9 mm; USNM 179622, 79.7 mm. Plate 160 Additional material. @ = Abudefduf sexfasciatus (La- cepede), USNM 221863, 68.4 mm; @ = Chromis amboinensis (Bleeker), USNM 338153, not mea- sured. Plate 161 Remarks. We also recorded limited data on the dorsal gill-arch musculature of several other species (data variably combined or assigned to individual species as warranted): Acanthochromis polyacanthus (Bleeker), USNM 275349, 309487; Amblyglyphido- don aureus (Cuvier), USNM 338213; Amphiprion al- lardi Klausewitz, MCZ 4489 (Stiassny and Jensen’s (1987) specimen), USNM 275417; A. melanopus Bleeker, USNM 338150; Chromis atrilobata Gill, USNM 321208, C. cyanea (Poey) (USNM 318909), C. iomelas Jordan and Seale (USNM 338155: C. ter- natensis (Bleeker), USNM 338158; C. viridis (Cu- vier), USNM 338160; Chrysiptera taupou (Jordan and Seale), USNM 338076; Dascyllus reticulatus (Richardson), USNM 338175; Lepidozygus tapeino- soma (Bleeker), USNM 97140, 265252, 275893, 348198; Mecaenichthys immaculatus (Ogilby), USNM 215191; Microspathodon chrysurus (Cuvier), USNM 194045; Plectroglyphidodon dickii (Liénard), USNM 338219; Pomacentrus vaiuli Jordan and Se- 191 ale, USNM 338226; Stegastes fasciolatus (Ogilby), USNM 338222. Description. LE1 broadly on Eb! uncinate process ventroanter- iorly. LE2 finely tendinously on bony tip of dorsally ex- panded posterior edge of Eb2. LE3 finely tendinously on tip of Eb3 uncinate pro- cess medially. LE4 on dorsal edge of Eb4 lateral to uncinate pro- cess, joining raphe with OP dorsal to level of Eb4; LE4 joined ventrolaterally by LP, which does not join OP. @ Lacks raphe with OP. ® Tendinously on dor- solateral edge of Eb4, does not join raphe ventrally with OP. (See also Discussion following Additional remarks.) LP finely tendinously on Eb4, joining LE4 pos- terolaterally. (See also discussion following Addi- tional remarks.) LI1 by slender tendon mainly on Pb2 dorsoanter- iorly just below broad cartilaginous tip of process articulating with IAC, tendon extending secondarily onto CT binding adjacent anteriormost end of Pb3 with Pb2. LI2 on Pb3 dorsolaterally ventral to medial end of Eb3. TD comprises TPb2a, TPb2 (see Discussion of TPb2 following Additional remarks), TEb2, and TPb3. TPb2a attached to anterior surfaces of Pb2s; attached mid-posteriorly to CT that passes between Pb3s. TPb2 and TEb2 each comprising a pair of mus- cles (TPb2 dorsal to TEb2) joined medially by broad area of CT. TPb2 attaching anteriorly to dorsoanterior tip of Pb2. TEb2 of each side attaches to ventrolateral edge of CT connecting TPb2s, muscle extends lat- erally and attaches on Eb2 dorsally anterior to LE2 insertion, joining medial end of Ad2. TPb3 on dor- soposterolateral end of Pb3 opposite medial end of Eb4. @ TEb2 may just fail to reach medial edge of LE2 or just reaches point anterior to LE2. ® Similar to Dischistodus, but has TPb3-Eb3 instead of TPb3; attachment to Eb3 is to posteromedialmost tip. Remarks. See discussion in remarks following de- scription of TD in Parahollardia (Triacanthodidae) for discussion of similarity of knob-like process at- taching Pb3s to ventral surface of skull in that taxon and pomacentrids, and difference from that of labrids, embiotocids, and cichlids. CPb fibers relative fine, closely bound and ob- scured by covering CT of pharyngeal area; originat- ing from longitudinal SO fibers passing posteriorly along medial surfaces of Pb2 and Pb3, branching posteriorly and extending along posterior surfaces of Pb3 and UP4 and attaching to posterolateral corner of UP4; another fine strap of muscle attaching to pos- terior surface of Pb2 and extending laterally around 192 Pb3 and UP4, and attaching to posterolateral corner of Pb4. @ CPb absent. Absent also in Lepidozygus tapeinosoma. Remarks. This muscle is very difficult to find, and may not be present in most genera. In large speci- mens of large species, such as Plectroglyphidodon dickii, it is relatively easy to find. OD3—4, OD3’ origin on lateral surface of Pb3 dor- sal articulating facet (facet partially ventral to TPb2), insertion on medial edges of Eb3 and Eb4 uncinate processes; OD3’ questionably identified here as thin layer of fibers originating on Pb3 laterally immedi- ately ventral to OD3—4 origin and inserting dorsally on medial end of Eb3. On only one side of one of four specimens, insertion extended dorsolaterally on Eb3 to point about halfway between medial end of Ad3 and base of uncinate process. Remarks. We also found OD3’ in Acanthochromis polyacanthus, Amphiprion allardi, and Lepidozygus tapeinosoma. It probably occurs in its shortened, spe- cialized state in most or all pomacentrids. The muscle is difficult to find as it is thin, completely overlain by OD3-—4, and easily damaged when cutting through OD3-—4 to expose the insertions of LI2 and TPb3. OP dorsally joining raphe with LE4, dorsoanter- iorly attached to Eb4 posteriorly beginning medial to uncinate process and extending laterally close to lat- eral end of Eb4, divided into lateral and medial sec- tions by long slender tendon, which extends from near dorsal end of muscle to dorsodistal end of Cb5; tendon joined laterally by ventromedial end of Ad5. @ Not continuous dorsally with LE4, comprising broad lateral and slender medial sections, which fuse ventrally and attach to Cb5 dorsoanterodistally, there also joining Ad5. Ad1-—3 each begin on dorsoanterior surface of re- spective Eb and extend broadly onto anterior surface of joint with respective Cb1, then continue as slender GFM along most of remaining Cb anterior edge. Ad4 dorsally broadly on ventral surface of Eb4 lateral to uncinate process, ventrally on Cb4 dor- soanteriorly medial to Eb4-Cb4 joint. Ad5 dorsally broadly on posterolateral surface of Cb4, ventrally broadly on posterolateral surface of Cb5 extending dorsally and joining tendinous exten- sion at ventrolateral end of OP. SOD absent. RDs well separated. Additional remarks. SCL free from Bb3. TV4 in two sections, ventral section relatively thin, contin- uous across fused Cb5s ventrally; dorsal section thick, interrupted, attaching to lateral surfaces of Cb5 keel. Pb4 absent, UP4 present. Eb4 levator process absent (present, at least, in Lepidozygus and Amphi- prion). See Table 8 for distribution of ACs in po- macentrids. Discussion. One of Stiassny and Jensen’s (1987: BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON 282) main synapomorphies of the Labroidei, is that the dorsal articulating Pb3 facets are completely na- ked and directly (“bone to bone”’) contact a ventrally rounded neurocranial apophysis on the base of the skull. They contrasted this condition with that of sev- eral families (Gerreidae, etc.) in which the central portion of TPb2 replaced by CT covering Pb3s dor- sally. They (1987:276) reported that pomacentrids lack cranio-pharyngobranchialis 2 (= our TPb2; their transversus pharyngobranchialis 2 = our TPb2a), presumably based on the alcohol preserved material they listed (* denotes taxa we also examined): Abu- defduf trochelii (Gill), A. saxatilis (Linnaeus); Am- phiprion allardi*; Chromis atrilobata*, C. cyanea*; Microspathodon chrysurus*; Neopomacentrus sin- densis; Pomacentrus otophorus (= Stegastes otopho- rus (Poey)), Pomacentrus* moluccensis Bleeker; Ste- gastes* acapulcoensis (Fowler). All the pomacentrids we examined have TPb2 (and TPb2a) similar to that illustrated in Plate 160), with the central portion com- prising a broad band of CT covering the Pb3 facets. We thought, perhaps, that Stiassny and Jensen might have considered the reduced pomacentrid TPb2 as part of TEb2, but other than the levator sling, they did not illustrate the pomacentrid muscles, citing il- lustrations in Stiassny’s (1980) Ph.D. dissertation, which is not readily available, and Kauffman and Liem (1982:figs. 2A, 2B). Stiassny kindly sent us a copy of her illustration of the muscles in Pomacen- trus, and it does not include TPb2, nor do Kaufman and Liem’s illustrations. Among the pomacentrids they examined, Stiassny and Jensen (1987:284 and fig. 8c) reported that LE4 is not continuous with OP in Chromis (based on C. atrilobata and C. cyanea), Neopomacentrus (based on N. sindensis (Day)) and Amphiprion (based on A. allardi). LE4 is continuous with OP in our specimen of C. cyanea, but not in our specimens of C. atrilo- bata and C. amboinensis. Considering the configu- ration of LE4 and LP in their illustration (the two muscles insert together very narrowly), Stiassny and Jensen appear to have based their description only on C. atrilobata. Stiassny and Jensen (1987:290) reported that LP “never contributes to the muscle sling”’ in pomacen- trids. We examined Amphiprion melanopus and A. allardi (Stiassny and Jensen’s specimen of A. allardi was in too poor condition to determine the state of LE4 and OP). In contrast to Stiassny and Jensen, we found that not only LE4, but also LP is broadly con- tinuous with OP in both species of Amphiprion, and that LP also contributes to the sling in Plectrogly- phidodon dickii and Stegastes fasciolatus. Among the four specimens of Lepidozygus tapei- nosoma we examined, a sling may be absent, LE4 and LP may be continuous with OP, or only one or the other may be continuous with OP, but only a few NUMBER 11 muscle strands of LP are involved when LP is con- tinuous with OP. The pomacentrids have not been analyzed cladis- tically, and we are unable to hypothesize the plesiom- orphic state for the muscle sling, which is important for corroborating inclusion of the Pomacentridae in the Labroidei of Kaufman and Liem (1982) and Stiassny and Jensen (1987). In any event, the po- macentrid sling, when present, differs considerably from the sling in the other labroids in being relatively simple. K.L. Tang is preparing a phylogeny of the pomacentrid genera, which was not available at the time we were preparing our study. We have arbitrari- ly selected Dischistodus, in which the sling consists of LE4 + OP, the most common state, and Amphi- prion melanopus (not illustrated), in which the sling comprises LE4 (not released from Eb4), LP, and OP, for inclusion in the cladistic analysis. The absence of a sling in Chromis and Lepidozygous, appears to be specialized in those two genera, because the other, more common character states are either shown by some species of Chromis, or other specimens of the monotypic Lepidozygous. EMBIOTOCIDAE Amphistichus argenteus Agassiz, USNM 132403, 3 specimens, ca. 100+ mm, of which one is cleared and stained, and one damaged. Plates 162.1, 162.2 Additional material. @ = Embiotoca lateralis Agas- siz, USNM 340889, 86.8 mm. Plate 162.1 Additional material. Cymatogaster aggregata Gib- bons, USNM 340888; Hysterocarpus traskii Gib- bons, USNM 61189; Phanerodon atripes (Jordan and Gilbert), USNM 337459; P. furcatus Girard, USNM 126843 & 337461; Rhacochilus vacca (Gi- rard), USNM 340884, Zalembius rosaceus (Jordan and Gilbert), USNM 337463. Most of the infor- mation on these taxa is found only on Table 9. Description. LE] broad based, on dorsalmost edge and surface of Eb1l near medial end of element (see Additional remarks for comment about presence of Eb! uncinate process). LE2 on mid-dorsoposterior bony edge of Eb2. LE3 dorsoanteriorly on Eb3 uncinate process, fus- ing anteroventrally with OD3-—4. LE4 complex, massive, essentially free from Eb4, fused posterolaterally with LP dorsal to Eb4, joining raphe with presumable middle section of OP. LP laterally on Eb4 posterolaterally, fused with LE4 ventroanteriorly, joining raphe ventrally with presumable middle section of OP. @ LP more distinct, 193 on levator process dorsally, continuous posteroven- trally by only a few muscle strands with OP. Remarks. There is a strong, bony process that bears the cartilage tip of the levator process in ®. The same process is present in Amphisticus, which lacks the cartilage (hence, uncinate process absent). LI1 on dorsomedial surface of Pb2 and dorsoan- terior surface of Pb3. @ On Pb2 dorsally. LI2 inserts by long tendon on Pb3 dorsopostero- laterally ventroanterior to medial end of Eb3. TD comprises TEb2 and TPb3-Eb4. TEb2 a pair of muscles joined medially by CT to ventrolateral surface of CT pad covering otherwise naked dorsal Pb3 articulating facets, extending laterally on dorsal surface of Eb2 anterior to LE2 insertion, joining ra- phe with medial end of Ad2. TPb3-Eb4 on Pb3 dor- soposterolaterally, a little medial to LI2 insertion, continuing posteriorly and attaching to medialmost edge of Eb4, thence continuing along posteromedial surface of Eb4 and joining raphe with dorsomedial edge of a branch of OP (possibly the same as either OP3 or OP4 of cichlids). TPb3-Eb4 posteromedially, continuous mid-posteriorly with SOD. @ TEb2 does not join raphe with Ad2. CPb absent, but, unusually, SO, which is covered ventrally by pharyngeal roof CT, spreads anterolat- erally (Plate 162.2) and lines the ventral surfaces of the epibranchials (see M. Pb3-UP4). Remarks. Muscle questionably distinct from loose mix of SO + CT fibers that spreads over pharyngeal roof, and is possibly homologous with one or all of the muscles identified as CPb in other putative la- broids and pholidichthyids, callionymoids, centro- geniids, lethrinids, pseudochromids, nemipterids, and sparids. Muscle fibers do not attach to the (reduced) Pb2s or pass around them anteriorly, as they do in the other taxa. M. Pb3-UP4 ventral to RD, on Pb3 ventroposter- iorly and UP4 ventroanteriorly (also present in @ and Cymatogaster, but not other genera examined). OD3-—4 originating on lateral surface of Pb3 dorsal articulating facet, inserting massively on dorsomedi- almost surface of Eb3 and anterior surface of unci- nate process (there joining LE3 insertion anteriorly) and medial edge of bony Eb4 uncinate process. OP complex, presumably comprising two or three sections (only two indicated here): posterior section joining raphes dorsally with LE4 and LP, ventrally on Cb5 distally, joining tendinous raphe laterally with Ad5; anterior section broad, on Eb4 posteriorly, fusing posteriorly with anterior section ventral to Eb4. Ad1-3, each broad, on anterolateral surfaces, cov- ering joint, of respective Eb and Cb. Ad4 thin muscle sheet dorsally on ventrolateral edge of Eb4 and ventrally narrowly on Cb4 dorso- 194 laterally, obscured from view by OP portion attach- ing broadly to Eb4 posteriorly. Ad5 dorsally on tiny AC4 (not visible in illustra- tion), narrowly on Cb4 distally, and broadly on Eb4 posterodistally; ventrally on Cb5 distal end, joining tendinous raphe with OP posterior section. @ Slight, conformational separation of Ad5 from OP along raphe. SOD present. @ Absent. See also Additional re- marks. RDs slightly separate. Additional remarks. SCL attached mid-dorsally by long tendon to posteroventral end of Bb3. TV4 com- plex, but dorsally attached to Cb5, ventrally contin- uous but interrupted by raphes, possibly comprising separate dorsal and ventral sets of muscles. Pb4 ab- sent, UP4 present. Pb1 cartilaginous. Pb2 edentate. The uncinate process of Eb1 appears to have moved medially and essentially become confluent with the medial end of the anterior arm of Eb! in all embioto- cids, hence, a synapomorphy for the family. Evidence for this assumption is that Pb] articulates with the rounded ventral portion of the cartilaginous medial end of Eb! and a ligament connects the dorsal, blade- like cartilaginous medial edge of Eb1 to Pb2. Tiny ACs present as follows (Table 8): Amphisti- cus argenteus, AC4 present on one side of each of two specimens (of which one cleared and stained; third specimen damaged), Ad5 attached to it in one specimen; Phanerodon atripes, AC4, Ad5 not at- tached to it; P. furcatus, AC1 and AC2 present or absent; Zalembius rosaceus, AC4, Ad5 attached to it. SOD in Amphistichus is distinct and readily rec- ognizable. In both species of Phanerodon, however, SOD arises from SO dorsolaterally as a short, fine strip of muscle that fuses with the ventrolateral surface of TEb4, and is easily overlooked. Stiassny and Jensen (1987), who examined Phanerodon, but not Amphis- tichus, reported that all labroids, in which they in- cluded embiotocids, lack SOD. Presence of SOD is generally plesiomorphic for percomorphs and possibly indicative that Amphistichus is one of the most ple- siomorphic genera in the family. The state of SOD in Phanerodon appears to be more specialized. Liem (1986) illustrated and described the gill-arch musculature of Embiotoca lateralis. Our observations are in general agreement with his; however, Liem de- scribed LI1 as inserting only on Pb2, and LP as com- pletely separate from LE4. We examined Liem’s ma- terial (MCZ 58890) and find that LI] and LP accord with the description of our specimen. Labroidea In order to facilitate discussion of the unquestion- ably monophyletic group comprising the Labridae, BULLETIN OF THE BIOLOGICAL SOCIETY OF WASHINGTON Table 10.—Distribution of certain characters in genera of labro- idean fishes. Dash (—) = absent; P = present; C = continuous; D = divided; F = fused; S = separate. Genera Labridae Achoerodus al Hla ell 9) ol Bodianus = VPS RAW Gaah Cheilinus = IP IyN € iF Cheilio =e JAC OE Choerodon Ries Awe CAE Clepticus = (RiP Comal Coris eo 1G SS Decodon Ee AG Cammy Halichoeres a =) (PAT CaaS: Hologymnosus Shh SPs Cees) Labroides Pe ts=