PROCEEDINGS OF THE California Academy of Sciences Volume 43 SAN FRANCISCO PUBLISHED BY THE ACADEMY 1982-1984 PUBLICATIONS COMMITTEE Sheridan Warrick, Editor Frank Almeda Daphne G. Fautin Tomio Iwamoto Frank H. Talbot (US ISSN 0068-547X) The California Academy of Sciences Golden Gate Park San Francisco, California 94 1 1 8 PRINTED IN THE UNITED STATES OF AMERICA BY ALLEN PRESS, INC., LAWRENCE, KANSAS CONTENTS OF VOLUME 43 Pages No. 1 . ROBERTS, TYSON R. The Southeast Asian freshwater pufferfish genus Chone- rhinos (Tetraodontidae), with descriptions of new species. Published June 15, 1982 1-16 No. 2. McCosKER, JOHN E., AND JOHN E. RANDALL. Synonymies of Indian Ocean eels, with the description of Gymnothorax enigmaticus, a moray previously known as G. Ruppeli. Published June 15, 1982 17-24 No. 3. PULAWSKI, WOJCIECH J. New species of North American Tachysphex wasps (Hymenoptera, Sphecidae). Published June 15, 1982 27-42 No. 4. ROBERTS, H. RADCLYFFE, AND CARLOS S. CARBONELL. A revision of the grass- hopper genera Chromacris and Xestotrachelus(Orthoplera, Romaleidae, Roma- leinae). Published November 4, 1982 43-58 No. 5. McCosKER, JOHN E. A new genus and two new species of remarkable Pacific worm eels (Ophichthidae, subfamily Myrophinae). Published November 4, 1982 59-66 No. 6. RAINBOTH, WALTER J. Psilorhynchus gracilis, a new cyprinoid fish from the Gangetic lowlands. Published July 6, 1983 67-76 No. 7. LE BOEUF, BURNEY J., DAVID AURICLES, RICHARD CONDIT, CLAUDIO Fox, ROBERT GISINER, RIGOBERTO ROMERO, AND FRANCISCO SINSEL. Size and distribution of the California sea lion population in Mexico. Published July 6, 1983 77-85 No. 8. TAYLOR, LEIGHTON R., L. J. V. COMPAGNO, AND PAUL J. STRUHSAKER. Megamouth— a new species, genus, and family of lamnoid shark (Megachasma pelagios, family Megachasmidae) from the Hawaiian Islands. Published July 6, 1983 87-110 No. 9. CRUMLY, CHARLES R. The cranial morphometry of Galapagos tortoises. Pub- lished January 1 7, 1 984 111-121 No. 10. PULAWSKI, WOJCIECH J. The status of Trypoxylon figulus (Linnaeus, 1758), medium De Beaumont, 1945, and minus De Beaumont, 1945 (Hymenoptera: Sphecidae). Published January 17, 1984 123-140 No. 11. ROBERTS, TYSON R., AND MAURICE KOTTELAT. Description and osteology of Thryssocypris, a new genus of anchovylike cyprinid fishes, based on two new species from Southeast Asia. Published January 17, 1984 141-158 No. 12. KAVANAUGH, DAVID H. Studies on Nebriini (Coleoptera: Carabidae), V. New Nearctic Nebria taxa and changes in nomenclature. Published July 12, 1984... 159-177 No. 13. ROBERTS, TYSON R. Skeletal anatomy and classification of the neotenic Asian salmoniform superfamily Salangoidea (icefishes or needlefishes). Published July 12, 1984 179-220 No. 14. TRICAS, TIMOTHY C., AND JOHN E. McCosKER. Predatory behavior of the white shark (Carcharodon carcharias), with notes on its biology. Published July 12, 1984 221-238 No. 1 5. NEWBERRY, ANDREW TODD. Dendrodoa (Styelopsis) abbotti, sp. nov. (Styelidae, Ascidiacea) from the Pacific Coast of the United States, and its impact on some gonadal criteria of its genus and subgenus. Published September 19, 1984 239-248 [Hi] Pages No. 1 6. MCMILLAN, CHARMION B., AND ROBERT L. WISNER. Three new species of seven- gilled hagfishes (Myxinidae, Eptatretus) from the Pacific Ocean. Published De- cember 11, 1984 249-267 No. 17. ALMEDA, FRANK. New and noteworthy additions to the Melastomataceae of Panama. Published December 11, 1984 269-282 No. 18. COMPAGNO, LEONARD J. V., AND TYSON R. ROBERTS. Marine and freshwater stingrays (Dasyatidae) of West Africa, with description of a new species. Pub- lished December 1 1 , 1 984 283-300 No. 19. FITCH, JOHN E., AND STEPHEN J. CROOKE. Revision of Eastern Pacific catalufas (Pisces: Priacanthidae) with description of a new genus and discussion of the fossil record. Published December 11, 1984 30 1-3 1 5 No. 20. ROBERTS, TYSON R. Amazonsprattus scintilla, new genus and species from the Rio Negro, Brazil, the smallest known clupeomorph fish. Published December 11,1984 3 1 7-32 1 Index to Volume 43 323-329 Instructions to Authors .... 331-332 [iv] PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES Vol. 43, No. 1, pp. 1-16, 10 figs. June 15, 1982 By Tyson R. Roberts California Academy of Sciences, Golden Gate Park, San Francisco, California 94118 ABSTRACT: The tetraodontid pufferfish genus Chonerhinos, restricted to fresh water in Southeast Asia, comprises five species, four of which are described as new. The species differ in adult size, coloration, orientation of squamation, depth of caudal peduncle, size of nasal organ, food habits, and geographical distribution. The most widely distributed, C. nefastus n.sp., occurs in southern, western, and northern Borneo, the Malay Peninsula, Thailand, Vietnam, Kampuchea, and Laos; it feeds mainly on fish fin rays and scales, and has a slender caudal peduncle and the smallest nasal organ. Chonerhinos modestus (Bleeker, 1850), in western Borneo and Sumatra, with perhaps the most varied diet, is the largest species and has the deepest caudal peduncle. The distinctively colored C. amabiUs n.sp., with the largest nasal organ, occurs in western Borneo and Sumatra and feeds almost exclusively on large aquatic insects. The two new species C. silus, with a moderately deep caudal peduncle, and C. remotus, with a slender caudal peduncle, have varied diets including insects, and are known only from northern and northeastern Borneo. INTRODUCTION The freshwater pufferfish genus Chonerhinos currently includes a single species, C. modestus (Bleeker, 1850), reported from localities throughout much of Southeast Asia. The nomi- nal species C. africanus Boulenger, 1909, known only from the holotype supposedly col- lected in the interior of the Congo basin, has been identified as a junior synonym of C. mo- destus with incorrect locality data (Roberts 1981; herein). The species formerly known as C. naritus (Richardson, 1848), from marine, brack- ish, and perhaps freshwater habitats along the coasts of the South China Sea and eastern Indian Ocean, has been placed in a monotypic genus, Xenopterus (Fraser-Bruner 1943; Tyler 1980; herein). I undertook this revision because three species of Chonerhinos were obtained during my ichthyological survey of the Kapuas basin in western Borneo (Kalimantan Barat, Indonesia) in 1976. MATERIAL EXAMINED AND METHODS More than 250 specimens of Chonerhinos from throughout the range of the genus were examined during this study. These are deposited in the British Museum (Natural History), Lon- don, BMNH; California Academy of Sciences, San Francisco (CAS), including material for- merly deposited at Stanford University, Stan- ford (SU); Field Museum of Natural History, Chicago (FMNH); Museum Geneve, Geneva (MG); Museum National d'Histoire Naturelle, Paris (MNHN); Museum Zoologicum Bogo- rense, Bogor, Indonesia (MZB); Musee Royal de 1'Afrique Centrale, Tervuren (MRAC); Nat- ural History Museum, Basel (NHMB); Rijks- [i] PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 1 museum van Natuurlijke Historic, Leiden (RMNH); University of Michigan Museum of Zoology, Ann Arbor (UMMZ); U.S. National Museum of Natural History, Smithsonian Insti- tution, Washington, D.C. (USNM); and Zoolog- ical Museum, Universiteit van Amsterdam, Am- sterdam (ZMA). Length of specimens is given as standard length unless total length is expressly indicated, and all proportional measurements are given as times in standard length (SL). Vertebral counts were made from radiographs prepared in the Department of Ichthyology, CAS. Sections of skin anterior and ventral to the pectoral fin were removed with a scalpel and stained in alizarin to facilitate illustration of the scales. Orientation of the scales is also obvious in radiographs and can be observed in whole specimens without special preparation. Chonerhinos Bleeker Chonerhinos BLEEKER, 1 854: 259-260 (type-species Tetraodon modest us Bleeker, 1850, by subsequent designation of Fra- ser-Bruner 1943:16). Chonerhinus BLEEKER, 1865:213 (unjustified spelling change). DESIGNATION OF TYPE-SPECIES. — Fraser- Bruner (1943) is apparently the first author to have properly designated a type-species for Chonerhinos. The original description of the ge- nus is as follows: "Chonerhinos Blkr [is gekenmerkt] door trechtervormige verdieping ter plaatse der neusopeningen met verhevene randen, lange rug- en aarsvinnen, zigtbare zijlijn en onegekielden rug . . . van Chonerhinos 2 t. w. Chonerhinos modestus Blkr = Tetraodon modestus Blkr olim (van Borneo, Sumatra), Chonerhinos naritus Blkr = Tetraodon naritus Richds (van Borneo)/' Thus, Bleeker included two species in his original account of Chone- rhinos and did not indicate a type-species. Hoi- lard (1857) defined Xenopterus (type-species X. belle ngeri = X. naritus, by monotypy) in such a way that it excludes Chonerhinos, which, however, he did not mention by name. Gill (1892) discussed the nomenclatural history of Chonerhinos (and Xenopterus} at length but oddly did not mention the lack of a type-species. Jordan (1919:256) incorrectly stated that Tetrao- don modestus Bleeker is the "orthotype" of Chonerhinos, meaning that Bleeker (1854) indi- cated or distinctly implied that this species is the type-species. DIAGNOSIS. — Chonerhinos and its close rela- tive Xenopterus differ from all other tetraodon- tids in having three lateral line canals on side of body instead of one, two, or none; dorsal fin with 22 or more rays; anal fin with 18 or more rays; at least 24 vertebrae; and prefrontal bones absent (Tyler, 1980). Chonerhinos differs from Xenopterus in its smaller adult size, less exten- sive squamation, less exposed olfactory lamel- lae, and fewer fin rays and vertebrae. The largest Chonerhinos I have examined is 106 mm; Xen- opterus attains at least twice this size. In Cho- nerhinos the scales are relatively small and re- stricted to the head and body ventral to the level of the pectoral fin; in Xenopterus the scales are relatively large and extend dorsally to the pec- toral fin. In Chonerhinos the olfactory lamellae are largely covered by nasal flaps in broad con- tact; in Xenopterus the nasal flaps are greatly reduced and the olfactory lamellae are conse- quently almost entirely exposed. Chonerhinos has 22-28 dorsal-fin rays, 18-22 anal-fin rays, 13-17 pectoral-fin rays, and 24-28 vertebrae; the same counts in Xenopterus are 32-38, 28-29, 18-19, and 29-30. REMARKS.— Tyler (1980) stated that Chone- rhinos and Xenopterus are highly specialized tetraodontids which have secondarily increased the number of dorsal- and anal-fin rays and ver- tebrae, elaborated the lateral line system, in- creased the number and size of the olfactory la- mellae, and increased the size of at least some of the scales; and that the greater numbers of vertebrae and fin rays in Xenopterus as well as the structure of the skull indicates that it is the more specialized of the two. In Chonerhinos, according to Tyler, apart from the absence of the prefrontal bones, the skull is not markedly different from that in many species of the tet- raodontid genera Monotreta, Chelonodon, and Tetraodon, whereas in Xenopterus the frontals are much more laterally expanded and thickened than in Chonerhinos, forming a large plate over most of the dorsal surface of the skull, and the supraoccipital crest is wider and heavier; in large specimens the two frontals may become indistinguishably fused to each other in the mid- dle of their lengths (Tyler 1980:340, fig. 274). I have examined two X. naritus from Sarawak, BMNH 1894.1.19.86-87, 71.2 and 108 mm. Ra- diographs reveal that the frontal bones, supra- occipital crest, supraneural bone, anteriormost ROBERTS: FRESHWATER PUFFERFISH FIGURE 1. Scales on side of body immediately anterior and ventral to pectoral fin (each square = 5x5 mm): (a) Xenopterus naritus, 71.2 mm, BMNH 1894.1.19.86; (b) Chonerhinos modestus, 48.4 mm, USNM uncatalogued; (c) Chonerhinos silus, 48.5 mm, FMNH 68815; (d) Chonerhinos remotus, 49.9 mm, FMNH 68475; (e) Chonerhinos nefastus, 48.3 mm, CAS 49507; (/) Chonerhinos amabilis, 48.7 mm, MZB 3973. anal-fin pterygiophore, and posteriormost neural and haemal spines are enormously thickened or hypertrophied, far out of proportion to neigh- boring bony elements. They appear to be hy- perosteotic (and in the case of the frontal bones, partially synosteotic), and therefore, I am du- bious about their phylogenetic significance and their being used as characters to distinguish Xenopterus from Chonerhinos. Other differ- ences between the two genera, cited above and in Tyler (1980), are sufficient to merit their sep- aration. Chonerhinos is known only from fresh water. Xenopterus, so far as I have been able to deter- mine, is marine or estuarine. There do not seem to be any museum specimens of Xenopterus with locality data from fresh water, and state- ments in the literature that Xenopterus occurs in fresh water (e.g., Cantor 1850:384; Weber and de Beaufort 1962:373) appear to be based at least partly on misinformation or confusion with Cho- nerhinos. In Chonerhinos and Xenopterus, as in many other tetraodontids, each scale has a spinelike distal portion which projects more or less straight out from the skin when erected, as usu- ally occurs when the fish inflates itself. When the scales are not erect, they are partially or wholly retracted beneath the skin, and the spines may be oriented dorsally, dorsoposte- riorly, or posteriorly, depending upon the species (Fig. 1). Size and shape of the jaw-teeth appear to be nearly identical in all species of Chonerhinos. One or two specimens of each species were dis- sected to permit observation of the gill rakers; all of the species have about 8-10 total gill rakers on each gill arch (sometimes fewer on the first arch). I have not attempted to distinguish the species by differences in the pathways of the lateral line canals. These are difficult to observe in many specimens, and they seem to be highly variable among individual specimens, often being irregularly interrupted or running into each other (Tyler 1980:fig. 223) and frequently differing in their courses on opposite sides of a specimen. Neither have I attempted to distin- guish the species by counts of olfactory lamel- PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 1 TABLE 1. PROPORTIONAL MEASUREMENTS IN Chonerhinos (expressed as times in standard length). C. amabilis C. modestus C. nefastus C. remotus C. silus n 20 13 54 31 33 SL mm 35.6-70.4 46.8-106 19.0-70.7 32.8-61.4 32.7-81.8 Eye 7.8-11.2 9.4-14.1 7.2-11.7 8.6-12.0 8.8-12.9 Nasal organ length 10.1-17.9 14.7-25.3 17.3-27.4 12.5-20.1 11.2-20.0 Snout length 6.0-7.2 6.4-7.5 5.8-7.1 6.6-8.7 6.3-8.1 Interorbital width 5.1-6.1 4.6-6.8 4.5-7.0 5.2-7.0 4.9-6.6 Pectoral-fin base length 10.5-12.2 9.5-11.5 9.9-13.5 8.9-11.1 9.1-12.4 Caudal peduncle depth 7.3-S.3 6.7-7.4 7.8-9.9 7.6-9.9 7.1-8.2 Caudal peduncle length 5.0-6.5 5.2-6.4 4.4-5.9 4.6-6.3 4.6-6.5 lae, the number of which seems to be highly variable within each species, as is the size of the nasal organ itself (Table 1). PROPORTIONAL MEASUREMENTS; MERISTIC FEATURES Proportional measurements, in most instances broadly overlapping and of little help in distin- guishing species, are presented in Table 1. Fre- quencies of counts of fin rays and vertebrae, diagnostic for the genus but differing slightly among species and of little or no help in identi- fying individual specimens, are presented in Ta- bles 2-3. Except in a few instances when counts or measurements are particularly useful for def- inition of species, these data are not repeated in the text. KEY TO SPECIES OF Chonerhinos la. Scales on side of body anterior and ven- tral to pectoral fin with spines directed posteriorly (Fig. le-f) 2 Ib. Scales on side of body anterior and ven- tral to pectoral fin with spines directed dorsally or dorsoposteriorly (Fig. \b-d) _ 3 2a. A roundish dark spot in middle of caudal peduncle; dorsal and anal fins always with angulated margins; upper lip not project- ing beyond lower lip; exposed portion of eye round; nasal organ relatively large, its length 10. 1-17. 9 (times in SL) C. amabilis 2b. No spot on caudal peduncle; dorsal and anal fins usually with rounded margins; upper lip usually projecting beyond lower lip; exposed portion of eye usually hori- zontally oval, especially in larger speci- mens; nasal organ relatively small, its length 17.3-27.4 C. nefastus 3a. Depth of caudal peduncle 6.7-7.4; upper and lower lips about equally projecting or lower lip slightly protruding; snout gently sloping; scales on side of body anterior and ventral to pectoral fin, very close-set with spines directed dorsally (Fig. Ib); anal-fin rays 20-22, modally 22 (Table 2); adult size to 106 mm C. modestus 3b. Depth of caudal peduncle 7.2-9.9; lower lip usually projecting beyond upper lip; snout strongly sloping; scales on side of TABLE 2. FREQUENCIES OF FIN RAY COUNTS IN Chonerhinos. Dorsal fin Anal fin Pectoral fin 22 23 24 25 26 27 28 18 19 20 21 22 13 14 15 16 17 C. amabilis - _ 1 12 6 1 _ _ _ 14 6 _ 3 16 1 _ C. modestus - - _ 4 12 8 1 _ 1 7 17 1 9 14 1 C. nefastus - 3 23 16 8 4 - 1 19 31 3 5 28 20 1 - C. remotus 5 30 40 8 2 - - 6 41 37 3 - 9 52 26 - C. silus - 1 8 18 27 3 1 1 2 19 34 3 _ 2 36 15 1 ROBERTS: FRESHWATER PUFFERFISH FIGURE 2. Chonerhinos amabilis, 45.2 mm, MZB 3972 (holotype). body anterior and ventral to pectoral fin not as close-set and with spines directed dorsoposteriorly (Fig. \c-d); anal-fin rays 18-22, rarely 22, modally 19 or 20 (Table 2); adult size to 82 mm 4 4a. Caudal peduncle moderately deep, its depth 7.2-8.2; dorsal-fin rays 23-28, av- erage 25.4 C. silus 4b. Caudal peduncle slender, its depth 7.6-9.9; dorsal-fin rays 22-26, average 23.6 __ C. remotus Chonerhinos amabilis new species (Figure 2) Chonerhinus naritus WEBER AND DE BEAUFORT, 1962:374 (specimens reported from "Labang hara, soengei Serawai"). Chonerhinus modestus WEBER AND DE BEAUFORT, 1962:fig. 84. HOLOTYPE. — MZB 3972, 45.2 mm, Kapuas R. 6 km w of Putussibau, Kapuas Ichthyological Survey, 9 Aug. 1976. PARATYPES. — CAS 49504, 45.0 mm, same data as holotype; MZB 3973, 48.7 mm, Kapuas basin, Sungai Landok at Nga- bang, 83 km ENE of Pontianak, Kapuas Ichthyological Survey, 15 July 1976; MZB 3974, 41.8 mm, Kapuas basin, Sungai Pi- noh 20-60 km upstream from Nangapinoh, Kapuas Ichthyo- TABLE 3. FREQUENCIES OF VERTEBRAL COUNTS IN Chonerhinos. C. amabilis C. modestus C. nefastus C. remotus C . silus 9 + 15 = 24(1) 9? + 16 = 25? (1) 9 + 16 = 25 (2) 9 + 16 = 25 (1) 9? + 16 = 25? (1) 10 + 15 = 25 (3) 10 + 15 = 25 (2) 9 + 17 = 26(1) 9 + 17 = 26 (1) 10 + 16 = 26 (9) 10 + 16 = 26 (3) 10 + 16 = 26(1) 10 + 16 = 26 (8) 10 + 16 = 26 (8) 10 + 17 = 27 (6) 10 + 17 = 27 (1) 11? + 16 = 27? (1) 11? + 16 = 27? (2) 11? + 16 = 271(1) PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 1 FIGURE 3. Chonerhinos modestus: (a) as illustrated in Bleeker 1865; length, locality, and present disposition of specimen unknown; (b) 78.7 mm, RMNH 26931 (neotype). logical Survey, 22-26 July 1976; MZB 3975, 38.3 mm, Kapuas R. near Kampong Nibung, 7 km NE of Selimbau, Kapuas Ich- thyological Survey, 5-6 July 1976; MZB 3976 and USNM 230359, 2:35.9-36.8 mm, Kapuas R. 53 km w of Putussibau, Kapuas Ichthyological Survey, 6-7 Aug. 1976; MZB 3977 and FMNH 94255, 2:35.6-46.0 mm, Kapuas R. about 23 km wsw of Putussibau, Kapuas Ichthyological Survey, 8-9 Aug. 1976; MNHN 91.216, 36.9 mm, Kapuas basin, M. Chaper, 1890; RMNH uncat., 2:40.9-41.2 mm, Kapuas basin, Sintang, July 1894; RMNH 7935, 4:55.5-68. 1 mm, Kapuas basin, Raun, Mar.-May 1894; ZMA 108.912, 3:56.3-70.4 mm, Kapuas ba- sin, Soengai Serawai, Lebang Hara, Witkamp, no date; UMMZ 171708, 2:36.2-38.3 mm, Sumatra, Moesi R. at Moera Klingi, A. Thienemann, 1913. DIAGNOSIS. — Chonerhinos amabilis is readily distinguished from all other members of the ge- nus by its highly distinctive coloration, almost all elements of which are visible in all specimens examined, including some century-old speci- mens which may have been dead for some time before being preserved. These unique features include a roundish dark spot in middle of caudal peduncle, visible in all specimens; a large, dis- tinctively shaped dark mark on dorsal surface of head extending uninterrupted from just behind upper lip to well behind the eyes, set off by pale coloration on the upper lip, sides of snout, nasal flaps, and skin dorsal to orbits; pale white or milky coloration on ventral and lateral surfaces of body extending very far dorsally; dark col- ROBERTS: FRESHWATER PUFFERFISH FIGURE 4. Chonerhinos modestus, 64.6 mm, CAS 49505. oration on dorsal surface of body markedly en- hanced around base of dorsal fin; and a small dark or dusky oval spot with indistinct margins near tip of chin (very faint or absent in some specimens). In addition, C. amabilis tends to have the largest nasal organ of any Chonerhi- nos, and thus of any tetraodontid (Tyler 1980:290); relatively large dorsal and anal fins with angulated (rather than rounded) margins; and scales on side of body anterior and ventral to pectoral fin relatively small, few in number, and with spines directed posteriorly (Fig. If). ETYMOLOGY. — Latin amabilis, lovely. Chonerhinos modestus (Bleeker) (Figures 3-5) Tetraodon (Arothron) modestus BLEEKER, 1850:16 (type-lo- cality "Banjermassing, in fluviis"). Chonerhinos modestus BLEEKER, 1854:260. Chonerhinus africanus BOULENGER, 1909:201 (type-locality "riv. Sankuru. a Kondue Kasai, Congo"). NEOTYPE.— RMNH 26931, 78.7 mm, Kapuas basin, Sang- gau, Westenenk, 1894. ADDITIONAL MATERIAL EXAMINED. — RMNH uncat., 2:49.2-59.2 mm, same data as neotype; RMNH 7934, 3:50.0-58.9 mm, Kapuas basin, Sintang, July 1894; CAS 49505 and MZB 3978, 2:64.6-106 mm, Kapuas R. about 23 km wsw of Putussibau, Kapuas Ichthyological Survey, 8-9 Aug. 1976; MZB 3979 and USNM 230360, 2:46.8-48.4 mm, Kapuas R. at Silat, Kapuas Ichthyological Survey, 17 Aug. 1976; BMNH 1846.6.22.75, 86.1 mm, Borneo, Frank Collection, no date; BMNH 1867.11.28.125, 87.3 mm, Borneo, Bleeker Col- lection, no date; RMNH 12004, 3:66.6-81.1 mm, Sumatra, Lahat, Bleeker Collection, 1850-60; NHMB 822-824, 3:44.7-73.5 mm, Sumatra, Indragiri, H. A. von Meckel, 1895; RMNH 7344 (part only), 8:47.9-62.0 mm, no locality data, Bleeker Collection, no date; MRAC 15306, 52.5 mm, "Congo, Sankuru River, Kasai" (holotype of C. africanus). SELECTION OF NEOTYPE. — Identification of C. modestus presented a difficult and taxonom- ically important problem which I have resolved by selecting a neotype. The holotype is lost or at least it cannot be positively identified, and the original description fits all five species of Cho- nerhinos about equally well. In order to facilitate the following discussion the original description (Bleeker 1850:16) is reproduced here in its en- tirety: PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 1 FIGURE 5. Chonerhinos modestus: (a) 52.5 mm, MRAC 15306 (holotype of C. africanus); (b) 48.4 mm, USNM 230360. Tetraodon (Arothron) modestus Blkr. Tetraod. corpore oblongo compresso, altitudine 4 circiter in ejus longitudine, latitudine 2 in altitudine; vertice, dorso, lateribus caudaque laevibus, pectore genisque scabris; ca- pite obtuso; lineo rostro-dorsali convexa; maxilla superiore paulo prominente; oculis paulo superis; tentaculis nasalibus 2 conicis obtusis loco narium; linea lateral! inconspicua; sacco pneumatico parvo; ano ante pinnam dorsalem sito; pinnis dorsali et anali obtusis angulatis angulis rotundatis, pectoralibus emarginatis, caudali truncata vel leviter emar- ginata 5 in longitudine corporis; colore corpore supra viridi infra argenteo, pinnis hyalino-viridescente. D. 5/20. P. 2/12. A. 3/20. C. 9 vel 11 et lat. brev. Habit. Banjermassing, in fluviis. Longitude speciminis unici 60'". Bleeker almost invariably recorded the length of his specimens as total length in millimeters (pers. commun. M. Boeseman, RMNH). Thus, the last two lines of the description indicate that it was based on a single specimen, the holotype, total length 60 mm, from riverine habitat at Bandjarmasin, southeastern Borneo (Barito ba- sin). Bleeker obtained in all 59 specimens which he identified as C. modestus, total lengths 46-126 mm, from Palembang (=Lahat?), Su- matra, and Sambas, Pontianak, and Sintang as well as Bandjarmasin in Borneo (Bleeker 1865:78). All extant "C. modestus" from Bleek- er's collection apparently are deposited in the BMNH, RMNH, and ZMA. The BMNH and ZMA each have a single Bleeker specimen, both of which are too large to be the holotype: BMNH 1867.11.28.125, 87.3 mm, Borneo, ex- amined by me, and ZMA 102.263, 104 mm, Bor- neo, examined for me by H. Nijssen. The RMNH has two lots, RMNH 12004, 3:66.6-81.1 ROBERTS: FRESHWATER PUFFERFISH mm, Sumatra, Lahat ( = Palembang?), and RMNH 7344, 52:29.0-74.0 mm (total lengths 37-88 mm), without locality data. If the holotype still exists, it presumably is in RMNH 7344. Among the 52 specimens are 4 which approxi- mate 60 mm in total length; thus, on the basis of length alone, the holotype cannot be identi- fied. Moreover, each of the four specimens dif- fers by one or two fin rays in at least two of the three counts reported by Bleeker for the dorsal, anal, and pectoral fins of the holotype. In my opinion, none of these specimens can reason- ably be identified as the holotype, and since their locality data are lost, a neotype should not be selected from among them. Unfortunately, I have been unable to find any specimen of Cho- nerhinos with locality data from Bandjarmasin or the Barito and do not know which of the species occur(s) there. As noted above, the original description of C. modestus fits all five species of Chonerhinos about equally well. All species of Chonerhinos normally have 1 1 caudal-fin rays, and all species are represented by specimens with 25 dorsal-fin rays and 14 pectoral-fin rays. On the other hand, none of the more than 250 specimens examined have 23 anal-fin rays. The highest number of anal-fin rays observed, 22, is usually found in the species herein identified as C. modestus, but also occurs in C. nefastus and C. situs. Color- ation and its variation in the species of Chone- rhinos are too poorly known at present to be of much help in their identification, and Bleeker' s description of coloration of the holotype cannot be accepted without reservation since he did not collect the specimen himself and could not have observed it until it had been in preservative for many days or weeks. Bleeker (1865:pl. 213, fig 8) published an excellent figure of a specimen which he identified as C. modestus. The length, locality, and date of collection of the specimen figured are not recorded, but it is not the holo- type. It is evidently a much larger specimen, with lateral line canals on the body plainly vis- ible, and differs also in fin-ray counts from the holotype as described by Bleeker. I have not tried to match up the figure with an extant spec- imen, although it may well be part of RMNH 7344. The figure does, however, show a number of features characteristic of the largest species of Chonerhinos, with which I unhesitatingly identify it. These features include its large size (indicated by the large size of the published il- lustration as well as by the relatively small eye); scales with dorsally oriented spines; relatively high counts of dorsal- and anal-fin rays; and deep caudal peduncle. All four specimens of to- tal length 60 mm in RMNH 7344 also belong to this species. Thus, there is every reason to identi- fy it as C. modestus, although we cannot be sure that this is the same species obtained for Bleeker at Bandjarmasin. In the absence of specimens with locality data from Bandjarmasin or the Ba- rito, a specimen from the Kapuas basin has been selected as neotype. This specimen bears a strong resemblance to Bleeker' s figure of C. modestus (Figs. 3a-b). DIAGNOSIS. — Chonerhinos modestus, attain- ing at least 106 mm, apparently is the largest species of Chonerhinos and has the deepest cau- dal peduncle. Depth of caudal peduncle 6.7-7.4 (vs. 7.2-9.9 in all other Chonerhinos). Scales relatively large and close-set, those on body an- teroventral to pectoral fin with spines directed dorsally, as in Xenopterus (vs. spines directed dorsoposteriorly or posteriorly in all other Cho- nerhinos). Upper and lower lips about equally projecting or lower lip slightly protruding. Ex- posed portion of eye round. Snout gently slop- ing. Nasal organ moderately large, its length 14.7-25.3. Dorsal-fin rays 25-28; and anal-fin rays 20-22 (generally fewer in other Chonerhi- nos). REMARKS ON SYNONYMY. — Most records of C. modestus in the literature other than those cited in the synonymy above refer in whole or in part to other species of Chonerhinos. Chonerhinos africanus was described briefly (and without a figure) on the basis of a single specimen supposedly obtained together with other fish specimens by E. Luja in the Sankuru River, Kasai, Congo basin, in 1908. No addi- tional specimens of Chonerhinos have been found in Africa, and the holotype has not been compared previously to Chonerhinos from Southeast Asia. I have examined the 52.5-mm holotype (Fig. 5a), comparing it directly with specimens of all five species of Chonerhinos, and conclude that it is conspecific with C. mo- destus. It has 26 dorsal-fin rays; 22 anal-fin rays; 15 pectoral-fin rays; 10+16 vertebrae; scales relatively large, those on sides of body antero- ventral to pectoral fin with spines directed dor- sally; lower lip slightly protruding; snout gently PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 1 FIGURE 5. Chonerhinos modestus: (a) 52.5 mm, MRAC 15306 (holotype of C. africanus); (b) 48.4 mm, USNM 230360. Tetraodon (Arothron) modestus Blkr. Tetraod. corpore oblongo compresso, altitudine 4 circiter in ejus longitudine, latitudine 2 in altitudine; vertice, dorso, lateribus caudaque laevibus, pectore genisque scabris; ca- pite obtuso; lineo rostro-dorsali convexa; maxilla superiore paulo prominente; oculis paulo superis; tentaculis nasalibus 2 conicis obtusis loco narium; linea lateral! inconspicua; sacco pneumatico parvo; ano ante pinnam dorsalem sito; pinnis dorsal! et anali obtusis angulatis angulis rotundatis, pectoralibus emarginatis, caudali truncata vel leviter emar- ginata 5 in longitudine corporis; colore corpore supra viridi infra argenteo, pinnis hyalino-viridescente. D. 5/20. P. 2/12. A. 3/20. C. 9 vel 11 et lat. brev. Habit. Banjermassing, in fluviis. Longitude speciminis unici 60'". Bleeker almost invariably recorded the length of his specimens as total length in millimeters (pers. commun. M. Boeseman, RMNH). Thus, the last two lines of the description indicate that it was based on a single specimen, the holotype, total length 60 mm, from riverine habitat at Bandjarmasin, southeastern Borneo (Barito ba- sin). Bleeker obtained in all 59 specimens which he identified as C. modestus, total lengths 46-126 mm, from Palembang (=Lahat?), Su- matra, and Sambas, Pontianak, and Sintang as well as Bandjarmasin in Borneo (Bleeker 1865:78). All extant "C. modestus" from Bleek- er's collection apparently are deposited in the BMNH, RMNH, and ZMA. The BMNH and ZMA each have a single Bleeker specimen, both of which are too large to be the holotype: BMNH 1867.11.28.125, 87.3 mm, Borneo, ex- amined by me, and ZMA 102.263, 104 mm, Bor- neo, examined for me by H. Nijssen. The RMNH has two lots, RMNH 12004, 3:66.6-81.1 ROBERTS: FRESHWATER PUFFERFISH mm, Sumatra, Lahat ( = Palembang?), and RMNH 7344, 52:29.0-74.0 mm (total lengths 37-88 mm), without locality data. If the holotype still exists, it presumably is in RMNH 7344. Among the 52 specimens are 4 which approxi- mate 60 mm in total length; thus, on the basis of length alone, the holotype cannot be identi- fied. Moreover, each of the four specimens dif- fers by one or two fin rays in at least two of the three counts reported by Bleeker for the dorsal, anal, and pectoral fins of the holotype. In my opinion, none of these specimens can reason- ably be identified as the holotype, and since their locality data are lost, a neotype should not be selected from among them. Unfortunately, I have been unable to find any specimen of Cho- nerhinos with locality data from Bandjarmasin or the Barito and do not know which of the species occur(s) there. As noted above, the original description of C. modestus fits all five species of Chonerhinos about equally well. All species of Chonerhinos normally have 1 1 caudal-fin rays, and all species are represented by specimens with 25 dorsal-fin rays and 14 pectoral-fin rays. On the other hand, none of the more than 250 specimens examined have 23 anal-fin rays. The highest number of anal-fin rays observed, 22, is usually found in the species herein identified as C. modestus, but also occurs in C. nefastus and C. silus. Color- ation and its variation in the species of Chone- rhinos are too poorly known at present to be of much help in their identification, and Bleeker' s description of coloration of the holotype cannot be accepted without reservation since he did not collect the specimen himself and could not have observed it until it had been in preservative for many days or weeks. Bleeker (1865:pl. 213, fig 8) published an excellent figure of a specimen which he identified as C. modestus. The length, locality, and date of collection of the specimen figured are not recorded, but it is not the holo- type. It is evidently a much larger specimen, with lateral line canals on the body plainly vis- ible, and differs also in fin-ray counts from the holotype as described by Bleeker. I have not tried to match up the figure with an extant spec- imen, although it may well be part of RMNH 7344. The figure does, however, show a number of features characteristic of the largest species of Chonerhinos, with which I unhesitatingly identify it. These features include its large size (indicated by the large size of the published il- lustration as well as by the relatively small eye); scales with dorsally oriented spines; relatively high counts of dorsal- and anal-fin rays; and deep caudal peduncle. All four specimens of to- tal length 60 mm in RMNH 7344 also belong to this species. Thus, there is every reason to identi- fy it as C. modestus, although we cannot be sure that this is the same species obtained for Bleeker at Bandjarmasin. In the absence of specimens with locality data from Bandjarmasin or the Ba- rito, a specimen from the Kapuas basin has been selected as neotype. This specimen bears a strong resemblance to Bleeker' s figure of C. modestus (Figs. 3a-b). DIAGNOSIS. — Chonerhinos modestus, attain- ing at least 106 mm, apparently is the largest species of Chonerhinos and has the deepest cau- dal peduncle. Depth of caudal peduncle 6.7-7.4 (vs. 7.2-9.9 in all other Chonerhinos). Scales relatively large and close-set, those on body an- teroventral to pectoral fin with spines directed dorsally, as in Xenopterus (vs. spines directed dorsoposteriorly or posteriorly in all other Cho- nerhinos). Upper and lower lips about equally projecting or lower lip slightly protruding. Ex- posed portion of eye round. Snout gently slop- ing. Nasal organ moderately large, its length 14.7-25.3. Dorsal-fin rays 25-28; and anal-fin rays 20-22 (generally fewer in other Chonerhi- nos). REMARKS ON SYNONYMY. — Most records of C. modestus in the literature other than those cited in the synonymy above refer in whole or in part to other species of Chonerhinos. Chonerhinos africanus was described briefly (and without a figure) on the basis of a single specimen supposedly obtained together with other fish specimens by E. Luja in the Sankuru River, Kasai, Congo basin, in 1908. No addi- tional specimens of Chonerhinos have been found in Africa, and the holotype has not been compared previously to Chonerhinos from Southeast Asia. I have examined the 52.5-mm holotype (Fig. 5a), comparing it directly with specimens of all five species of Chonerhinos, and conclude that it is conspecific with C. mo- destus. It has 26 dorsal-fin rays; 22 anal-fin rays; 15 pectoral-fin rays; 10 + 16 vertebrae; scales relatively large, those on sides of body antero- ventral to pectoral fin with spines directed dor- sally; lower lip slightly protruding; snout gently PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 1 FIGURE 6. Chonerhinos nefastus, 47.0 mm, MZB 3980 (holotype). sloping; eye 10.7; olfactory organ 18.7; snout 7.2; interorbital width 5.25; depth caudal pedun- cle 7.4; length caudal peduncle 5.9; and pectoral- fin base 10.8. The caudal peduncle depth and anal-fin ray count, while not conclusive, agree best with C. modestus. Presumably, the speci- men originated somewhere in Southeast Asia and somehow became mixed with Luja's mate- rial when it was sent on loan to the BMNH for Boulenger to study. Chonerhinos nefastus new species (Figure 6) Chonerhinos modestus D'AUBENTON AND BLANC, 1966:561 (Mekong basin, Kampuchea); TAKI 1974:199-200, fig. 187 (Mekong basin, Laos); IMAKI et al. 1978:29, pi. 18 (Kapuas R. at Sintang); TYLER 1980 (fig. 223?). HOLOTYPE.— MZB 3980, 47.0 mm, Kapuas R. 29 km w of Putussibau, Kapuas Ichthyological Survey, 11 Aug. 1976. PARATYPES.— CAS 49506 and MZB 3981, 4:34.3^3.3 mm, Kapuas R. near Kampong Nibung, 7 km NE of Selimbau, Ka- puas Ichthyological Survey, 5-6 July 1976; BMNH 1982.3.29. 254-5 and MZB 3982, 3:36.6-43.7 mm, Kapuas R. 53 km w of Putussibau, Kapuas Ichthyological Survey 6-7 Aug. 1976; IRSNB 632, MZB 3983, ROM 38601, and USNM 230361, 6:32.9-60.2 mm, Kapuas R. about 23 km wsw of Putussibau, Kapuas Ichthyological Survey, 8-9 Aug. 1976; MZB 3984, 51.7 mm, Kapuas basin, small tributary of Sungai Mandai 17 km wsw of Putussibau, Kapuas Ichthyological Survey, 10 Aug. 1976; MZB 3985, 64.9 mm, Kapuas basin, Sungai Mandai Kechil, 18 km wsw of Putussibau, Kapuas Ichthyological Sur- vey, 11 Aug. 1976; CAS 49507 and MZB 3986, 3:36.7-57.8 mm, Kapuas basin, Sungai Tawang near Danau Pengembung, Kapuas Ichthyological Survey, 14-15 Aug. 1976; RMNH 7936, 61.8 mm, Kapuas basin, Sibau, June 1894; RMNH un- cat., 25.5 mm, Kapuas basin, Sintang, July 1894; ZMA 1 10.220, 65.8 mm, Kapuas basin, Bunut, H. A. Lorentz, 26 June 1909; FMNH uncat. 3:38.2-50.1 mm, Sarawak, Niah R., T. Harrisson, 1 Apr. 1963; FMNH uncat., 2:64.8-70.7 mm, Sarawak, Niah, T. Harrisson, no date; FMNH uncat., 3:36.9-41.8 mm, Sarawak, Rejang basin, Baleh R. between Sungai Mujong and Sungai Gaat, R. F. Inger, 3 Aug. 1956; RMNH 7933, 2:56.6-68.0 mm, Mahakam basin, Tepoe, A. W. Nieuwenhuis, 1896-97; MG 2058.94, 34.9 mm, Kalimantan Tengah, Mentaya basin near Sampit, Pfeuffer, May 1980; UMMZ uncat., 50.1 mm, Sumatra, Moesi R. at Moera Klingi, A. Thienemann, 1913; SU 36040, 41.7 mm, Malay Peninsula, Per- ak, Chandra dam, A. W. Herre, 18 Mar. 1923; UMMZ 197038, 43.7 mm, Thailand, Songkhla Lake off Patalung, K. F. Lagler, 6 Jan. 1965; UMMZ uncat., 48.0 mm, Thailand, Mekong ba- sin, Ubon Ratchtani, Huay Phai, 16 Oct. 1975; UMMZ uncat., 38.9 mm, Thailand, Mekong basin, Ubon Ratchtani, Huay Kwang, 1 Oct. 1976; UMMZ uncat., 42.5 mm, Thailand, Me- kong basin, Huay Kwang s of Khong Chiam, Arden, 7 Oct. 1975; UMMZ uncat., 30.9 mm, Thailand, Mekong basin, Mun R. at Khong Chiam, Songrad and Buskirk, 19 July 1975; UMMZ uncat., 3:15.4-32.2 mm, Thailand, Mekong R. and ROBERTS: FRESHWATER PUFFERFISH FIGURE 7. Chonerhinos remotus, 52.7 mm, FMNH 68475 (holotype). tributaries from Ban Dan to Nakon Phanom, Mekong fish sur- vey, Mar.-Apr. 1975; MNHN 1966.55-56, 9:21.6-48.1 mm, Kampuchea, Mekong basin, Prek Tasom, F. d'Aubenton, 5 June and 9 Nov. 1961; MNHN 1966.57, 12:19.0-47.5 mm, Kampuchea, Mekong basin, Prek Andor, F. d'Aubenton, 2 Dec. 1961. DIAGNOSIS. — Chonerhinos nefastus differs from all other species of Chonerhinos in having upper lip usually projecting beyond lower lip; nasal organ relatively small (Table 1); and ex- posed portion of eyeball usually horizontally oval rather than round or vertically oval. It dif- fers from all other species except C. amabilis in having scales on side of body anterior and ven- tral to pectoral fin usually with spines directed posteriorly (Fig. le), and from all except C. re- motus in its slender caudal peduncle (Table 1). Body usually without distinct color marks ex- cept for a slightly darkened spot on dorsal sur- face of head posterior to eyes. COMMENTS. — The exposed portion of the eyeball is distinctly horizontally oval in more than half of the specimens examined. It is usu- ally round in very small specimens, however, and sometimes round in large specimens (in- cluding the holotype). Most specimens have the scales on the side of the body anterior and ven- tral to the pectoral fin with the spines directed posteriorly, as in Figure le. This character is variable, however, and in a few specimens the spines are directed posterodorsally or almost dorsally. This is most noticeable in the sample of 12 specimens from Prek Andor, 4 of which have the spines more dorsally directed than is usual in C. nefastus. The rest of the specimens in the sample have the spines directed poste- riorly or posterodorsally. Specimens from the Mekong River differ from C. nefastus from oth- er localities in having a dark transverse mark on the dorsal surface of the snout between the up- per lip and the nostrils. ETYMOLOGY. — Latin nefastus, wicked, abominable, in reference to the food habits (see below). Chonerhinos remotus new species (Figure 7) Chonerhinos modestus HERRE, 1940:55 (Sandakan District, Sungei Segaliud and Sungei Sibuga); INGER AND CHIN 1962: 190-191, fig. 101 (Kinabatangan District). HOLOTYPE. — FMNH 68476, 52.7 mm, Kinabatangan basin, mouth of Sungai Deramakot, R. F. Inger and P. K. Chin, 27 Apr. 1956. PARATYPES. — FMNH uncat., 9:32.8-54.4 mm, same data as holotype; CAS 49743 and FMNH 68475, 61:29.1-56.8 mm, Kinabatangan R. below mouth of Malubok R., R. F. Inger and P. K. Chin, 25 Apr. 1956; FMNH 68474, 3:47.2-54.4 mm, Kinabatangan R. at Deramakot camp, R. F. Inger and P. K. Chin, 24 Apr. 1956; FMNH 44931, 38.3 mm, Kinabatangan District, N. Borneo Fisheries Dept., 20 Jan. 1949; SU 33487, 2:60.5-61.4 mm, Sandakan District, Sibugal R. (=Sungai Si- buga), A. W. Herre, 19 Apr. 1938; SU 33563, 10:30.5-40.4 mm, Sandakan District, Segaliud R., A. W. Herre, 4 Feb. 1937. DIAGNOSIS. — Chonerhinos remotus is most similar to C. silus, from which it differs in hav- ing a more slender caudal peduncle (Table 1); fewer dorsal- and anal-fin rays on the average PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 1 FIGURE 8. Chonerhinos silus, 44.9 mm, FMNH 68477 (holotype). (Table 2); an even more strongly sloping snout; lower lip almost always strongly projecting be- yond upper lip (vs. lower lip slightly projecting or equal to upper lip); and eye vertically oval or round (vs. usually round). Scales anterior and ventral to pectoral fin moderately large and close-set, with spines directed posterodorsally (Fig. \d). No distinctive color marks except for a well-defined dark blotch on dorsal surface of head posterior to eyes. ETYMOLOGY. — Latin remotus, remote, in ref- erence to the type-locality. Chonerhinos silus new species (Figure 8) HOLOTYPE. — FMNH 68477, 44.9 mm, Sarawak, Rejang ba- sin, Sungai Baleh between Sungai Mujong and Sungai Gaat, R. F. Inger, 3 Aug. 1956. PARATYPES.— CAS 49744, FMNH uncat., 36:35.0-60. 1 mm, same data as holotype; FMNH 62987, 44.0 mm, Sarawak, Niah R., Niah, Lord Medway, 22 Aug. 1959; FMNH 68813, 81.8 mm, Sarawak, Niah, T. Harrisson, no date; FMNH 68814; 2:44.1-68.7 mm, Sarawak, Niah, Niah R., Pengkalan Lobang, T. Harrisson, 2-11 Nov. 1960; FMNH 68815, 16:37.1-72.9 mm, Sarawak, Niah R., T. Harrisson, 1 Apr. 1963; SU 33610, 32.7 mm, Sarawak, 16 miles [ca. 26 km] E of Kuching, A. W. Herre, 16 Feb. 1937. DIAGNOSIS. — Chonerhinos silus is most simi- lar to C. remotus and C. modestus. Differences between C. silus and C. remotus are set forth above in the diagnosis of C. remotus. It differs from C. modestus in attaining smaller adult size (largest specimen examined 82 mm vs. 106 mm); snout more strongly sloping; lips equally pro- jecting, or lower lip variably protruding, fre- quently much more so than in C. modestus; scales anterior and ventral to pectoral fin with spines projecting dorsoposteriorly (Fig. Ic) rath- er than dorsally (Fig. \b); and caudal peduncle relatively slender, its depth 7.2-8.2 (vs. 6.7-7.4). C. silus tends to have fewer dorsal-, anal-, and pectoral-fin rays than C. modestus (Table 2), but the counts are broadly overlapping and of little help in identifying individual specimens to species. ETYMOLOGY. — Latin silus, pugnosed. COLORATION IN LIFE Most of the specimens of Chonerhinos col- lected during the 1976 Kapuas Ichthyological Survey were caught at night and preserved be- fore their coloration in life could be properly observed. Colors of the 106-mm C. modestus, gill-netted at night and removed the next morn- ing, are recorded in my field notes and in a 35- mm Kodachrome slide. It was pale blue dorsal- ly, white on the sides and abdomen, and with a reddish eye. It is my impression that the three smaller C. modestus collected during the survey were similarly colored. C. amabilis is described in my field notes as lime-green dorsally, with a darkened area along the base of the dorsal fin, and a reddish eye; the round spot on the caudal peduncle, so evident in preserved specimens, was not observed during life (at least it is not recorded in my field notes, and I do not recall ROBERTS: FRESHWATER PUFFERFISH 13 having seen it in the live specimens). I suspect that some C. amabilis were blue dorsally but this is not recorded in my field notes. My impression is that all C. nefastus caught during the survey were pale green dorsally; at least this was so in several specimens observed during the day. I doubt that any of them were blue dorsally. D'Aubenton and Blanc (1966) reported color- ation of C. nefastus (as C. modestus) from the Mekong basin in Kampuchea as green on the back and white on the flanks and belly, while Taki (1974) reported specimens from the Me- kong in Laos as having "back and upper surface of head and body olivaceous golden, underside pale yellow to white. Dorsal and caudal fins greenish yellow; anal fin pale yellow; pectoral fins hyaline." SEXES Secondary sexual dimorphism is unknown in Chonerhinos. I have examined ripe males and gravid or ripening females in all five species. Ovaries of the left and right sides are about equally well developed. The following approxi- mate counts of eggs and measurements of egg diameters contained in the right ovary were made; C. amabilis, 57.4 mm, 180 eggs, 1.1-1.9 mm; C. modestus, 106 mm, 800 eggs, 1.5-2.1 mm; C. nefastus, 56.5 mm, 100 eggs, 1.4-1.5 mm, 57.8 mm, 80 eggs, 1.3 mm, and 64.9 mm, 230 eggs, 1.3-1.6 mm; C. remotus, 54.2 mm, 85 eggs, 1.9-2.3 mm; and C. situs, 58.7 mm, 200 eggs, 1.5-2.1 mm. All of these specimens are gravid except the three C. nefastus, which are nearly ripe. In C. remotus I observed two gravid females, 54.2 and 54.4 mm, and three spent fe- males, 48.8, 51.4, and 52.7 mm (the holotype, Fig. 7), with genitoanal areas much swollen. Such swelling, perhaps present only in females just before or after spawning, has not been ob- served in other species. FOOD HABITS Food habits of Chonerhinos, determined by complete or partial examination of gut contents in more than 100 specimens, may be summa- rized as follows: C. amabilis feeds almost ex- clusively on large aquatic insects; C. modestus feeds mainly on terrestrial insects, shrimps, seeds, and to a less extent on whole fish, fin rays, or scales; C. nefastus feeds mainly on fish fin rays and scales, and to a lesser extent on insects (aquatic and terrestrial); C. remotus and C. silus feed mainly on insects aquatic and ter- restrial), but also ingest vegetable matter and other items. No fish remains were found in C. amabilis, C. silus, or, excepting a single fish scale in one specimen, C. remotus. Pieces of clam flesh and gills were found in several C. silus, and numerous small, whole clams in a sin- gle C. nefastus, but otherwise molluscs were absent. The food of the five species may be de- scribed in more detail as follows. In C. amabilis, 18 of 20 specimens contained more or less abundant remains of insects, mainly large aquatic forms; partial examination of the gut contents of these specimens failed to reveal any other food items. Of the remaining two specimens, one contained moderate amounts of an unidentified flocculent material, and one had empty guts. This species is noteworthy in that nearly all individuals had much food in their guts, and in being the most stenophagic of any species of Chonerhinos. In C. modestus, guts were examined in 10 specimens, half of which had empty guts. Of the remaining five, four con- tained moderate to large amounts of insects (mainly terrestrial), two had prawns, two had seeds, two had fish scales, one had fish fin rays, and one had the remains of a small whole cobitid fish (identified by its Weberian apparatus). The last C. modestus, the 106-mm specimen, is of particular interest because of its large size and because of the circumstances of its capture. It was gill-netted together with a large catfish, Pangasius polyuranodon (Fig. 9), which had much of its abdominal wall and portions of its anal and caudal fins and caudal peduncle bitten away. I suspected that part of the damage may have been done by the C. modestus, but careful examination of its gut contents failed to reveal any material from the Pangasius. While the C. modestus may have regurgitated, its stomach did contain other food items, and it seems more likely that the Pangasius was ravaged by some other predator, possibly C. nefastus. Of 31 C. nefastus in which the gut contents were exam- ined, 1 1 had more or less substantial amounts of fish fin rays, six had fish scales, three contained small pieces offish flesh, six had small to mod- erate amounts of insects (terrestrial and aquat- ic), two had unidentified debris or detritus, one had numerous small, whole bivalves, and one had a large amount of sand and grit; seven had PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 1 FIGURE 9. An 106-mm Chonerhinos modestus gill-netted together with an 80-cm Pangasius polyuranodon catfish ravaged by an unknown predator, possibly C. modestus or C. nefastus (Kapuas River near Putussibau). empty guts. The Latin name nefastus refers to the predominantly pterygophagous and lepido- phagous habits of this species. Inger and Chin (1962:191) reported gut contents of 11 C. re- motus (as C. modestus) as follows: bits of leaves (6); parts of terrestrial insects (6); Plecoptera nymphs (3); Trichoptera larval cases (1); un- identified insect larvae (3); Acarina (2); unspec- ified parts of fishes (2). Of 21 C. remotus I ex- amined, 18 had guts containing food items: 14 with insects (aquatic and terrestrial), 4 with parts of higher plants, 1 with a mite, 1 with a fish scale, and several with unidentified debris or detritus. In 33 C. silus, 22 had guts containing insects (aquatic and terrestrial), 6 contained higher plant material (fine rootlets, leaf, seeds, or seed pulp?), 1 had several pieces of a large, spinulose oligochaete, and 1 had chunks of spiny or hairy flesh (mammalian?); the remainder had empty guts. INTRASPECIFIC BITING Intraspecific biting, although infrequently documented, probably occurs in many members of the family Tetraodontidae. In Fugu niphobles (Jordan and Snyder, 1901), biting is an integral part of spawning behavior: egg laying occurs on the beach at high tide after a female has been bitten on the sides by two to four males (Uno 1955). Many of the specimens of Chonerhinos examined exhibited characteristically shaped bite marks on the flanks and, even more fre- quently, had portions of the median fins bitten off. I suspect that much of the biting, at least in C. nefastus, is inflicted by conspecifics. More than half of the specimens examined of this species had bite marks on the flanks or had por- tions of the dorsal, anal, or caudal fins missing. In many specimens these fins appear to have been bitten repeatedly, as evidenced by scar tis- sue and imperfect regeneration of fin rays. It is noteworthy that this species feeds predominant- ly upon fish fin rays (see above under Food Hab- its). C. modestus and C. silus, both of which occur sympatrically with C. nefastus, also ex- hibit high frequencies of specimens with bite marks and bitten fins, but it is unclear whether this is a result of intraspecific attacks, attacks by C. nefastus, or a combination of both. In all three species the bite marks and fin damage ap- pear to be about equally distributed between the sexes, and between gravid and nongravid fe- males. None of the specimens of C. amabilis and C. remotus examined exhibited bite marks on the flanks, and their fins were relatively un- damaged, with little or no indication of fin-nip- ping. Perhaps the generally pterygophagous and lepidophagous feeding behavior of C. nefastus was preceded by the evolution of an exception- ally aggressive intraspecific biting and fin-nip- ping behavior. GEOGRAPHICAL DISTRIBUTION Tetraodontidae is the only one of the nine families of the large order Plectognathi or Te- traodontiformes which has representatives that occur in fresh water. About 25 of the approxi- mately 140 described tetraodontid species are ROBERTS: FRESHWATER PUFFERFISH 15 -30° -20C -10C °C. amabilis *C. modestus •C. nefastus *C. remotus °C. silus -oc 11(11 FIGURE 10. Geographical distribution of species of Chonerhinos. endemic to fresh water. Carinotetraodon and Chonerhinos, both from Southeast Asia, are the only tetraodontid genera restricted to fresh water. Other genera with freshwater species in- clude Tetraodon or Monotreta in India, South- east Asia, and New Guinea; Tetraodon in Afri- ca; and Colomesus in South America. Two features of the geographical distribution of freshwater Tetraodontidae merit comment. First, although marine tetraodontids extend into high latitudes in the Northern and Southern hemispheres, freshwater species occur only within tropical latitudes. Second, the tropical rivers with endemic tetraodontids generally have rich ichthyofaunas dominated by primary freshwater fishes. Geographical distributions of the species of Chonerhinos, based mainly on material exam- ined in this study, are illustrated in Figure 10. Two of the species, C. amabilis and C. modes- tus, have distributions lying within the hydro- graphic limits of the ancient Central Sundaland River basin, now fragmented by the Java and South China seas. I suspect that C. modestus also occurs in Thailand but have not examined specimens from there. The most widely distrib- uted species, C. nefastus, occurs throughout the area occupied by the Central Sundaland River basin; it also occurs in northern and southern Borneo and in the Mekong basin. Whether the Mekong River once also formed part of the Cen- tral Sundaland drainage is a matter under inves- tigation. C. silus and C. remotus, in northern and northeastern Borneo, have restricted distri- butions entirely outside the limits of the Central Sundaland drainage area. C. amabilis, C. mo- destus, and C. nefastus occur sympatrically in the Kapuas River and probably also in some rivers in Sumatra including the Indragiri and PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 1 Moesi. C. nefastus and C. silus occur sympatri- cally in Sarawak (Rejang and Niah basins). ACKNOWLEDGMENTS It is a pleasure to thank the following individ- uals for their help during this study: Oliver Crimmen, Gordon Howes, and Alwyne Wheel- er, BMNH; Marie-Louise Bauchot, Jaques Dag- et, and Martine Desoutter, MNHN; Marinus Boeseman, Peter van Helsdingen, and M. J. P. van Oijen, RMNH; Han Nijssen, ZMA; Volker Mahnert, MG; Donald J. Stewart, FMNH; Wal- ter Rainboth, UMMZ; Dirk Thys van den Au- denaerde, MRAC; Maurice Kottelat, Universite de Neuchatel; and Lillian Dempster, Madeleine Graham, W. I. Follett, Michael Hearne, and James Jackson, CAS. Photography is by Al- phonse Coleman, Museum of Comparative Zo- ology, Harvard University, and Orrin Moon, The Darkroom, San Rafael. The ichthyological survey of the Kapuas basin was sponsored by the Museum Zoologicum Bo- gorense, Indonesian National Research Council, and Smithsonian Tropical Research Institute. Soetikno Woerjoatmodjo, Leo Poerwadi, and Rajali assisted in the field. Research was done during visits to the BMNH, MNHN, RMNH, and ZMA, and at the California Academy of Sci- ences and Tiburon Center for Environmental studies, and was supported by National Science Foundation grant DEB77-24759. LITERATURE CITED BLEEKER, P. 1850. Bijdrage tot de kennis der ichthyologische fauna van Borneo, met beschrijving van 16 nieuwe soorten van zoetwatervisschen. Nat. Tijdschr. Ned. Ind. 1:1-16. . 1854. Vijfde bijdrage tot de kennis der ichthyolo- gische fauna van Celebes. Nat. Tijdschr. Ned. Ind. 7:225-260. . 1865. Atlas ichthyologique des Indes Orientales Neer- landaises, vol. 5. BOULENGER, G. A. 1909. Catalogue des poissons du Congo du Musee d'Histoire naturelle de Luxembourg. Faune de Sankuru a Kondue (Collection Ed. Luja). Monatsber. Ge- sell. Luxemburg. Naturf., n. sen, 3:189-202. CANTOR, T. E. 1850. Catalogue of Malayan fishes. J. Thomas, Calcutta, xii + 461 p., 24 pis. D'AUBENTON, F., AND M. BLANC. 1966. Poissons tetraodon- tiformes du Cambodge. Bull. Mus. Natl. Hist. Nat. ser. 2, 38:554-561. FRASER-BRUNER, A. 1943. Notes on plectognath fishes. — VIII. The classification of the suborder Tetraodontoidea, with a synopsis of the genera. Ann. Mag. Nat. Hist. ser. 11, 10:1-18. GILL, T. N. 1892. Note on the genus Chonerhinos or Xe- nopterus. Proc. U.S. Natl. Mus. 14:696-699. HERRE, A. W. 1940. Additions to the fish fauna of Malaya and notes on rare or little known Malayan and Bornean fishes. Bull. Raffles Mus. 16:27-61. HOLLARD, H. 1857. Etudes sur les Gymnodontes et en par- ticulier sur leur osteologie et sur les indications qu'elle peut fournir pour leur classification. Ann. Sci. Nat. (Paris), zool., ser. 4, 8:275-328. IMAKI, A., A. KAWAMOTO, AND A. SUZUKI. 1978. A list of freshwater fishes collected from the Kapuas River, West Kalimantan, Indonesia. The Institute for Breeding Re- search, Tokyo University of Agriculture, 50 p. INGER, R. F., AND P. K. CHIN. 1962. The fresh-water fishes of North Borneo. Fieldiana: Zool. 45: 1-263. JORDAN, D. S. 1919. The genera of fishes, 2. Stanford Univ. Publ., univ. ser., i-x + 163-284 + i-xiii p. ROBERTS, T. R. 1981. Identification of the presumed African freshwater fishes Micracanthus marchei (Belontiidae) and Chonerhinos africanus (Tetraodontidae). Cybium, ser. 3, 5:91-92. TAKI, Y. 1974. Fishes of the Lao Mekong basin. USAID Mission to Laos, Agric. Div., vi + 232 p. TYLER, J. C. 1980. Osteology, phylogeny, and higher classi- fication of the fishes of the order Plectognathi (Tetraodon- tiformes). NOAA Tech. Rep. NMFS Circ. 434, 422 p. UNO, Y. 1955. Spawning habit and early development of a puffer, Fugu (Torafugu) niphobles (Jordan et Snyder). J. Tokyo Univ. Fish. 41:169-183. WEBER, M., AND L. F. DE BEAUFORT. 1962. The fishes of the Indo-Australian Archipelago, vol. 11. E. J. Brill, Lei- den, ix + 481 p. CALIFORNIA ACADEMY OF SCIENCES Golden Gate Park San Francisco, California 94118 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES Vol. 43, No. 2, pp. 17-24, 7 figs. June 15' 1982 SYNONYMIES OF INDIAN OCEAN EELS, WITH THE DESCRIPTION OF GYMNOTHORAX ENIGMATICUS, A MORAY PREVIOUSLY KNOWN AS G. RUPPELI By John E. McCosker Steinhart Aquarium, California Academy of Sciences, Golden Gate Park, San Francisco, California 94118 and John E. Randall Bernice P. Bishop Museum, Honolulu, Hawaii 96818 ABSTRACT: The common, banded Indo-Pacific morays called Gymnothorax peteUi (Bleeker, 1856) and G. ruppeli (McClelland, 1845) by recent authors are recognized as G. rueppelliae (McCleUand, 1845) and G. enigmaticus n.sp., respectively. They are separable on the basis of coloration, vertebrae, and morphology and have different geographic ranges. G. signffier Bliss, 1883, is placed in the synonymy of G. rueppelliae, along with Muraena umbrofasciata Ruppeli, 1852; M. interrupta Kaup, 1856; Sideria chlevastes Jordan and Gilbert, 1883; G. leucacme Jenkins, 1904; and G. waialuae Snyder, 1904. The moray Uropterygius xanthopterus Bleeker, 1859, is recognized as distinct from V. marmoratus (Lacepede, 1803), and V. alboguttatus Smith, 1962, is synonymous with it. Ophichthus retifer Fowler, 1935, from Durban, South Africa, is a synonym of (). erabo (Jordan and Snyder, 1901), an ophichthid also known from Hawaii, Japan, and Taiwan. INTRODUCTION the snout tip to the posterodorsal margin of the In preparation for the publication of the eel gill opening; trunk length is taken from the end section of the revised Sea Fishes of Southern of the head to mid-anus; maximum body depth Africa (McCosker and Castle, MS), we assign does not include the median fins. Vertebral several poorly known taxa to synonymy and counts (which include the hypural) were taken provide a description for a common, conspicu- from radiographs. Materials used in this study ously banded Indo-Pacific moray, Gymnothorax are housed at the following institutions: Acad- ruppeli of earlier authors, which lacks a holo- emy of Natural Sciences of Philadelphia type and scientific name. (ANSP); Bernice P. Bishop Museum (BPBM); British Museum of Natural History (BMNH); METHODS California Academy of Sciences (CAS); U.S. Measurements are straight-line, made either National Museum of Natural History (USNM); with a 300-mm ruler with 0.5-mm gradations (for Museum of Comparative Zoology, Harvard Uni- total length, trunk length, and tail length) and versity (MCZ); J. L. B. Smith Institute of Ich- recorded to the nearest 0.5 mm, or with dial cal- thyology, Rhodes University (RUSI); Natur- ipers (all other measurements) and recorded to Museum Senckenberg (SMF); and the Scripps the nearest 0. 1 mm. Body length comprises head Institution of Oceanography (SIO). Paratypes of and trunk lengths. Head length is measured from the new species will also be deposited at the [17] 18 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 2 FIGURE 1. Paratype of Gymnothorax enigmaticus n.sp., BPBM 9491, 335 mm TL, from Palau. Museum National d'Histoire Naturelle, Paris (MNHN), the BMNH, and the USNM. FAMILY MURAENIDAE Gymnothorax enigmaticus, new species (Figure 1) Holotype.— CAS 48815, 303.2 mm total length, a male (?) collected with rotenone in a 0-1-m tidal flat, off southern cor- ner Ngatchab Beach, Angaur I., Palau, Western Caroline Is., by H. DeWitt, Sumang, and Sengjch, 21 Oct. 1957. Paratypes. — Collected in shallow coral reef flats and tide- pools (0-3 m) using rotenone ichthyocides. PALAU: CAS 48823 (8; 51-299 mm), Angaur I., H. DeWitt et al., 22 Oct. 1957. CAS 48817 (293 mm), Auluptagel I., H. Fehlmann et al., 10 Aug. 1955. CAS 48816 (335.5 mm), Ngethil I., Sumang and R. Johaness, 13 July 1959. CAS 48822 (2; 238-257 mm), Aulong I., Sumang, 5 Nov. 1959. CAS 48826 (301.5 mm), Urukthapel I., H. Fehlmann et al., 19 Aug. 1955. BPBM 9491 (335 mm), Malakal Harbor, A. Emery, 21 Apr. 1970. K.APINGAMARINGI: CAS 48818 (163.8 mm), Thokotaman, R. Harry, 12 July 1954. IFALUK ATOLL: CAS 48819 (174 mm), Falarik Islet, R. Harry, 26 Sep. 1953. GUAM: CAS 48820 (214.4 mm), N of Cocos Is., Nangauta and H. Fehlmann, 8 Oct. 1958. ENEWETAK ATOLL: CAS 42377 (144 mm), Runit I., R. Nolan and L. Taylor, Jr., 23 Feb. 1974. BPBM 8184 (127 mm), Enewetek I., J. Randall, 1 Dec. 1967. BPBM 22339 (2; 219-233 mm), Enjebi I., J. Ran- dall et al., 27 Apr. 1978. BIKINI ATOLL: BPBM 12354 (310 mm), Eman I., V. Brock et al., 18 June 1947. LINE Is.: CAS 48825 (302 mm), Palmyra I., E. Herald et al., 16 Aug. 1951. BPBM 7715 (2; 310-393 mm), Cooper I., J. Randall, 13 Nov. 1968. HONG KONG: CAS 48821 (3; 79-88 mm), Santa Cruz Is., Vanikoro I., R. Bolin, 30 Sep. 1958. INDONESIA: BPBM 20890 (2; 103-383 mm), Bali, Sanur Beach, J. Randall, 18 July 1977. THAILAND: BPBM 22827 (460 mm), Similan I., Ko Miang, J. Randall, 14 Feb. 1979. PHILIPPINES: CAS 48824 (2; 508-518.5 mm), Negros Oriental, D. Empero, 28 July 1958. DIAGNOSIS. — A moderate-length species of Gymnothorax with anus before midbody; tubu- lar anterior nostrils; uniserial jaw and vomerine teeth; and cream body coloration with 17-21 dis- tinctive brown bands encircling head and body and extending onto fins. DESCRIPTION OF HOLOTYPE (followed paren- thetically by mean and range of the condition of holotype and nine paratypes). — Greatest depth McCOSKER & RANDALL: INDIAN OCEAN EELS 19 FIGURE 2. Gymnothorax rueppelliae, BPBM 18412, 339 mm TL, from Enewetak. of body 16.8 (19.0; 15.4-22.7) times in total length (TL). Tail longer than body, its length 1.76 (1.76; 1.71-1.82) in TL. Head 7.94 (7.69; 7.19-8.19) and trunk 3.26 (3.30; 3.19-3.55) in TL. Dorsal fin low, its origin ahead of gill open- ings, arising above fourth vertebra. Snout 6.37 (5.76; 5.29-6.37), upper jaw 3.01 (2.78; 2.65-3.01) times in head length (HL). Eye 9.5 (9.4; 8. 3- 10.4) in HL and 1.5(1.63; 1.4-1.9) in snout, closer to rictus than to tip of snout. Fleshy in- terorbital width 7.8 (8.4; 7.7-9.9) in HL. Gill openings nearly horizontal, their centers slightly below midbody, their length about equal to di- ameter of eye. Anterior nostril tubular, elongate, slightly less than eye diameter in length. Posterior nostril a hole above eye, beginning in a line with eye. Jaws subequal, the mouth closing completely. Teeth in jaws uniserial, stout, pointed and slight- ly retrorse. Six pairs of intermaxillary canines form a U-shaped margin around three central canines, the third the largest. Approximately six uniserial, small vomerine teeth. About 12 upper jaw teeth pairs, 18 lower jaw pairs; 3 pairs of depressible canines behind mandibular sym- physis. Number of vertebrae 130 (129.7; 128-131), 50.5 (50.8; 50-51.5) before anal fin. First dorsal pterygiophore arises above fourth vertebra. Head pores present but not obvious. A single pore anterior and proximal to, and a second pore below base of anterior nostril. Six pores along the mandible, the second through fifth the larg- est. Four equally spaced pores along upper jaw, the first beneath nostril base, the last beneath rear of eye. A single pore between anterior and posterior nostrils. Color in isopropyl alcohol cream, overlain with 17-21 distinctive brown bands which com- pletely encircle head and body and extend onto 22 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 2 FIGURE 5. Uropterygius xanthopterus , CAS 35254, 245 mm TL, from Kapingamaringi, displaying white cephalic puncta- tions. Arrows indicate the location of the anterior lateral line pores. the Red Sea. Its synonyms also include Mu- raena interrupta Kaup, 1856; Sideria chlevastes Jordan and Gilbert, 1883; Gymnothorax signifer Bliss, 1883; G. leucacme Jenkins, 1904; and G. waialuae Snyder, 1904. Whereas previous literature has suggested that most morays are nocturnal, it now appears that many, and possibly the majority of, moray species are diurnal but rarely observed due to their secretive nature (fide Chave and Randall 1971; Hobson 1974). On the basis of material collected and our observations while diving, we presume that G. enigmaticus is a shallow-water, nocturnally active piscivore (Fig. 4). It is note- worthy that G. rueppelliae is also a nocturnal predator (Hobson 1974, as G. petelli), as is G. undulatus (our observations), both of which are also strongly banded species. Uropterygius xanthopterus Bleeker, 1859 Uropterygius xanthopterus Bleeker, 1859, has had a sketchy taxonomic history. We have lo- cated the type-specimen, recognize it as a valid species, and include U. alboguttatus Smith, 1962, in its synonymy. Weber and de Beaufort (1916:397), without comment, included U. xanthopterus in the syn- onymy of Gymnomuraena marmorata La- cepede, 1803, a wide-ranging, elongate Indo-Pa- cific species of Uropterygius which possesses a single anterior lateral line pore and lacks white spotting on its head. Schultz (in Schultz et al. 1953:154) and Gosline (1958:226), on the basis of central Pacific specimens, recognized U. xan- thopterus as a distinct small species (the largest of 2 1 3 specimens from 76 CAS rotenone collec- tions in the Indian and central Pacific oceans which we examined was 345 mm) which pos- sesses two anterior lateral line pores and white cephalic punctations (Fig. 5). Smith (1962:427) again synonymized U. xanthopterus with U. marmoratus and described U. alboguttatus on the basis of Indian Ocean and Schultz' s central Pacific specimens. In describing U. kamar McCosker and Randall, 1977, we considered U. alboguttatus to be a possible synonym of U. xanthopterus. One of us (JEM) has subsequently examined the complete type-series of U. albo- guttatus and was unable to find differences in coloration, meristic features, or morphometry. The type-specimen of U. xanthopterus has not been clearly identified; however, through correspondence with Alwyne Wheeler, we have located the 275-mm specimen in the British Mu- seum (cat. no. 1867.11.28.271) received from Bleeker and labeled "Muraena xanthopterus."'' In that no specimen similar to Bleeker' s type exists in the Rijksmuseum (M. Boeseman, in litt.), we presume that this is the type, and the specimen which Bleeker illustrated and described in his Atlas (1864:pl. CLXIV, fig. 4). A radiograph of the British Museum speci- men clearly indicates that it is not U. marmo- ratus, a species which possesses obvious, large intramuscular bones. McCOSKER & RANDALL: INDIAN OCEAN EELS 23 FIGURE 6. Uropterygius marmoratus, BPBM 12336, 701 mm, Nuku Hiva, Marquesas. FAMILY OPHICHTHIDAE Ophichthus retifer Fowler, 1935 Fowler (1935) described and illustrated Oph- ichthus retifer on the basis of a 718-mm speci- men from Durban, Natal. Eugenie Bohlke has kindly examined the holotype (ANSP 63915) for us and compared it with a syntype (ANSP 26224) of O. erabo (Jordan and Snyder, 1901) from Ja- pan. They do not significantly differ in color- ation or proportions, yet there is a vertebral dif- ference. A radiograph of the holotype of O. retifer shows 143 vertebrae, with 73 before the anal opening. McCosker (1979) reported that six specimens of O. erabo from Japan, Hawaii, and Taiwan had 152-155 vertebrae (x = 154). Fow- ler (1935) suggested that O. retifer was "greatly like Microdonophis fowleri Jordan and Ever- mann 1903" (=O. erabo fide McCosker 1979) "and its synonym Ophicthys garretti Giinther 1910" (a valid species). We consider O. retifer to be conspecific with O. erabo, and account the vertebral difference to clinal variation. ACKNOWLEDGMENTS We thank the following individuals: Susan Middleton for photographic assistance; Michael Hearne for the preparation of the radiographs; M. Boeseman (Rijksmuseum van Natuurlijke Historic), Eugenie and James Bohlke (ANSP), William Fink (MCZ), W. Klausewitz (SMF), Margaret M. Smith (RUSI), and Alwyne Wheel- er (BMNH), for assistance with museum speci- mens and records; Lillian Dempster and W. I. Follett (CAS) for nomenclatural advice; and the 24 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 2 FIGURE 7. Adult Ophichthus erabo, from Jordan and Snyder (1901). curators and staffs of many museums for allow- ing us to examine specimens under their care. Randall's collections were made possible in part by grants from the National Geographic Society and the American Philosophical Society. A por- tion of McCosker's work was supported by funds from the Charline Breeden Foundation. LITERATURE CITED BLEEKER, P. 1859. Over eenige vischsoorten van de Zuid- kustwateren van Java. Nat. Tijdschr. Neder. -Indie 19:329- 352. . 1864. Atlas ichthyologique des Indes Orientales Neer- landaises. Vol. 4. Amsterdam. 132 p. BLISS, R. 1883. Descriptions of new species of Mauritian fish- es. Trans. Soc. Roy. Arts Sci., Maurice 13:45-63. BLOCH, M. E. 1795. Naturgeschichte der Auslandischen Fische. Vol. 9. CHAVE, E. H., AND H. A. RANDALL. 1971. Feeding behavior of the moray eel, Gymnothorax pictus. Copeia 1971 (3):570-574. FOWLER, H. W. 1935. South African fishes received from Mr. H. W. Bell-Marley in 1935. Proc. Acad. Nat. Sci., Phila- delphia 87:361^*08. . 1956. Fishes of the Red Sea and southern Arabia. 1. Branchiostomida to Polynemidae. Weizmann Sci. Press, Jerusalem. 240 p. GOSLINE, W. A. 1958. Central Pacific eels of the genus Urop- terygius, with the descriptions of two new species. Pac. Sci. 12(3):22 1-228. GUNTHER, A. 1910. Andrew Garrett's Fische der Siidsee, . . . Heft IX. J. Mus. Godeffroy, Hamburg 17:389-515. HOBSON, E. S. 1974. Feeding relationships of teleostean fish- es on coral reefs in Kona, Hawaii. Fish Bull., U.S. 72(4):915-1031. JENKINS, O. P. 1904. Report on collections of fishes made in the Hawaiian Islands, with descriptions of new species. U.S. Bur. Fish. Fish. Bull. 22:417-511. JORDAN, D. S., AND C. H. GILBERT. 1883. Description of a new muraenoid eel from the Galapagos Islands. Proc. U.S. Natl. Mus. 6:208-210. , AND J. O. SNYDER. 1901. A review of the apodal fishes or eels of Japan, with descriptions of 19 new species. Proc. U.S. Natl. Mus. 23:837-890. KAUP, J. 1856. Uebersicht der Aale. Arch. Naturges. 22(l):41-77. KLUNZINGER, C. B. 1871. Synopsis der Fische des Rothen Meeres. II. Theil. Verh. Zool.-Bot. Ges. Wien 21:441-668. MCCLELLAND, J. 1845. Apodal fishes of Bengal. J. Nat. Hist. Calcutta 5: 150-226. McCosKER, J. E. 1979. The snake eels (Pisces, Ophichthidae) of the Hawaiian Islands, with the descriptions of two new species. Proc. Calif. Acad. Sci., ser. 4, 42(2):57-67. , AND J. E. RANDALL. 1977. Three new species of Indo-Pacific moray eels (Pisces: Muraenidae). Proc. Calif. Acad. Sci., ser. 4, 41(3): 161-168. , AND R. H. ROSENBLATT. 1975. The moray eels (Pisces: Muraenidae) of the Galapagos Islands, with new records and synonymies of extralimital species. Proc. Calif. Acad. Sci., ser. 4, 40(13):4 17-427. RANDALL, J. E. 1973. Tahitian fish names and a preliminary checklist of the fishes of the Society Islands. Occ. Pap. Bernice P. Bishop Mus. 24(11): 167-214. RUPPELL, W. P. E. S. 1852. Verzeichniss der in dem Museum der Senckenbergischen . . . Fische und deren Skelette. Frankfurt-a-M. SCHULTZ, L. P., AND COLLABORATORS. 1953. Fishes of the Marshall and Marianas islands. Families from Asymme- trontidae through Siganidae. U.S. Natl. Mus. Bull. 202, 1. 685 p. SMITH, J. L. B. 1962. The moray eels of the western Indian Ocean and the Red Sea. Ichthyol. Bull. Rhodes Univ. 23:421-444. SNYDER, J. O. 1904. A catalogue of the shore fishes collected by the steamer "Albatross" about the Hawaiian Islands in 1902. U.S. Bur. Fish. Fish. Bull. 22:513-538. WEBER, M., AND L. F. DE BEAUFORT. 1916. The fishes of the Indo-Australian Archipelago. Vol. 3. Leiden. 455 p. PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES Vol. 43, No. 3, pp. 27-42 June 15, 1982 NEW SPECIES OF NORTH AMERICAN TACHYSPHEX WASPS (HYMENOPTERA, SPHECIDAE) By Wojciech J. Pulawski California Academy of Sciences, Golden Gate Park, San Francisco, California 94118 ABSTRACT: The following new North American species of Tachysphex are described (type-localities are given in parentheses): acanthophorus (Arizona: Willcox), apricus (California: Borrego Valley), arizonac (Arizona: 2 mi. NE Portal), armatus (Nevada: Sandy), bohartorum (California: Boca), idiotrichus (Arizona: 5 mi. W Portal), irregularis (California: Hallelujah Junction), krombeiniellus (Florida: Levy County), lamellatus (Mexico: Sonora: Alamos), menkei (California: Borrego Valley), mirandus (California: Palm Springs), musciventris (California: Borrego), occidentals (California: 12 mi. E Lone Pine), papago (Arizona: Nogales), Solaris (California: Borrego Valley), spatulifer (California: Arroyo Seco Camp), verticalis (California: 9 mi. W Beaumont), yuma (Mexico: Baja California: La Paz), and vo/o (California: Davis). INTRODUCTION For several years, I have been working on a monographic revision of North American Tachysphex. Because of the size of this under- taking, it will be some time before it is finished. Therefore, I am describing some of the new species now so their names will be available to those persons working on Tachysphex behavior. Furthermore, many hundreds of paratypes have been deposited in 34 collections in the USA and abroad, and it is desirable to validate these manuscript names now to avoid their possible use as nomina nuda in the works of others. The descriptions given below are restricted to those features which enable unambiguous recognition of each species. More complete characteriza- tions will be given when my revision is pub- lished. The terminology used below is based mainly on Bohart and Menke (1976). A few terms which need clarifications are the following: clypeus: the clypeus has a midsection and two lateral sections; the midsection usually has a densely punctate, setose basomedian area, a sparsely punctate shiny bevel, and a marginal lip. scutum: this term is used here for brevity's sake instead of mesoscutum. tergum, sternum: short terms for gastral tergum, gastral sternum. Many collectors are cited numerous times in the lists of material examined. Their names have been abbreviated to initials, as follows: ASM, A. S. Menke; BV, B. Villegas; DRM, D. R. Miller; EEC, E. E. Grissell; EIS, E. I. Schlin- ger; GEB, G. E. Bohart; GDB, G. D. Butler; FDP, F. D. Parker; FGW, F. G. Werner; FXW, F. X. Williams; HKC, H. K. Court; JCH, J. C. Hall; JAP, J. A. Powell; JMD, J. M. Davidson; JWMS, J. W. MacSwain; LAS, L. A. Stange; MAC, M. A. Cazier; MEI, M. E. Irwin; MSW, [27] 28 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 3 M. S. Wasbauer; PDH, P. D. Hurd; PMM, P. M. Marsh; PFT, P. F. Torchio; RGB, R. C. Bechtel; RMB, R. M. Bohart; ROS, R. O. Schuster; RRD, R. R. Dreisbach; TG, Terry Griswold; WJP, W. J. Pulawski. In the geographic names below, the following words have been abbreviated: County, Co.; Creek, Cr.; Highway, Hwy.; Island, I.; miles, mi.; Mountain(s), Mt(s).; River, R.; Station, Sta. The name Lower California has been used for the peninsula rather than Baja California, be- cause the latter may refer either to the peninsula or to a state in Mexico. Altitudes and distances are given as they appear on the original labels — in feet and miles. Multiplying the distances in miles by 1.609 and the elevations in feet by 0.3048 will convert them into kilometers and meters, respectively. SOURCES OF MATERIAL AND ACKNOWLEDGMENTS The specimens described in this paper came from institutional and private collections listed below. The initials preceding the names are the abbreviations by which institutions or private collections are referred to in the text. AMNH: American Museum of Natural History, New York, New York (M. Favreau) ANSP: Academy of Natural Sciences of Philadelphia, Phila- delphia, Pennsylvania (D. C. Rentz) ASU: Arizona State University, Department of Zoology, Tempe, Arizona (F. F. Hasbrouck) BMNH: British Museum (Natural History), London, England (C. R. Vardy) CAS: California Academy of Sciences, San Francisco, Cali- fornia (P. H. Arnaud, Jr., T. J. Zavortink, W. J. Pulawski) CIS: California Insect Survey, Division of Entomology, Uni- versity of California, Berkeley, California (H. Daly) CNC: Canadian National Collections of Insects, Arachnids and Nematodes, Biosystematics Research Institute, Otta- wa, Ontario (J. R. Barron) CSDA: California Department of Food and Agriculture, Sac- ramento, California (M. S. Wasbauer) CSU: Colorado State University, Department of Zoology and Entomology, Fort Collins, Colorado (H. E. Evans) CU: Cornell University, Department of Entomology and Lim- nology, Ithaca, New York (L. L. Pechuman) FSCA: Florida State Collections of Arthropods, Gainesville, Florida (E. E. Grissell) HKT: H. K. Townes, American Entomological Institute, Ann Arbor, Michigan INHS: Illinois State Natural History Survey, Urbana, Illinois (W. E. LaBerge) KU: University of Kansas, Snow Entomological Museum, Lawrence, Kansas (G. W. Byers) KVK: K. V. Krombein, Arlington, Virginia (private collec- tion), now in USNM LACM: Natural History Museum of Los Angeles County, Los Angeles, California (R. R. Snelling) MCZ: Museum of Comparative Zoology at Harvard Univer- sity, Cambridge, Massachusetts (J. Lawrence, J. C. Scott, M. K. Thayer) MPM: Milwaukee Public Museum, Milwaukee, Wisconsin (J. K. Lawton) NYSU: New York State University, College of Environmen- tal Sciences and Forestry, Department of Environmental and Forest Biology, Syracuse, New York (F. E. Kurczews- ki) OSDA: State of Oregon Department of Agriculture, Salem, Oregon (R. L. Westcott) OSU: Oregon State University, Department of Entomology, Corvallis, Oregon (P. Oman, G. R. Ferguson) TG: Terry Griswold, % Bee Biology & Systematics Labora- tory, Utah State University, Logan, Utah (private collec- tion) UAE: University of Alberta, Department of Zoology, Ed- monton, Alberta (A. L. Steiner) UAT: University of Arizona, Department of Entomology, Tucson, Arizona (F. G. Werner) UCD: University of California, Davis, Department of Ento- mology, Davis, California (R. M. Bohart, R. O. Schuster) UCR: University of California, Riverside, Department of Bi- ological Control, Riverside, California (S. Frommer) UFG: University of Florida, Department of Entomology and Nematology, Gainesville, Florida (B. Saffer) UGA: University of Georgia, Department of Entomology, Athens, Georgia (R. W. Matthews, C. L. Smith) UIM: University of Idaho, Department of Entomology, Mos- cow, Idaho (W. F. Barr) UMSP: University of Minnesota, Department of Entomology and Zoology, St. Paul, Minnesota (P. J. Clausen) USNM: United States National Museum of Natural History (Smithsonian Institution), Washington, D.C. (A. S. Menke, K. V. Krombein) USU: Utah State University, Department of Zoology, Logan, Utah (G. E. Bohart, F. D. Parker, Terry Griswold) WJP: Wojciech J. Pulawski, % California Academy of Sci- ences, San Francisco, California (private collection) WSU: Washington State University, Department of Entomol- ogy, Pullman, Washington (M. T. James, R. Zack) I express my sincere thanks to the curators and other persons who kindly submitted speci- mens for study. I feel especially indebted to R. M. Bohart, A. S. Menke, K. V. Krombein, and F. F. Kurczewski who helped me in many ways. SPECIES GROUPS Sixteen species groups are recognized in Tachysphex (see Pulawski 1971, 1974, 1977), but only four of them are represented in North America. They are: the pompiliformis, termi- natus, brullii, and julliani groups. The species described in this paper belong to the pompili- formis and brullii groups which are defined as follows: The pompiliformis group lacks peculiarities which characterize other groups and thus pos- sibly is a heterogenous assemblage of conve- nience. The propodeal hindface in this group is inclined, the female pygidial plate is not broad- NEW SPECIES OF NORTH AMERICAN TACHYSPHEX 29 ened and without peculiar microsculpture, the preapical bristles on the female gastral segments are not thickened, and the male sterna are pru- inose (except in mirandus). By comparison, in the julliani group the propodeal hindface is ver- tical or nearly so, male sterna are glabrous or sparsely pruinose, and in the females of most species the preapical bristles of gastral segments IV and V are thickened, and the pygidial plate is broadened or has a peculiar microsculpture. The vertex is simple in the pompiliformis group, while in the terminatus group a swelling is pres- ent behind each hindocellus. Unlike the brullii group, the apical female tarsomeres are simple (see that group for details). The pompiliformis group is cosmopolitan. Its species prey upon acridid nymphs, but the Palearctic species ful- vi tars is collects tettigonids. The following new species are members of the pompiliformis group: apricus, arizonac, bohartorum, idiotri- chus, irregularis, lamellatus, mirandus, musci- ventris, occidentalis, papago, Solaris, spatuli- fer, verticalis, yolo, and yuma. The brullii group is characterized by the pe- culiar apical female tarsomeres: dorsum convex, apicoventral margin produced into a lobe or at least convex, and vertex variously modified (covered with erect setae except glabrous ba- sally, or angulate basally in lateral view, or densely spinose). In other groups the dorsum is scarcely convex, the apicoventral margin is straight or nearly so, and the venter is evenly covered with setae which are usually inclined (but erect in verticalis), and it may have a few spines in some species. Furthermore, the pro- podeal dorsum setae are erect or inclined back- wards in most species of the brullii group, but only laterally so in acanthophorus, alayoi, ar- matus, many individuals of mundus, and some Australian species. Setae are inclined obliquely cephalad in the Australian species brevicornis and in most species of other groups. The brullii group is widespread throughout all zoogeo- graphic regions. Some species prey upon tetti- gonids, while others are blattid collectors. The following new species are members of this group: acanthophorus, armatus, krombeiniel- lus, and menkei. SPECIES OF THE POMPILIFORMIS GROUP Tachysphex apricus sp.n. ETYMOLOGY. — The specific name apricus is a Latin word meaning exposed to the sun. DIAGNOSIS. — Tachysphex apricus differs from other species of the pompiliformis group by the setal pattern of its propodeal dorsum: median setae are inclined cephalad, but the lateral setae are directed obliquely backwards and join api- comesally. Some species of the brullii group (e.g., acanthophorus) have an identical pattern, but the unspecialized apical female tarsomere of apricus is distinctive. The male of apricus can be distinguished by the compressed femoral notch whose glabrous bottom forms an obtuse, longitudinal crest. T. idiotrichus has a similar crest, but unlike that species the body vestiture is short in apricus. Unlike most species of the pompiliformis group, the propodeal side of apri- cus is alutaceous, shiny, impunctate or minutely punctate. GEOGRAPHIC DISTRIBUTION. — Xeric areas between southern Texas, southern Nevada, and southern California, and also Lower California. MATERIAL EXAMINED. — HOLOTYPE: d, California, San Diego Co., Borrego Valley, 3 May 1956, P. D. Kurd (UCD). PARATYPES: 38 9 , 60 6 , 31 Mar. to 3 July, 10 and 31 Aug., 9 Sep. Specimens for which institution is not indicated below are all in UCD. UNITED STATES OF AMERICA Arizona. Cochise: 6 mi. N Apache, collector unknown (1 d, NYSU). Coconino: 4.5 mi. E Moenkopi, JMD & MAC (1 9, ASU). Maricopa: 10 mi. E Gila Bend, GDB (2 d); 3 mi. sw Wickenburg, PFT & GEB (Id, USU). Mohave: 4 mi. w Chloride, PFT, GEB, FDP (1 d, USU); 8 mi. E Mesquite (Nevada), FDP & PFT (19, USU). Pfma: Organ Pipe Cactus National Monument, J. L. Sperry (19,1 d); Tucson, W. Benedict (19, NYSU), Bryant (1 9 , 1 d , CAS), FDP, LAS (1 9, 2 d; 1 9, WJP). Final: w Stanfield, GDB & FGW (1 9). Yavapai: Bloody Basin, collector unknown (1 9); 10 mi. NW Congress, FDP & LAS (19). California. Imperial: Glamis, RMB (1 9), FDP(1 d); Palo Verde, ROS (19); Pinto Flat, FXW (Id, CAS). Inyo: An- telope Springs, HKC (19,5 d); Big Pine Cr., RMB (1 9), FDP (1 d); 2 and 5 mi. E Big Pine, EEG (19,1 d); Little Lake, BV (1 9); 3 mi. w Lone Pine, RMB (2 d); Tuttle Cr. (2 mi. sw Lone Pine), JAP (1 9 , CIS). Kern: Kernville, T. R. Haig (Id). Riverside: 18 mi. w Blythe, RMB (Id); 3.5 and 4 mi. s Palm Desert, MEI, S. Frommer & R. M. Worley (2 9 , UCR); San Andreas Canyon, RMB (Id); Shavers Summit, MSW (1 9 , UCD); San Timoteo Canyon, MSW, R. McMaster (1 9, CSDA). San Bernardino: 1 mi. s Adelanto, MEI (1 d); Colton Hills, TG (2 9 , TG), Kramer, MSW (29,3d, CSDA); 3 mi. s Kramer Junction, MEI (1 9); sand dunes 7 mi. sw Kelso, MSW & J. S. Wasbauer (1 9 , CSDA), Mitchells Cav- erns, TG (1 9 , TG), 36 road mi. E Twenty nine Palms, TG (1 d , TG). San Diego: Borrego Valley, RMB (4 d ; 1 9 , 2 d , USNM), JCH (Id), PDH (4 d), G. A. Marsh (19), EIS (2 d; 2 d, WJP), MSW (Id, CIS), MSW, J. Slansky (19,1 d , CSDA), FXW (9 d , CAS); Scissors Crossing, J. C. Down- ey (1 9), H. & M. Townes (Id, HKT). Ventura: Sespe Can- yon, R. W. Sporne (Id). Nevada. Clark: Jean, GEB (3 d); 30 mi. s Searchlight, PFT, R. Rust, Youssef (1 d, USU). 30 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 3 New Mexico. Dona Ana: Las Cruces, RMB (19, UCD). Otero: Alamogordo, collector unknown (19); Alamo Canyon near Alamogordo, MEI (Id, UCR). Texas. Brewster: Big Bend National Park (Nine Point Draw), W. R. M. Mason (19, CNC). MEXICO Lower California. San Angel, N. Leppla, JMD, J. Bigelow, M. Bentzien, W. Fox, S. William, MAC (19, ASU); 16 mi. N Puertocitos, MEI (19, UCR). Tachysphex arizonac sp.n. ETYMOLOGY. — The specific name arizonac is an Indian word meaning little spring. DIAGNOSIS. — Tachysphex arizonac is charac- terized by the well-defined mesopleural punc- tures, median scutal setae transversely oriented, and sternum I with a horizontal depression api- cally. The females of arizonac and lamellatus have a peculiar clypeus whose free margin is undulate; they can scarcely be distinguished from each other. A useful character is the hy- postomal carina which is low in arizonac, but high in many lamellatus. Furthermore, in some arizonac the middle projection of the clypeal lip is markedly larger than the sublateral one (pro- jections about equal in lamellatus). The male of arizonac has a peculiar clypeus: bevel semilu- nate, lobe forecorner prominent (acutely in some specimens), lip usually with obtuse pro- jection mesally. The clypeus is somewhat simi- lar in texanus, but in that species setae are ap- pressed or nearly so beneath the mesopleural scrobe (setae suberect in arizonac). GEOGRAPHIC DISTRIBUTION. — Southern Utah, Arizona, and adjacent areas of California; So- nora State in Mexico. MATERIAL EXAMINED. — HOLOTYPE: » i i- +, .1 • tne world. A generic revision. University of Ca hfornia National Park (Nine Point Draw), R. Mason, J. F. McAlpme /-, n -T » /-vr^ HUTO r>- n j XT .- i ™ i /o .1-1 Press, Berkeley, Los Angeles, London. 1 color pi., ix + (3 9 , 7 n/-™ E-I n PuLAWSKi, W. J. 1971 . Les Tachysphex Kohl (Hym., Spheci- (near Doughout Well), B. J. Adelson (2 1 1 .5 ON to ^ £ -r :=: c * s £ d ~~ U 1 i ^ •••« "^35;'^: o Q "15 O -5; GJ -s: ft, -c •*' 'C 5- ^ O JC •i; « S & vicina vandenbl it i £P S .g |-|| Chatham 1 1 •§ a t T3 C t — , eg c^ ^ 2 | a s t t a t a §• :g | 1 1 Wermut Mert< Os 1 1 1 •§ R •« O « ft. O -5: -^ -s; ft. ^ s si -s: -s: ?• ?• -Si -Si 1 K -s: -I 1 -H J "a & i 5 a !« « 1 S TS s: S) a eg ."2 ^ 2 § 3 g c B E n C ^ ^. t 1 1 1 ^. ~s k iff <5 S.b S 5 microphyes elephantopi elephantopi nigra microphyes « •~ «3 .bo ,&c S S <3 S e O "S 3 C "8 I s J TJ. £ C -i : 1 c r-" o Os a .§ 1 1 "1 1 3 t 'S 2 -S inctive ut not na It's s .§ 41 thamensi. 1 -5 S t a -c irtoises. lever occi n O Q Ij o -s: «u -s: ft. -o •i * 5 So •~ •3 « §• s a -§ -§ a & su o *- « -i1 O X> oo eg O. O .Co eg C T3 rs •/"> '^ £5 "i n — C Os Co 2 :~ E c Clj C <1< a s 0 C — O ON 1 - xS ^£ Q ^J •« -0 -s: 'c ft, su 2 -s: •~ 1 5 & vicina 1 1 '5 1-1^ 1 11 Chatham, .5 -c ^ ^ a o "<3 ft. ! §> nomy of se tortois 0 3 •a T3 X eg eg o tu t- E E C -o » 2 X c OO 1 :5 si la « % 1 1| 1 1 eg .X C 5 •S c I !S c 00 1 .S | 0 C -£ 3 •« i_ .£ 3 1 §;» ^s 'n at 5 .c s 5 Q £ "^ •3 "I ° 1 "° t cd ^J 0 ll Q •O T3 -O *O T3 "O o o tu o w .£2 TO GALAP/ t. g e a •£ -^ -^ -^ 000 X! X) XI _CU _U _U 13 -a -o •O T3 T3 Jo J? ^ eg eg •3 "3 £ £ Intermedia Domed •ej "ej .2 0> cu C Intermedia 2 •3 1 I 1 1 Saddleback of revising shown tha Q l/l -— * u y— ^ y* &o j >^ ^_^ ^-^ Ci ^* CL a, tr- ^ — iJJ i o o (U I.S iple NAMES i name "ob c "1^1 S Js S III' X Z. < Z. t- di CO & > £ 0 X a° | (Villamil S. Albemarl U u Abingdon Duncan Jervis Chatham .0 c/5 ,cg o> **- £v u •0 eg c Charles 55) is in the \ •s have conv u •o C o K c on •2 s: 3 a f- 'c a eg f 1 eg 1 §1 C CB si-"2 n Cristoba O ' — ^ •0 <= N l.i? a 1 s u C^ eg « c «£. c nta Maria (Floreana) i- f VI U .2 u £ H * * * t/3 m " PJ u. J2 cu '£ i eg C/5 J? 5 $ 114 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES. Vol. 43. No. 9 FIGURE 2. Measurements taken from Galapagos tortoise skulls (see Table 2 for explanations of abbreviations). G. hoodensis: CAS 8121, 8122. G. microphyes: CAS 8158. G. nigrita: CAS 8381, 8289, 8286, 8385; MVZ 67613-67615, 59528, 67624-67629, 67631- 67633; USNM 104330-104331. G. phantastica: CAS 8101. G. vandenburghi: CAS 8141. G.vicina: CAS 8 179, 8 193, 8 177; USNM 129247. G. wallacei (probably an invalid form, fide MacFarland et al. 1974a): CAS 8134. Geochelone sp. (but definitely Galapagos tortoise species): AMNH 7288, 42961, 63415, 36420, 36568-36570, 63416; CAS 8298, 8404, 841 1, 8409, 8402, 8377, 8407, 8410, 8403, 8414, 8397, 84 1 2, 8272; Calif. State Univ., Fullerton Coll. 3 uncat.; FMNH 13523, 1 uncat.; LACM (Vert. Paleo.) pr 63, pr 58, pr 64; MCZ 46606, 11070, 11069, 32098, 1905, 4668; MVZ 80075; SDSNH 56605, 55458; USNM 65896, 102904, 129393, 15192, 29338, 29305, 29254, 29252, 15190, 15193,29256. Means and standard deviations were calculat- ed for each of the 1 6 measurements and corre- lation coefficients were also calculated. At the recommendation of Fritts (pers. comm.), I followed the last thorough taxonomic review (Van Denburgh 1 9 1 4) in which the different forms were given species-level designations. The species names used by Van Denburgh (1914) are fol- lowed with one exception; G. porteri is consid- ered a junior synonym of G. nigrita (fide Fritts in press). Statistical comparisons between island TABLE 2. SKULL MEASUREMENTS RECORDED FOR GALAP- AGOS TORTOISES. (All measurements taken with dial calipers and recorded to nearest 0.01 mm.) Variable— Description B— Basicranial length WAT— Width of skull at anterior tympanic opening WO— Width between orbits HN — Height of external narial opening WN— Width of external narial opening LB— Length of basisphenoid WB— Width of basisphenoid WZ— Width of quadratojugal WP- Width of postorbital WS-Width of jugal DPV — Distance (greatest) from prepalatine foramina (or fo- ramen, if only one present) to vomer LP— Length of prootic WFS — Width of prootic at stapedial foramen PW— Width of pterygoid waist APW — Width of anterior premaxillae PC— Length of sagittal contact of prefrontals CRUMLY: TORTOISE SKULLS 115 TABLE 3. MEANS AND STANDARD DEVIATIONS FOR 16 VARIABLES IN FIVE GALAPAGOS TORTOISE SPECIES. Measurements are illustrated in Figure 2 and abbreviations are listed in Table 2. Most sample sizes are small; all measurements are in millimeters. G. ephippium (N = 9) G. guntheri (N= 15) G. nigrita (N=18) G. vicina (N = 4) G. chathamensis (N = 6) Variable X SD X SD X SD X SD X SD B 96.7 11.4 128.0 21.4 121.5 39.2 109.0 49.2 98.1 27.3 WAT 73.9 9.2 106.6 19.9 98.4 31.8 86.0 38.9 80.4 25.0 WO 25.1 3.1 35.4 7.3 37.0 13.3 28.4 12.6 28.4 7.7 HN 12.5 2.0 18.6 3.4 18.6 6.0 16.1 7.3 13.9 4.1 WN 17.0 2.2 25.1 4.4 23.1 7.7 21.3 9.2 18.5 4.7 LB 13.3 3.2 18.7 4.2 14.7 4.3 18.1 8.9 14.7 5.6 WB 14.6 2.2 19.1 4.3 17.1 4.6 15.8 7.2 13.8 3.4 WZ 9.3 3.6 14.3 4.4 13.3 4.7 12.6 6.8 10.1 3.5 WP 7.0 2.2 9.5 2.5 9.0 3.5 8.8 4.5 7.3 3.0 WS 7.3 2.3 12.0 3.0 9.5 4.0 9.6 5.1 7.9 3.1 DPV 3.2 0.8 4.2 0.8 4.2 1.5 3.7 2.4 3.1 1.0 LP 14.1 2.1 21.0 5.6 18.1 6.5 14.8 6.6 15.2 3.7 WFS 10.0 2.0 16.0 6.1 12.8 6.1 8.9 6.3 12.5 6.2 PW 19.2 1.8 25.9 5.1 26.1 8.4 21.9 8.6 19.0 4.0 APW 10.5 1.8 15.2 2.4 14.1 4.9 11.8 8.0 10.5 3.0 PC 8.6 1.5 10.6 4.0 13.3 5.4 8.5 3.9 8.2 3.9 populations were hampered by incomplete lo- cality data; 50 of 1 1 6 specimens (43%) examined possessed doubtful or unknown locality data. The specimens without locality data were readily identified as Galapagos tortoises, but could not be identified to species without locality data. These specimens were used in the computation of correlation coefficients and in factor analysis, but could not be used in other statistical proce- dures. To facilitate my analyses, populations were combined based on the shell types advocated by Van Denburgh (1914) and Fritts (in press). Thus, the saddlebacked forms (G. abingdonii [N = 2], G. phantastica [N = 1], G. becki [N = 1], G. hoodensis [N = 2], and G. ephippium [N = 9]) were combined, yielding a sample of 1 5 individ- uals. The non-saddlebacked forms (intermediate and domed shells of Van Denburgh 1914) were also combined, forming a larger sample of 48 individuals (G. chathamensis [N = 6], G. dar- wini [N = 2], G. guntheri [N = 15], G. micro- phyes [N = 1], G. nigrita [N = 18], G. vicina [N = 4], and G. vandenburghi [N = 1]). These larger samples were then compared to determine whether cranial variation mirrored the already well known shell variation. Comparisons were also made among G. ephippium, G. guntheri, and G. nigrita to determine whether noncombined and combined samples contained the same mag- nitude of variation. The Statistical Package for the Social Sciences (SPSS) was used on the WYLBUR facility at the Campus Computer Information Service (CCIS) at Rutgers— The State University for initial data examination. Final statistical analyses were ac- complished using SPSS programs available TABLE 4. MEANS AND STANDARD DEVIATIONS FOR SADDLE- BACKED TORTOISES REPRESENTED BY SPECIMENS OF FIVE SPECIES AND NON-SADDLEBACKED TORTOISES REPRESENTED BY SPECI- MENS OF SEVEN SPECIES. Measurements are illustrated in Figure 2 and abbreviations are listed in Table 2; all measurements are in millimeters. Saddlebacked (N= 15) Nonsaddlebacked (N = 48) Variables X SD X SD B 98.9 15.9 116.5 37.1 WAT 75.8 13.0 96.0 30.4 WO 26.4 4.6 32.9 11.2 HN 13.2 2.9 17.1 5.6 WN 17.5 2.4 22.6 7.2 LB 13.8 3.2 16.0 5.7 WB 14.5 2.0 17.1 5.5 WZ 9.5 3.4 13.1 5.3 WP 7.1 2.2 8.8 3.5 WS 7.3 2.3 9.9 3.9 DPV 2.9 1.2 3.9 1.4 LP 14.4 2.7 18.4 6.2 WFS 10.1 1.8 13.7 6.4 PW 20.1 2.9 24.6 7.8 APW 10.5 1.7 13.2 4.5 PC 7.9 1.4 10.2 5.2 116 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 9 TABLE 5. CORRELATION COEFFICIENTS BETWEEN ALL THE SKULL MEASUREMENTS ILLUSTRATED IN FIGURE 2 AND ABBREVIATED IN TABLE 2. All specimens measured are combined into a single sample. Nevertheless, all coefficients are significant to at least the P = 0.05 level. Variable B WAT WO HN WN LB WB WZ WP WS DPV LP WFS PW APW WAT .98 WO .94 .94 HN .96 .94 .93 WN .97 .96 .93 .95 LB .88 .86 .79 .84 .88 WB .92 .92 .87 .88 .92 .87 WZ .84 .86 .86 .81 .87 .78 .84 WP .87 .86 .88 .83 .87 .79 .80 .88 WS .91 .92 .87 .88 .91 .86 .89 .84 .84 DPV .69 .69 .69 .64 .72 .60 .68 .64 .66 .66 LP .95 .94 .91 .91 .94 .85 .91 .86 .84 .91 .63 WFS 90 89 87 86 89 84 .85 .78 .77 .87 55 91 PW 92 91 90 .90 .91 .78 .88 .85 .83 .83 .69 90 82 APW .93 .93 .89 .91 .94 .84 .89 .82 .83 .89 .68 .86 .82 85 PC 60 56 55 63 60 38 .49 43 48 48 49 55 50 54 59 through the Office of Computer Services (OCS) at the Smithsonian Institution. Simple descrip- tive statistics, linear regression, factor analysis, and stepwise discriminant analyses were used to summarize observed cranial variation. RESULTS Geochelone ephippium appears to have the smallest skull and G. guntheri the largest skull of Galapagos tortoises (Table 3), but when max- imum basicranial lengths (mean plus two stan- dard deviations) are compared, G. nigrita ap- pears to possess the largest skull (Bmax = 171 mm for G. guntheri, 200 mm for G. nigrita). The efficacy of this procedure is in some doubt since the Bmax for G. vicina exceeds that of G. nigrita, even though no skull of the former is anywhere near as large as the latter. This may be the prod- uct of a small sample size for G. vicina, repre- sented by only four specimens. The largest skulls in these samples are G. ephippium, 1 14.0 mm; G. guntheri, 157.7 mm; G. nigrita, 157.6 mm; and G. vicina, 142.7 mm. The G. nigrita sample includes the two smallest tortoises measured, which depresses the mean basicranial length and elevates the standard deviation. Combined samples clearly show a size differ- ential between saddlebacked and domed tortois- es; saddlebacked tortoises have smaller skulls. This is supported by all 16 variables (see Table 4). All correlation coefficients were significant to at least the P = 0.05 level (Table 5). Some vari- ables, however, did not correlate as highly with other variables. Examples include PC, DPV, and LB. Because intervariable correlation was so high, linear regression showed slight, if any, tendency toward curvilinearity. The intercepts for saddle- backed forms were lower than the intercepts for non-saddlebacked forms, reflecting the differ- ence in size between the two groups. Slopes, how- ever, were practically identical. As an example, linear equations relating WO to LB for saddle- backed and non-saddlebacked tortoises have slopes of 1.38 and 1.37, respectively, whereas intercepts are 7.92 and 1 1.74, respectively (r = 0.75 for saddlebacks and 0.69 for nonsaddle- backs, P < 0.005 for both). Factor analysis yielded three factors, the first accounted for almost 95% of the data variance (see Table 6). Before rotation all 16 variables correlated most highly with this first factor. Ro- tation simplifies vectors derived by the analysis procedure and is necessary because factor anal- ysis problems have more than one solution. There are two general rotation techniques: orthogonal and oblique. Orthogonal rotation solutions de- rive vectors along axes of data variation that are perpendicular to one another and thus uncorre- lated. Oblique techniques, on the other hand, do not require that vectors be orthogonal, so vectors can be correlated. Even after varimax rotation, an orthogonal technique that simplifies the col- umns of a factor matrix by maximizing factor- variable loadings, 12 of the 16 variables correlate most highly with factor one. Varimax rotation CRUMLY: TORTOISE SKULLS 117 C© 00, 0 Factor 2 O o -2- Factor 3 FIGURE 3. A plot of factor scores for factors two and three. Geochelone nigrita (solid circles), G. guntheri (cross-hatched circles) and G. ephippium (open circles). When factor scores for all tortoises are plotted there is a prominent trend from the lower-left to upper-right quadrant. Although this general trend for all tortoises is suggestive of a positive trend toward increased snout elongation with increased robustness (as illustrated by G. nigrita), the points for G. ephippium and G. guntheri show a negative relationship between robustness and snout elongation. was chosen because it maximizes the variation accounted for by the factor vectors without all the variables loading highly on the same factor, as occurs in quartimax rotation. Identifying vectors of data variation is spec- ulative; but it seems likely that factor one sum- marizes variation in size. Thus, 95% of the vari- ation in Galapagos tortoise skulls may be the result of variation in size. The other two factors are more difficult to interpret, partly because so little variation (only 5%) is summarized by these factors. Factor two summarizes variation in cra- TABLE 6. STATISTICS PRODUCED BY FACTOR ANALYSIS USING VARIMAX ROTATION. All specimens were included in this anal- ysis. Abbreviations used in the summarized factor matrix are listed in Table 2. Eigenvalues are measures of the relative importance of the factors. Factor 1 2 3 Eigenvalue 13.19 0.47 0.28 % Variation 94.6 3.4 2.0 Summarization of WFS 0.82 WP 0.63 PC 0.75 Factor Matrix LB 0.80 DPV 0.60 HN 0.51 LP 0.78 PW 0.53 WAT 0.76 WO 0.53 WS 0.76 WZ 0.72 B 0.76 WB 0.75 WN 0.73 HN 0.72 nial width and the width of skull arches, em- phasizing WO, WP, WZ, PW, and DPV. There- fore, factor two could be identified as some measure of robustness. Factor three, emphasiz- ing PC and HN, suggests there is variation in the anterior part of the skull. A high factor three score results from an increase in PC and HN. This results from elongating the anteromedial portion of the triturating surface, which concom- itantly yields a longer skull. A bivariate plot of the second and third factor scores for G. nigrita, G. guntheri, and G. ephip- pium (Fig. 3) indicates that as skulls become more robust, the anterior nasal part of the skull elon- gates; as robustness increases the skull becomes relatively longer. However, examining the indi- vidual points for G. guntheri and G. ephippium suggests just the opposite; as robustness increases elongation decreases. This negative relationship seems more pronounced in G. guntheri. Three separate discriminant function analyses were done: one for G. nigrita, G. ephippium, and G. guntheri; one for the combined samples; and one comparing small samples to larger samples. In the first analysis, the three forms were distin- guished by two factors (Table 7). Factor one sum- marized variation in 14 of the 16 variables but accounted for only 54.9% of the data variance. A high canonical correlation coefficient and a low Wilks's lambda indicate that this factor is good 118 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES. Vol. 43, No. 9 Jill im -6 -2 2ee 4 9 99 _2 9 a9 9 FIGURE 4. A plot of the discriminant scores derived from an analysis that included Geochelone guntheri, G. nigrita, and G. ephippium. Discriminant factor one is the horizontal axis and discriminant factor two is the vertical axis. Statistical sep- aration of these three populations is marked. High positive scores along the horizontal axis indicate small size, whereas high positive scores along the vertical axis indicate a poorly exposed basisphenoid. Upper case letters indicate group cen- teroids. at distinguishing groups. Geochelone ephippium is separated from the other two forms by this factor. The discriminating variables are nega- tively correlated to factor one, and the species with a small skull is differentiated from the two species with large skulls, suggesting that factor TABLE 7. STATISTICS PRODUCED BY A DISCRIMINANT ANALYSIS OF G. ephippium, G. guntheri AND G. nigrita. Ab- breviations are listed in Table 2. Eigenvalues are measures of the relative importance of the factors; high canonical corre- lation coefficients (near 1) and low Wilks's lambdas (near 0) indicate that factors are good discriminators. Discriminant function 1 2 Eigenvalue 7.06 5.82 % Variation 54.8 45.2 Canonical correlation 0.94 0.92 Wilks's lambda 0.02 (P = 0.003) 0.15(/>=0.02) Pooled within groups HN -0.22 LB -0.21 correlations bet. LP -0.20 WB -0.17 canonical WN -0.19 discr. fncts. & WAT-0.19 discr. variables APW-0.18 WO -0.18 Groups delineated G. ephippium G. nigrita from others from G. guntheri FIGURE 5. A histogram illustrating the results of a discrim- inant analysis of the saddlebacked and non-saddlebacked forms. The saddlebacked species are in the upper histogram, the non- saddlebacked species are in the lower histogram. The arrows indicate the median in each class. The discriminant scores (high positive scores indicate large size) are on the lower axis and the number of individuals are represented by left-hand axis. Although the saddlebacked and non-saddlebacked forms are clearly different sizes, there is significant overlap. one is an inverse measure of size. Factor two, which also has a high canonical correlation coef- ficient and a low Wilks's lambda, distinguishes G. nigrita from G. guntheri and accounts for the remaining variation in the data. Two variables are highly correlated with this second factor, LB and WB. Geochelone nigrita has high positive values for discriminating factor two, indicating that the basisphenoid is poorly exposed. Figure 4 graphically illustrates the completeness of sep- aration. Standardized canonical discriminant function coefficients are available upon request. These coefficients can be used to calculate discriminant scores for individual specimens whose identity is unknown; but choices are restricted to the pop- ulations originally compared (in this case G. ephippium, G. guntheri, or G. nigrita). The second discriminant analysis applied to the combined samples. Because only two groups were analyzed, a single discriminating factor was computed. The Wilks's lambda was not low, sug- gesting that the two groups cannot be easily dis- tinguished. The size differential between saddle- backed and non-saddlebacked tortoises is readily apparent (Fig. 5). Standardized canonical dis- criminant function coefficients are available upon request. The third discriminant analysis compared CRUMLY: TORTOISE SKULLS 19 TABLE 8. CLASSIFICATION RESULTS OF A DISCRIMINANT ANALYSIS CLASSIFICATION PROCEDURE. Individual specimens were classified to one of three species: G. ephippium (a saddlebacked species), G. guntheri (an intermediate form) or G. nigrita (a domed form). Asterisk indicates invalid taxon (fide MacFarland et al. 1974a). Shell type Trivial name Sex Mus. no. Classified as domed vandenburghi 9 CAS 8141 ephippium intermediate chathamensis 7 CAS 8 133 ephippium intermediate chathamensis 9 CAS 8131 ephippium intermediate chathamensis 9 USNM 29255 ephippium intermediate chathamensis male CAS 8 127 ephippium intermediate chathamensis 9 CAS 8 130 ephippium intermediate chathamensis ? CAS 8 128 ephippium intermediate darwini female CAS 8 106 ephippium intermediate darwini male CAS 8 108 guntheri intermediate microphyes male CAS 8 158 guntheri intermediate vicina male CAS 8 179 ephippium intermediate vicina female CAS 8 193 ephippium intermediate vicina 9 USNM 129247 ephippium intermediate vicina male CAS 8 177 guntheri intermediate wallacei* male CAS 8 134 guntheri saddlebacked abingdonii 9 USNM 29269 guntheri saddlebacked abingdonii male CAS 81 12 guntheri saddlebacked becki female CAS 8 120 ephippium saddlebacked hoodensis male CAS 8121 ephippium saddlebacked hoodensis female CAS 8 122 ephippium saddlebacked phantastica male CAS 8101 guntheri small samples of tortoise species to large sam- ples. Small samples were classified by the dis- criminant function classification procedure to one of three species (G. guntheri, G. ephippium, G. nigrita). The results of this procedure are sum- marized in Table 8. Some species with inter- mediate shell types (fide VanDenburgh 1914) were classified as saddlebacked species (e.g., G. chath- amensis was classified as G. ephippium), whereas other species with intermediate shell types were classified as G. guntheri, an intermediate form. No species was classified as a domed form. Skull variation did not parallel shell variation in any meaningful way. DISCUSSION Small sample sizes and the paucity of accurate locality data limit the utility of this study. There- fore, samples were combined. (Thorpe, 1976, discusses the ramifications of such procedures.) Because most of the specimens in the United States were examined, this limitation cannot be overcome without costly and time-consuming removal of skulls from skins and stuffed speci- mens of known provenance. The choice of a putative ancestral morphotype makes an enormous difference in how one in- terprets evolutionary processes, patterns, and mechanisms. The size of the ancestral Galapagos tortoise is not known. Auffenberg (1971) be- lieved that the fossil Geochelone hesterna was a likely ancestral candidate for Galapagos tortoises as well as Geochelone chilensis from Argentina. The skull of G. hesterna is very much like a Galapagos tortoise skull. Although it is not as large as that of the largest of Galapagos domed tortoises, it is larger than that of the small sad- dlebacked tortoises. Thus, I favor an interme- diate-sized ancestor for Galapagos tortoises, per- haps something smaller than G. guntheri. If so, then G. nigrita is the result of continued gigan- tism and G. ephippium is the result of dwarfism. Why is there such flimsy coincidence between shell variation and cranial variation in Galapa- gos tortoises? Zangerl and Johnson (1957) and Zangerl (1969) have intimated that much of the shell variation observed in most species has little effect on an individual's survival or fitness. Fritts (in press) has shown the contrary for Galapagos tortoises. But this selection on shell morphology does not seem to apply to skull morphology. What other selective factors could be molding skull morphology? I tend to agree with Bramble (1971), who felt that biomechanical constraints on chewing are the primary sources of selection upon turtle skulls. 120 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43. No. 9 How strong are these selective forces? Selective factors imposed by diet are known to be espe- cially strong in other animals from Galapagos. For example, Boag and Grant (1981) discovered that finches in Galapagos experienced intense se- lection upon beak size and shape as the result of a one-year drought. Because of the long life of tortoises and their ability to survive long periods without food or water, however, short-term en- vironmental changes such as those described by Boag and Grant are unlikely to affect tortoises as severely. Another source of selective pressure is possi- ble. During intraspecific agonistic encounters (Fritts, pers. comm.), the victor is the individual capable of raising its head the highest. Could apparent head width also affect the outcome of these battles? Interestingly, the relative head width of Geochelone guntheri and G. ephippium increases with size. These tortoises inhabit low dry islands (or parts of islands) where carrying capacities of the habitat may be lower and in- traspecific competition therefore higher. In con- trast, relative head width in G. nigrita decreases with size. This tortoise lives on a higher moist island where carrying capacities may be higher and intraspecific competition may not be as in- tense. Also, this apparent decrease in relative width actually accompanies an increase in the length of the masticatory surface area, perhaps allowing more efficient mastication. ACKNOWLEDGMENTS I thank L. Barnes (LACM), R. Crombie (USNM), R. Drewes (CAS), G. Foley (AMNH), T. Fritts (SDSNH), A. Leviton (CAS), H. Marx (FMNH), W. Presch, A. Resetar (FMNH), J. Ro- sado (MCZ), H. Voris (FMNH), D. Wake (MVZ), E. Williams (MCZ), G. Zug (USNM), and R. Zweifel (AMNH) for making specimens avail- able. I also thank R. Crombie, C. Ernst, T. Fritts, S. McDowell, E. Meyer, K. Miyata, D. Stead- man, and G. Zug for reading parts or all of the manuscript and providing helpful suggestions. Funding, for which I am most thankful, came from a Sigma Xi Grant-In-Aid of Research, the Alvarado Community Hospital Research Foun- dation, the Vertebrate Zoology Reserve Fund of San Diego State University, and the Theodore Roosevelt Memorial Fund of the American Mu- seum of Natural History. Data analysis, begun at Rutgers— The State University with Depart- ment of Zoology and Physiology support, was completed with the assistance of the Office of Computer Services during my tenure as a Smith- sonian Predoctoral Fellow at the National Mu- seum of Natural History. LITERATURE CITED ARNOLD. E. N. 1979. Indian Ocean giant tortoises: their sys- tematics and island adaptations. Phil. Trans. R. Soc. Lond. (8)286:127-45. AUFFENBERG, W. 1971. A new fossil tortoise, with remarks on the origin of South American testudinines. Copeia 1971: 106-17. BOAG, P. T., AND P. R. GRANT. 1981. Intense natural selec- tion in a population of Darwin's finches (Geospizinae) in the Galapagos. Science 214:82-85. BRAMBLE, D. M. 1971. Functional morphology, evolution, and paleoecology of gopher tortoises. Ph.D. thesis, Univ. Calif., Berkeley. 341 p. CRUMLY, C. R. 1980. The cranial osteology and evolution of the tortoise genus Geochelone (Testudines, Testudinidae). M.S. thesis, San Diego State Univ. 338 p. . 1982. A cladistic analysis of Geochelone using cranial osteology. J. Herpetol. 16(3):2 15-34. DUELLMAN, W. E., T. FRITTS, AND A. LEVITON. 1978. Mu- seum acronyms. Herp. Rev. 9(l):5-9. FRITTS, T. H. [in press]. Morphometrics of Galapagos tor- toises: evolutionary implications. GARMAN, S. 1917. The Galapagos tortoises. Mem. Mus. Comp. Zool. 30:262-96. GUNTHER, A. 1875. Description of the living and extinct races of gigantic tortoises. Parts I and II: The tortoises of the Galapagos Islands. Phil. Trans. R. Soc. Lond. (B) 165: 251-84. . 1877. The gigantic land tortoises (living and extinct) in the collection of the British Museum. British Museum. London. 96 p. MACFARLAND, C. G., J. VILLA, AND B. TORO. 1974a. The Galapagos giant tortoises (Geochelone elephantopus). Part I: The status of the surviving populations. Biol. Conserv. 6: 118-33. . 1974b. The Galapagos giant tortoises (Geochelone elephantopus). Part II: Conservation methods. Biol. Con- serv. 6:198-212. MARLOW, R., AND J. L. PATTON. 1981. Biochemical rela- tionships of the Galapagos tortoises (Geochelone elephan- topus). J. Zool., London 195:413-22. PRITCHARD, P. C. H. 1979. Encyclopedia of turtles. T. F. H. Publishers. 895 p. ROTHSCHILD, L. 1901. A new land tortoise from the Gala- pagos Islands. Nov. Zool. 8:372. . 1902. Description of a new species of gigantic land turtle from the Galapagos Islands. Nov. Zool. 9:619. . 1903. Description of a new species of gigantic tortoise from Indefatigable Island. Nov. Zool. 10:119. . 1915. The gigantic land tortoises of the Galapagos Islands in the Tring Museum. Nov. Zool. 22:403-17. THORPE, R. S. 1976. Biometrical analysis of geographic vari- ation and racial affinities. Biol. Rev. 51:407-52. VAN DENBURGH, J. 1907. Expedition of the California Acad- emy of Sciences to the Galapagos Islands, 1905-1906. Part CRUMLY: TORTOISE SKULLS 1 2 1 I: Preliminary descriptions of four new races of gigantic land — . 1977. Liste der rezenten Amphibien un Reptilien. tortoises from the Galapagos Islands. Proc. Calif. Acad. Sci. Testudines, Crocodylia, Rhyncocephalia. Das Tierrich 100: 1:1-16. 1-174. 1914. Expedition of the California Academy of Sci- ZANGERL, R. 1969. The turtle shell. Pages 31 1-339 in Gans, ences to the Galapagos Islands, 1905-1906. Part X: The C, A. d'A. Bellaris, and T. S. Parsons. Biology of the Rep- gigantic land tortoises of the Galapagos Archipelago. Proc. tilia, vol. 1 , Academic Press, New York. Calif. Acad. Sci., ser. 4, 2:203-374. — , AND R. G. JOHNSON. 1957. The nature of shield WERMUTH, H., AND R. MERTENS. 1961. Schildkroten, kro- abnormalities in the turtle shell. Fieldiana: Geol. 10:341- kodile, und bruckenechsen. Gustav Fischer Verlag, Jena. 62. 422 p. CALIFORNIA ACADEMY OF SCIENCES Golden Gate Park San Francisco, California 941 18 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES Vol. 43, No. 10, pp. 123-140, 8 figs., 1 table January 17, 1984 THE STATUS OF TRYPOXYLON FIGULUS (LINNAEUS, 1758), MEDIUM DE BEAUMONT, 1945, AND MINUS DE BEAUMONT, 1945 (HYMENOPTERA: SPHECIDAE) By Wojciech J. Pulawski California Academy of Sciences, Golden Gate Park, San Francisco, California 94118 ABSTRACT: Trypoxylon figulus (Linnaeus, 1758), medium de Beaumont, 1945, and minus de Beaumont, 1945, currently confused under the name figulus, are separated on the basis of newly discovered characters. /. majus Kohl, 1883, figulus barbarum de Beaumont, 1957, and figulus yezo Tsuneki, 1956, are newly synonymized with figulus, and figulus koma Tsuneki, 1956 is newly synonymized with minus. Neotypes are designated for Sphex fuliginosus Scopoli, 1763, and Trypoxylon majus Kohl, 1883, both synonyms of figulus, and a lectotype is designated for Trypoxylon rubi Wolf, 1959, a synonym of medium. INTRODUCTION De Beaumont (1945) was first to observe that Trypoxylon figulus of European authors actually consisted of three phena. Their status has been controversial over the years. De Beaumont ( 1 945, 1964a) and Richards (1980) called them vari- eties, Bliithgen (1951) gave them species rank, and Wolf (1959) and Bohart and Menke (1976) treated them as subspecies. The last interpreta- tion is untenable, since the three phena are large- ly sympatric. Tsuneki (1981) regarded medium as a good species, characterized by both external and genitalic characters, and considered minus as a simple form of figulus. According to Valkeila (1961), specimens reared from one nest mostly are one phenon, but he reported that two phena (e.g., majus and minus) are found in some nests. He concluded that all three are individual variants of one species. Un- fortunately, Valkeila's data cannot be verified. I have examined all of his specimens, which are presently kept at Helsinki University. Some specimens have identification labels by de Beau- mont, but not a single label refers to nests or cells from which specimens were reared. Possibly Valkeila misidentified some specimens, but this cannot be determined because his identification labels give the name figulus only, without ref- erence to form or varietal names. Another pos- sible explanation is that offspring of two nests were accidentally confused. A thorough examination of the three phena, based on more than 3800 specimens from many countries, convinced me that actually they are good species. My opinion is based on the follow- ing evidence: 1. Morphology. Although some males of fi- gulus and minus cannot be distinguished with certainty, females are separated by structural gaps and do not intergrade; also the male of medium is easily recognized by its peculiar gonoforceps. Some previously unnoticed characters (antero- [123] 124 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 10 ventral mesothoracic process, length of thoracic vestiture, female hindcoxal pit) are especially helpful in recognizing the three species. 2. Rearing. Of 9 1 specimens reared from trap nests in Liege, Belgium, by A. Jacob-Remacle, and examined by me, all are minus (7 2, 5 3 reared in 1976, 44 2, 35 $ reared in 1981). This result contradicts Valkeila's conclusions: If the three phena really are variants of one species, then one would expect some variation of phena in the Liege trap nest material. 3. Geographic distribution. All three species are largely sympatric, but only figulus and me- dium have been found in Great Britain, only figulus and minus in the Iberian Peninsula, and only figulus in North Africa and North America. If the three phena were just individual forms of one species, some variation should have oc- curred in North American populations, and all three phena should have been observed in Great Britain (all three do occur in the Netherlands, where climatic and ecological conditions are practically identical to those in England). 4. Habitat preference. Of 429 specimens col- lected by J. Leclercq in waterbowl traps in Liege, Belgium, in 1980 and 1981 (and examined by me), 4 are medium and the remainder are minus (202 2, 223 <5). Not a single figulus was trapped, in spite of the fact that this species is common in Belgium. I fully agree with Tsuneki (198 1) that some of the previously used characters are not reliable. For example, the mesopleural punctures of fi- gulus are dense and well denned according to de Beaumont (1945, 1964a), but in the smallest males they actually are as sparse and minute as in most minus. Such characters have not been used here. Proper mounting is critical for studying the species considered. For example, the anteroven- tral mesothoracic region must not be damaged by the pin. When pinning the specimens, one should insert the pin so that it passes through the membrane between the mesothorax and fore- coxae (which then extend laterad). In this posi- tion, the anteroventral mesothoracic region is easily visible. Unfortunately, many European collectors mount their specimens on cardboard rectangles with glue or minutiae (venter down rather than on a side), and such specimens must be relaxed and remounted before examination. In the text below the locality records are ar- ranged according to current administrative di- visions for each country except Sweden and Fin- land, for which biogeographic provinces have been used. Localities given on specimen labels but not found on available maps or in gazeteers have not been considered. An exclamation mark preceding the word Ho- lotype or Neotype in the bibliographic citations indicates that the type has been examined. SOURCES OF MATERIAL The specimens examined came from institu- tional and private collections listed below. The acronyms preceding the names are the abbrevi- ations by which these collections are referred to in the text. AKM: Aimo K. Merisuo, Turku, Finland AWE: Father Andreas W. Ebmer, Linz, Austria BB: Padre Bruno Bonelli, Cavalese, Italy BMNH: British Museum (Natural History), London, England (Mr. C. R. Vardy) CAS: California Academy of Sciences, San Francisco, Cali- fornia (W. J. Pulawski) CNC: Canada National Collection of Insects, Arachnids and Nematods, Biosystematics Research Institute, Ottawa, On- tario CU: Cornell University, Department of Entomology and Lim- nology, Ithaca, New York (Dr. L. L. Pechuman) DBB: Major Donald B. Baker, Ewell, Surrey, England DEI: Institut fur Pflanzenschutzforschung der Akademie der Landwirtschaftswissenschaften der DDR, Zweigstelle Ebers- walde, Abteilung Taxonomie der Insekten (formerly Deutsches Entomologisches Institut), Eberswalde-Finow (Dr. J. Oehlke) FIS: Forschungsinstitut Senckenberg, Frankfurt am Main, Fed- eral Republic of Germany (Dr. J.-P. Kopelke) FJS: Seftor Francisco J. Suarez, Almeria, Spain FSAG: Faculte de Sciences Agronomiques, Gembloux, Bel- gium (Dr. J. Leclercq) GP: Signer Guido Pagliano, Turin, Italy GVR: Mr. Gerard van Rossem, Wageningen, The Netherlands HD: Dr. Holger Dathe, Forschungsstelle fur Wirbeltierfor- schung, Berlin, German Democratic Republic HW: Herr Heinrich Wolf, Plettenberg, Federal Republic of Germany HY: Helsingin Yliopisto (=University of Helsinki), Depart- ment of Agricultural and Forest Zoology, Finland, including E. Valkeila collection (Dr. Martti Koponen) IEE: Institute Espanol de Entomologia, Madrid, Spain (Dr. E. Mingo Perez) JG: Dr. Joseph Gusenleitner, Linz, Austria KMG: Mr. Kenneth M. Guichard, % British Museum (Natural History), London, England K.S: Professor Dr. Konrad Schmidt, Zoologisches Institut der Universitat, Karlsruhe, Federal Republic of Germany KT: Professor Katsui Tsuneki, Mishima, Japan LEM: Lyman Entomological Museum & Research Laboratory, Ste. Anne de Bellevue, Quebec, Canada (Dr. A. Finnamore) MGA: Muzeul de Istorie Naturala Grigore Antipa, Bucharest, Rumania (Mrs. X. Scobiola Palade) PULAWSKI: THE STATUS OF TRYPOXYLON FIGULUS, MEDIUM, AND MINUS 125 TABLE 1. ANCESTRAL AND DERIVED CHARACTER STATES OF THREE SPECIES IN THE GENUS TRYPOXYLON. Character Ancestral Derived 1 . Thoracic pilosity 2. Antero ventral mesothoracic process 3. Free margin of female clypeus 4. Female hindcoxal pit 5. Sete of hindcoxal pit 6. Male apical flagellomere 7. Gonoforceps process shorter (as in medium) absent straight or sinuate circular evenly distributed shorter (as in medium) absent longer (as in figulus) present concave oblong channel-like structure longer (as in figulus) present MHNG: Museum d'Histoire Naturelle de Geneve, Switzerland (Dr. Cl. Besuchet) MSNM: Museo Civico di Storia Naturale, Milano, Italy (Dr. C. Leonardi) MCZ: Museum of Comparative Zoology, Harvard University, Cambridge, Massachusetts NHMB: Naturhistorisches Museum Bern, Bern, Switzerland (Dr. H. D. Volkart) NHMV: Naturhistorisches Museum, Vienna, Austria (Dr. M. Fischer) NRS: Naturhistoriska Riksmuseet, Stockholm V, Sweden (Mr. S. Erlandsson) RMNH: Rijksmuseum van Natuurlijke Historic, Leiden, The Netherlands, including collections of J. P. van Lith and P. M. F. Verhoeff (Dr. K. van Achterberg) SFG: Dr. Severiano Fernandez Gayubo, Departamento de Zoologia, Universidad de Salamanca, Spain SMT: Staatliches Museum fur Tierkunde, Dresden, German Democratic Republic (Dr. Regine Eck) TMB: Termeszettudomanyi Muzeum, Budapest, Hungary (Dr. J. Papp) TN: Mr. Toshiaku Nambu, Yorii-machi, Saitama Prefecture, Japan USNM: United States National Museum (Smithsonian Insti- tution), Washington, D.C. VH: Dr. Volk Haeseler, Universitat Oldenburg, Oldenburg, Federal Republic of Germany VLK: Dr. Vladimir L. Kazenas, Zoological Institute, Kazakh Academy of Sciences, Alma Ata, USSR WJP: Wojciech J. Pulawski, San Francisco, California WSU: Washington State University, Department of Ento- mology, Pullman, Washington ZMB: Museum fur Naturkunde an der Humboldt Universitat zu Berlin, German Democratic Republic (Dr. F. Koch) ZMH: Zoologisches Institut und Zoologisches Museum der Universitat Hamburg, Federal Republic of Germany (Dr. R. Abraham) ZMK: Zoological Museum, Copenhagen, Denmark (Dr. O. Lomholdt) ZMMU: Zoological Museum, Moscow State University, Mos- cow, USSR (Dr. L. V. Zimina, via Dr. A. P. Rasnitsyn) ZMUB: Zoological Museum, University of Bergen, Norway (Dr. Lita Greve Jensen) ZSM: Zoologische Staatssammlung Milnchen, Federal Repub- lic of Germany (Dr. E. Diller) ACKNOWLEDGMENTS Study of the Linnean type of Sphex figulus at the Burlington House, London, was possible ow- ing to the kind assistance of Michael C. Day of the British Museum (Natural History). North American specimens of figulus belonging to var- ious U.S. and Canadian institutions listed above (except for LEM) were kindly forwarded by Rolin C. Coville, University of California, Berkeley, who had them on loan. Arnold S. Menke and Eric E. Grissell critically reviewed the manu- script and made many valuable suggestions. Da- vid H. Kavanaugh commented on the phyloge- netic trees. Mary Ann Tenorio drew the phylogenetic schemes and distributional maps, and Donald J. Becker took the photographs with a Hitachi S-520 scanning electron microscope. PHYLOGENETIC RELATIONSHIPS Reconstructing phyletic relationships between three isolated species of a large genus like Try- poxylon is precarious, because polarities of mor- phological transformations can easily be misin- terpreted. With this restriction in mind, I nevertheless think it worthwhile to analyze the relationships between figulus, medium, and mi- nus. Their ancestral and derived character states, based on outgroup comparisons, are shown in Table 1. Based on the above table, the three possible phylogenetic trees (only dichotomic trees are considered) are as shown in Fig. 1 . Tree B is the most probable, since no single derived character state is shared by any two of the three species in the schemes A and C. Furthermore, trees A and C imply a parallel development of the elongate male flagellomere XI in figulus and minus, an unlikely event. KEY TO THE SPECIES 1 . Female: clypeal free margin evenly concave between orbit and median projection (Fig. 7A); mesopleural setae around scrobe shorter than midocellar diameter; hindcox- 126 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 10 FIGURE 1 . Hypothetical phylogenetic trees of Try poxy Ion figulus (fig), medium (med), et minus (min). Open circles: generalized character states. Solid circles: derived character states. Numbers refer to characters listed on p. 125. al pit nearly circular, its setae forming pe- culiar, channel-like structure (Fig. 7B-D). Male: maximum length of flagellomere X 0.75-0.9 times its width; maximum length of flagellomere XI generally 2.0-2.2 times its basal diameter (occasionally 2.4 times); gonoforceps with externoventral expansion at about midlength (Fig. 7E, F) medium de Beaumont - Female: clypeal free margin (Fig. 2A, 5A) sinuate to nearly straight, except concave in occasional western European figulus in which mesopleural setae around scrobe are generally longer than midocellar diameter, and hindcoxal pit is generally oblong; setae of hindcoxal pit not forming channel-like structure. Male: maximum length of fla- gellomere X 0.5-0.8 its width; maximum length of flagellomere XI 2.2-3.6 times its basal diameter; gonoforceps not expanded exteroventrally (Fig. 2E, F) 2 2. Mesothorax without anteroventral process1; mesopleural setae around scrobe in most specimens longer than midocellar diameter (shorter in some individuals). Female: clyp- eal free margin sinuate between lobe and orbit (Fig. 2A), hindcoxal pit oblong (Fig. 1 In occasional males the anteroventral mesothoracic process is absent and the mesopleural setae are shorter than the mid- ocellar diameter. Such specimens can be either figulus with unusually short mesopleural setae, or minus without meso- thoracic process. I cannot find characters for distinguishing them. 2B-D) or (some specimens) circular, eva- nescent in Japanese specimens figulus (Linnaeus) - Mesothorax with anteroventral process (Fig. 5B, C) in more than 95% of specimens; mesopleural setae around scrobe shorter than midocellar diameter. Female: clypeal free margin almost straight between lobe and orbit (Fig. 5A), hindcoxal pit circular (Fig. 5D-F) minus (de Beaumont) Trypoxylon figulus (Linnaeus) Sphex figulus LINNAEUS, 1758:570. ! Holotype: 9, Sweden, Uppsala (Linnean Society, London). — DAY, 1979:62. — In Trypoxylon: LATREILLE, 1802:79; TSUNEKI, 1981:15 (rede- scription, geographic variation). — In Apius: JURINE, 1807: 142. Sphex fuliginosus SCOPOLI, 1763:292 (as fuliginosa, incorrect original spelling). Holotype or syntypes: Carniolia (formerly in Austria, since 1919 part of Italy and Yugoslavia), lost, see ROGENHOFER UNO DALLA TORRE, 1882:599. ! Neotype: 2, Austria: "Carinthia, Ostkarawanken, Ebriach, 580-750 m, 21-29. VII. 1964, G. van Rossem," present designation (CAS). — As probable synonym of figulus: VANDER LINDEN, 1829:42. — As synonym of figulus: subsequent authors. Trypoxylon figulus var. majus KOHL, 1 883:657, 9, $ (as major, incorrect original spelling). Holotype or syntypes: 9, Swit- zerland: no specific locality (originally NHMV, Vienna, now lost). ! Neotype: 2, Switzerland, "P. 3 VIII 84" and "Cn. Tournier" (=Peney near Geneva, collection Tournier), pres- ent designation (MHNG). New synonym. — DE BEAUMONT, 1945:477 (var. major); BLUTHGEN, 1951:234 (var. major); DE BEAUMONT, 1958:206 (forma major), 1959:30 (same); WOLF, 1959:15, 1 6 (figulus major); VALKEILA, 1961:244 (var. major); DE BEAUMONT, 1964a:290, 1 964b:84 (forma major), 1965:56 (same), 1967:338 (same); BOHART AND MENKE, 1976: 346 (ssp. major); LOMHOLDT, 1976:267 (figulus major); RICHARDS, 1980:45 (var. major). Trypoxylon apicale W. Fox, 1891:1 42, 2 (as apicalis, incorrect PULAWSKI: THE STATUS OF TRYPOXYLON FIGULUS, MEDIUM, AND MINUS 127 original spelling). Lectotype: 9, Canada (ANSP, Philadel- phia), designated by CRESSON, 1928:52. — SANDHOUSE, 1940: 156 (apicale). Synonymized by PATE, 1943:16. Trypoxylon figulus barbarum DE BEAUMONT, 1957: 9, $. Ho- lotype: 3, Morocco: Marrakech (Mus. Zool. Lausanne). New synonym. — BOHART AND MENKE, 1976:346. Trypoxylon figulus yezo TSUNEKI, 1956:29, 9, $. Holotype: 2, Japan: Hokkaido [=Yezo]: Jozankei (K. Tsuneki collection, Mishima). New synonym. — BOHART AND MENKE, 1976: 346; TSUNEKJ, 1981:21 (summary of faunistic data). Trypoxylon fieuzeti GINER MARI, 1959:389, 6. ! Holotype: 3, Morocco: Fez (IEE, Madrid). Synonymized with figulus bar- barum by SUAREZ, in GINER MARI', 1959:400. COMMENTS ON NEW SYNONYMS. — T. figulus barbarum was based mainly on the elongate male flagellomere XI. Because flagellomere XI varies in length (see Geographic Variation below) this subspecies is not recognized here. I also feel that a formal name for the Japanese populations (fi- gulus yezo) is unwarranted on morphological or other grounds. COMMENTS ON NEOTYPES.— The identity of Sphex fuliginosus has never been satisfactorily established, because the original description is inadequate and the original material is lost (Ro- genhofer und Dalla Torre, 1882). Consequently, the name can only be denned by designation of a neotype. In selecting a specimen of Trypoxylon figulus as a neotype of Sphex fuliginosus I have followed the traditional interpretation of the last name. The original material of Trypoxylon majus cannot be found in the Vienna Museum (Dr. M. Fischer's letter of 21 October 1982) and must be lost. However, this name indicates a large body size, and figulus averages larger than either me- dium or minus. A neotype of majus has also been designated. DIAGNOSIS. — Most specimens of figulus differ from medium and minus in having the meso- pleural setae around the scrobe slightly longer than the midocellar diameter. However, the setae length is slightly less than this diameter in some specimens from southern France and the Iberian Peninsula (as they are in the other two species). The anteroventral mesothoracic carina is sin- uate, curved posterad mesally, but unlike most minus it has no process. The free margin of the female clypeus (Fig. 2A) is usually sinuate be- tween orbit and the median projection (free mar- gin concave in medium, almost straight in mi- nus). However, the free margin is almost evenly concave in certain specimens from Spain (almost like medium, which is unknown from Spain), in a specimen from Zirbelwald, Austria, and one from Balderschwang, Federal Germany. The fe- male hindcoxal pit is mostly oblong (Fig. 2B-D) in western palearctic specimens, but occasionally it is nearly circular, as in medium and minus; it is evanescent in Japanese females. In the male, the maximum length of flagellomere X equals 0.65-0.8 of its width (the lowest ratios are ob- served is specimens in which flagellomere XI is short, and vice versa); the maximum length of flagellomere XI usually is 2.4-3.6 times the basal diameter instead of 2.0-2.2 in most medium, but only 2.2 times in occasional specimens (which differ from medium in having a longer meso- pleural vestiture and a shorter flagellomere X). Body length 9-12 mm in female, 7.5-10 mm in male. GEOGRAPHIC VARIATION. — In most males (in- cluding the two males seen from Portugal), the maximum length of flagellomere XI equals 2.4- 2.7 times its basal diameter, but in occasional specimens it is only 2.2 times (e.g., in a male from Wachseldornmoos, Switzerland); it is 2.7- 3.2 times its basal diameter in Spanish individ- uals, and 3.3-3.6 times in Moroccan individuals. LIFE HISTORY.— Many specimens of figulus (voucher specimens examined by me) were reared from nests established in wood (Wolf, 1959). Six females and 1 7 males examined were reared by O. Lomholdt from nests in reed stems which had been used for thatching roofs at Tisvilde Hegn, Denmark. GEOGRAPHIC DISTRIBUTION (Figs. 3, 4).— Most of the Palearctic Region between Great Britain and Japan, and also eastern North America (east- ern Canada and northeastern USA). RECORDS (Old World). —Algeria ( 1 <3): El Harrach (as Maison Carree, apical flagellomeres missing, BMNH). Austria: (102 9, 50 6, NHMV if not indicated otherwise): Karnten: Afritzer See (WJP), Ebene Reichenau (RMNH), Ebriach in Ostkarawanken (WJP), Eisenkappel, Mallnitz (ZMB), Mauthen (ZMB), NOtsch, Waidisch bei Ferlach (FSAG, JG). Niederosterreich: Bisamberg near Vienna (NHMV, CU), Buck- lige Welt S Vienna, Briihl, Dornbach (CAS, NHMV), Eichkogel near Vienna (RMNH), Guntramsdorf (DEI), Hainbach (FSAG), Hamburg an der Donau (ZMB), Herzogenburg, Herzograd (JG), Kalksburg near Vienna, Krumbach, Lobau near Vienna (NHMV, ZMH), Marchfeld (ZMB), Mistelbach (ZMH), Mo- dling (ZMH), Oberweiden (DBB), Piesting, Purk (W Krems), Rappendorf bei Molk (AWE), Roggendorf bei Melk (JG), Rohr im Gebirge, Schneeberg, Stillfried (ZMH), Traismauer, Weid- lingsbach (ZMH), Wien (NHMV, FSAG, ZMH) including Donauauen, Kahlenberg and Turkenschanze. Oberosterreich: Frauenstein (JG), Gemeinde Reichenthal (AWE), Gutau (FSAG, JG), Hofkirchen (FSAG), Innerbreitenau (FSAG, JG), Kalten- 128 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 10 FIGURE 2. Trypoxylon figulus: A— female clypeus, B— female hindcoxa ventrally, C— female hindcoxal pit, vertical view, D— same, oblique view, E— male genitalia (arrow: area shown in F), F— same, portion of gonoforceps. berg (AWE), Linz (HY, NHMV), Molln, MUhlviertel (JG), Neumarkt (JG), Oberwallsee bei Miillachen (JG), Riedegg bei Gallneukirchen (JG), Sankt Willibald (AWE), Ternberg (JG), Welserheide, Zeissberg bei Freistadt (FSAG, JG). Salzburg: Koppl bei Aschach (JG), Salzburg (BMNH). Steiermark: Leut- schach (JG), Sankt Ulrich (JG), Tragoss-Oberort (DBB), Wein- burg (FSAG). Tirol: Huben in eastern Tirol (ZMB), Innsbruck (NHMV, ZMB), Iselsberg (DBB), Lienz (RMNH), Obladis, Hopfgarten, Salvenberg (CAS), Zirbelwald near Obergurgl, 1 km SW Zwiselstein in Otztal. Voralberg: Ittensberg. Belgium (95 9, 63 $, FSAG if not indicated otherwise): Bra- bant: Evere, Genval, Gistoux, Grez-Doiceau, Mont-Saint-Gui- bert, Nethen, Rhode-Sainte Agathe, Thorembais-Saint-Trond, Uccle, Waterloo. Hainaut: Aiseau, Athis, Barry, Binche, Bous- su, Bouvignies, Fleurus, Orcq, Seneffe, Taintignies, Velaines, Wanfercee. Liege: Acosse, Aubel, Barchon, Ben-Ahin, Beyne: ca 15 km SE Liege (BMNH), Beyne-Heusay, Cerexhe, Char- neux, Chevron, Clermont-sur-Berwinne, Fleron, Foret de Grunhault, Francorchamps, Henri-Chapelle, Hombourg, Ju- pille, La Calamine, La Reid, Lontzen, Montzen, Pepinster, Queue-du-Bois, Romsee, Spa, Xhendelesse, Welkenraedt. Limburg: Berg pres de Tongres, Bocholt, Godsheide, Tongres. Luxembourg: Amonines, Hotton, Les Epioux, Lomprez, Ozo, Saint-Medard, Sampont, Smuid, Waharday, Wibrin. Namur: Aische, Alle (RMNH), Andenne, Baillonville, Belgrade, Bievre, Branchon, Champion, Eghezee, Ernage, Feschaux, Gembloux, Gesves, Grand Leez, Ham-sur-Sambre, Lonzee, Mount-Gau- thier, Saint-Aubin, Saint-Gerard, Saint-Marc, Sorinnes, Sau- veniere, SombrefTe, Winenne. Bulgaria (1 $): Rila Mts. (DEI). PULAWSKI: THE STATUS OF TRYPOXYLON FIGULUS, MEDIUM, AND MINUS 129 130 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 10 FIGURE 4. Trypoxylon figulus: geographic distribution in the New World. Czechoslovakia (1 2, 3 3): Jihomoravsky Kraj: Vranov W Znojmo (as Frain, NHMV). Vapadoslovensky Kraj: Sturovo (SMT). Denmark (41 9, 33 6, ZMK if not indicated otherwise): Al- borg: Vegger. Bornholm: Bastemose. Frederiksborg: Asserbo, Hillerod, Hundested, Jaegerspris Nordskov (tip of Horn- sherred peninsula), Sorte Mose near Farum, Store Karlsminde near Lynaes, Tisvilde Hegn (FSAG, ZMK). Holbsek: Kongsore. Maribo: Maribo. Kgbenhavn: Holte. Odense: vCbelo. Randers: Glatved Strand on Djursland Peninsula, Mols Bjerge. Ring- kgbing: Gindeskov, Kjelstrup (E Skjern). Sdnderborg: S0nder- borg. Sor«: AgersO. Svenborg: Langeland Island: Hellenor. lander: Rome (VH), Stensbaek Plantage. Vejle: Klattrup. Finland: (47 9, 48 <3, HY unless indicated otherwise): Abo: Lohja, PerniO (AKM), Ryma'ttyla (AKM), Turku (AKM). Alandia: Ecker6(AKM, HY), Finstrom (AKM), Hammarland, Jomala (AKM), Saltvik (AKM). Karelia Borealis: PyhSselka (Hammaslahti). Nylandia: Helsinki (CU, HY), Parvoo = Bor- ga, Pernaja. Ostrobotnia Australis: Koivulahti. Satakunta: Loimaa (AKM), Yla'ne (AKM). Tavastia Australis: Hattula, Hameenlinna, Janakkala, Nastola (AKM), PalkSne (AKM, HY), Somero (AKM), Urjala, Vanaja, Yloja'rvi. France (35 9, 16 <3): Alpes-Maritimes: Guillaumes (FSAG). Ariege: Ax-les-Thermes (ZMB). Basses-Alpes: Allos (FSAG), Annot (FSAG), Fugeret (FSAG), Les Dourbes (KMG), Mon- tagane de Lure (ZMK), Peyresq (FSAG), Saint-Andre-les-Alpes (FSAG). Bouche-du-Rhone: Marseille (FIS). Calvados: Lisieux (FSAG). Corse: Corte (KMG). C6tes-du-Nord: Saint-Rieul. Haute-Savoie: Dent d'Oche (MHNG), Mont Jorat (RMNH), Val de Charmy (RMNH). Haute-Vienne: Rochechouart (FSAG). Jura: Arbois. Loire-et-Cher: Blois (FSAG). Loire-Atlantique: Foret de la Roche Bernard (RMNH), Herbignac (RMNH). Saone-et-Loire: Uchizy (FSAG). Seine-et-Oise: Poissy (IEE). Van Frejus (KMG), Gonfaron (FSAG), Montouroux (RMNH). Vaucluse: Carpentras (RMNH). Yonne: Foissy-sur-Vanne (FSAG). Germany, Democratic (81 2, 29 $, DEI if not indicated oth- erwise): Berlin: Berlin (DEI, HD, ZMB, ZSM). Cottbus: Alt Dobern (ZMB), Muskau, Neu Zauche (ZMB), Schlieben (ZMB). Dresden: Daubitz (SMT), Gersdorf near Kamenz (SMT). Er- furt: Erfurt (CU), Gotha (ZSM). Frankfurt: Biesental, Ebers- walde area. Gera: Blankenburg (ZMB), Jena (NHMV, ZMB). Halle: Gernrode (ZMH), Halle (DEI, ZMB), KyftTiauser, See- burg, Naumburg (TMB). Leipzig: Winkelmiihle. Magdeburg: Arendsee (SMT). Neubrandenburg: Faule Ort, Naturschutz- gebiet Muritzhof. Potsdam: Furstenberg (TMB), Zechlin (ZMB), Zootzen. Rostock: Prerow, Rostock, Stralsund (DEI, ZMB). Riigen: Hiddensee Island (DEI, SMT), Riigen Island: Monch- gut (SMT) and Ummanz. Schwerin: Campow (ZMB), Schwerin (ZMB), Wendeltorf near Schwerin. PULAWSKI: THE STATUS OF TRYPOXYLON FIGULUS, MEDIUM. AND MINUS 131 Germany, Federal (93 2, 51 3): Baden-Wiirtemberg: Enz- klosterle (KS), Heidelberg (ZSM), Hochwacht (HW), Isny (NHMB), Kaiserstuhl (ZMB), Karlsruhe (KS, ZMH), Kiissa- berg (KS), Radolfzell (ZMH), Schwarzwald (SMT), Tiengen in Wutach Valley (KS). Bayern: Abensberg (ZSM), Allach (ZSM), Aschaffenburg (FIS), Balderschwang (KS), Bamberg (ZSM), Ebenhausen (ZSM), Erdweg (ZSM), Horgersthausen near Moosburg (ZSM), Ingolstadt (ZSM), Kahl (FIS), Miinchen (FSAG, ZSM), Nurnberg(ZSM), Rotwand area (ZSM), Schlier- see (ZSM), Tegernsee (ZSM). Hamburg (ZMH): Ochsenwarder, Warwisch. Hessen: Battenfeld near Biederkopf (ZMH), Griln- dau E Frankfurt (HW, HY), Marburg (HW, WJP). Nieder- sachsen: 2 km NW Dotlingen (VH), Dorpen: 14 km SW Pa- penburg (VH), Elbe Islands (VH), 5 km S Oldenburg (VH), Pevestorf: 72 km SE Lauenburg (VH), Wobeck (ZMH). Nord- rhein-Westfalen: Ahaus (ZSM), Neheim (FSAG), Leverkusen (ZMH), Plettenberg(HW), Siegen (HW). Rheinland-Pfalz: low- er Ahr valley (FIS), Mainz (KS), Nattenheim (FSAG), Worms (FIS). Schleswig-Holstein: Amrum Island (VH), Eutin (KS), Ihlkathe 2 km SE Kiel (VH), Lutjenburg (KS), Ratzeburg (ZMH), Schierensee SW Kiel (VH), Schleswig (VH). Great Britain (92 2, 106 3; BMNH unless stated otherwise): Berkshire: Reading. Buckingham: Iver, Slough. Devon: Paign- ton. Dorset: Wareham. Essex: Brentwood, Colchester, Epping Forest. Gloucester: Chalford. Hampshire: Brockenhurst, Fleet, New Forest, Wickham. Isle of Wight: Sandown, Shanklin. Kent: Cobham, Darenth, Faversham (WJP), Goudhurst. London: Hampstead, Mill Hill, Mitcham Common, Norwood, Putney. Northampton: Ashton Wold (Oundle). Oxford: Goring, Ox- ford, Tubney near Oxford. Somerset: Dunster. Suffolk: Arger Fen, Bury St. Edmunds, Dunwich. Surrey: Byfleet, Esher, Hor- sell, Weybridge. Sussex: Midhurst: Ambersham Common. Greece (2 2, 4 3): Peloponnesus (de Beaumont, 1965): Mega Spilaion, Pirgos, and Taygetus. Sterea Ellas: Karpenission (KMG). Thessalia: Aspropotamos near Kalabaka (KMG). Hungary (13 2, 3 <3): Bacs-Kiskun: Kalocsa (TMB), Tabdi (TMB). Gyor-Sopron: Neusiedlersee (NHMV). Somogy: Bal- atonszemes (TMB). Szolnok: Jaszbereny (TMB). Tolna: Si- montornya (NHMV). Veszprem: Tihany Peninsula on Balaton Lake (HD, TMB, WJP). Italy (19 2, 8 3): Emilia- Romagna: Cattolica (RMNH). Lom- bardia: Pavia: Cignolo Po (MSNM), Sondrio: Valtellina (KS). Piemonte: Alpignano (GP), Colle di Sestriere in Alpi Cozie (GP), Murazzano (GP), San Benedetto Belbo 20 km S Alba (GP), Val d'Angrogna in Alpi Cozie (WJP). Valle d'Aosta: Bresson near St. Vincent (GP). Venezia Giulia: Trieste (CU, NHMV). Trentino-Alto Adige: Bolzano (as Bozen, NHMV), Cavalese (BB), Collalbo (de Beaumont, 1959), Ortisei (NRS), Trafoi (NHMV). Japan (Tsuneki 1981): western Hokkaido (Esashi, Hakodate, Jozankei, Kamikawa) and central Hondo (Prefectures: Fukui, Ishikawa, Kyoto, Nagono, Niigata, Saitama, and Yamanashi). Specimens studied: 4 2, 4 • They have been referred to as Nudel~ Vietnam (Tonkin). The greatest concentration of ^^ . in Ge™an'. but noodlefishes> a "* aP' , ...._. - propnate and distinctive name, seems not to have genera and species is in China and Korea. Of 1 1 , . „ ,. , , ^ ., . " • appeared in English except m a translation of a species herein recognized, eight occur in China, Russian wQrk (Berg { %2.480) Thg fle§h i§ tasty5 eight or nine m Korea, and four m Japan. Only whether cooked as a SQU^ eaten whh vin£gar Qr Salangichthys microdon occurs along the outer scrambled eggs, or fried (Okada 1955:60). The coast of Korea and in Siberia, and only Salanx species most commonly eaten in Japan is Sa- reevesi and Neosalanx brevirostris have been re- langichthys microdon, and in China probably ported as far south as Tonkin (or Haiphong). Neosalanx brevirostris or N. jordani. Protosa- Members of the Salangidae have almost al- lanx and Salanx are also consumed, but I doubt ways been referred to in English as icefishes. In that tiny Sundasalanx has ever been dined upon. [179] 180 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 13 FIGURE 1 . Axial skeleton, (a) Protosalanx chinensis, CAS-SU 6306, 85.5-mm adult male; (b) Salanx cuvieri, CAS-SU 32454, 6 1.7-mm juvenile sex undetermined; (c) Salangichthys ishikawae, CAS 6780, 74-mm adult female; (d) Neosalanx jordani, CAS 52028, 38.3-mm adult male; (e) Sundasalanx microps, CAS 44220, 17-mm adult sex undetermined. Despite their standing as a delicacy— sufficient for them to be imported by the Chinese and Jap- anese communities of San Francisco and served in the city's sushi bars— relatively little is known about the systematics and biology of noodlefish- es. An impression of their morphological diver- sity can be obtained from Figures 1 and 2. The present study was undertaken in connec- tion with the discovery of some minute, scaleless, and transparent fishes during my fieldwork in the Malay Peninsula (1 97 1 , 1973) and on the Kapuas River in Kalimantan Barat, Indonesia (1976). When first found, although in fresh water, they were living close to the sea and were mistaken for elopoid leptocephali, which they resemble only superficially. In the Kapuas River, however, they were living 800 km upriver in the midst of a rich riverine fish fauna dominated by Ostar- iophysi and with no elopoids. The observation that the maxillary bones curved inwards below ROBERTS: SALMONIFORM SUPERFAMILY SALANGOIDEA 181 TABLE 1. REPRODUCTIVE BIOLOGY OF SALANGOIDS (Wakiya and Takahasi 1937, Okada 1960, Senta 1973a, b, Roberts 1981, and pers. obs.). Larg- Smallest Larg- est Diam- mature est fe- eter male male male egg (mm) (mm) (mm) (mm) Fecundity Remarks Protosalanx chinensis 82 168 146 1.15 Breeds January-February (Korea) Salanx ariakensis — — 147 0.75 Breeds October-November (Korea) Salanx cuvieri — — 144 — — — Salanx prognathus 100 Ill 119 0.85 Breeds April-May Salanx chinensis 130 130 153 — — — Neosalanx andersoni 79 100 95 — Breeds April-May (Korea) Neosalanx brevirostris — 64 60 0.7 — — Neosalanx jordani 34 56.5 59.5 0.5 Breeds March-May Neosalanx reganius — 56 58 0.9 Breeds February-March Salangichthys ishikawae — 71 74 0.95 Breeds April-May Salangichthys microdon 65 90 100 0.91-0.99 1300-2700 Breeds March-May Sundasalanx microps — — — — Largest specimen (sex unknown) 19.9 mm Sundasalanx praecox 14.9 18.3 17.3 0.20-0.25 50 Both sexes ripe in June the head led to an hypothesis that they are sa- langoids, and observations of their skeletal anat- omy and particularly the suspensorium con- firmed this (Roberts 198 1). These fishes differ in a number of respects from Salangidae and con- stitute a separate family, Sundasalangidae, with one genus, and two or more species, one in the Malay Peninsula and one or two in the Kapuas River (Roberts 1981). Sundasalanx also occur in the Mekong basin, as reported herein. This is the only truly tropical genus in the entire order Salmoniformes. Sundasalanx praecox, with males and females sexually ripe at only 1 4.9 mm, is the smallest member of the order, and provides a striking example of a minute secondary fresh- water fish living in the midst of a rich freshwater ichthyofauna dominated by primary freshwater Ostariophysi. Interest in Sundasalangidae and its relation- ships led me to examine other salangoids but my observations and drawings quickly became too extensive to incorporate in the original descrip- tion of the new taxa; hence the present mono- graph. Food Habits All salangoids, including tiny Sundasalanx, appear to be predators. The largest species, Pro- tosalanx chinensis and Salanx reevesi, both with well-developed teeth on the tongue and jaws, ap- parently feed mainly on fishes. Salangichthys microdon taken in the Takahashi River had fed on larvae of the goby Chaenogobius sp. and on the mysid shrimp Neomysis sp. (Senta 1973b). Other species of Salanginae and Salangichthyi- nae feed mainly on small Crustacea (in marine environments) or on insects (in fresh water). Sundasalanx are known only from fresh water and feed on tiny insects (Roberts 1981). Reproduction While some species are primarily marine or at least brackish water inhabitants (e.g., Protosa- lanx chinensis), and many spend part of their lives in the sea, others are restricted to fresh water or have populations which presumably repeat their life cycle without leaving fresh water. Basic information on salangoid reproductive biology is summarized in Table 1 . Fecundity ranges from several thousand eggs in Protosalanginae and Salanginae (no precise numbers available) down to only about 50 in Sundasalangidae. The external egg membrane is adhesive, eggs becoming attached to any solid object at the spawning site. Wakiya and Takahasi (1937, pi. 21) published drawings of the basal portion of the adhesive strands on the eggs of Protosalanx chinensis, Salanx ariakensis and S. prognathus, Salangichthys microdon and S. ishikawae, and Neosalanx jordani. The eggs illustrated are pre- sumably ovarian, since the adhesive strands are not detached. For photomicrographs of the 182 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 13 FIGURE 2. Radiographs, (a) Salanx cuvieri, MNHN 9900, 1 12 mm, holotype; (b) Neosalanx andersoni, NRM 10287, 79 mm, holotype. spawned eggs with detached adhesive threads of unfertilized and fertilized eggs of Salangichthys microdon see Okada (1960, pi. 18). Spawning ecology of this species is described by Senta (1973a). According to Wakiya and Takahasi (1937:269), after spawning "the body becomes very lean and the vertebrae become visible through the skin, whence it is generally assumed that death then ensues." I suspect that this is true in Salanginae as well as Salangichthyinae but not in Protosalanx. Sexual Dimorphism A notable feature of salangoids is their unique sexual dimorphism. In all Salangidae except Neosalanx, sexually mature males have the pec- toral fins longer and more pointed (falcate) and the pelvic fins larger. In all adult male Salangidae the anal fin is larger than in females and has modified rays. The anterior rays of the anal fin are greatly enlarged, the middle rays thin and strongly curved, and the posterior rays short and widely separated at the base. The morphology of the anal fin is very similar in sexually mature males of all of the genera and species of Salan- gidae. In all Salangidae, mature males have a row of large, tightly adherent scales on the body par- allel to the anal fin base (sometimes extending posteriorly a short distance beyond the anal fin base onto the caudal peduncle). The number of anal scales ranges from 14 to 28. Sexual dimor- phism has not been observed in Sundasalangi- dae. Although salangids differ greatly in the size of adult males, the morphology of the modified male anal fin is remarkably uniform (Fig. la, d). The total range of anal fin-rays is 23-32. The first two or three rays are simple, the first one or two small or minute. The last simple ray and the first four to six branched rays are greatly enlarged and somewhat thickened; near the base of each of these rays is a very large lateral projection. The next 1 2 or so rays are noticeably thinner and are deflected backwards near the middle of their length, so that their distal portions lie close to- gether. In Protosalanx these rays are simple, but in other Salangidae they are branched. The pos- teriormost rays may be simple or branched, are reduced in size and not modified, except that their bases tend to be relatively wide apart (much more so than the bases of the preceding rays or of the corresponding rays in females), especially in Salanginae. The proximal pterygiophores, es- pecially for the anterior portion of the anal fin, are also enlarged in males. In alcian-alizarin preparations the anal fin-rays and pterygiophores of sexually mature males are deeply stained with alizarin, whereas those of females tend to be less well stained with alizarin or in some instances stained only with alcian. Near the middle of the rays in the most mod- ified part of the male anal fin, a tough, almost tendonlike membrane arises from each ray and ROBERTS: SALMONIFORM SUPERFAMILY SALANGOIDEA 183 extends obliquely and posteroventrally across the densely webbed portion of the fin to end in a thickened, obliquely oriented non-muscular pad of tissue. The distal portion of this oblique pad is free from the surface, so it can be readily lifted, and even when not lifted forms a sort of groove for the length of the pad. This portion of the anal fin can be flexed in such a way that it forms a slight concavity. The fin may be expanded man- ually by pulling on the anteriormost rays; when released, it snaps back into a less expanded con- dition. Spawning behavior has not been reported upon, but presumably the male's anal fin remains in contact with the vent region of the female in such a way that it temporarily retains eggs and sperm in proximity while fertilization occurs ex- ternally. In addition to the modified anal fin, sexually mature males of all Salangidae bear a row of large, cycloid scales on the side of the body above and co-extensive with the anal fin or extending a short distance beyond it onto the caudal pe- duncle. The scales are tightly adherent and broadly overlapping (more so anteriorly than posteriorly). In addition to the main row of anal scales, some specimens exhibit two or three smaller scales in a separate row overlying the vent. These usually have been overlooked by previous authors, and are not included in the counts of anal scales in Table 2. Breeding tubercles and other forms of tem- porary sexual dimorphism have not been re- ported previously in salangoids. I have observed breeding tubercles in adult males and females, apparently in spawning condition, of Protosa- lanx chinensis, and in adult males of Salangich- thys microdon and Neosalanx jordani. This pre- sumably temporary tuberculation is most extensive and easily observable in an 120-mm male Protosalanx (CAS-SU 36025). In this spec- imen breeding tubercles occur on the anal, pec- toral and pelvic fins, abdominal keel, and head. The strong lateral projections on the anterior face of the first nine branched anal fin-rays are en- tirely or almost entirely covered by a thickened, longitudinal band of thickened skin 9 mm long and 1.2 mm high. The surface of this spongy band of skin is covered with hundreds of small, overlapping, scale- or leaflike breeding tuber- cles, with their raised free margins projecting an- teriorly. There are about 1 2-20 of these tubercles in a vertical series. Discrete pads of similarly thickened skin covered with similar breeding tu- bercles extend obliquely posteroventrally on the basal third of the first five branched anal fin-rays. There are up to eight tubercles across each ray. The skin on the middle third of the same rays appears to be only slightly thickened and bears only a few, small widely spaced, low-lying round (not scalelike) tubercles. The distal third or branched portion of the first eight branched rays is covered with thick skin densely coated with scalelike tubercles. There are up to about eight tubercles across each ray-branch. The leading edge of the third (enlarged) simple anal fin-ray bears a thick, lamellar projection of skin, 11.5 mm long and up to 2.2 mm wide, covered with widely scattered, low-lying round tubercles without free margins. The midventral abdominal keel is also notably thickened, and covered with minute, closely spaced round or granular tubercles which extend for a short distance onto the abdomen and sides of the body just anterior to the anal fin. The pelvic and pectoral fins bear round tu- bercles dorsally and ventrally; these are most no- ticeable on the enlarged outermost pectoral fin- ray. The dorsal fin is slightly tuberculate, the adipose and caudal fins non-tuberculate. The dorsal, lateral, and ventral surfaces of the head bear irregularly scattered, round, low-lying tu- bercles without free margins. These are largest and most numerous on its ventral surface. The skin of the oral margin of the upper and lower jaws and gular margin of the lower jaw is thick- ened and tuberculate. Fine granular projections, which may be minute breeding tubercles, extend in a dorsomedian longitudinal band from the dorsal fin origin anteriorly halfway to the occi- put. In the two gravid females the skin is less modified, and although tuberculation is very much lighter, there are small, low-lying round tubercles on the anal, pelvic, and pectoral fins and on the head. In one of them the skin on the jaws is thickened as in the male; in the other it is not. The first female has the median abdominal fold somewhat thickened, suggestive of the more pronounced thickening of this fold seen in the male; the other female does not. Tubercles have not been observed in females of any other sa- langoid. In other salangoids breeding tubercles have been observed only on the anal fin of males. An 8 3.1 -mm male Salangichthys microdon (CAS 52033) has small scalelike breeding tubercles on 184 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 13 the branched portion of the anteriormost branched anal fin-rays. These are arranged uni- serially on each fin-ray branch. A 47.5-mm male Neosalanx jordani (AMNH 51704) has very similar scalelike tubercles on thickened skin sur- rounding the lateral projection at the base and on the basal half of the first four branched anal fin-rays (which are enlarged); fin-rays 6-14, which are bent, each have three to four melanophores on the basal one-fourth of their length. (Similar coloration has been observed on the anal fin in occasional males of Salangichthys microdon.) Breeding tubercles apparently do not occur in Sundasalangidae, in which neither secondary sexual dimorphism nor dichromatism has been observed. Pigmentation The only pigmentation known to be exhibited by salangoids, apart from that of their eyes, is in melanocytes or melanophores, which tend to oc- cur as widely separated single cells or isolated clumps of relatively few cells. In life all, or almost all, salangoids (except Protosalanx) are trans- parent or translucent, except for the prominent eyes. The most constant pigmentary feature of the salangoids is a row of melanophores at the interface of the ventral myotomic musculature and the non-segmentally muscularized ventral abdominal wall. This series of melanophores, with a single cell at about the middle of the ven- tral end of each myotome, from the most anterior myotome to the anal fin origin, is present in near- ly all salangoid specimens examined. Usually these melanophores are longitudinally elongate, giving the appearance of a series of widely spaced thin black dashes. A second pigmentary feature found in many salangoids is a ventromedian row of widely spaced melanophores, one for each body segment. These melanophores tend to be den- dritic when expanded or round when contracted, and may extend the entire length of the abdomen; sometimes they are restricted to the preanal membranous keel. These two pigmentary fea- tures of salangoids occur in many teleost larvae and in adults of other neotenic teleosts. Some salangoids exhibit a row of melano- phores along the anal fin base, one between each anal fin-ray. This row of melanophores, lying deep in the body and median rather than paired, may be the continuation of the midabdominal row of melanophores described above. This row usually extends the length of the anal fin; some- times it continues beyond the anal fin onto the caudal peduncle near its ventral margin. Clusters of a few melanophores occur just an- terior to the bases of the pectoral and pelvic fins in most salangoids, at the tip of the snout and chin, especially in Salangichthys, and infre- quently on the dorsal surface of the head over- lying the fore- and hind-brain. In sexually mature (spawning?) males of Salanx and Salangichthys there may be a cluster of melanophores on the proximal portion of the middlemost anal fin- rays. The dorsal, anal, pectoral, and pelvic fins are otherwise usually devoid of melanophores, but the caudal fin lobes frequently are dark or dusky due to numerous fine melanophores. The anal scales of the males are always entirely de- void of melanophores. In most salangoids the entire dorsal and most of the lateral body surfaces are devoid of mela- nophores. Protosalanx chinensis and Neosalanx andersoni provide notable exceptions. Young of Protosalanx and. Neosalanx exhibit very few me- lanophores. Large and sexually ripe individuals of these two species, however, may have the dor- sal and lateral surfaces of the body with numer- ous melanophores. Those on the dorsal body sur- face are fine, exceedingly numerous, and generally scattered over the entire musculature, but those on the sides are few and peculiarly restricted along the course of the myotomal septae. About a doz- en melanophores lie on each myotomal septa; the melanophores of successive septae are more or less parallel to each other; the cells are oblique- ly elongate, conforming to the thinness and obliquity of the septae and thus forming a series of widely spaced thin black slashes. This pattern, sometimes barely evident or absent in P. chi- nensis, is very well developed in two gravid fe- males of 1 29-132 mm (USNM 1 20746). Wakiya and Takahasi (1937) show it well developed in female P. chinensis (not gravid?) and N. ander- soni (gravid); and relatively weakly developed in males of both species. It is present only on the upper part of the body in the relatively small male holotype of N. andersoni (NRM 10287, 79 mm). Chyung (1961) shows it well developed in a gravid N. andersoni. I have seen clupeomorphs but no osmeroids or other salmoniforms with similarly distributed melanophores. ROBERTS: SALMONIFORM SUPERFAMILY SALANGOIDEA 185 Some Misconceptions Some misconceptions about Salangidae should be noted. These concern the reported presence of scales other than anal scales, presumed ab- sence of the swimbladder, and reputed oral brooding of eggs. The most persistent misinfor- mation concerns the occurrence of scales on the body other than the anal scales. Gunther (1866: 205) stated that the body is "naked or covered with small, exceedingly fine, deciduous scales (?)" and added in a footnote, "There is no trace of scales in specimens preserved in spirits for some time; but others, which we received lately, show scattered fragments of scales, without any regular arrangement." He was unaware of the anal scales of males. Regan (1908b:444), in diagnosing Sa- langinae (= Salangidae), stated simply "scales de- ciduous" but described the anal scales of males in a footnote. Fang ( 1 934a:239) stated body "na- ked or with a few exceedingly thin, large, scat- tered, deciduous scales, without any regular ar- rangement" in addition to the anal scales of males. Nichols (1944) referred to several species with "scales small, deciduous, little evident." Nelson (1976:104) cautiously stated "body generally scaleless" without referring to the anal scales of males. As noted by Wakiya and Takahasi (1937) all salangids are totally scaleless except for the anal scales of sexually mature males; as noted above, the anal scales are large and strongly ad- herent. Reports of scales on other parts of the body are all attributable to dislodged scales from other fishes. Various authors, including Gunther (1866: 205), Fang (1934a:239), and Nelson (1976:104) have stated that salangids lack a swimbladder. Wakiya and Takahasi (1937:268, fig. 1) reported a physostomous swimbladder in Protosalanx chinensis, Salanx ariakensis, S. prognathus, Neosalanx jordani, Salangichthys ishikawae, and S. microdon. In P. chinensis and S. ishikawae the swimbladder is depicted as relatively large and oval, and in the others as equally long but almost uniformly slender for its entire length. The condition of the swimbladder in Sundasa- langidae is unknown. Fang (1934a:238, 252, fig. 7) suggested that Salangidae are oral brooders. In a series of 6 1 males and 27 females identified as Hemisalanx (=Salanx) prornathus collected at Chinkiang in April 1933, Fang found 6 males and 19 females with 1-21 eggs in the mouth. He also reported one Protosalanx (sex not mentioned) with eggs in its mouth. I have also observed a few speci- mens of both sexes, especially of Salanginae, with small numbers of eggs in the mouth; this is at- tributable to rupture of the ovaries and spillage of eggs after the fish had been caught. There is no information indicating that salangoids prac- tice oral brooding or any other form of parental care. This introduction to salangoids concludes with a key for their identification. Key to Salangoidea 1 a. Pelvic fin with 5 rays; adipose fin absent; pectoral fin rayless throughout life; sex- ually mature males without anal scales or enlarged anal fin; vertebrae 37-43; standard length to 22 mm (Sundasalan- gidae) 11 1 b. Pelvic fin usually with 7 rays (rarely 6 or 8); adipose fin present; pectoral fin with rays except in larvae; sexually mature males with a row of large anal scales and enlarged anal fin; vertebrae 48-79; adults at least 35 mm in standard length (Sa- langidae) 2 2a. Teeth on palatal toothplate and lower jaw in two rows; teeth on tongue in two marginal rows or widely spread over ba- sihyal toothplate (Protosalanginae) Protosala nx ch inensis 2b. All oral teeth in single rows 3 3a. Head extremely depressed; snout very elongate and relatively pointed; cranial fontanel entirely closed in juveniles and adults; premaxillae larger than maxillae, those of opposite sides meeting broadly in front of snout; premaxillary teeth rel- atively large; supramaxilla absent; ver- tebrae 66-79 (Salanginae) 4 3b. Head moderately depressed; snout mod- erately elongate and broadly rounded; cranial fontanel with anterior and pos- terior portions open throughout life, pre- maxillae smaller than maxillae, more or less separated from each other in front of snout; premaxillary teeth relatively small, tiny, or absent; supramaxilla pres- 186 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 13 ent; vertebrae 48-65 (Salangichthyinae) 7 4a. Tongue with a median row of conical teeth (subgenus Leucosomd) Salanx reevesi 4b. Tongue toothless 5 5a. Head strongly pointed; lower jaw not projecting beyond upper jaw; presym- physeal fleshy appendage, bone, and teeth frequently present in adults; vertebrae 72-79 (subgenus Salanx) 6 5b. Head less strongly pointed; lower jaw projecting slightly beyond upper jaw; no presymphyseal fleshy appendage, bone, or teeth; vertebrae 70-73 (subgenus Hemisalanx) Salanx prognathus total rakers on first gill arch 9-15 (main- land Asia) Neosalanx jordani lOc. Vertebrae 52-56, average 53.75 (after Wakiya and Takahasi 1937); standard length to 58 mm; total rakers on first gill arch 15 (known only from Ariake Bay, Kyushu, Japan) Neosalanx reganius I la. Horizontal diameter of eye less than 4% of standard length; ceratobranchial 5 with 0-3 small conical teeth; total rakers on first gill arch 0-2; vertebrae 41-43 Sundasalanx microps I 1 b. Horizontal diameter of eye more than 5% of standard length; ceratobranchial 5 with about 8-10 large conical teeth; total rakers on first gill arch 10-12; vertebrae 37-41 Sundasalanx praecox 6a. Presymphyseal bone usually present in specimens over 1 00 mm standard length, MATERIAL EXAMINED relatively elongate and with up to 1 7 teeth Salangoid specimens deposited in the follow- on each side; vertebrae usually 77-78 ing institutions have been examined for this study: (rarely 76 or 79) Salanx cuvieri American Museum of Natural History, AMNH: 6b. Presymphyseal bone usually absent, or British Museum (Natural History), BMNH; Cal- relatively short and with no more than ifornia Academy of Sciences, CAS, including 6 teeth on each side; vertebrae 72-75 ... specimens formerly deposited at Stanford Uni- Salanx ariakensis versity, CAS-SU; Museum national d'Histoire _ „ . . , naturelle, Paris, MNHN; Naturhistoriska Riks- 7a. Palatal toothplate with minute teeth; '^ , ' XTr,A(, »„ r~ , ... f, . museet, Stockholm, NRM: Museum of Zoology, premaxilla with numerous small or mi- TT . . r»*-il- T™™*^ o • u A, 1^-11 University of Michigan, UMMZ; Smithsonian nute teeth, snout relatively elongate; ver- T . . TT0^Tnk, \ ~ , . , », cn' - /0 , . ,,, ' o Institution, USNM; and Zoologisch Museum, tebrae 59-65 (Salangichthys) 8 TT . ' „,?,. -,, r» , , , , ... „ Umversiteit van Amsterdam, ZMA. 7b. Palatal teeth absent; premaxilla usually . , , .. ,.. , r . . . , ,. , ,. iU . -.!_ 1 c • .L A detailed list of material examined (including toothless or with 1-5 minute teeth; snout . . .. .... . . , , „. . , , , alcian-alizann preparations) is given under each relatively short except in Neosalanx an- . . . -o,c/,, , , ~ species in the systematic account. aersom; vertebrae 48-65 (Neosalanx) 9 8a. Pectoral fin-rays 14-19 SKELETAL ANATOMY Salangichthys microdon Salangoid skeletal anatomy cannot be ob- 8b. Pectoral fin-rays 20-28 served adequately from alizarin preparations be- Salangichthys ishikawae cause it is largely cartilaginous, and even ossified 9a. Snout relatively short, standard length to Portions (induding dermal bones) often fail to 64 mm, males with 14-21 anal scales, stam wflth allzann' The only Previous obs^rva- vertebrae fewer than 60 10 tlons of salangoid skeletal anatomy are brief and 9b. Snout relatively elongate, standard length relatively uninformative. The only general ac- to 100 mm, males with 20-28 anal scales, count' that of McDowall (1969:815), is limited vertebrae 63-65 Neosalanx andersoni to three Paragraphs, one on the cranium, one on the jaws, and one on the remainder of the skel- lOa. Vertebrae 55-59; standard length to 64 eton emphasizing the median fins. Wakiya and mm; total rakers on first gill arch 15-19 Takahasi (1937) figured toothed portions of the (mainland Asia) Neosalanx brevirostris jaws, palate, and tongue of various salangids. 1 Ob. Vertebrae usually 50-53, rarely 49 or 54; Nelson ( 1 970) described and figured the gill arch- standard length usually less than 50 mm; es in Salanx reevesi and Neosalanx brevirostris ROBERTS: SALMONIFORM SUPERFAMILY SALANGOIDEA 187 parietal dilatator fossa supraethmoid frontal nasal cranial fontanel trabecula communis ethmoid plate basioccipital lamina orbitonasalis parasphenoid 4 mm FIGURE 3. Dorsal and ventral view of cranium. Protosalanx chinensis, CAS-SU 6306, 158 mm. (his Salanx chinensis and Salangichthys micro- don). He particularly noted the well-developed fourth hypobranchials, "which so far as known are absent from all other adult teleostean fishes." My own observations and drawings of salangid gill arches agree closely with Nelson's. Rosen (1974; figs. 16g, 26a & b) figured and com- mented briefly upon the caudal skeleton and por- tions of the gill arches of Neosalanx brevirostris (his Salangichthys microdori). 188 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 13 frontal supraethmoid parasphenoid 1 mm FIGURE 4. Dorsal and ventral view of cranium. Salanx cuvieri, CAS-SU 32454, 69.4 mm. The advent of a technique for staining whole specimens with alcian and alizarin (Dingerkus and Uhler 1977) made the present relatively ex- tensive observations possible but even so there have been difficulties. Some specimens stained well with alcian but not with alizarin, or vice versa, and in some specimens that otherwise stained well with both stains there are still por- tions of the skeleton which failed to take up no- ticeable amounts of either stain. Such difficulties could not always be made up for by staining additional specimens. In general, alizarin stains only bone. Alcian stains cartilage but also stains some skeletal fea- tures which are obviously bony and have no car- tilaginous precursors, such as fin-rays. Cartilag- inous structures, however, often stain much more deeply with alcian than such non-cartilaginous structures. Thus the salangoid hyopalatine is al- most always stained deep blue and the opercle ROBERTS: SALMONIFORM SUPERFAMILY SALANGOIDEA 189 ethmoid plate supraethmoid frontal 2 mm parietal FIGURE 5. Dorsal view of cranium and membrane bones on dorsal surface of cranium. Salanx prognathus, CAS 51439, 1 10 mm. appears variably pale blue and/or red. In a few of my figures such differences are indicated by the intensity of stippling, but in general the dis- tribution of stain is far too complex to permit its representation in black-and-white illustrations. Some idea of the difficulty involved may be gained from Figure 20 (pelvic girdle of Protosalanx), in which the distribution of stain is indicated. In the cranium the distribution is far more com- plicated and could be conveyed only by illustra- tions in full color. CRANIUM (Figures 3-8) The cranium of all salangoids is depressed, very strongly in Salanginae and almost as strong- ly in Protosalanginae, but relatively moderately in Salangichthyinae and Sundasalangidae. Some other features correlated with the cranial depres- sion are the peculiarly underslung maxilla, ven- trolateral eye position (especially in Salanginae), and perhaps the posteriorly recurved jaw teeth (especially in Salanginae and Protosalanginae). 190 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 13 tectum synoticum auditory capsule cranial fontanel foramen ' magnum supraethmoid taenia marginalis epiphyseal bar semicircular canals pila prooptica nasal recess basioccipital lamina orbitonasalis trabecula communis hyomandibular fossa parasphenoid 1 mm FIGURE 6. Dorsal and ventral views of cranium. Neosalanx jordani, CAS 52028, 38.3 mm. The development of the cranial fontanel ex- hibits considerable differences. The fontanel ap- parently remains open anterior and posterior to the epiphyseal bar throughout life in Salangich- thyinae and Sundasalangidae, although the an- terior portion may be greatly reduced in larger Salangichthyinae. In Protosalanginae the ante- rior portion closes while the posterior portion always remains open, albeit much reduced in the largest specimens examined. In Salanginae the cranial fontanel is entirely closed in all specimens in which skeletal preparations have been ex- amined. Young Osmeridae in which the cranium is still ROBERTS: SALMONIFORM SUPERFAMILY SALANGOIDEA 191 auditory fenestra taenia marginalis lamina orbitonasalis parachordals trabecula communis hypophysial fenestra notochordal groove 1 mm FIGURE 7. Dorsal and ventral views of cranium. Sundasalanx microps, CAS 44220, 17 mm. cartilaginous have a median bar (taenia tecti me- dialis) separating the anterior and posterior por- tions of the cranial fontanel into left and right halves. Such a feature is usually but not invari- ably absent in salangoids. In a series often Neo- salanx jordani (39.7-45.7 mm), nine have the cranial fontanel entirely undivided, but one (4 1 .0 mm) has a median cartilaginous bar dividing both the anterior and posterior portions of the fon- tanel. The bar is slender posteriorly, but ante- riorly it is much wider, so that the anterior por- tion of the fontanel is represented by two widely 192 a PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 13 4 mm 1 mm cranial fontanel ethmoid plate anterior myodome lateral fenestra pila prooptica trabecula communis lamina orbitonasalis pila prooptica 1 mm 1 mm FIGURE 8. Lateral view of cranium, (a) Protosalanx chinensis, CAS-SU 6306, 158 mm; (b) Salanx cuvieri, CAS-SU 32454, 69.4 mm; (c) Neosalanx jordani, CAS 52028, 38.3 mm; (d) Sundasalanx microps, CAS 44220, 17 mm. separated and relatively small openings. The epi- physeal bar in this specimen is also larger than usual. The condition of the cranial fontanel in this specimen closely resembles that observed in osmerid chondrocrania. In Protosalanx of 85- 89 mm, the anterior portion of the cranial fon- tanel is similarly divided into greatly reduced left and right openings, which become entirely closed in specimens slightly larger. The ethmoid plate is greatly enlarged in all salangoids. In Salangichthyinae and Sundasa- langidae it is broad and moderately elongate, while in Protosalanginae and Salanginae it is broad and extremely elongate. Ossification of the chondrocranium is relatively poor in all salan- goids but varies greatly. The greatest amount of cranial ossification is observed in the skulls of the largest Protosalanx, in which the supraeth- moid, frontals, parietals, parasphenoid, and basi- occipital are all stained more or less deeply with alizarin. In large Protosalanx the posterior por- tion of the parasphenoid has broad lateral wings and the basioccipital has small thin lateral wings (largely obscured by the overlying parasphe- noid). Neither of these features has been ob- served in other salangoids. In all other salangoids the basioccipitai ossification is apparently re- stricted to the basioccipital centrum. In Protosalanginae the outline of the cranium is more irregular, suggesting a more primitive condition; while in Salanginae it is relatively smooth and streamlined, suggesting a more de- rived or specialized condition. The auditory cap- sules are most pronounced or laterally prominent in Salangichthyinae. The interorbital septum is relatively open in Sundasalangidae and Salangichthyinae, almost as open in Protosalanginae, but greatly reduced in Salanginae. In Salangichthyinae the anterior- medial portion of the orbit is occupied by very large pilae proopticae arising from the ventral surface of the taenia marginalis or anterior su- praorbital cartilage. In Sundasalangidae the pilae proopticae are rudimentary. A number of cranial features that occur in Sun- dasalangidae have not been observed in the other (mostly juvenile and adult) salangoids examined. Thus the lamina orbitonasalis, which appears as a single apparently simple entity in other sa- langoids, has two components in Sundasalanx: a dorsoanterior contribution from the taenia marginalis and a ventroposterior contribution from the trabecular communis or posteroventral portion of the ethmoid plate. The ethmoid plate is separated by the anterior myodome into dorsal and ventral portions; the anterior myodome ex- tends anteriorly almost to the tip of the snout. In other salangoids the anterior myodome lies much farther posterior, and the ethmoid plate is relatively thin and more or less greatly depressed (least so in Salangichthyinae). In Sundasalanx the base of the cranium is largely occupied by the hypophysial fenestra, a character of all developing teleost chondrocrania usually lost at an early stage. In all other salan- goids the hypophysial fenestra is closed off by cartilaginous growth and the area it once occu- pied may be overlaid by the parasphenoid. In Sundasalangidae the passage for the internal ca- rotid artery is represented by an anterolateral ROBERTS: SALMONIFORM SUPERFAMILY SALANGOIDEA 193 hyopalatine opercle premaxilla supramaxilla maxj||a/ preopercle hyomandibular dentary Meckel's cartilage coronoid process 4 mm articular quadrate interopercle subopercle FIGURE 9. Lateral view of jaws, suspensorium, and opercular bones. Protosalanx chinensis, CAS-SU 6306, 158 mm. extension of the hypophysial fenestra; in all other salangoids the passage for this pair of arteries is isolated and widely separated. In Sundasalanx the base of the cranium ex- hibits a median groove on either side of which extends a slight ridge. This groove probably rep- resents the pathway of the embryonic cranial no- tochord before its absorption (complete in all other salangoids examined) into the basioccipital centrum. The ridges on each side may be rem- nants of the parachordal cartilages. JAWS (Figures 9- 13) The jaws of salangoids are relatively general- ized, in that the jaw bones, their shape, and the distribution of teeth on them are similar to those in many lower teleosts. In all salangoids the max- illa is toothed and enters broadly into the gape. All salangoids have a single supramaxilla, except Salanginae, in which this element is lacking. In some Salanginae the bony tip of the lower jaw is formed not by the dentaries, but by a median presymphyseal bone (usually tooth-bearing). Due in part to poor quality of alcian-alizarin staining of the lower jaw in salangoids, the relationships of bones that constitute it have not been ade- quately observed. The premaxillae and maxillae are somewhat variable (see remarks in system- atic account). SUSPENSORIUM (Figures 9- 13) The outstanding feature of the salangoid sus- pensorium is the union of the hyomandibula (hy- osymplectic) and pterygoquadrate, which are united into a single continuous cartilaginous ele- ment, here called the hyopalatine (=palatohyo- mandibuloquadrate of Roberts 1981). Only in Sundasalanx praecox is the hyopalatine divided into anterior and posterior portions, but the di- vision apparently is more anterior than the prim- itive division between hyomandibula (or hy- osymplectic) and pterygoquadrate. In developing vertebrates the rudimentary mandibular arch divides into two cartilages where it bends around the corner of the mouth: the pterygoquadrate bar (dorsal) and the mandibular bar or Meckel's cartilage (ventral). The rudi- mentary hyoid arch divides into the hyoman- dibular (dorsal) and hyoid bar (ventral). All sa- langoids except Sundasalanx praecox show the 194 nasal PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 13 supraorbital hyopalatine pseudobranch . 1 mm , FIGURE 10. Lateral and medial views of jaws, suspensorium, and opercular bones. Salanx cuvieri, CAS-SU 32454, 69.4 mm. most unusual condition of having the dorsal por- tions of the mandibular and hyoid arches fused into a single element. This salangoid element has readily definable features corresponding to the palatine or pterygoid, quadrate, and hyoman- dibula of teleosts in which these elements are separate, but it is unclear whether a portion rep- resenting the symplectic is present. No separate symplectic has been detected in any salangoid; the symplectic may be represented by a thickening or ridge near the ventral margin of the quadrate portion of the hyopalatine. In Sundasalangidae and some Salanginae and Salangichthyinae the suspensorium consists sole- ly of the cartilaginous hyopalatine, but in other Salanginae and Salangichthyinae and in Proto- salanginae a number of perichondral, endochon- dral, or dermal ossifications develop on the sus- pensorium. The elements most often added are the mesopterygoid and an anterior palatal tooth- plate (=ectopterygoid?), which may or may not bear teeth. The suspensorium exhibits more os- sification in large Protosalanx than in any other salangoids examined: heavily toothed palatal toothplate, mesopterygoid, and partial ossifica- tion of quadrate and hyomandibula. Whether the dorsal portions of the mandibular and hyoid arches are similarly fused in any other fishes is unknown. In the few fishes for which the development of these arches has been adequately observed it would appear they are separate, in- cluding Salmo (DeBeer 1937), Flops (pers. obs.), Hepsetus (Bertmar 1959). In young salmoni- forms I examined (including Salmo, Galaxias, Lepidogalaxias, Hypomesus, and Spirinchus) cartilaginous pterygoquadrate and hyomandib- ular or hyosymplectic are always separate. Circumorbital Bones (Figure 12) A supraorbital bone is seen in all Salangidae but is absent in Sundasalangidae. The dermo- sphenotic or sixth infraorbital appears to be ab- sent in all salangoids. An isolated infraorbital (fourth or fifth?) is seen in some Salangichthyinae but is greatly reduced (Fig. 1 2). Gill Arches (Figures 14-17) The upper elements of the gill arches of sa- langoids are relatively generalized and, except in Sundasalangidae, so are the lower elements. Ex- cept for the upper and lower pharyngeal tooth- plates the salangoid gill arches apparently are entirely cartilaginous. Four basibranchials are probably present in all salangoids but in none are all of them separate. In Protosalanginae, Sa- ROBERTS: SALMONIFORM SUPERFAMILY SALANGOIDEA 195 , 1 mm | FIGURE 1 1. Lateral view of jaws, suspensorium, and opercular bones. Salanx prognathus, CAS-SU 51439, 1 10 mm. langinae, and Salangichthyinae basibranchials 2 and 3 are indistinguishably fused to each other, and in some Salanginae basibranchials 1 , 2, and 3 may be so fused. All salangoids have four hypobranchials; hy- pobranchial 4 is always separate and relatively large. The basibranchial series in salangoids is en- tirely cartilaginous. Not only do the basibran- chials themselves not ossify, but basibranchial toothplates, a characteristic feature of many sal- moniforms including salmonids, osmerids, and galaxiids, are absent. The basic basibranchial ar- rangement in Salangidae appears to be basi- branchial 1 separate, basibranchials 2 and 3 fused, and basibranchial 4 separate. A basibranchial 5 is fused to basibranchial 4 in various salmoni- forms, and is apparently usually present in many salmonoids, osmeroids, and galaxioids (includ- ing Lepidogalaxias) as a thin cartilaginous shaft projecting posteriorly between the fifth cerato- branchials. In some instances there is a clear de- marcation between basibranchials 4 and 5, and they may be separate or at least not completely fused. Basibranchial 5, fused with basibranchial 4, is indicated in Salangidae by Nelson (1970), but in Salangidae I have examined there is no indication of a fusion or demarcation between the presumed basibranchial 5 and basibranchial 4. Basibranchial 5 does not project so far pos- teriorly nor is it slender and rodlike as in other Salmoniformes in which its presence is less doubtful. I therefore tentatively consider basi- branchial 5 absent in Salangidae. That it is absent in Sundasalangidae seems highly likely. Gill rakers are poorly ossifed (never stained with alizarin) and edentulous (frequently dentig- erous in salmonoids, osmeroids, esocoids). Those on the trailing (inner) face of the arches usually are fewer and smaller than those on the leading (outer) face (Figs. 14-17). Total number of gill rakers on leading face of first gill arch is 8-19 in Salangidae and 0-10 in Sundasalangidae (Table 2). Dentition The most complete and presumably most primitive dentition in salangoids is observed in 1 mm FIGURE 12. Lateral view of jaws, suspensorium, and opercular bones. Neosalanx jordani, CAS 52058, 35.1 mm. 196 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 13 LJ MX PL HQ FIGURE 13. Lateral view of jaws, suspensorium, and opercular bones, (a) Sundasalanx praecox, CAS 52031, 17 mm; (b) Sundasalanx microps, CAS 44290, 17 mm. HQ = hyomandibula + quadrate, LJ = lower jaw or Meckel's cartilage, MX = maxilla, P = premaxilla, PHQ = hyopalatine cartilage, PL = palatine, OP = opercle, SO = subopercle. hypobranchials urohyal branchiostegal rays basibranchials ceratobranchials 1 mm FIGURE 14. Dorsal and ventral views of hyoid and branchial arches. Protosalanx chinensis, CAS-SU 6306, 153 mm. ROBERTS: SALMONIFORM SUPERFAMILY SALANGOIDEA -basihyal dorsohyal anterohyal posterohyal interhyal basihyal toothplate accessory cartilage uncinate process epibranchials 197 infrapharyngobranchials upper pharyngeal toothplate lower pharyngeal toothplate 4 mm 2mm FIGURE 1 5. Dorsal view of hyoid and branchial arches and ventral view of upper pharyngeal elements. Salanx cuvieri, CAS- SU 32454, 69.4 mm. , 1 mm | FIGURE 16. Dorsal view of hyoid and branchial arches; ventral view of infrapharyngobranchial 4 and upper pharyngeal toothplate. Neosalanx jordani, CAS 52058, 38.3 mm. 198 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 13 basibranchial 1 + 2 + hypobranchial 1 + 2 basibranchial 3 + hypobranchial 3 hypobranchial 4 basibranchial 4 , 1 mm FIGURE 17. Dorsal view of hyoid and branchial arches. Sundasalanx microps, CAS 44220, 17 mm. Protosalanx, with numerous large, conical teeth on the premaxilla, maxilla, palatal toothplate (=ectopterygoid?), tongue (basihyal toothplate), and upper and lower pharyngeal toothplates. As in all Salangidae, there are only two pairs of pha- ryngeal toothplates: the upper, on infrapharyn- gobranchial 4, and the lower, on ceratobranchial 5; the branchial arches are otherwise entirely toothless. In Protosalanx the teeth on the palatal tooth- plate and lower jaw are in two rows, as in many other salmoniforms, but in all other salangoids the teeth on each tooth-bearing element except those in the pharynx are restricted to single rows. In largest specimens of Protosalanx the tongue teeth are more or less widely scattered over the surface of the basihyal (as in Fig. 9), but in small- er ones they are restricted to two marginal rows, as in salmoniforms generally. The only other sa- langoid with tongue teeth, Salanx (Leucosoma) reevesi, has them in a single median row on the basihyal toothplate, a unique specialization for salmoniforms. This character is diagnostic of the subgenus Leucosoma. The maxilla and lower jaw are well-toothed in all salangoids; the palate is toothless in Neosa- lanx and Sundasalanx. In Neosalanx the teeth on the premaxilla, maxilla, and lower jaw are very small, and frequently the premaxilla and lower jaw are entirely toothless. In Sundasalanx bony pharyngeal toothplates apparently fail to form, and the pharyngeal teeth appear to be di- rectly attached to the cartilaginous infrapha- ryngobranchial 4 and ceratobranchial 5. The only bony tooth-bearing elements in Sundasalanx ap- pear to be the premaxilla and maxilla; the lower jaw teeth are loosely attached to Meckel's car- tilage. PECTORAL GIRDLE (Figures 18-19) All salangoids have a secondary pectoral girdle (connecting the primary girdle to the back of the cranium) consisting of three dermal bones: post- temporal, supracleithrum, and cleithrum. Post- cleithra are absent except in Salanginae, in which there is a single postcleithrum. In Salangidae, the primary shoulder girdle consists of the entirely cartilaginous paired scapulocoracoids and one or two series of radials. The basic number of pri- mary radials appears to be five in all Salangidae. The first primary radial, associated with the out- ermost (enlarged) pectoral fin-ray, is relatively simple; it is largest in males of Protosalanginae and Salanginae. The other primary radials are complex, with numerous deep divisions approx- imately corresponding in number to the fin-rays. These divisions are most numerous in Salangich- thyinae, particularly Neosalanx, but are well de- veloped in all Salangidae. Comparable divisions or fimbriae occur in the pectoral basal plate of ROBERTS: SALMONIFORM SUPERFAMILY SALANGOIDEA 199 post temporal supracleithrum proximal radials1-5 distal radials postcleithrum 1 mm 1 mm 1 mm FIGURE 18. Left half of pectoral girdle, (a) Protosalanx chinensis, CAS-SU 6306, 158 mm (medial view); (b) Salanx cuvieri, CAS-SU 32454, 69.4 mm (lateral view); (c) Neosalanxjordani, CAS 52058, 43. 1 mm (dorsal view); (d) Salangichthys ishikawae, CAS 6780, 74 mm (lateral view). the salmoniform Dallia pectoralis but are not present in other salmoniforms I have examined and do not seem to have been reported in any other teleosts. Secondary radials, more or less corresponding in number to the pectoral fin-rays, are small and simple. The mesocoracoid is lack- ing in all salangoids except that Protosalanx has a process on the median surface of the scapu- locoracoid that may represent the ventral portion of the mesocoracoid (Fig. 1 8a, medial process). In Sundasalangidae the primary pectoral girdle consists of a U-shaped median scapulocoracoid and a basal plate. Fin-rays are absent. PELVIC GIRDLE (Figure 20) The left and right halves of the pelvic girdle develop in the ventral myotomic wall, and, as the ventral myotomic progression is arrested in Salangidae while the myotomes are still widely separated, the pelvic girdle halves remain widely apart and fail to form any sort of ligamentous or cartilaginous connection between each other. As pointed out by Klyukanov (1975), in Salmoni- formes the two halves of the pelvic girdle are usually joined at least anteriorly for a short dis- tance by strong cartilaginous or ligamentous tis- sues. AXIAL SKELETON (Figure 1) All salangoids have a pair of small dorsal car- tilages straddling the intervertebral disc between the basiocciptal and first vertebral disc; such car- tilages occur in many (perhaps most or all) Sal- moniformes. In all Salangidae the neural arches of vertebrae 1 and 2 are fused dorsally; this condition has not been observed in Osmeridae or any other sal- moniforms I have examined. In Sundasalangidae the neural arches of vertebrae 1 and 2 are sep- arate from each other and morphologically sim- ilar to those of the vertebrae immediately suc- ceeding them. In salangoids the mineralized portion of each centrum is relatively elongated and hourglass shaped, so that the intervertebral joints are nar- row and the notochord greatly constricted. In 200 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 13 PP 1 mm 1 mm FIGURE 19. Pectoral girdle, (a) Sundasalanx microps, CAS 44220, 17-mm adult (posterior view); (b) Elops hawaiiensis, CAS 52035, 30-mm leptocephalus larva (posterior view); (c) Sardina pilchardus, 20-30 mm (ventral view?, after Goodrich 1922); (d) Dallia pectoralis, (lateral view, after Starks 1904; apparently based on CAS-SU 12615, 125 mm, Nushagak River, Alaska); AP = ascending process, CL = cleithrum, F = fin margin, PP = posterior process, PT = posttemporal, R = basal plate, SCL = supracleithrum, SCO = scapulocoracoid. In (b) and (c) the first primary radial has pinched off from the basal plate. salmonids, osmerids, galaxiids, and other Sal- moniformes, especially in the young stages, the mineralized portion of each centrum tends to be relatively short and cylindrical, so that the in- tervertebral space is much larger and the verte- bral section of the notochord is entirely intact. A comparable condition is not present in any salangoid skeletal material I have examined. Ribs are absent or weakly developed and stain poorly. They are small, weakly stained with al- cian when present (Fig. Ib). Gosline (1960) and others have pointed out that neural and hemal spines of most Salmoni- formes, especially posteriorly, may be flattened or laminar, even to the extent of resembling a continuous keel. The neural and hemal spines of salangoids are always relatively slender, espe- cially posteriorly. A round, oval, or elongate and splintlike adi- pose fin cartilage lies at the base of the adipose fin in all Salangidae. A survey of lower teleosts for the adipose fin cartilage by Matsuoka and Iwai (1983) revealed its presence in Salangidae, Osmeridae, Plecoglossidae, Myctophidae, and Neoscopelidae; it was not observed in other low- er teleosts with an adipose fin including Sal- ROBERTS: SALMONIFORM SUPERFAMILY SALANGOIDEA 201 radials1-4 blue 1 mm blue parapelvic cartilages 1 mm 1 mm FIGURE 20. Ventral view of left half of pelvic girdle, (a) Protosalanx chinensis, CAS-SU 6306, 158 mm; (b) Salanx cuvieri, CAS-SU 32454, 69.4 mm; (c) Neosalanx jordani, CAS 52028, 43.1 mm; (d) Sundasalanx microps, CAS 44220, 17 mm (with lateral view of pelvic girdle and parapelvic cartilages above). monidae, Retropinnidae, Prototroctidae, Aulo- podidae, Synodontidae, Chlorophthalmidae, Argentinoidei, Characoidei, or Siluriformes. The similar morphology of the adipose fin cartilages in Salangidae and Osmeridae, as noted by Mat- suoka and Iwai, is possibly indicative of rela- tionship between these two families. Caudal Fin Skeleton (Figure 21) The caudal fin is more or less deeply forked, and the upper and lower lobes are about equal. Principal caudal fin-rays are invariably 10 + 9; upper and lower procurrent caudal fin-rays are moderately numerous (to 14). The complex ural or hypural centrum apparently consists of three centra and uroneural 1 (sometimes also uro- neural 2?) fused into a single unit. The three cen- tra involved are the terminal centrum and post- terminal centra 1-2, according to the nomenclature of Gosline (1960), or preural cen- trum 1 and ural centra 1-2, according to Rosen ( 1 974). In none of the skeletal material examined is there any indication of separate centra poste- rior to the complex hypural centrum. Epurals 0- 3. A separate uroneural 2 is sometimes present, but uroneural 1 is apparently always fused to complex hypural centrum. Free radial or ptery- gial cartilages are sometimes present, usually be- tween ray halves at the base of the anteriormost 2-3 upper or lower procurrent rays and the low- ermost upper and uppermost lower principal rays. Hypurals six. Six separate hypurals occur in Sa- langichthys microdon (Rosen 1 974, Fig. 26). Pro- tosalanx chinensis occurs with hypurals 1-2 and 5-6 separate, but with 3-4 fused near the base. The hypurals are more fused in Neosalanx, Sa- lanx, and Sundasalanx. In Salanx parhypural and hypurals 1-2 are fused near the base; hy- purals 1-2 and 3-4 are fused for their entire length except for oblong basal foramina where fusion evidently failed to complete. In Sundasalanx parhypural and hypurals 1-3 are evidently fused into a single element. SYSTEMATICS In the present account the salangoids are rec- ognized as a salmoniform superfamily separate from osmeroids, which they superficially resem- ble. There are two families, Sundasalangidae, with 202 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 13 uroneural epura| ural centrum opisthural hypurals1-6 neural spines hemal spines hypurapophysis accessory cartilages 1 mm parhypural 1 mm FIGURE 21. Lateral view of caudal fin skeleton, (a) Protosalanx chinensis, CAS-SU 6306, 158 mm; (b) Salanx cuvieri, CAS- SU 32454, 61.7 mm; (c) Neosalanx jordani, CAS 52028, 43.1 mm (note: hypurals 2 and 3, normally separate from each other in all salangoids, are fused in this specimen); (d) Sundasalanx microps CAS 44220, 17 mm. In a-c left half of fin rays removed to facilitate observation of median structures. only a single genus and two species, and Salangi- dae. Salangidae is further divided into three subfamilies, four genera, and eleven species. The genus Salanx is further divided into three sub- genera; this taxonomic category is not employed in the other genera of salangoids. In addition to the new superfamily Salangoidea, the new subfamily Salangichthyinae is proposed for Neo- salanx and Salangichthys, leaving the subfamily Protosalanginae with only the genus Protosa- lanx. No new genera or species are proposed. Some previous workers, particularly Regan (1908b) and Fang (1934a, b) recognized far more species than I have, especially in the subgenus Salanx (genera Salanx and Parasalanx of Re- gan). This is attributable in part to their basing species on only one or a few type-specimens and utilizing characters such as cranial proportions, body depth, and relative position of dorsal and anal fins which vary considerably within the species. Neither Regan nor Fang utilized verte- bral counts, which I find extremely useful in dis- tinguishing species. My extensive data on ver- tebral counts of types and other material are presented in Table 2. My counts of vertebrae, fin-rays (except pelvic) anal scales, branchiostegal rays, and gill rakers are presented in Table 2. This table includes all species of salangoids herein recognized as valid except Neosalanx reganius, which I have not ex- amined. Pelvic fin-ray counts are excluded be- cause they are invariably 5 in Sundasalangidae and almost invariably 7 in Salangidae (6 in one observed specimen of Neosalanx jordani, 8 in two specimens of Salangichthys microdori). Pre- vious authors have presented data on most of the species but have often lumped data from various localities (and frequently of two or more ROBERTS: SALMONIFORM SUPERFAMILY SALANGOIDEA 203 species). In order to minimize this problem, my data are presented separately for each locality. Although Table 2 includes meristic data ob- tained from whole specimens, radiographs, and cleared and stained specimens, the stained spec- imens provide the best material for accurate count of fin-rays, teeth, and gill rakers. It is difficult to observe the jaw teeth and lowermost pectoral fin- rays in Neosalanx except in stained material. In dorsal and anal fin-ray counts the last "two" rays ("divided to base") are counted as one ray. In vertebral counts the basioccipital centrum is not counted and the hypural complex centrum is counted as one. In salangoids, especially in fe- males, the anteriormost anal-fin pterygiophore does not provide a ready basis for distinguishing abdominal and caudal vertebrae. In order to ob- tain additional meristic data from the vertebral column and at the same time obtain more precise data on the relative position of fin origin, I have taken data on the vertebrae parallel to the origins of the pelvic, dorsal, and anal fins. The number of vertebrae posterior to a vertical line through the base of the last anal fin-ray is also recorded. Radiographs are usually satisfactory for ob- taining vertebral counts of salangoids and some- times for fin-ray counts. Sometimes the verte- brae may show up very faintly but it is almost always possible to obtain a count repeatable to within one vertebra. Fin-rays, however, fre- quently cannot be accurately counted on radio- graphs, and I have only incorporated data on fin-ray counts taken from radiographs when the radiographs seemed reliable. Some characters utilized by other workers to distinguish species are not emphasized here be- cause they do not seem useful. This particularly applies to pectoral fin-ray counts in Neosalanx and to the elongation of the head or cranium, relative position of the dorsal- and anal-fin bases, and body depth, especially in Salanx. In salan- gids the number of pectoral fin-rays generally continues to increase slightly with growth, es- pecially so in those such as Neosalanx, in which the rays are exceptionally numerous. The elon- gation of the cranium (particularly its anterior portion) is extremely variable in Salanx, as not- ed also by Wakiya and Takahasi (1937:289). This variation is individual and is probably enhanced by growth. The position of the dorsal and anal fins relative to each other is also highly variable in salangids, subject to individual variation as well as sexual dimorphism. In defining species of Salanx too much reliance has been placed on slight differences in fin positions based on only one or two specimens. Salanx, Salangichthys, and other salangids vary enormously in body depth due to sex-related body changes and non- sexual factors of condition and preservation. In discussing salmonoid classification, Gosline (197 1:1 19) stated: The suborder Salmonoidei as here recognized (Families Sal- monidae, Osmeridae, Plecoglossidae, Salangidae, Retropin- nidae, Aplochitonidae, and Galaxiidae) is a group of highly diverse inshore and freshwater salmoniform fishes. Though the included families no doubt should be divided into su- perfamily groupings, inadequate knowledge of the Salan- gidae and the Southern Hemisphere forms would seem to make any formal superfamily classification premature at the present time. Informally, the members may be divided between Northern and Southern Hemisphere forms. The diverse forms from the Southern Hemisphere seem to be most closely related to the northern osmerids. . . . The Northern osmeroids are represented by four quite distinct lines: Salangidae, Plecoglossidae, Osmeridae, and Salmon- idae. Rosen (1974) divided the suborder Salmo- noidei into two superfamilies, Salmonoidea— in- cluding the Southern Hemisphere families (ex- cept Retropinnidae) and Salmonidae — and Osmeroidea (with four families listed as incertae sedis: Osmeridae, Plecoglossidae, Retropinni- dae, and Salangidae). I have not investigated Retropinnidae or the highly aberrant Plecoglos- sidae but suspect that Retropinnidae (particu- larly Prototroctes) and Plecoglossus may indeed be closely related to each other and perhaps to Osmeridae. But I have not been able to find any good evidence (in the form of shared specializa- tions or derived characters) between Salangidae and any one or combination of these families. I have therefore designated the new superfamily Salangoidea, which is coequal with the superfam- ilies Osmeroidea and Salmonoidea (and Galax- ioidea, if this is also to be recognized). SALANGOIDEA, NEW SUPERFAMILY This superfamily apparently differs from all other Pisces in having a suspensorium in which the cartilaginous palatine and pterygoid (of the mandibular arch) and quadrate and hyomandib- ular (of the hyomandibular arch) are fused into a single element, the hyopalatine. Gill arches with well-developed fourth hypobranchials— so far as known absent from all other adult teleosts (Nel- 204 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 13 son 1970). Pharyngobranchials 4. Anterior pha- ryngobranchial modified as an elongate "suspen- sory pharyngeal"; only fourth pharyngobranchial bears teeth (teeth absent in Sundasalanx). Fifth ceratobranchial with well-developed teeth (ab- sent in osmeroids; McAllister 1963:4). Bran- chiostegal rays 2-5 (7-19 in salmonoids, 5-10 in osmeroids, 3-9 in galaxioids). Cranium mod- erately to excessively flattened (more so than in any other salmoniforms). Maxillary bone, bear- ing teeth for its entire length, with its posterior half abruptly curved medially beneath head (so that teeth on posterior half of maxillary are di- rected anteriorly rather than ventrally). Scales entirely absent except for a row of strongly ad- herent anal scales in adult male Salangidae. Dermosphenotic and circumorbital bones ab- sent, except for a single small troughlike bony element observed in Neosalanx, which may rep- resent a fifth or sixth circumorbital (not dermo- sphenotic). Supraocciptal bone absent (present in most other salmoniforms). Pectoral fins pedunculate throughout life (with pectoral radials in a fleshy pedestal separate from body). Pelvic fin-rays usually 5 or 7 (rarely 6 or 8; 8 in osmeroids). Principal caudal fin-rays in- variably 10 + 9 (as in most lower teleosts includ- ing salmoniforms with generalized caudal fins; galaxioids have fewer). Salangoids apparently have no laterosensory canals on the body. The cephalic laterosensory canals, although well developed, are superficial (i.e., not enclosed in bony tubules) and often dif- ficult to observe in their entirety. Those of Sa- lanx chinensis, illustrated by Nelson (1970, Fig. 1 5), do not exhibit any particularly unusual fea- tures for lower teleosts. There are preopercular, mandibular, supraorbital, infraorbital, and ex- trascapular canals. The mandibular is not con- tinuous with the preopercular. The supraorbital and infraorbital extend anteriorly only a short distance in front of the nostrils, i.e., not signifi- cantly onto the greatly depressed and enlarged snout. The infraorbital has 8 pores, the preoper- cular 6, and the mandibular 5. Alimentary canal a relatively simple, straight tube. Pyloric caecae absent. Gonads paired. Salangidae Jordan and Snyder, 1 902 Pelvic fin almost invariably with 7 rays (8 ob- served in one specimen of Salangichthys ishi- kawae and two S. microdon, 6 in one Neosalanx jordani). Pelvic girdle without parapelvic carti- lages. Pectoral fin-rays 8-34. Pectoral girdle with five proximal radials; distal ends of one or more proximal radials with more or less numerous branches; adult males with a series of anal scales and enlarged, modified anal fins; total vertebrae 49-79. The family Salangidae comprises three subfamilies: Protosalanginae, Salangichthyinae, and Salanginae. Protosalanginae Wakiya and Takahasi, 1937 This subfamily, here restricted to the mono- typic genus Protosalanx, differs from all other salangoids in having the premaxilla, palatal toothplate (=ectopterygoid?), and dentary with two rows of teeth instead of at most a single row; the basihyal toothplate of the tongue also has the teeth in two marginal rows (a primitive condition for salmoniforms) or irregularly scattered over its entire surface; the only other salangoid with basihyal teeth has them in a single median row. Pelvic fins relatively larger and more anterior than in any other salangoids (see Fig. 1, Table 2). Cranium strongly depressed (almost as much as in Salanginae); adults with anterior portion of cranial fontanel closed, posterior portion of cra- nial fontanel greatly reduced but remaining open throughout life (both portions closed in adult Salanginae, open throughout life in Salangich- thyinae and Sundasalangidae). Lower jaw weakly projecting beyond upper jaw; premaxillae pro- jecting anteriorly beyond snout tip as in Sa- langinae but failing to form a membrane-covered space through which symphyseal teeth of lower jaw project. Lower jaw without enlarged sym- physeal teeth (present in Salanginae), sometimes with a weakly developed fleshy presymphyseal process but without presymphyseal teeth or bony process. Adults attaining slightly greater stan- dard length (Table 1) and heavier-bodied than any other salangoids. Dorsal fin-rays 16-18 and anal fin-rays 30-32 (vs. 10-15 and 14-32 in all other salangoids); vertebrae 66-70 (Table 2). Protosalanx Regan, 1 908 Eperlanus BASILEWSKY, 1855:242. Salanx ABBOTT, 1901:490. Protosalanx REGAN, 1908b:444 (type-species, by monotypy, Salanx hyalocranius ABBOTT, 1 90 1 = Eperlanus chinensis BASILEWSKY, 1855). Paraprotosalanx FANG, 1934a:246 (type-species, by mono- ROBERTS: SALMONIFORM SUPERFAMILY SALANGOIDEA 205 typy, Protosalanx andersoni FANG, 1934a (non RENDAHL, 1923) = Protosalanx chinensis BASILEWSKY, 1855). Protosalanx chinensis (Basilewsky, 1855) Eperlanus chinensis BASILEWSKY, 1855:242 (type-locality "in sinu Tschiliensis habitat" [not "Pekin" as usually cited]). Salanx hyalocranius ABBOTT, 1901:3490 (type-locality Pei-ho at Tien-tsin). Protosalanx hyalocranius REGAN, 1908b:445. Paraprotosalanx andersoni FANG, 1934a:246 (Figs. 4-6, text in part [non Paraprotosalanx andersoni RENDAHL, 1923]). Protosalanx chinensis CHYUNG, 1961:163. MATERIAL EXAMINED. -BMNH 1929.2.5.2.-3, 61-65 mm, Kiangyin; CAS 52026, 60:48.2-70.7 mm, no locality (pur- chased in San Francisco); CAS-SU 6306, 25:80.5-163 mm, Pei-ho at Tien-tsin, paratypes of Salanx hyalocranius (7:85.5- 158 mm alcian-alizarin); CAS-SU 23639, 1:137 mm, Seoul; CAS-SU 36025, 3:120-136 mm, no locality; UMMZ 180096, 2:127-129 mm, Korea; USNM 120746, 2:129-132 mm, Ko- rea. Protosalanx appears to be the most primitive salangoid. There is no indication that it com- prises more than a single species. Although Ab- bott's account begins "Salanx hyalocranius new species," it concludes "this species is probably identical with Eperlanus chinensis Basilewsky, from Pekin, but the name chinensis is already used for the 'whitebait of Makao' " (Abbott 1 90 1 : 490-491). In Abbott's time Salangidae were so poorly known it was reasonable for him to as- sume that his material might represent an un- described species, but even so it is clear from this statement that Abbott was really proposing a re- placement name. Now that Salangidae are better known it seems Basilewsky's account could only refer to this species, as explicitly recognized by Wakiya and Takahasi (1937), although they re- tained the name P. hyalocranius. The holotype of P. chinensis cannot be found (Barsukov, pers. comm. 1983). Since the "whitebait of Makao" has been referred to as Leucosoma or Salanx chinensis but never as Eperlanus or Protosalanx chinensis, the epithet chinensis is available for a species of Protosalanx. As this is also the earliest name proposed it must replace hyalocranius, and the species should be known as Protosalanx chi- nensis. The only publication to come to my at- tention in which this name is correctly applied is by Chyung(1961). Wakiya and Takahasi (1937) correctly iden- tified Paraprotosalanx andersoni Fang, 1934a with this species. Fang's figures agree in every respect with P. chinensis. The fleshy presymphy- seal appendage, presumed by Fang to differen- tiate his Paraprotosalanx from Protosalanx, is also present in some of Abbott's type-specimens of S. hyalocranius. Fang's figures presumably are based upon the single large male, "S. 4374," 153 mm (total length according to Table 4, but stan- dard length according to p. 247) from Nanking. All or almost all of the other specimens referred to Paraprotosalanx andersoni by Fang are prob- ably Neosalanx. It should be noted that small specimens in museum collections identified as Protosalanx are usually Neosalanx and that all or almost all pub- lished reports of smaller Protosalanx up to the present time are based on Neosalanx. For ex- ample, I find that all of the small specimens in Abbott's type-series of S. hyalocranius are Neo- salanx. Young P. chinensis are relatively rare in collections. Those I examined (smallest 48.2 mm) closely resemble the largest adults in every way except they lack the sexually dimorphic char- acters of adult males. The strongly pointed snout and large teeth arranged in two rows on the pal- ate, tongue, and lower jaw are easily observable. Neosalanx have no teeth on the tongue or palate, and the jaw teeth except on the maxillary are absent or minute and difficult to observe, while the males are sexually mature and provided with greatly enlarged anal fins and anal scales at rel- atively small size. The smallest male Protosalanx with anal scales is probably considerably larger than any Neosalanx. Protosalanx chinensis appears heavier-bodied at all sizes and to attain a greater size than any other salangoid. The 163-mm specimen is the largest that has been reported. Salanginae Regan, 1908b Cranium and especially ethmoid plate very strongly depressed and elongate, more so than in any other salmoniforms. Adults with cranial fon- tanel entirely closed (posterior and sometimes also anterior portion of cranial fontanel open throughout life in all other salangoids). Upper and lower jaws with strongly pointed or project- ing tips. Teeth relatively large and few in num- ber. Premaxillae projecting beyond concave an- terior margin of ethmoid plate to form a membrane-covered space penetrated by enlarged symphyseal teeth of lower jaw. Lower jaw often with a fleshy or bony presymphyseal process and presymphyseal teeth (Wakiya and Takahasi 1937, pi. 20, figs. 31-34). Supramaxilla absent 206 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 13 (present in all other salangoids). Pectoral fin-rays 7-1 1 (20-32 in all other Salangidae). Pectoral girdle with a single postcleithrum (absent in all other salangoids). Body extremely elongate, more so than in any other salmoniforms. Basal portion of neural and hemal arches expanded, frequently covering centrum laterally and fusing with each other. Distal portion of neural arches with an- terior and posterior projections, those of succes- sive vertebrae articulating with each other. Ver- tebrae 68-79 (37-70 in all other salangoids). The enlarged symphyseal teeth of the dentary and the membrane in the roof of the mouth through which they project presumably form a puncturing device to kill prey. Similar devices, with foramina in the roof of the mouth through which fanglike lower jaw teeth pass, occur in other piscivorous or predatory teleosts, including the characoids Hepsetus, Hop lias, and Acestro- rhynchus (Roberts 1969). In many respects Salanginae appear to be the most highly specialized members of the family. Even the low pectoral fin-ray count, which might be considered primitive, may be secondarily evolved, since primary pectoral radials 2-4 of Salanginae exhibit the distally fimbriate or dig- itate condition that is probably associated with the retention of pedunculate pectoral fins (prob- ably a neotenic character) and evolution of large numbers of pectoral fin-rays (a specialization shared by all other Salangidae). SalanxOken, 1817 "Les Salanx" CUVIER, 1817:185 (French vernacular; not avail- able for zoological nomenclature). Salanx OKEN, 1817:1183 (Latinization of Cuvier's "Les Sa- lanx," and the earliest name available for zoological no- menclature; see ICZN Declaration 87, paragraph 12. Type- species, by monotypy, Salanx cuvieri VALENCIENNES, 1 849). Leucosoma GRAY, 1831:4 (type-species, by monotypy, Leu- cosoma reevesi GRAY, 1831). Hemisalanx REGAN, 1908b:444 (type-species, by monotypy, Hemisalanx prognathus REGAN, 1 908b). Parasalanx REGAN, 1908b:444 (type-species, by subsequent designation of FANG, 1934a:259, Parasalanx gracillimus REGAN, 1 908b = ISalanx cuvieri VALENCIENNES, 1 849). Reganisalanx FANG, 1934b:509 (type-species, by monotypy, Reganisalanx normani FANG, I934b = Salanx ariakensis KJSHINOUYE, 1901). Metasalanx WAKIYA AND TAKAHASI, 1937:293 (type-species, by monotypy, Metasalanx coreanus WAKIYA AND TAKAHASI, 1937, a nomen nudum). The four species herein recognized as consti- tuting the genus Salanx have been placed by other authors in three genera, Salanx, Hemisa- lanx, and Leucosoma. Wakiya and Takahasi (1937) even placed Hemisalanx in a subfamily of its own, Hemisalanginae, regarded by them as intermediate between Protosalanginae and Sa- langinae. Because these four species differ strik- ingly from all other salangids in several features of skeletal anatomy but agree closely with each other in conformation of the cranium and jaws, distribution and size of jaw teeth, number of pectoral fin-rays, and the peculiar modification of their neural and hemal arches and high ver- tebral counts, I prefer to recognize them as be- longing to three subgenera in the sole genus of the subfamily Salanginae. Salanx (Salanx) ariakensis (Kishinouye, 1901) Salanx ariakensis KJSHINOUYE, 1 90 1 :359 (type-locality Ariake Bay, Kiushiu). Salanx acuticeps REGAN, 1908a:360 (type-locality Lake Can- didius, Formosa). Parasalanx acuticeps REGAN, 1908b:446. Parasalanx longianalis REGAN, 1 908b:446 (type-locality Liao- ho, northern China). Parasalanx annitae VAN DAM, 1926:342 (type-locality Pei- taiho, China). Reganisalanx normani FANG, 1 934b:509 (type-locality Ichang, as herein restricted). MATERIAL EXAMINED. -AMNH 10327, 7:125-147 mm, Hunan; BMNH 1888.5.15.1 1-12, 2:141-143 mm, Ichang (lec- totypeandparalectotypeofT?. normani); BMNH 1898.2.8.20- 23, 4:114-123 mm, Liao-ho, northern China (syntypes of P. longianalis); BMNH 1904.4.2835-36, 2:116-118 mm, Lake Candidius, Formosa (syntypes of S. acuticeps); BMNH 1927.3.26.3, 125 mm, Nanking; BMNH 1928.6.22.6, 1 15 mm, Wenchow; CAS-SU 8574, 2:99.1-104 mm, Ariake Sea (iden- tified by Kishinouye); CAS-SU 23103, 107 mm, Maruyama, Taihoku, Formosa; ZMA 1 12.587, 128 mm, Peitaiho, China (holotype of P. annitae). In vertebral counts and in all other respects so far as known the four syntypes of P. longianalis agree well with other material herein referred to as Salanx ariakensis, except for their consis- tently high anal fin-ray counts of 30-32 (reported by Regan 1908b:446). Most samples of S. ari- akensis examined have only 26-29 anal fin-rays, but two specimens from Ariake Bay have 27 and 31. Reganisalanx normani is based primarily on the description by Regan (1908b) and supple- mentary notes by Fang (1934b:509) of two spec- imens from Ichang (BMNH 1888.5.15, 11-12), identified by Regan (ibid.) as Salanx cuvieri. Fang declared that the specimens represented a dis- tinct genus but did not provide a proper generic ROBERTS: SALMONIFORM SUPERFAMILY SALANGOIDEA 207 diagnosis or description; apparently he distin- guished it from Salanx based on the lack of a presymphyseal bone. In my opinion the char- acter cannot be used to split the genus Salanx. I have reidentified these specimens as S. ari- akensis, a species in which the presymphyseal bone may be present or absent. I have not seen the third specimen referred to R. normani by Fang (ibid.). It is clear from Fang's account that he did not compare this specimen directly with the two specimens from Ichang, and it might not be conspecific. In order to fix the identity of this nominal taxon, the 141 -mm undamaged speci- men from Ichang (BMNH1888.5.15.11)is here- by designated the lectotype. The 143-mm spec- imen, with the body damaged just behind the head and at mid-abdomen, is a conspecific para- lectotype (BMNH 1888.5.15.12). Fang (1934a) reported 1 1 specimens (as Para- salanx longianalis) with the following anal fin- ray counts: 28(5), 29(2), 30(3), 32(1). The ver- tebral counts are unknown for these specimens but it seems likely from Fang's account that they are all S. cuvieri. The holotype of P. annitae has the head rel- atively short and broad (for the subgenus Salanx) and in this respect is more like S. ariakensis than S. cuvieri. A presymphyseal bone is present, but it is short considering the large size of the spec- imen, and has only 2 teeth on each side. The premaxilla has 7 teeth, maxilla 1 2, and dentary about 10. Salanx (Salanx) cuvieri Valenciennes, 1 849 Salanx cuvieri VALENCIENNES in CUVIER AND VALENCIENNES, 1849:360 (type-locality unknown). ^Parasalanx gracillimus REGAN, 1 908b:446 (type-locality Shanghai). Parasalanx angusticeps REGAN, 1908b:446 (type-locality China). Parasalanx cantonensis HERRE, 1932:425 (type-locality Can- ton). MATERIAL EXAMINED. -AMNH 51689, 3:88.6-106 mm, Canton; BMNH 1855.9.19.1539, 144 mm (holotype of P. an- gusticeps); BMNH 1891.1.31 .20, 1 1 1 mm, Shanghai (holotype of P. gracillimus); BMNH 1936.10.7.13, 1 19 mm, Sharp Peak, Fukien; CAS 52057, 4:76.5-98.0 mm, Hong Kong (1 alizarin); CAS-SU 225732, 112 mm, Canton (holotype of P. cantonen- sis); CAS-SU 32454, 18:56-66 mm, Chuan Is. (4:61.7-69.4 mm alcian-alizarin); CAS-SU 32943, 117 mm, near Pakhoi, SW Kwangtung; MNHN 9900, 1 12 mm, no locality (holotype). So far as I have been able to determine, vari- ation in the presymphyseal bone within each species, including its presence or absence and its length or amount of dentition, is correlated chief- ly with size and is not sexually dimorphic. NOTES ON HOLOTYPE.— The holotype (Fig. 2a) is dried but complete and in fair condition. The body immediately posterior to the head is badly damaged and fin-rays brittle, so it must be han- dled with care. Cranial width (at anterior margin of eyes) 3.5 in cranial length. Presymphyseal bone, 2.1 mm long, with 1-2 moderately large teeth basally and at least 2 minute teeth distally. Pre- maxilla considerably elongated anteriorly, with 7-8 teeth. Maxilla with about 7 teeth. Dentary with about 1 3 teeth of variable size. Palatal teeth 7, very small and in a single row. The following proportional measurements are expressed as times in standard length. Length of cranium about 7; length of head (to end of gill cover) 4.7; length from anterior midline of ethmoid plate (concave) to anterior rim of orbit 16; length from tip of upper jaw (premaxilla) to anterior rim of orbit 1 0; diameter of eye (slightly shrunken) approx- imately 28. NOTES ON SYNONYMY.— P. angusticeps is dis- tinguished by Regan primarily on the basis of its exceptionally elongate head: "head nearly 4 times as long as broad; snout a little longer than post- orbital length of head" versus head 3 times or a little more than 3 times as long as broad, and snout only as long as or a little shorter than post- orbital length of head in all other Parasalanx and Salanx (Regan 1908b:445-446). The den- tition of the holotype of P. angusticeps, a gravid female of 144 mm, is complete and undamaged. Presymphyseal bone elongate with 5-6 teeth on each side; premaxilla with 7 teeth; maxillary teeth 10 or 1 1; dentary with a small tooth anteriorly (just behind symphysis), then an enormous ca- nine tooth, followed by 7 small teeth and 6 mod- erately large teeth; palatine with 8 small teeth in a single straight row. The holotype of P. gracillimus is in poor con- dition, dried, twisted, and slightly shrunken. Its body depth, reported as 18 times its length, is attributable to the poor condition (emaciation) of the specimen. Its dentition is as follows: pre- symphyseal bone with 3 teeth on each side, pre- maxillary 5, maxillary 8, dentary with 1 mod- erately large, 6 small, and 5 moderately large, and palatal 7 moderately large. The vertebral column is broken anteriorly, making all of the counts based on vertebrae doubtful. Wakiya and Takahasi (1937:288) tentatively placed P. gra- 208 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 13 c 13 < 8 < 2 i , N ,, (N l r->\o« c^ \o t— r— X O — 1 'V *V '^000 O (N — O OvO " 00^0x1-- • ci 1 | o ^ u 2 ca ca E V3 0. EX _ G ~ ^ VO — • o j^. Tf r~ VO rs ^ 3| £ ^-2^^^-12:71 u 7 "~~ 7 7 os — ' Tt — ' 2 070 — o ~ TT .- I £ 2 2 1 "> 1 1 1 1 1 "> 1 1 '3 ' J I ' 1 m m m m m 03 — VO ca ^j c •« 7 I I I I i i i £ I r-j 1 I 1 1 I' ^ 1 O OOO o r-~ *O (N OO OO »O t> >% •^- i i i j i i ^ OO 7 1 ^V 1 *^ *Y ^O 1 1 o ooo £ S vo ^ (N (N (N /"» "O oo >/~i r^ ^^ Ov — vo c ^ o 7^^^-7^§-^'7 ^7^7 7 £; 777 (N — (N — < s (N ^ fN <^ ^ (N fN (N fN fN n TJ- »n m Tt m m m m ^ m u^ TJ- o 15 •** m "J'm'^-"YrfTt~j'm~ m ~j" TJ- ~j" ~j" m 777 7 777 Q 2 2 2 2 — "" ~ "" 2 2 "" (N O (N (N 3 -a c ^^^•7^,^,7 oo °f 7 <* a, 0, Ov ON * TTT u "° > VO VO VO *n w-t oo oo oo m CO r**i m (N C-4 Tf (N Tf r^ r-j oo to •^- Tj- Tj- r^ OO "— " Csl fT\ • 03 O ^ ^T ^ | r*"> OV (N 1 ^ 1 vA «N i c o a . 4? 2 o ° P; J.fJjr^^^mmH^ (N 7 (N 7 7 00 ^0^0 r~ "** ^ vo Ov «-, J«f- TT 70 — 07 — — 77 (N (N ""> fl- m' rr^ P m -^ 2 £ rr, — mr^w-iSooS^^S 2 m £ oo ™ vo — £ VO vo £ m ? £ H 1 vo S"1^ " S^^ S£ K C^ ^ ^ ^ £ S "° 2 8S X S8§ > g ^ g ™ C-C- S" 5" ^ "^ Ci vO^ vo Cl m m C. «n m o' 0^ c c -^ 00 O e "> 'S I c - •^ ^ c e'CJ O^^*^ O-^ '^^ C 3fg 3 Ofcq &o^^^. ? 00 ' S Illl -. ca 8*J. 1 5' 1 S' i - s' § 5 1 1 ^ 1 g's'jjjfj! ill 1 3SS {°- § 0 0 0 | S ||||||^^cA| ^ II M M c2 '| ^ ^ '| £ £ S 1 S 1 £ £ £ |z !<<<< "15^ % <&*-< FIGURE 8. Wounds inflicted upon pinnipeds by white sharks. It is possible to hypothesize the posture of prey and the attack behavior of the shark from bite scars. Elephant seals (Mirounga angustirostris): (A) Subadult survivor at Afio Nuevo Island, California. Note lower jaw puncture wounds and tearing caused by upper jaw teeth. Photo by R. Bandar. (B) Adult female survivor at Southeast Farallon Island, California. Again note lower and upper teeth wounds. Photo by S. H. Morrell. California sea lions (Zalophus californianus): (C) Adult survivor at Afio Nuevo Island. Photo by R. Bandar. (D) Subadult male carcass (left) and Richard Ellis (right) at Afio Nuevo Island. Photo by Pam Wing. (E) Subadult male carcass along the central California coast. Photo by R. Bandar. Southern fur seal (Arctocephalus doriferus): (F) Large adult male survivor at South Neptune Island, South Australia. Photo by J. McCosker. particularly when one considers the euryphagic diet of the fish. PREDATOR-PREY RELATIONSHIPS.— The stom- ach contents of nine white sharks (193-51 1 cm total length) captured in northern and central California waters are presented in Fig. 10. Seven- ty-eight percent of the sharks had recognizable food items in their stomachs. The most frequent prey was the California bat ray (Myliobatis cali- fornica), found in four stomachs; other fish prey were less frequent in the diet. Fifty-six percent of the sharks examined contained elasmo- TRICAS AND McCOSKER: PREDATORY BEHAVIOR OF THE WHITE SHARK 231 FIGURE 9. Sea otters (Enhydra lutris) from the central California coast. Above, adult in normal feeding or basking posture along the edge of a kelp bed in Monterey Bay. Photo by J. McCosker. Below, lacerated carcass from which several white shark tooth fragments were removed, suggesting that the animal was bitten at the surface while in a belly up, prone position. Pismo Beach. Photo by J. Ames. branchs, and 44 percent contained teleost prey species. No evidence of predation on marine mammals was found in the nine sharks. Although the white sharks took prey that nor- mally occur in both pelagic and inshore habitats, the two most frequent prey are generally asso- ciated with demersal inshore communities. The California bat ray (M. californicd) is common in bays and inshore sandy habitats 2-50 m deep, where it feeds on benthic sand-dwelling inver- tebrates. The spiny dogfish (Squalus acanthias) is also demersal, being found in both shallows and deeper offshore waters. Other prey species that live on the bottom in inshore areas are the lingcod (Ophiodon elongatus) and the cabezon (Scorpaenichthys marmoratus). These latter two species are relatively sedentary, have small home ranges, and show cryptic coloration. Limbaugh 232 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 14 Myliobatis California IE) Squ.lu. acanthi, 3 IE) Cetorhinus maxlmut IE) Cynoscion noblll* Galeorhintis zyopterus (E) Ophlodon elongatus % Occurrence FIGURE 10. Stomach contents of nine sharks captured in northern California waters on record at California Academy of Sciences. % occurrence = percentage of the nine shark stom- achs that contained that prey item. E = elasmobranch, all oth- ers are teleosts. (1963) reported cabezon from the stomachs of three immature sharks captured at La Jolla, Cal- ifornia, and described a number of incidents that indicate 5". marmorafws is an important prey for young white sharks. Earlier researchers ques- tioned how sharks could detect and capture such inconspicuous and apparently inaccessible prey; our studies suggest that weak electric fields might be involved in prey detection (see Sensory Bi- ology section below). The white sea bass (/l/rac/osr/on (=Cynosciori) nobilis) also occurs in shallow rocky inshore hab- itats, and is often found among canopies of giant kelp (Macrocystis pyrifera). Unlike the majority of the other inshore prey species, however, it occasionally swims in the water column as well as on the bottom. White sharks have been reported to feed on the carcasses of captured basking sharks (Fast 1955), although we know of no published ac- counts of predation under natural circumstances. However, potential vulnerability of basking sharks to large predators was suggested by Lim- baugh (1963) in an account of a dead basking shark with a large wound probably inflicted by killer whales. Basking sharks, which reach lengths of more than 1 1 m, are found seasonally in off- shore waters of central and northern California. From aerial surveys made over a 2.5-yr period near Monterey, California, Squire (1967) found that basking sharks were most common from September through May, when water tempera- tures were generally below 14°C. White shark Fish Pinnipeds Cetaceans Other prey LJ Elasmobranchs LJ Actinopterygians % Occurrence FIGURE 11. Stomach contents of 33 white sharks. Data combined from this study and other published records. % oc- currence = percentage of the 33 sharks that contained the prey category. Fish prey subdivided into elasmobranchs and rayed- fin fishes (teleosts and sturgeons). Other prey include birds, crustaceans, and sea turtles. sightings, however, were most common in the warmer-water months of May through August, when water temperatures neared or exceeded 1 4°C. The cause of the seasonal disappearance of basking sharks from the coastal waters of Cal- ifornia remains unknown. Other prey that in- habit pelagic waters include the soupfin shark (Galeorhinus zyopterus), the Pacific sardine (Sar- dinops sagax), and occasionally bat rays (Myl- iobatis californicd) (Roedel and Ripley 1950; Federetal. 1974). Combined data on the food habits of 33 white sharks from this study and other published rec- ords are shown in Fig. 1 1 . Here again, fish were the most frequent prey items, occurring in over half of white sharks in the analysis. Elasmo- branchs and rayed-fin fishes (teleosts and stur- geons) comprised equal proportions (each oc- curred in 30 percent of sharks analyzed) of the piscine prey. Pinnipeds were also a major com- ponent in the diet of sharks, while cetaceans and other prey groups were less common. Bass et al. (1975) provided the only other gut content data from white sharks useful for comparison. They too found both elasmobranchs (40 percent of sharks examined) and teleost fishes (25 percent) as the most common prey items, although little information was given on specific identification. Figure 1 2 shows the distribution of fish and mammal prey in relation to shark size. Fish prey predominated in the diet of sharks approxi- mately 3 m or less (TL), while pinnipeds and cetaceans predominated in those of larger sharks. This shift in diet may occur for a number of reasons. For example, larger sharks are less agile and would be less successful in chasing and cap- TRICAS AND McCOSKER: PREDATORY BEHAVIOR OF THE WHITE SHARK 233 O Pinnipeds & Cetaceans • Fish o oooooo oo o o 100 200 300 400 500 600 Total Length (cm) FIGURE 12. The relationship between white shark length and prey type. Data taken from stomach contents of the 33 specimens in Fig. 1 1 . turing smaller fish prey that dart about when pursued. Larger sharks may thus switch to dif- ferent prey types and associated new hunting modes. In addition, the energetic requirements of large, warm-bodied sharks may be better met by prey high in fat content (i.e., high-energy- density prey). Carey et al. (1982) estimated the metabolic rate for a 4.6-m white shark, and con- cluded that the animal could survive for ap- proximately 1.5 months on 30 kg of whale blub- ber (a conservative meal size). They suggest this to be adaptive during long intervals between en- counters with prey. Although little is known of the movements of white sharks, they do show seasonal peaks in abundance in California waters (Squire 1967; Ainley et al. 1981), which might indicate some sort of regional or long-distance movement. Morphological differences between large and small sharks may also account for different pred- atory tactics. Fig. 1 3 shows the relationship be- tween tooth shape and shark total length. Smaller sharks have a relatively long, narrow tooth shape that is better adapted for grasping prey like small fishes. This feature is so well developed in small white sharks that they are often incorrectly iden- tified as mako sharks (hums spp.) (Smith 1951, 1957). At about 3 m TL, the teeth broaden at the base and take on the diagnostic triangular serrated form. Unlike the long narrow teeth, this shape is well-suited for gouging and cutting pieces from prey too large to swallow whole. Le Boeuf et al. (1982) found evidence that marine mam- mals were the only prey of large white sharks they examined from California. Of seven spec- imens examined, all but one were approximately 4 m or longer and had evidence of marine mam- mals in their stomachs. The only exception was the smallest shark (2.4 m TL), which had only a 10-cm patch of pinniped pelage in its stomach. Total Length ( m) FIGURE 1 3. The relationship between shark total length and tooth shape. Tooth shape expressed as the ratio of width of enamel base to medial height of enamel for the first tooth, right side, upper jaw of 16 sharks. Low ratio indicates a long narrow tooth shape; higher ratio indicates relatively broad triangular shape. Perhaps this shark's teeth were too narrow to excise a portion of flesh. In California waters, elephant seal populations at offshore rookeries peak in both the spring and winter months (Le Boeuf et al. 1 974), but almost no predation occurs during the spring peak. Hy- pothetical explanations advanced to explain this seasonal discrepancy in predation include either: 1 ) sharks fasting while breeding; 2) water too cold for sharks to feed; or 3) emigrations of sharks from the area. Even though sharks occur in Cal- ifornia waters during the spring (Miller and Col- lier 1981), the decrease in shark attacks is prob- ably due to emigrations of large sharks from coastal areas (see Squire 1967). Adult male seals are more susceptible to shark predation because they spend more time in the water near the rook- ery during the breeding season than do females (Le Boeuf et al. 1982). It is possible that the loss of peripheral males to sharks may not adversely affect the population because of the polygynous mating system of the elephant seal, where rela- tively few dominant males do the majority of the breeding. Although it is clear that white sharks do nor- 234 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 14 FIGURE 14. Underwater photo of a male white shark (ap- proximately 3.5 m TL) in a "tail stand" posture with snout directly over zinc anode on rudder of study vessel. Photo by T. Tricas. mally prey upon elephant seals, the significance of the interaction is not evident. Ainley et al. (1981) reported an increase in the number of attacks on elephant seals at the Farallon Islands between 1970 and 1979, but their data indicate a density-dependent relationship between num- ber of attacks and numbers of elephant seals. More data are needed on the mortality rates of attacked seals and on numbers in the shark pop- ulation before any effects of shark predation on elephant seal populations can be quantitatively assessed. SENSORY BIOLOGY.— Our cursory field exper- iments and observations qualitatively indicate white sharks are sensitive to electric fields. In the pulsed electric field tests, sharks took the exper- imental bait 8 times (73 percent) and the control 3 times (27 percent). In the constant current (DC) tests the experimental was taken 4 times (44 per- cent) and the control 5 times (56 percent). Al- though our sample size was too small to show any statistically significant preference for baits with either type of electric field, sharks did take baits with the pulsed electric field almost three times more often than the control. The sharks also appeared to be more responsive to pulsed fields than to continuous fields. Kalmijn (1971, 1974) reported that sharks were most responsive to weak electrical fields at frequencies from 0 (DC) to 8 Hz. We also observed the behavior of sharks to metallic objects attached to the bottom of the boat. On three occasions one of us (TCT) watched from underwater a 3.5-m shark approach zinc plates attached to the boat's rudder and assume a near vertical "tail stand" posture (Fig. 1 4). The shark remained upright for approximately 10- 20 s as it waved its snout approximately 5-10 cm above the zinc. Sharks were also observed several times to swim back and forth with their snouts very near a 1 0-m-long copper grounding strip on the bottom of the boat's hull. We interpret these observations as a response by sharks to the galvanic currents produced by the electrochemical interaction between the me- tallic plates and seawater. White sharks have a well-developed system of ampullae of Lorenzini (Fig. 15), and although the role of electric detec- tion of prey by sharks is well demonstrated (see Kalmijn 1978, 1982), the degree of importance for such a sensory modality in white sharks re- mains unknown. It is noteworthy, however, that electric fields produced by large mammals (e.g., humans and presumably pinnipeds) in seawater are well within the sensory range of elasmo- branchs (Kalmijn 1971). Perhaps young white sharks are able to detect electrically sedentary camouflaged fish prey like the cabezon (Scor- paenichthys marmoratus). It also seems reason- able that the ampullae would be particularly use- ful to detect: 1) the location of a marine mammal at the moment just prior to attack; 2) any change in position or escape attempts by the prey; and 3) any change in the prey's condition, such as bleeding, which might alter the strength or sig- nature of the electric field. TELEMETRY.— Two sharks were tagged with temperature-sensing transmitters during this study. The first shark (a 4.5-m male) carried a unit that monitored ambient water temperature only. After tagging, the shark remained around the boat even after all baits were removed from the water. The boat was then moved away from the area and the shark began to move westward; parallel to the north shore of Dangerous Reef. Once past the island the shark moved offshore in a northwesterly direction. Contact was lost with the animal approximately 4 h after initial tagging, due to its rapid speed and bad seas that created poor tracking conditions. During this time the shark swam in waters 20-2 1°C as indicated by the temperature sensor on the transmitter. The second shark was tagged on 22 January 1980. The body temperature probe was placed 31 cm deep into the lateral musculature, ap- TRICAS AND McCOSKER: PREDATORY BEHAVIOR OF THE WHITE SHARK 235 FIGURE 15. C. J. Slager. Distribution of the ampullae of Lorenzini on the head of a young female white shark (CAS 37917). Figure by proximately 25 cm below the first dorsal fin. This shark was monitored continuously near the boat for approximately 2 hr, until it swam out of range. It returned to the anchored boat near midnight, and then again departed. Results of the thermal data are presented in Table 1 and Fig. 16. The shark swam in water ranging from 20.9° to 2 1 .5°C. Mean difference between ambient and body tem- perature was 3.7°C, and ranged from 3.2° to 4.3°C. TABLE 1. EPAXIAL MUSCLE TEMPERATURES OF A 3.5 M (TL) MALE WHITE SHARK MONITORED AT DANGEROUS REEF, SOUTH AUSTRALIA ON 22 JANUARY 1980. Mean (AT) = 3.7°C. SD = 0.37. Measure- ment Temp (°C) Water Body Difference (AT) 1 21.2 25.2 4.0 2 21.5 24.7 3.2 3 21.2 24.7 3.5 4 21.2 25.2 4.0 5 21.2 25.2 4.0 6 21.2 25.2 4.0 7 21.2 24.7 3.5 8 20.9 25.2 4.3 9 20.9 24.7 3.8 10 20.9 24.2 3.3 11 20.9 24.2 3.3 12 20.9 24.2 3.3 Largest and smallest differences were recorded when the shark entered water of a different tem- perature, before internal temperatures could con- form. This time lag to thermal equilibrium and variation in muscle temperature indicate that the shark did not thermoregulate. Carey et al. (1982) found that a 4.6-m white shark had a body tem- perature 3-5°C higher than the surrounding water. Their shark swam over deeper waters, and for the most part remained in the thermocline. Tem- peratures were lower in their study, ranging ap- proximately from 5° to 19°C ambient, and 18° WHITE SHARK MUSCLE TEMPERATURE White Shark Epaxial Musculalu FIGURE 1 6. Temperature difference between ambient sea- water and epaxial musculature of a 3.5 m TL white shark, monitored on 22 January 1 980 at Dangerous Reef, South Aus- tralia. Question marks (?) indicate time interval when shark swam away from anchored study vessel and out of telemetry range. Figure by K. O'Farrell. 236 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 14 FIGURE 17. (Right) Silhouette of a surfer on a contemporary surfboard. (Left) Silhouette of an adult female (TL = 1.7 m) harbor seal (Phoca vitulina). Photo by Al Giddings. to 23°C muscle temperature. Our study took place in relatively shallow waters around Dangerous Reef (< 30 m), and we found no sign of a marked thermocline. The water and shark-muscle tem- peratures we recorded were generally higher (20.9°-2 1.2° and 24.2°-25.2°C, respectively), but they are consistent with the values for body tem- perature elevation over ambient recorded by Ca- rey et al. One of the primary advantages of being warm- bodied is thought to be related to the changes in muscle physiology as temperature increases. It is known that a 10°C increase in temperature may result in a three-fold increase in the contraction- relaxation rate of frog muscle (Hartree and Hill 1921). For fish, this may be translated to an in- crease in potential tail-beat frequency and a re- lated increase in sustained swimming speed. Higher speeds may be selectively advantageous when chasing prey or fleeing from predators. In addition, conservation of heat theoretically al- lows for more total energy conversion to work, thus enabling an animal to swim longer distances on a given meal. Being warm-bodied might also allow for temporary excursions into colder or deeper waters. This thermal inertia (see Neill et al. 1976) would not only expand the range of environments which the animal could exploit, but would also permit increased swimming ef- ficiency for predation at otherwise limiting en- vironmental temperatures. ON WHITE SHARKS AND SURFBOARDS. — In con- clusion, we comment on the increasing attacks by white sharks upon humans who surf in the north Pacific. Since 1972, there have been 11 recorded white shark attacks upon surfers in Cal- ifornia and Oregon (Miller and Collier 1981) and one such attack in Hawaii in 1959 (Balazs and Kam 1981). The similarity in appearance of the silhouette of a prone human on a surfboard or "belly board" to a large surface-basking pinniped is clear (Fig. 1 7), and observations of attacks by sharks upon surfers fit well with our assessment of the feeding strategy of white sharks. Attacks have occurred in the vicinity of pinniped rook- eries, such as the much-publicized death of Lewis Boren on 19 December 1981 at Spanish Bay, Monterey, Calfornia. TRICAS AND McCOSKER: PREDATORY BEHAVIOR OF THE WHITE SHARK 237 Since the early 1970s, the trend in surfboard design has been toward an increase in flotation, reduction in board length, multiple posterior- fixed rudders ("skegs"), and bifurcated or "V" tails. All of these modifications have enhanced the similarity between the silhouette of a surfer and that of a pinniped, and we suggest that this may increase the probability of attack of surfers encountered by white sharks. We feel it advisable that those who surf be aware of and consider the potential risks of surfing in coastal waters known to be frequented by white sharks. ACKNOWLEDGMENTS We are particularly grateful to Al Giddings, President of Ocean Images, Ltd., for financial assistance, for providing access to his film library for cinematographic analyses, and for sharing his observations of shark behavior with us. We also thank Terry Thompson, Ocean Images, Ltd., for his assistance. Additional shark data were pro- vided by W. I. Follett (CAS), L. J. V. Compagno (CAS), J. Randall (Bernice P. Bishop Museum), and the staff of the Department of Ichthyology of the CAS. A. Dizon (NMFS, Honolulu) and I. Cooke (Bekesy Laboratory, U. of Hawaii) pro- vided facilities during construction of our trans- mitters. We thank H. Tricas, C. J. Slager, K. O'Farrell, S. Middleton, and S. Nakamura for assistance with our figures; J. Ames, R. Bandar, R. Dunne, A. Giddings, S. Morrell, P. Romano, and P. Wing for allowing us to use their pictures; and Bob Britcher and Chico Chingwidden, the Master and the mate of the Nenad. We give spe- cial thanks to Rodney Fox, for his guidance in the field in South Australia and for helpful dis- cussions concerning shark behavior, and to Leighton Taylor, Jr. (Waikiki Aquarium), and Phil Motta (Univ. of Montana) for their critical reading of this manuscript. Senior authorship of this paper was deter- mined by the outcome of a pinball match played at Port Lincoln, South Australia, in January 1 980. REFERENCES AINLEY, D. G., C. S. STRONG, H. R. HUBER, T. J. LEWIS, AND S. H. MORRELL. 1981. Predation by sharks on pinnipeds at the Farallon Islands. Fish. Bull., U.S. 78:941-945. ALEXANDER, R. McN. 1967. Functional design in fishes. Hutchinson and Co., London. 160 pp. AMES, J. A., AND G. V. MOREJOHN. 1980. Evidence of white shark, Carcharodon carcharias, attacks on sea otters, En- hydra lutris. Calif. Fish Game 66:196-209. BACKUS, R. H., S. SPRINGER, AND E. L. ARNOLD, JR. 1956. A contribution to the natural history of the white-tip shark, Pterolamiops longimanus (Poey). Deep-Sea Res. 3: 1 78-188. BALAZS, G. H., AND A. K. H. KAM. 1981. A review of shark attacks in the Hawaiian Islands. 'Elepaio. 41(10):97-105. BASS, A. J., J. D. D' AUBREY, AND N. KISTNASAMY. 1975. Sharks of the east coast of southern Africa. IV. The families Odontaspididae, Scapanorhynchidae, Isuridae, Cetorhini- dae, Alopiidae, Orectolobidae, and Rhiniodontidae. Invest. Rep. Oceanogr. Res. Inst. no. 39. 102 pp. BIGELOW, H. B., AND W. C. SHROEDER. 1948. Sharks. In Fishes of the western north Atlantic. J. Tee- Van, C. M. Breeder, S. F. Hildebrand, A. E. Parr, and W. C. Schroeder, eds. Part one. Mem. Sears Found. Mar. Res., Yale Univ. 1. 576 pp. BONE, Q., AND B. L. ROBERTS. 1969. The density of elas- mobranchs. J. Mar. Biol. Assn. U.K. 49:913-937. BONHAM, K. 1 942. Records of three sharks on the Washing- ton coast. Copeia 1942:264-266. BUDKER, P. 1971. The life of sharks. Columbia Univ. Press, New York. 222 pp. CAREY, F. G., G. GABRIELSON, J. W. KANWISHER, AND O. BRAZIER. 1982. The white shark, Carcharodon carcharias, is warm-bodied. Copeia 1982:254-260. COLLIER, R. S. 1964. Report on a recent shark attack off San Francisco, California. Calif. Fish Game 50:261-264. COMPAGNO, L. J. V. 1977. Phyletic relationships of living sharks and rays. Amer. Zool. 17:303-322. DAY, L. R., AND H. D. FISHER. 1954. Notes on the great white shark, Carcharodon carcharias, in Canadian waters. Copeia 1954:295-296. ELLIS, R. 1975. The book of sharks. Grosset and Dunlap, New York. 320 pp. . 1983. Chiller from the depths. Geo Magazine 5:91- 97. FAST, T. N. 1955. Second known attack on a swimmer in Monterey Bay. Calif. Fish Game 41:348-351. FEDER, H. M., C. H. TURNER, AND C. LIMBAUGH. 1974. Ob- servations of fishes associated with kelp beds in southern California. Calif. Dept. Fish Game, Fish. Bull. 160. 144 pp. FOLLETT, W. I. 1974. Attacks by the white shark, Carchar- odon carcharias (Linnaeus), in northern California. Calif. Fish Game 60: 192- 198. GILBERT, P. W. 1963. The visual apparatus of sharks. Pages 283-326 in Sharks and survival. P. W. Gilbert, ed. D. C. Heath and Co., Boston. GUITART, D., AND J. F. MILERA. 1974. El monstruo de Co- jimar. Mar y Pesca 104:10-1 1. GUNTHER, A. 1870. Catalogue of fishes in the British Mu- seum. Taylor and Francis, London. HALLER, G. 1926. Uber die Entwicklung, den Bau und die Mechanik des Kieferapparates des Dornhais (Acanthias vul- garis). Z. mikrosk. ant. Forsch. 5:749-793. HARTREE, W., AND A. V. HILL. 1921. The nature of the isometric twitch. J. Physiol. 55:389-41 1. KALMIJN, A. J. 1971. The electric sense of sharks and rays. J. Exp. Biol. 55:371-383. . 1974. The detection of electric fields from inanimate and animate sources other than electric organs. In Handbook of sensory physiology. A. Fessard, ed. Vol. HI/3. Springer- Verlag, New York. -. 1978. Electric and magnetic sensory world of sharks, skates, and rays. Pages 507-528 in Sensory biology of sharks, skates, and rays. E. S. Hodgson and R. F. Mathewson, eds. Off. Nav. Res., Arlington. 238 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 14 . 1982. Electric and magnetic field detection in elas- mobranch fishes. Science 218:916-918. LAUDER, G. V. 1980. Hydrodynamics of prey capture by teleost fishes. Biofluid Mechanics 2:161-181. LE BOEUF, B. J., D. G. AINLEY, AND J. T. LEWIS. 1974. El- ephant seals on the Farallones: population structure of an incipient breeding colony. J. Mammal. 55:370-385. LE BOEUF, B. J., AND R. S. PETERSON. 1969. Social status and mating activity in elephant seals. Science 163:91-93. LE BOEUF, B. J., M. RIEDMAN, AND R. S. KEYES. 1982. White shark predation on pinnipeds in California coastal waters. Fish. Bull., U.S. 81:80(4):891-895. LEMIER, E. H. 1951. Recent records of the great white shark, Carcharodon carcharias, on the Washington coast. Copeia 1951:249. LIEM, K. F. 1978. Modulatory multiplicity in the functional repertoire of the feeding mechanism in cichlid fishes. 1. Pis- civores. J. Morph. 158:323-360. LIMBAUGH, C. 1963. Field notes on sharks. Pages 63-94 in Sharks and survival. P. W. Gilbert, ed. D. C. Heath and Co., Boston. LUTHER, A. F. 1909. Untersuchungen uber die vom N. tri- geminus innervierte Musculator der Selachier. (Haie und Rochen) unter Berucksichtigung ihrer Beziehungen zu be- nachbarten Organen. Acta Soc. Sci. Fenn. 36:1-176. MCCOSKER, J. E. 1981. Great white shark. Science 81 2:42- 51. MILLER, D. J., AND R. S. COLLIER. 1981. Shark attacks in California and Oregon, 1926-1979. Calif. Fish Game 67: 76-104. Moss, S. A. 1972. The feeding mechanisms of the sharks of the family Carcharhinidae. J. Zool. Lond. 167:423-436. . 1977. Feeding mechanisms of sharks. Am. Zool. 17: 355-364. NEILL, W. H., R. K. CHANG, AND A. E. DIZON. 1976. Mag- nitude and ecological implication of thermal inertia in skip- jack tuna, Katsuwonnus pelamis (Linnaeus). Env. Biol. Fish. 1:61-80. OSSE, J. W. M. 1969. Functional morphology of the head of the perch (Perca fluviatilis L.): an electromyographic study. Neth. J. Zool. 19:289-392. PRATT, H. L., J. G. CASEY, AND R. B. CONKLIN. 1982. Ob- servations on large white sharks, Carcharodon carcharias, off Long Island, New York. Fish. Bull., U.S. 80:153-156. RANDALL, J. E. 1973. Size of the great white shark (Car- charodon). Science 181:169-170. RICKER, W. E. 1973. Linear regression in fishery research. J. Fish. Res. Board Can. 30:409-434. ROEDEL, P. M., AND W. E. RiPLEY. 1950. California sharks and rays. Calif. Div. Fish Game, Fish. Bull. 75. 88 pp. ROYCE, W. F. 1963. First record of white shark (Carcharodon carcharias) from southeastern Alaska. Copeia 1963:179. SCATTERGOOD, L. W. 1962. White sharks, Carcharodon car- charias, in Maine, 1959-1960. Copeia 1962:446-447. SCHROEDER, W. C. 1938. Records of Carcharodon carcharias (Linnaeus) and Pseudopriacanthus altus (Gill) from the Gulf of Maine, summer of 1937. Copeia 1938:46. SMITH, J. L. B. 1951. A juvenile of the man-eater, Carchar- odon carcharias Linn. Ann. Mag. Nat. Hist. 4(12):729-736. . 1957. Sharks of the genus Isurus Rafinesque, 1810. Ichthy. Bull. 6:91-96. SPRINGER, S. 1961. Dynamics of the feeding mechanism of large galeoid sharks. Am. Zool. 1:183-185. SQUIRE, J. L., JR. 1967. Observations of basking sharks and great white sharks in Monterey Bay, 1948-50. Copeia 1967: 247-250. THOMSON, K. S., AND D. E. SIMANEK. 1977. Body form and locomotion in sharks. Amer. Zool. 17:343-354. TRICAS, T. C. 1979. Relationships of the blue shark, Prionace glauca, and its prey species near Santa Catalina Island, Cal- ifornia. Fish. Bull. 77:175-182. PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES Vol. 43, No. 15, pp. 239-248, 5 figs., 1 table September 19, 1984 DENDRODOA (STYELOPSIS) ABBOTTI, SP. NOV. (STYELIDAE, ASCIDIACEA) FROM THE PACIFIC COAST OF THE UNITED STATES, AND ITS IMPACT ON SOME GONADAL CRITERIA OF ITS GENUS AND SUBGENUS By Andrew I odd Newberry Cowell College, University of California, Santa Cruz, California 95064 ABSTRACT: Dendrodoa (Styelopsis) abbotti, a newly described styelid ascidian from the central and northern California coast, the San Juan Islands of Washington, and southwestern Vancouver Island, resembles I). cornea but differs in branchial and gonadal traits. Inclusion of D. abbotti in the genus Dendrodoa requires modification of the gonadal criteria of the genus to accommodate styelan gonadal resemblances (non-encap- sulation of the testis-lobes with the ovary) and styelan or cnemidocarpan spermiducal resemblances (gonad's single vas deferens and spermipore). INTRODUCTION The tunicate named and described in this pa- per, Dendrodoa (Styelopsis) abbotti, is a styelid ascidian that has long been collected along the central and northern California coast. Donald P. Abbott, who first found this ascidian in 1948 near Point Arena (Mendocino County), included it as "Alloeocarpa sp." in the urochordate key of the second edition of Light's Manual (Light et al. 1954) but, for want of more certain identifi- cation, omitted it from that handbook's third edition (Smith and Carlton 1975). This ascidian's aggregative habit does create an appearance of budding (Fig. 1 A), but adjacent zooids' tests are unfused and easily separated from one another; no evidence of budding has been found in several hundred zooids from sev- eral sites and all seasons. Apparently, then, this is a solitary ascidian and cannot be placed in the genus Alloeocarpa. It shows Dendrodod's restric- tion of the single, elongate ovary to the zooid's right side. The ovary's unbranched shape and the pharynx's simplicity place the species in the subgenus Styelopsis of Dendrodoa. The specific name, abbotti, honors Professor Donald P. Ab- bott, of the Hopkins Marine Station of Stanford University, who has shared with his students and colleagues a singular keenness of intellect and generosity of spirit, and it expresses the esteem and affection of his fellow ascidiologists. MATERIALS AND METHODS This report is based principally on specimens collected intertidally at Pigeon Point, San Mateo County, California (lat. 37°11'0"N, long. 122°23'10"W), at intervals of roughly six weeks throughout 1977. 1 have also drawn on material taken over the past two decades from there; from Point Pinos, Monterey County, California (lat. [239] 240 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 15 B FIGURE 1 . A) Living zooids of Dendrodoa abbotti, including one (upper left-center) that has been wounded or severely disturbed and is extremely contracted while the others remain relaxed. Zooids are about 1 cm long. B) Closer view of two slightly disturbed zooids, showing siphons in the process of bilabial closure. Brood pouch is detectable in the left zooid by inflated aspect of the posterior region of the zooid (to left in photo). Zooids are about 1 cm long. NEWBERRY: DENDRODOA (STYELOPSIS) ABBOTTI, SP. NOV. 241 36°38'0"N, long. 121°56'0"W); and subtidally from Peavine Pass (lat. 48°35'4"N, long. 122°45'48"W) in the San Juan Islands, Wash- ington. I have used, as well, D. P. Abbott's un- published notes and drawings of specimens from northern and central California. In all, I have examined approximately 30 spec- imens thoroughly. I have examined several doz- en more in a cursory way to verify the criteria that characterize the species. All specimens were relaxed with MgCl2 or MgSO4, and menthol, then fixed in seawater Bouin's fluid or 10% formalin, and all were preserved in 70% ethanol. The Bouin's-fixed material provided excellent serial sections but brittle dissections. Formalin always 'fixed adequately for dissections but rarely well enough for close scrutiny by serial section (which was required, for example, to trace the very fine spermiducts). Specimens were dissected in 70% ethanol. Dissected specimens usually were stained, once opened, with Grenacher's borax carmine; serially sectioned specimens were either prestained, often for prior dissection, in Gren- acher's alcoholic borax carmine or stained in sec- tion with "standard alum hematoxylin" (Galigh- er and Kozloff 1964) and eosin. Prestaining proved satisfactory for general examination, but staining in section was necessary to reveal finer structural details or to take advantage of the bet- ter fixation achieved with Bouin's fluid than with formalin. COORDINATES.— The endostyle designates the anterior-posterior axis and the ventral midline. Thus, the dorsal midline extends from the- oral siphon through and beyond the atrial siphon. By these coordinates, the ovary lies against the right ventral margin of the zooid, and the loop of the gut dominates the left posterior region of the zooid (Fig. 2). DESCRIPTION OF SPECIES Dendrodoa (Styelopsis) ahbotti, sp. nov. TYPE-SPECIMENS. — Holotype at California Academy of Sci- ences, San Francisco, Calif. (CAS #034790). Paratypes at Cal- ifornia Academy of Sciences, San Francisco, Calif. (CAS #034791). TYPE-LOCALITY.— North side of Pigeon Point, San Mateo County, California (lat. 37°1 1'0'N, long. 122°23'10"W). OTHER RECORDS. — Intertidal records from Point Pinos and Hopkins Marine Reserve (Monterey County), Pigeon Point and Moss Beach (San Mateo County), Point Arena (Mendocino County), California, and near Sooke, Vancouver Island, British Columbia; subtidal records from Peavine Pass (San Juan Coun- ty), Washington. EXTERNAL APPEARANCE (Fig. 1 ). — Zooids round or oval (lengthened antero-posteriorly) low hemispheres; entire sub-endostylar surface ap- plied to substrate; attached surface extends be- yond ovary on right and gut-loop on left. Spec- imens including test reach 8 to 1 2 mm length, 6 to 10 mm width, 2 or 3 mm height when relaxed; zooids removed from test reach 8 to 10 mm length, 6 to 8 mm width, 2 to 3 mm height. Test clean, thin, and parchment-like, spreading as a thin apron 1 to 2 mm wide on the substrate around the zooid. Ventral test extremely thin. Color in life translucent gray tinted with ochre or very pale brownish pink, with borders of si- phonal apertures sometimes slightly darker. Zooids fixed in formalin become plain translu- cent white-gray. Alive or fixed, zooid's branchial sac, gut, ovary, and mass of brooded young are faintly visible through dorsal and lateral regions of test. Oral siphon far anterior; atrial siphon placed centrally atop hemispheric zooid; both siphons fairly evident in relaxed living animals but reduced to obscure slits in contracted ones. Relaxed zooids have circular siphonal apertures; disturbed zooids close their siphons bilabially into transverse slits (Fig. IB) and flatten them- selves against the substrate within a delicately crumpled test. Zooids are simple and non-bud- ding but often aggregate in pairs or trios (rarely groups of more) with young ones often settling adjacent to or even on the test "apron" around older zooids (but not on zooidal surfaces them- selves). Mature zooids, even when tightly adja- cent to one another, attach entirely to the sub- strate itself; they do not form clumps of zooids growing thickly one upon another. Adjacent zooids often are oriented similarly on the sub- strate. VASCULAR ELEMENTS OF THE TEST.— Test- ves- sels not prominent; as revealed by staining, branching systems of test- vessels ramify toward the margin of the test. Test- vessel ramifications connect to zooid by one or more sub-zooidal circulatory junctions; tips of all branches of test- vessel ramifications end peripherally in slender, bulbous vascular ampullae. MANTLE.— Thin, lightly muscled mantle ex- cept for extensive arrays of fibers radiating from each siphon and controlling its bilabial closure; fairly conspicuous concentric musculature sur- rounding oral siphon, less developed concentric musculature around atrial siphon. About a dozen endocarps project from the mantle into the atrium 242 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 15 FIGURE 2. Dorsal view of zooid, with pharynx removed to show disposition of ovary and testis-lobes (including far posterior group), vasa efferentia and vas deferens (including tiny spermiporal papilla near atrial siphon), mid- ventral endostyle and heart beneath it, and gut-tract (stomach, pyloric duct, caecum, intestine). Position of oral and atrial siphons indicated by ovals. FIGURE 3. Zooid opened by mid-ventral cut to show ovary, testis-lobes (including posterior lobes), larvae in brood pouch, several endocarps (stippled), gut-tract, siphons with neural complex between them. Pharynx removed. of most zooids; particularly large endocarps usu- ally protrude from the atrial mantle anterior to the ovary and in the region of the gut-loop. ORAL TENTACLES. — 36 to 40 filiform oral ten- tacles of three sizes; largest ones most abundant (24-30), others about half their size intercalated irregularly, a few to many tiny papillae evident upon close examination of the band of oral ten- tacles. Just distal to this circle of tentacles is a siphonal flange that marks the inward limit of the test that lines the oral siphon. ATRIAL TENTACLES.— 40 to 50 tiny filiform atrial tentacles in band analogous to that of the circle of oral tentacles. Just distal to this incon- spicuous circle is the atrial siphonal flange that marks the inward limit of the test lining the atrial siphon. DORSAL TUBERCLE (Fig. 4).— Simple C-shaped slit atop a short, stout projection; concavity of the C faces posteriorly (toward the dorsal lami- na). The dorsal tubercle is set slightly to the right of the dorsal midline. NEURAL COMPLEX.— In dorsal or ventral sil- houette, whole complex forms a rectangle elon- gated antero-posteriorly and extended somewhat at each corner. Like the dorsal tubercle, the neu- ral complex is set slightly to the right of the dorsal midline. BRANCHIAL SAC (PHARYNX) (Fig. 5).— Folds absent, perhaps represented by internal longi- tudinal branchial vessels. In dissection, 4 inter- nal longitudinal vessels are evident on each side of the pharynx; in transverse serial sections, a fifth internal longitudinal vessel is sometimes discernible on each side close to the endostyle, and in a few specimens even a sixth vessel on each side may run only some length of the sac. Usually 9 or 10 stigmata lie between these in- ternal longitudinal vessels. Ten to 1 2 transverse vessels separate the rows of longitudinally ori- ented stigmata, and there are about 1 0 parastig- matic vessels partly or entirely traversing each side of the pharynx. Along the ventral midline the branchial sac connects with the body wall by widely spaced sub-endostylar vascular trabecu- lae, not by a continuous sub-endostylar mem- brane. Other vascular trabeculae connect the branchial sac abundantly in all directions to the NEWBERRY: DENDRODOA (STYELOPSIS) ABBOTTI, SP. NOV. 243 FIGURE 4. Dorsal tubercle in relation to peripharyngeal groove and dorsal lamina. FIGURE 5. Right side of pharynx, showing several rows of stigmata and the four internal longitudinal branchial vessels of the pharynx's right side. Dorsal lamina at top, endostylar groove at bottom. Drawing based in part on unpublished notes of D. P. Abbott, in part on freshly dissected specimens. atrial surface of the mantle and to the atrial ep- ithelium around the gut. DORSAL LAMINA. — Prominent, continuous, smooth-bordered dorsal lamina, without lan- guets. GUT (Figs. 2, 3). — Esophageal aperture far dorso-posterior in pharynx; stout esophagus bends sharply ventrally into stomach; stomach empties anteriorly into fore-intestine, which bends to left and passes posteriad on the lateral side of the stomach. Hind-intestine then curves sharply dorsad and follows the left mantle to the anus, which lies slightly to the left-posterior of the atrial siphon. Stomach has 16 to 18 mod- erately evident external folds corresponding to well-developed internal gastric septa. The gastric septa are reduced to low ridges in the left pyloric region of the stomach, near the pyloric caecum. Pyloric caecum is small, sometimes absent. A highly vascularized pyloric duct joins the sinus- oidal sheath surrounding the stomach with that surrounding the fore-intestine. Intestine com- prises a fore-intestine with a large typhlosole-like longitudinal plication of its wall and a thick si- nusoidal jacket between the gut wall and its sheath of atrial epithelium, and a hind-intestine of more simply tubular section whose atrial sheath is much closer to the gut wall. Anus lies dorso-medial or slightly to the left, above the stomach; anus is cut square to the axis of the rectum; anal margin is scalloped into usually 5 lobes that fit together when the anus is tightly closed. HEART.— Fairly straight within a somewhat curved and inflated pericardium; set at about 45° obliquely to the endostyle, oriented right-ante- rior to left-posterior, centered roughly beneath the endostyle in the posterior half of the zooid (site and orientation in Fig. 2). OVARY (Figs. 2, 3). — Single, unbranched, sau- sage-shaped ovary along the right ventral margin of the zooid, extending almost the entire length of the zooid, curving sharply dorsad posteriorly and following the right mantle to arch halfway over the atrium, recurving dorsally to terminate in an oviduct directed posteriorly toward the brood pouch and away from the atrial siphon. Oviduct lies lateral (away from atrium) to main mass of ovary, with its lumen penetrating among the ripening gametes; lateral surface (away from germinal tissue) of oviduct heavily ciliated, other 244 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 15 oviducal surfaces apparently not ciliated. Ova- ries of all specimens examined by dissection or serial section show all stages of o vogenesis pres- ent, regardless of season. TESTIS (Figs. 2, 3).— A dozen to more than 20 separate lobate sacs, not encapsulated with the ovary but instead lying in the mantle wall ad- jacent to but clearly outside the ovary's delim- iting membrane. Most sacs lie medial to the ovary; some lie anterior to the ovary; few are lateral; many lobes lie partly "beneath" the ovary, in the mantle between the ovary and the ventral surface of the zooid. In many specimens, but not all, a few testis-lobes lie far posteriorly and on the left side of the ventral midline, but their spermiducts join the vas deferens of the testis-lobes that lie beside the ovary. All sacs join by vasa efferentia to a single, long vas deferens that lies between the ovary and the atrial epithelium. This duct follows the ovary to the region of the atrial si- phon, and there leaves the ovarian surface to project toward the atrial siphon from the dorsal roof of the atrium while the ovary bends pos- teriad toward its ovipore. The ciliated vasa ef- ferentia are extremely thin, visible only in serial section; the vas deferens, also scarcely visible except in serial section, is a compressed, ciliated channel terminating in a tiny, spermipore-bear- ing papilla pointing toward the atrial aperture. In all specimens examined from all seasons for gametic condition, many testis-lobes have tailed sperm, but the spermiducts contain only scat- tered sperm. BROOD CHAMBER AND BROODED YOUNG (Figs. 1, 3).— The posterior region of the atrium serves as a brood chamber, occluded anteriorly by the branchial sac, on the left by the gut-loop, on the right by the ascending limb of the ovary. All specimens examined were brooding young in all stages of development from (relatively rarely) fertilized eggs and cleavage stages to (usually) tadpoles that were still curled (although many of these straightened upon removal from the brood chamber during dissections). Quantities of brooded young vary greatly— fewest (20 to 30) in midwinter specimens, most (100 to 200) in late spring to midfall specimens. The brood chamber often is so swollen with young that it is readily apparent in living animals. Young are crammed tightly into the chamber; external study of zooids divested of test may suggest only a few larvae, but dissection then reveals many dozens. The brood chamber is criss-crossed by many vas- cular trabeculae that connect the branchial sac and the atrial and gut wall and may keep loose young from being swept from the brood chamber by atrial water-currents. ECOLOGICAL DISTRIBUTION IN CALIFORNIA.— Intertidal. At Point Pinos and Pigeon Point, peak abundance is at about +0.3 m, and I have found no specimens at either site above +0.6 m or below 0.0 m. This is an open-coast ascidian in California, inhabiting surf-swept rocky habitats where the full force of the waves is broken by surrounding rocks and reefs. Zooids congregate on horizontal undersides of large boulders, usu- ally well back from the boulders' edges. Such boulders restrict waterflow underneath, so much so that at Point Pinos the rocks that harbor Den- drodoa abbotti may lie partially in sand that by its odor and color appears to be virtually anoxic. At Pigeon Point most rocks with this ascidian are slightly propped up by their neighbors, so that oxygen remains plentiful in waters perco- lating or flowing underneath. Large boulders that do not have D. abbotti on them may shelter smaller rocks that do. Many rocks that seem ap- propriate for this species do not harbor speci- mens. This spotty distribution of aggregated in- dividuals may indicate a short swimming period and quick settlement by brooded larvae, or as yet unclear ecological restrictions on the adults. At Pigeon Point, other invertebrates found on surfaces with Dendrodoa abbotti include the anemone Epiactis prolifera, the polyclad Noto- plana acticola, the polychaetes Spirorbis and Salmacina, the barnacle Balanus glandula (and sometimes Chthamalus dalli), porcelain crabs such as Petrolisthes, several encrusting bryozo- ans such as Eurystomella bilabiata, the asteroid Leptasterias pusilla, and the aplousobranch as- cidian Aplidium californicum. But none of these associated invertebrates seems so severely kept back from the margins of boulders, so cryptic in its under-rock habitat, as Dendrodoa abbotti. ECOLOGICAL DISTRIBUTION IN WASHINGTON.— Subtidal. At Peavine Pass, San Juan Islands, specimens were dredged from 10 to 12m. The species has been sought elsewhere in rocky areas, but only Peavine Pass, which is swept to the bottom by strong tidal currents, has proved a reliable site for collecting by this method, and even there the species is rarely taken. Debris har- boring Dendrodoa abbotti contains, as well, Bal- NEWBERRY: DENDRODOA (STYELOPSIS) ABBOTTI, SP. NOV. 245 TABLE 1. DENDRODOA CARNEA AND D. ABBOTTI: CONSISTENT DIFFERENCES. Feature Dendrodoa carnea Dendrodoa abbotti Color in life Siphonal apertures Dorsal tubercle Transverse branchial vessels and rows of stig- mata Internal longitudinal branchial vessels and folds (DL = dorsal lamina, (#) = number of vessels in fold, E = endostyle) Endocarps Margin of anus Ovary Testis Spermiduct Brood chamber Brooded young Bright pink to blood red. Bilabial. Narrow ovoid slit whose axis is oriented almost anterior-posterior. 1 7 or more. Left: DLO(1)0(1)0(1)0(1)OE as in D. abbotti. Right: DLO(4-5)0(1)0(1)0(1)OE; prominent low fold carrying at least 4 vessels on right pharyngeal wall. Many, small, widely scattered over entire atrial wall. "Reflected but not lobed, often somewhat two- lipped" (van Name 1912, p. 587). Straight along right ventral margin of body; ovi- duct continues so. Not clearly encapsulated with ovary, testis-lobes extend somewhat into body wall, predominantly ventro-lateral to ovary; all testis-lobes close to ovary. (?) as in D. grossularia, many short spermiducts converge in multiple spermipores on atrial sur- face of ovary (?) Extensive, including right-posterior region be- yond oviduct there. (From a small sample) only a few dozen embryos brooded at a time. Gray to ochre, occasionally reddish around si- phonal apertures. Bilabial, somewhat more pronouncedly so than in D. carnea. Fairly sharply bent "C" whose long axis is ori- ented laterally. Ca. 12. Left: DLO(1)0(1)0(1)0(1)OE as in D. carnea. Right: DLO( 1)0(1)0(1 XX 1)OE; no multi-vessel fold on right pharyngeal wall. Fewer, larger, more (but not entirely) confined to ventral atrial surface. Scalloped into usually 5 lobes. Along right-anterior ventral margin of body, then bends sharply into ascending limb, recurves be- hind atrial siphon into dorsal oviduct that pro- jects posteriorly. Clearly not encapsulated with ovary, testis-lobes lie wholly in body wall, predominantly ventro- medial to ovary; often one posterior group of testis-lobes far from ovary. Single, long vas deferens on atrial surface of ovary receives vasa efterentia of all testis-lobes, ends mid-dorsally in spermipore-bearing papilla pointing at atrial siphon. More restricted to far posterior part of body. Many dozens to more than 1 00 embryos brooded at a time. anus nubilis (one of the best indicators that the ascidian may be present) and the hydrocoral Al- lopora. The ascidian occurs especially around the husks of dead barnacles and in crannies in large rocks. But dredging of course destroys the set of surfaces and actual relationships among mem- bers of the fauna at the site, and so no compar- ison can yet be made between the subtidal hab- itat of Dendrodoa abbotti at Peavine Pass and its intertidal circumstances at Pigeon Point. The bathymetric contrast between California and Washington (San Juan Islands) records of Dendrodoa abbotti is striking. The species may occur subtidally in California; its inaccessibility, beneath large boulders, could account for the cur- rent lack of such records by dredging or even by diving. But D. abbotti does not occur in the very low intertidal zone in California, below about mean low-low tidal levels. Thus, if it does occur subtidally, there is not a continuous distribution of the species from those depths to the low- to mid-tidal habitats where it characteristically is found. In contrast, in the San Juan Islands, I have not found the species at all intertidally in habitats that resemble California's coastal sites— except, of course, for the lack of surf in the San Juans. Dendrodoa abbotti appears to be only a subtidal species in that archipelago. But to the west of the San Juan Islands, on the southwest coast of Van- couver Island, B.C., Dr. Ivan Goodbody has 246 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 15 found this species "on the underside of boulders at extreme low tide ... on the open coast north of Sooke." Dr. Goodbody reports (pers. comm.) that the site there is "a rough boulder strewn shore with many large rounded boulders indi- cating heavy wave action." His record thus ex- tends the intertidal range of D, abbotti into those Canadian habitats where surf does resemble Cal- ifornia's. Dr. Goodbody's Canadian specimens of D. abbotti are now in the collection of the California Academy of Sciences. DISCUSSION A. Comparison of Dendrodoa abbotti with D. carnea A comparison of Dendrodoa abbotti with the western North Atlantic species D. carnea— the styelopsid dendrodoan that most closely resem- bles D. abbotti— indicates an array of differences, some trivial, some marked, but all consistent. This comparison sets a great many specimens of D. abbotti against necessarily only a few dissected specimens of D. carnea (from the USNM col- lection) and others' reports on D. carnea (see van Name 1912, 1945). But differences that emerge even in this perhaps unbalanced sampling of these species gain force as they become elements in a consistent pattern of distinctions between the two taxa, and this pattern has become more persua- sive with each examination of new specimens. Table 1 summarizes the comparison. Even arguably minor distinctions (for exam- ple, the character of the brood chamber or of the endocarps) take on significance in Table 1's ar- ray. The two species are most effectively distin- guished, however, by the following criteria: 1 . shape and disposition of the ovary; 2. testis-ovary relationship, including D. abbot- ti's posterior group of testis-lobes; 3. structure of the spermiducts, especially of the vas deferens; 4. arrangement and number of internal longi- tudinal vessels of the right side of the pharynx; 5. number of transverse branchial vessels and rows of stigmata on both pharyngeal walls; 6. shape and orientation of the dorsal tubercle; 7. color in life. B. Generic Traits By most accounts and diagnoses, in the genus Dendrodoa the testis and ovary are "encapsu- lated" within a common sheath (Monniot and Monniot 1972), and the testis comprises many lobes that do not lie in the body wall but rather hug the parietal (away from the atrium) surface of the ovary (Huntsman 1913). Most of the go- nad of Dendrodoa grossularia shows this con- dition of encapsulation and testis-ovary juxta- position clearly, although some of the anterior testis-lobes do lie more in the body wall than wholly against the ovary. Dendrodoa carnea ex- hibits a somewhat looser gonadal arrangement: the testis-lobes apparently are still encapsulated with the ovary and lie largely against its parietal surface, but they extend into the adjacent body wall much more than do the testis-lobes of D. grossularia, especially to the lateral (right) side of the ovary. Dendrodoa abbotti carries this loos- ening of the testis-ovary bond further still: the testis-lobes of D. abbotti lie "beneath" the ovary or close by on the medial (left) side of the ovary, but they lie in the body wall itself, not against the ovarian mass, and there is no sheath enclos- ing these gonadal elements into a single structural unit of intimately juxtaposed parts. And al- though most of the testis-lobes of D. abbotti lie very close to the ovary, there is often a group of testis-lobes lying in the far posterior atrial floor of the zooid, and actually on the left side of the zooid, although even this separated and isolated group is still connected by a vas deferens to the common spermiduct of all the other, "ovary- affiliated" testis-lobes. Dendrodoa carnea is so much like D. grossu- laria (Traustedt's (1883) type species of his genus Styelopsis, now a subgenus of Dendrodoa} that Arnback (1922) and Hartmeyer ( 1 903) have both suggested these could be merely geographic vari- ants of a single species— a view not held, how- ever, by van Name (1945). The main distinction between these two species is their different num- ber of internal longitudinal branchial vessels, more numerous in D. grossularia than in D. car- nea. But the slight gonadal contrast reported here also seems to be a consistent one. The difference takes on added taxonomic significance when D. abbotti joins the comparison, because the genus thereby shows a series of testis-ovary juxtapo- sitions from a tightly joined one to an appreci- ably looser one— from the condition "character- istic" of the genus Dendrodoa to one rather akin to that of the genus Styela. Perhaps the perplexing Dendrodoa uniplicata NEWBERRY: DENDRODOA (STYELOPSIS) ABBOTTI, SP. NOV. 247 Hartmeyer 1903, which Millar (1966) redesig- nates Styela uniplicata Bonnevie 1896 because "the structure of the gonad agrees better with Styela," extends the grossularia-carnea-abbotti series of gonadal arrangements further while re- taining dendrodoan features of the pharnyx. Un- fortunately, the meager remnants currently available of Dendrodoa (or Styela) uniplicata will not by themselves resolve this question. Another dendrodoan trait from which Den- drodoa abbotti diverges involves the spermiduct. In the genus Dendrodoa, testis-lobes empty in groups into very short vasa deferentia or even more cloaca-like pits on the atrial surface of the ovary, and there are several such spermiporal loci on the ovary (Berrill 1950). The repetition of short vasa deferentia, each emptying a group of testis-lobes, is not usually as striking in D. grossularia as in the somewhat stylized depiction of this trait by Lacaze-Duthiers and Delage (1892), from which work many accounts of the species have been partly drawn. But Riedlinger (1902) indicates in his careful study how slight or even absent the vasa deferentia may be in that species, in place of which spermiporal loci serve the converging vasa efferentia of groups of testis- lobes. Dendrodoa carnea also appears to have multiple spermipores along the atrial surface of the ovary (again, though, a condition difficult to discern in dissections). In contrast, the gonad of D. abbotti has a single, long vas deferens, as in Cnemidocarpa and Styela (Fig. 2). All the sper- miducts of this species are exceedingly fine, and their disposition difficult to trace except in serial sections. Such a close scrutiny of D. carnea would seem appropriate, to find out if that species is intermediate between D. grossularia and D. ab- botti in this trait, as it is in testis-ovary juxta- positions. Dendrodoa (Styelopsis) abbotti is placed in Dendrodoa by its possession of a single gonad, and in Styelopsis because of its unbranched ovary and its simple pharynx, which lacks folds and possesses few internal longitudinal vessels. Den- drodoa abbotti is so much like D. carnea, which in turn is so much like D. grossularia, that this placement of the new species seems indisputable. But the consequence is to relax and modify long- held gonadal criteria of Dendrodoa, recognizing that species with styelan gonadal patterns or cnemidocarpan spermiducal patterns occur in the genus. ACKNOWLEDGMENTS A grant from the Faculty Research Committee of the Academic Senate of the University of Cal- ifornia, Santa Cruz has supported much of the research reported in this paper. I appreciate the assistance of Linda Cole, U.S. National Museum of Natural History, who guided me through the collection there, with the consequence that Den- drodoa carnea came into consideration at a crit- ical moment in this study. Professor Ivan Good- body has shared with both Professor Abbott and me several Canadian specimens of Dendrodoa abbotti and ecological information about their site; I am grateful for his help and for his read- iness to include this important northern inter- tidal record in this initial paper about the new species. Donald P. Abbott, without realizing at the time the nomenclatural consequence of his generosity, shared with me his notes and draw- ings of many years' acquaintance with the species described in this paper, and I am most grateful for these and for many other ways in which he has encouraged me. LITERATURE CITED ARNBACK-CHRISTIE-LINDE, A. 1922. Northern and arctic in- vertebrates in the collection of the Swedish State Museum. 8. Tunicata. 1 . Styelidae and Polyzoidae. In Kungl. Svenska Vetenskapsakad. Handlingar 63(2): 1-62, pis. 1-3. BERRILL, N. J. 1950. The Tunicata, with an account of the British species. London: Ray Society. 354 pp. BONNEVIE, K. 1896. Ascidiae simplices og Ascidiae Com- positae fra Nordhavs Expeditionen. In Norske Nordhavs- Expedition 23(2): 1-16, pis. 3, 4. GALIGHER, A. E. AND E. N. KOZLOFF. 1964. Essentials of practical microtechnique. Philadelphia: Lea & Febiger. 484 pp. HARTMEYER, R. 1903. Die Ascidien der Arktis. In ROmer, F. and F. Schaudinn, Fauna Arctica 3(2):91-412, pis. 4-14. HUNTSMAN, A. G. 1913. The classification of the Styelidae. Zool.Anz. 41:482-501. LACAZE-DUTHIERS, H. DE AND Y. DELAGE. 1892. Etudes sur les ascidies des cotes de France. Faune des Cynthiadees de Roscoff et des c&tes de Bretagne. M6m. Acad. Sci. France (ser. 2)45:1-323. LIGHT, S. F., R. I. SMITH, F. A. PITELKA, D. P. ABBOTT AND F. M. WEESNER. 1954. Intertidal invertebrates of the cen- tral California coast. 2nd ed. Berkeley: Univ. Calif. Press. 446 pp. MILLAR, R. H. 1966. Tunicata Ascidiacea. Marine inverte- brates of Scandinavia, No. l.Oslo:Universitetsforlaget. 123 pp. MONNIOT, C. AND F. MoNNioT. 1972. C16 mondiale des gen- res d'ascidies. Arch. Zool. Exp. Gen. 1 13:31 1-367. RIEDLINGER, R. 1902. Untersuchungen ilber den Bau von Styelopsis grossularia des Ostsee. Nova Acta Akad. Leop.- Carol., Halle 81:1-62, pis. 1-6. 248 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 1 5 SMITH, R. I. AND J. T. CARLTON, eds. 1975. Light's manual: VAN NAME, W. G. 1912. Simple ascidians of the coasts of intertidal invertebrates of the central California coast. 3rd New England and neighboring British provinces. Proc. Bos- ed. Berkeley: Univ. Calif. Press. 716 pp. ton Soc. Nat. Hist. 34:439-619, pis. 43-73. TRAUSTEDT, M. P. A. 1883. Vestindiske ascidiae simplices. . 1945. The North and South American ascidians. 2. Molgulidae og Cynthiadae. Vid. Medd. Naturhist. Kj5- Bull. Amer. Mus. Nat. Hist. 84:1-476, pis. 1-31. benh., ann. 1882:108-136, pis. 5, 6. PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES Vol. 43, No. 16, pp. 249-267, 6 figs., 6 tables December 11, 1984 THREE NEW SPECIES OF SEVEN-GILLED HAGFISHES (MYXINIDAE, EPTATRETUS) FROM THE PACIFIC OCEAN By Charmion B. McMillan and Robert L. Wisner Marine Biology Research Division, A -01) 2, Scripps Institution of Oceanography, La Jolla, California 92093 ABSTRACT: Three new species of hagfishes (Myxinidae, Eptatretus) from the Pacific Ocean are described, and compared with /-.. cirrhatus. All four species have seven pairs of gill pouches and associated external openings. Of the new species, /:. carlhubbsi is known from Molokai to Guam, north-central Pacific, K. laurahubbsi from off south-central Chile, and /:. strahani from near Lubang Island, Philippines, South China Sea. Eptatretus cirrhatus occurs in the Australian-New Zealand area. Methods used in examination of hag- fishes are described, and sensory (lateral line) canals are delineated and discussed briefly. INTRODUCTION This study of seven-gilled hagfishes (genus Ep- tatretus) from the Pacific Ocean is one of a series resulting largely from the specimens and data accumulated under direction of the late Carl L. Hubbs. Herein we describe three new species, present new data on E. cirrhatus (Bloch and Schneider 1801), offer suggestions for initial preservation of myxinids to provide good study material, and discuss methods useful in the taxo- nomic study of hagfishes. We also offer figures and a brief description of the sensory canals found in the ocular regions of two of the four species. DISCUSSION Our examinations have shown that species of Eptatretus from the Pacific Ocean have six to fifteen pairs of gill pouches and corresponding external apertures. The three new species de- scribed below, with Eptatretus cirrhatus, com- prise a group having seven pairs of gill pouches. One aberrant specimen has eight pouches on each side, but with corresponding apertures arranged abnormally. Our rather limited counts (22 pairs) from the three new species may not reflect ex- tremes of variation, but the number of gill ap- ertures in Eptatretus cirrhatus appears to be con- stant—seven pairs in 48 specimens. In 44 counts from 22 specimens of the three new species, the only variation from seven was the specimen cited above (further discussed and figured below). Counts of six apertures for Eptatretus cirrhatus recorded in the literature apparently resulted from a confusion of species. Giinther (1870) stated that the species had "six or seven gill openings on each side," but he listed specimens from South Africa (E. hexatrema Miiller, 1834) and Japan (E. burgeri Temminck and Schlegel, 1850). Species from these areas commonly have six pairs of gill openings. Referring to Eptatretus cirrha- tus, Waite (1909) stated, "The gill-openings ap- pear to be seven in number, but I have seen an [249] 250 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 16 example in which there were but six openings on one side, though seven were present on the other." A variation of one per side is common in species having ten or more gill pouches, such as E. deani (Evermann and Goldsborough 1907) and E. stoutii (Lockington 1 878). Also, it is possible that Waite had an abnormal specimen in which two pouches shared the same opening (see above). Strahan's (1975) finding of "seven (rarely six) pairs of branchial apertures" for Eptatretus cir- rhatus may have been influenced by Giinther's or Waite's accounts. Regan (1912) listed a species with "7 gill open- ings: on each side two rows of 8 teeth. Southern Pacific" as Heptatretus banksii, and placed in its synonymy Homea banksii Fleming 1822, and Bdellostoma heptatrema Miiller (1834). Regan's total count of 32 teeth is much lower than that of any of the four species treated herein (Table 6), and may indicate an erroneous count or an undescribed species. Regan may have counted three fused median teeth (multicusps) on each row as one, thus reducing the count to 32 from a possible 40. This would have been much nearer our minimal count of 43 for Eptatretus cirrhatus, under which we synonymize the above three names. Species of Eptatretus having seven gill aper- tures are not restricted to the Pacific Ocean. Fern- holm and Hubbs (1981) listed a species having seven apertures from the Caribbean Sea. Fern- holm (1982) has further described it as new. In general, we concur with Fernholm and Hubbs in terminology, with but minor varia- tions. We believe the term "dental muscle" is more appropriate than "tongue," "lingual," or "club-shaped muscle" in reference to the firm elongate complex of muscles and cartilages which constitutes the feeding mechanism of myxinids. Apparently the term "tongue" was first used by Miiller (1834), but we concur with Ayers and Jackson (1900) that the entire apparatus in no way resembles a tongue. They stated, "The ho- mology of this organ with the vertebrate tongue has never been discussed, nor do we know of any effort to determine the true nature of this organ." Dawson (1963:248, fig. 11) provided a detailed analysis and figure of the structure, and of the "teeth" and "jaw apparatus." She concluded (p. 253) that it was unwise to make any definite assumptions concerning homologies of the car- tilages and muscles. There are two pairs of anterior and posterior sets (series) of sharply pointed, laterally flattened, horny structures in the oral cavity which are embedded in a dental plate. These structures cut and scrape food into ingestible portions when everted and retracted by the dental muscle. Al- though the term "teeth" has been widely used in reference to these structures, they are unlike the teeth of other vertebrates, being composed en- tirely of keratin and devoid of calcification. Daw- son (1963:247) concluded that, "It is most likely that there is no phylogenetic connection between these teeth and calcified teeth, and that they are an individual adaptation to a parasitic mode of life."1 For descriptive and statistical purposes, we prefer the terms unicusps and multicusps to differentiate between single and composite teeth— the latter formed by the fusion of two or three unicusps. We consider the number and arrange- ments of both the multicusps and unicusps to be a significant species character. MATERIALS Collection data and disposition of specimens examined in this study are listed in the treatment of each species. Institutions which have fur- nished study material, or in which type speci- mens have been deposited, are: Bernice P. Bish- op Museum, Honolulu, Hawaii (BPBM); United States National Museum, Washington, D.C. (USNM); Scripps Institution of Oceanography, La Jolla, California (SIO); California Academy of Sciences, San Francisco (CAS); Museum Na- tional d'Histoire Naturelle, Paris, (MNHN); University of the Philippines Zoological Mu- seum, Diliman, Quezon City, Philippines (UPZM); Australian Museum, Sydney (AMS); Zoological Institute, Academy of Sciences, Len- ingrad (ZIN). METHODS The methods of measuring and counting de- scribed herein represent original methods as well as some used by prior authors including Dean (1904), Nani and Gneri (1951), Richardson (1953), and Strahan (1975). Fernholm and Hubbs (1981) reported many of these methods in their study of the eastern Atlantic Eptatretus. When 1 Hagfishes are not parasitic; they scavenge dead or mori- bund fishes and invertebrates. MCMILLAN AND WISNER: NEW SPECIES OF PACIFIC HAGFISHES 251 the senior author, in collaboration with the late Carl L. Hubbs, began work on the myxinids (in 1969), it was obvious that no standard criteria existed for the study of hagfishes, which lack the jaws, opercula, rayed fins, scales, gill rakers, and bones found in most fishes. Early workers ap- plied different names to the same anatomical characters, defining them differently or not at all, and often not mentioning the methods used in measuring and counting. Therefore, it was dif- ficult to correlate or compare data of different authors, and taxonomic confusion resulted. We hope that the methods proposed and defined be- low will provide future investigators with a stan- dard by which hagfish species and specimens may be readily compared and identified. Proper treatment immediately after capture is of particular importance in rendering specimens suitable for study. Often too many live hagfish are crowded in jars of preservative, resulting in coiled or bent bodies, usually heavily coated with slime (mucus) and difficult to measure or count. The copious secretion of slime, characteristic of the family Myxinidae, is dramatically curtailed by prompt immersion in fresh water, preferably warm. This rapidly kills the hagfish and prevents further extrusion of slime, which otherwise con- tinues for several minutes even in formalin. Any remaining slime may be removed with paper or cloth towels, and the specimens should then be laid straight in a suitably large container of for- malin until fixed. If a specimen is too large for a flat pan, it should be coiled smoothly in a 3- 5 -gallon container, taking care not to deform the snout or twist the body, and covered with for- malin. This treatment produces fairly straight specimens with a minimal coating of slime, and greatly facilitates accurate counts and measure- ments. Since fresh hagfishes deteriorate rapidly, pres- ervation should be prompt. Color photos or notes should be made to record pigmentation, and tis- sue or blood desired for biochemical or chro- mosomal studies should be taken prior to im- mersion in formalin. We find that initial freezing prior to chemical preservation may cause soft- ening of the tissue and collapse of eggs and in- ternal organs, but it may be preferable to crowd- ing into a too-small container. Due to the many body openings, we consider it unnecessary to slit the skin or to inject preservatives; hagfishes are so soft that the skin may tear and some under- lying tissues may come apart, causing difficulty in subsequent measures and counts. ABBREVIATIONS PCD: external opening of the pharyngocuta- neous duct; ordinarily confluent with the pos- teriormost left gill aperture, and much larger than all other apertures. GA: gill (branchial) aperture; external opening of the efferent duct leading from a gill pouch. GP: gill pouch; rounded, serially arranged structures along and posterior to the dental mus- cle. DM: dental muscle; the firm, elongate, cylin- drical complex of muscles and cartilages that moves the dental plates and sets of cusps during feeding. Posterior portions of DM are shown in Figure 3. VA: ventral aorta; the portion between the heart (ventricle) and where it branches to each side of DM. ABA: afferent branchial artery; one of the small blood vessels that lead to each gill pouch from VA or its branches. MEASUREMENTS If the specimen is distorted due to preserva- tion, it should be moderately straightened to ap- proximate its normal form. Measurements are taken from the left side with the fish lying on a meter stick; dividers or dial calipers are advisable for shorter lengths. We arbitrarily divided the body into four major sections (Fig. 1): prebranchical, branchial, trunk, and caudal. These are particularly apropos to genera Eptatretus and Paramyxine, as each has more than one GA, thus a branchial section. In Myxine, Neomyxine, and Nemamyxine, there is only one GA on each side, that on the left being confluent with PCD. Synonymous terms appearing in the literature are: "head" or "pectoral" for prebranchial, "gill" for branchial, and "abdominal" for trunk. The term "mucus" has often been used for slime, "teeth" for cusps, "tongue" or "lingual muscle" for dental muscle, and "outer" and "inner" for posterior and anterior in referring to the series of cusps. Body measurements we have found particu- larly useful are: Total length (TL): snout (anterior tip of ros- trum, excluding barbels) to posteriormost mar- gin of tail or caudal fin. 252 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 16 FIGURE 1. A-B: Sketches of an Eptatretus and a Myxine, showing regions of body used in study of myxinids: 1 to 5, total length; 1 to 2, prebranchial; 2 to 3, branchial; 3 to 4, trunk; 4 to 5, caudal. C: sketch of head region of a myxinid, showing barbel pairs 1 , 2, and 3, and nasopharyngeal opening, 4. Preocular length: snout to center of eyespot, unpigmented area (if present) marking the ocular region. Prebranchial length: snout to front of first, or only, GA. Branchial length: front of first to front of last GA (PCD). The anterior edge of the last GA is used because the posterior margin is often too vague and poorly defined to provide a definite reference point. Trunk length: front of PCD to origin of cloaca. Body width: maximum dimension about mid- way between rostrum and PCD. Body depth: maximum vertical depth in trunk region, including finfold if present; depth ex- cluding finfold should be taken at the same place. In both width and depth measurements the body should be molded into a seemingly natural shape if necessary. Depth at cloaca: vertical depth at origin of cloaca. Tail depth: maximum vertical depth of flat- tened tail, with any roll-up or fold of the thin tail margin uncurled and flattened. Barbel length: from center of base to tip of each barbel (Fig. 1). The distance between bases of each pair may be measured from the inside edge of each base. Barbels are often curled and difficult to measure accurately, but in certain species barbel length may be a significant char- acter, and is worth measuring. Dental muscle length (DM): snout to tip of DM, as revealed by a midventral incision in the prebranchial region. Dental muscle width: measured at a straight- sided portion well anterior to tapering end. Dental muscle depth: measured at same place as width measure. Rather than using the total MCMILLAN AND WISNER: NEW SPECIES OF PACIFIC HAGFISHES 253 length, we have found it convenient to compare the length (or width) with the unbranched por- tion of the VA with measurements of the DM. This is a significant ratio in certain species, but varies greatly between specimens of other species. Weight: may be taken, but we have not found it to be a reliable or useful character, principally because of the uncertainty in determining if all the entrapped fluid was drained, and because of dehydration of body fluids during preservation. COUNTS Ordinarily the branchial openings (GA) are the first items examined to ascertain the genus and possible species. The gill pouches are usually counted after the teeth (cusps) when the oral cav- ity incision is extended midventrally to the re- gion of the PCD. Before counting the slime pores, we gently scraped away any coagulated slime overlying the line of pores; an air jet greatly fa- cilitated location of pores. Because so few spec- imens were available for this study, both sides were counted to obtain wider range of variation. Counts we have found particularly useful are: Slime pores: Prebranchial— from anteriormost slime pore to last one before first GA. Branchial— those pores in immediate associ- ation with (usually below and to the right of) each GA; often one less than GA count in Ep- tatretus, and much less, or absent entirely, in Paramyxine. There is usually no slime pore as- sociated with PCD, but this varies with species and individual specimens. In this study all species except E. strahani have a branchial pore count equal to or higher than the number of GA; the extra pores vary in location and number. Trunk— the series posterior to PCD and ter- minating anterior to end of cloaca, distinctly sep- arate from cloacal series. Cloacal— the pores distinctly before a vertical from posterior end of cloaca, usually starting somewhat anterior to and elevated from origin of cloaca. Caudal— from first pore distinctly behind a vertical from posterior end of cloaca to last pore on tail. For statistical purposes we combine counts of cloacal and caudal pores under the heading "tail pores" (Table 2). Cusps (teeth): We refer to a single "tooth" as a cusp, or unicusp, if it is not fused to one or more adjoining cusps. A unit of two or more cusps fused together at some point prior to its FIGURE 2. Cusps and basal plates, in excised and spread condition, ofE. carlhubbsi, paratype USNM 233742, 955 mm TL. embedment in the cartilaginous dental plate is a multicusp. The two paired sets of cusps (the outer and inner rows of Fernholm and Hubbs [1981] and Fernholm [1982]) are examined from the ventral aspect. They are revealed by a midline incision from the base of the oral cavity through the car- tilaginous pharynx until the sets are free and eas- ily turned outward for viewing. There are dis- advantages to this method. It is easy to misjudge the midline (if the "face" has been distorted in preservation) and cut through the median teeth, making counts difficult; also, the resulting view presented to the observer is a reversed image of the actual arrangement. The inner left row ap- pears on the outer right side and vice versa. To avoid this confusion, the incision may be made from either side of the oral cavity to just under the third barbel, then extended laterally down- ward through the thin membrane, exposing the paired sets of cusps which, when spread apart, appear as shown in Figure 2. On most specimens the count of multicusps may be determined by placing a dissecting or air jet needle under the first two cusps and gently 254 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 16 I br. VA FIGURE 3. Ventral view of branchial region of: 1, E. carlhubbsi: 2-3, E. laurahubbsi, showing diversity in afferent branchial arteries (ABA) leading off from branches of ventral aorta (VA); 4, E. strahani; 5, E. cirrhatus. lifting; the multicusp usually lifts and separates from the adjacent unfused cusp. However, in the nine largest specimens examined by us (E. carl- hubbsi), lifting often raised the entire dental plate and set of cusps. Even if cusps are unquestion- ably fused, a line may extend among the fusion to the plate or "gum line;" in such instances per- haps the only valid criterion for separating mul- ticusps from unicusps is the distinctness of this line as seen under magnification. Such lines are in marked contrast to the condition shown by scanning electron microscopy of E. springeri (Fernholm and Hubbs 1981: fig. 2), wherein no lines are evident in the multicusps. Gill apertures and pouches: In genera Myxine, Neomyxine, Nemamyxine, and Notomyxine, dissection is necessary to determine the number of gill pouches, since only one pair of efferent ducts leads to the exterior. A midventral incision is made from the single pair of GA anteriorly MCMILLAN AND WISNER.: NEW SPECIES OF PACIFIC HAGFISHES 255 until all pouches are revealed (Fig. 3). The cut should be deep enough to expose VA and ven- tricle, taking care not to sever branches of VA or any ABA, or to destroy the origin of the ventral finfold if it is present anterior to PCD. There are multiple, readily visible GA in gen- era Eptatretus (5-15 pairs) and Paramyxine (5- 7 pairs). Although the number of internal pouch- es ordinarily is the same as the external apertures, there may be variation; thus, it is desirable to count the pouches and examine the arrangement of the GP relative to DM and branched and un- branched portions of VA (Fig. 3). The arrange- ment is often of taxonomic importance, although variation occurs (see E. laurahubbsi). Sensory canals (lateral lines): Ayers and Wor- thington (1907:331, figs. 5-10), in a study of the skin-end organs of the trigeminal and lateralis nerves of Bdellostoma dombeyi (=Eptatretus stoutii [Lockington 1 878]), described and figured lateral line canals, associated dermal grooves, and nerve endings. They showed the canals as short lines occurring only dorsally and somewhat lat- erally on the "head" and in two groups, one be- fore and one behind the eyespots. Plate (1924: 66, fig. 6 1 D) accepted the interpretation by Ayers and Worthington that the short lines constituted lateral line canals, but considered the dermal grooves to be artifacts. Ross (1963:155) cited both these studies and stated that the lateral lines had not been described in Myxine glutinosa. To our knowledge these are the only prior references to lateral line canals of hagfishes. We concur with Ayers and Worthington that the canals occur only on the head (in the ocular area of the prebranchial region). However, they are lateral only in that a few occur on the side of the head, with most on the dorsal surface (Fig. 4), and none at all on the rest of the body. As- suming that the canals are indeed sensory in function, we prefer the term "sensory" to "lat- eral." Sensory canals occur in only two of the four species discussed here (E. strahani and E. cirrhatus, Fig. 4), but not on all specimens, and are irregular in number and form. The erratic occurrence in location and in numbers of canals is intriguing, as is their total absence in two of the four species. Due to the limited number of specimens avail- able, it is difficult to draw any firm conclusions regarding the taxonomic value of sensory canals. Ayers and Worthington (1907) stated that these canals were difficult to find because they were A E. cirrhatus \ E . strahani FIGURE 4. Sketches (not to scale) of head regions of Ep- tatretus cirrhatus and E. strahani showing arrangements of sensory canals. The first two pairs of barbels are omitted. very small and the surface indications faint, and that any apparent erratic appearance might be due to the observer. However, on the specimens examined by us the canals, when present, were readily visible under adequate magnification and lighting, and often by the unaided eye. They ap- pear as thin lines, about 1-3 mm long, variably straight or curved (Fig. 4), often very slightly raised above the skin, and sometimes covered with a coating of coagulated slime. Histological examination was not done, nor have we attempt- ed to observe these canals on unpreserved fishes. Old, healed scars are often present in areas occupied by the sensory canals, and elsewhere on the body, mostly anteriorly. These are iden- tifiable as shallow depressions, usually wider and longer than the sensory canals. Many scars occur singly, but often they are in groups of parallel lines, the spacing closely resembling that of the anterior cusps. Possibly this scarring occurs when many hagfishes are feeding in close proximity competing for food, or when crowded in a trap. 256 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 16 TABLE 1 . AVERAGES AND RANGES (IN THOUSANDTHS OF TOTAL LENGTH) OF SELECTED BODY PROPORTIONS FOR FOUR SPECIES OF SEVEN-GILLED HAGFISHES (GENUS EPTATRETUS) FROM THE PACIFIC OCEAN. E. carlhubbsi E. laurahubbsi E. strahani E. cirrhatus N (size range in mm) 9(813-1160) Avg. (range) 8 (240-375) Avg. (range) 5 (265-520) Avg. (range) 8(481-655) Avg. (range) Preocular length Prebranchial length Branchial length Trunk length Tail length Tail depth Body depth with finfold Body depth without finfold Body depth at cloaca 38 (36-54) 184(168-197) 68 (55-77) 602 (577-623) 160(145-176) 97(89-105) No finfold 93 (78-106) 73 (65-85) 50 (44-59) 193(184-204) 55 (52-59) 561 (545-585) 198(181-213) 90 (82-99) 89 (74-97) 81(73-91) 70(61-80) 63 (57-68)* 220(210-231) 77 (69-83) 521 (500-537) 182(174-202) 117(109-125) 111 (101-117) 98 (94-105) 87 (77-94) 60 (52-67) 225(214-239) 76 (69-89) 550 (525-563) 154(135-168) 83(77-91) 89 (69-102) 88 (69-102) 67 (57-75) * Due to lack of visible eyespots, the preocular length was taken from center of uncovered pupil. Waite (1909) placed three adult E. cirrhatus in a bucket of formalin and observed them savagely attacking each other. One was bitten at least 1 5 times by the other two. KEY TO SEVEN-GILLED SPECIES OF Eptatretus FROM THE PACIFIC OCEAN la. Slime pores of trunk 60-70, low, well be- low mid-lateral aspect. Total cusps 61-71. Eyespots present 2 Ib. Slime pores of trunk 45-53, high, near mid-lateral aspect. Total cusps 43-53. Eyespots present or absent 3 2a. Ventral finfold absent. Two (rarely three) fused cusps on anterior multicusps, three on the posterior. Eyespots large, promi- nent E. carlhubbsi n.sp. 2b. Ventral finfold prominent. Two (rarely three) fused cusps on each of the four mul- ticusps. Eyespots present E. laurahubbsi n.sp. 3a. Ventral finfold readily visible. Eyespots absent. Ventral margin of tail forming a nearly straight line from cloaca to abrupt beginning of curve around tail. Anterior few gill apertures small, slitlike. No pale rings around slime pores or gill apertures. Three fused cusps on each of the four mul- ticusps E. strahani n.sp. 3b. Ventral finfold vestigial. Eyespots present. Tail margin smoothly ovate. All apertures rounded. Pale rings around slime pores and gill apertures. Three fused cusps on each of the multicusps E. cirrhatus Eptatretus carlhubbsi new species HOLOTYPE.— SIO 68-473, mature female, 96 1 mm TL, taken at 19°18'N, 166°33.5'E, near Wake Island, in a free-vehicle trap on bottom at 1574 m, 12-13 Sept. 1968. PARATYPES.— SIO 68-473, female, 810 mm TL, taken with the holotype; SIO 82-63 (formerly BPBM 27850), female, 1 1 25 mm TL, taken at Brooks Banks, between French Frigate Shoals and Gardner Pinnacles, Leeward Islands, Hawaii, Nov. 1981, Mokihana Cruise 81-12, set 35, shrimp trap, depth not given; BPBM 27848, male, 1 160 mm TL, taken at 12°56'N, 166°22'W, French Frigate Shoals, Leeward Islands, Hawaii, 7 Nov. 1981, shrimp trap at 684 m; BPBM 27851, male, 830 mm TL, taken off the north shore of Molokai Island, Hawaii, 26-27 Dec. 1981, shrimp trap at 659 m; USNM 227440, male, 900 mm TL, taken at 24°48'N, 167°14'W, R/V Cromwell Cruise 80-05, Station 57, in a shrimp trap at 835 m; USNM 233742 (formerly NMFS P-0289), male, 955 mm TL, taken at 14°59'N, 145°13'E, Esmeralda Bank, Guam, 5-6 April 1981, Cruise Typhoon 81- 01, Station 151, in a shrimp trap at 1061 m; CAS 50705 (formerly BPBM 27847), male, 1064 mm TL, Leeward Islands, Hawaii, Nov.-Dec. 1981, depth and method of capture not given; CAS 50706 (formerly BPBM 27849), male, 980 mm TL, taken at French Frigate Shoals, East Plateau, north side, Leeward Islands, Hawaii, 19 Nov. 1981, in a shrimp trap at 481 m. DIAGNOSIS.— A seven-gilled Eptatretus having no ventral finfold, very large eyespots, two (rarely three) fused cusps on the anterior multicusps and three on the posterior. DESCRIPTION. —Counts: Those of holotype giv- en first (left and right sides), followed by ranges for all specimens in parentheses: gill apertures 7, 7 (all); prebranchial slime pores 15, 16 (12-17); branchial pores 7, 7 (6-8); trunk pores 60, 61 (60-70); cloacal pores 2, 2 (1-3); caudal pores 11, 11 (11-13); tail pores 13, 13 (12-16); total slime pores 95, 97 (93-1 10). Cusps on anterior multicusps 2, 2 (rarely 3); posterior multicusps 3, 3 (all); anterior unicusps 16, 16 (15-17); pos- MCMILLAN AND WISNER: NEW SPECIES OF PACIFIC HAGFISHES 257 TABLE 2. PREBRANCHIAL, BRANCHIAL, AND TAIL SLIME PORES OF FOUR SPECIES OF SEVEN-GILLED HAGFISHES (GENUS EPTA- TRETUS) FROM THE PACIFIC OCEAN. Prebranchial slime pores 12 13 14 15 16 17 18 19 20 E. carlhubbsi 1 1 2 5 2 1 18 E. laurahubbsi 2 3 8 3 16 E. strahani 1 4 2 3 10 E. cirrhatus 4 17 17 4 1 43 Branchial slime pores 6 7 8 N E. carlhubbsi 4 12 2 18 E. laurahubbsi 1 6 3 16 E. strahani 10 10 E. cirrhatus 7 34 2 43 Tail slime pores' 10 12 13 14 15 16 Tail count is the total of the cloacal and caudal slime pores. N E. carlhubbsi 2 4 3 8 1 18 E. laurahubbsi 2 12 2 16 E. strahani 3 4 2 9 E. cirrhatus 1 6 16 14 6 43 tenor unicusps 12, 13 (11-13); total cusps 68 (64-71). Morphometry: In thousandths of total length; values for holotype given first, followed by ranges for all specimens: preocular length 38 (36-54); prebranchial length 184 (168-197); branchial length 68 (55-77); trunk length 602 (577-623); tail length 160 (145-176); tail depth 97 (89-105); body depth 90 (78-1 12); depth at cloaca 74 (62- 86). All specimens from Hawaii were frozen ini- tially; the body proportions of these may not be closely comparable to the other collections, which were initially preserved in formalin. It is not known what effect freezing may have on subse- quent shrinkage, but it is possible that the soft tissues of hagfishes are greatly affected by the expansion of cells in freezing. It is known that length of time in preservative significantly affects the total length; a shrinkage of 10% is not un- common. However, to our knowledge no study has been done showing the changes in other body proportions. Body proportions (Table 1) and counts (Tables 2-6) are compared with similar data for other seven-gilled Eptatretus from the Pacific Ocean. Body robust; prebranchial region slightly deeper than wide; body increasingly compressed laterally to tail, varying in greatest depth from 8% to 1 1% of TL. Two to four GP anterior to tip of DM, which is somewhat flattened poste- riorly. Length of DM 19% (17-21%) of TL, its width 15% (9-19%) of its length, its depth 66% (57-87%) of its width. VA short, wide, its width TABLE 3. TRUNK SLIME PORES OF FOUR SPECIES OF SEVEN-GILLED HAGFISHES (GENUS EPTATRETUS) FROM THE PACIFIC OCEAN. Trunk slime pores 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 N E. carlhubbsi 1 2 3 145 2 18 E. laurahubbsi 3 1 22 125 16 E. strahani 2223 9 E. cirrhatus 5 2 10 11 3 5 2 5 43 258 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 16 £$-~ a S a 5 a :§ 5 -5 J « § §£ ^j -. margaritella. Dasyatis ukpam, a large thick-bodied species with a vestigial sting previously known only from fetal specimens obtained at Old Calabar (without precise information on habitat), is now reported from the Ogooue and the lower Zaire or Congo rivers. It is perhaps related to the genus Urogymnus. INTRODUCTION Ogooue basin in 1978. We have also identified This study was undertaken to clarify the sys- a specimen of this species collected in the lower tematics, distribution, and relationships of West Zaire (Congo) River in 1937. African freshwater stingrays. Although widely The other West African freshwater dasyatid, distributed and familiar to local fishermen, sting- D. garouaensis, was described originally as a rays inhabiting the larger rivers of West Africa species of Potamotrygon, a genus of the otherwise are poorly known scientifically. There are at least exclusively Neotropical freshwater family Po- two species. One, Dasyatis ukpam, was de- tamotrygonidae. Evidence that it is actually a scribed more than a century ago, but the two member of the family Dasyatidae was advanced fetal type-specimens obtained at Old Calabar by Thorson and Watson (1975). Our own ob- lacked precise habitat data, and the species was servations fully support this conclusion. Reid and not reported again or recognized as valid until Sydenham (1979) suggested that D. garouaensis the junior author obtained specimens in the may be identical with the coastal species D. mar- [283] 284 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 18 garita. Our studies, however, indicate that it is a valid species, albeit a close relative of D. mar- garita and the previously undescribed D. mar- garitella. Dasyatis garouaensis was known only from the Benue and lower Niger, but the junior author collected a specimen in the Cross River, Cameroun, in 1980, and we have also identified a specimen from Lagos, Nigeria (habitat un- known). So far as we have been able to deter- mine, D. margarita and D. margaritella occur only in marine and estuarine habitats. The key below should permit identification of all stingrays now known from West Africa. Fol- lowing the key we present a definition of the genus Dasyatis and detailed descriptions of D. margarita, D. margaritella, D. garouaensis, and D. ukpam. MATERIALS AND METHODS This paper is based on material deposited in the American Museum of Natural History (AMNH); British Museum (Natural History) (BMNH); California Academy of Sciences (CAS and CAS-SU); Institut fur Seefischerei, Hamburg (ISH); Museum National d'Histoire Naturelle, Paris (MNHN); Musee Royale de 1'Afrique Cen- trale, Tervuren, Belgium (MRAC); and Smith- sonian Institution (USNM). Disc width (measured at widest point) is our standard measure of size, and proportional mea- surements (unless otherwise indicated) are ex- pressed as percent of disc width. Definitions or explanations of some other terms are as follows: Disc length — midline measurement from snout-tip to a transverse line parallel to poste- riormost extension of pectoral fins Disc depth— greatest height or depth of disc (usually at scapulocoracoid) Preorbital length— midline measurement from snout-tip to a transverse line parallel to anterior margin of eyes Prenarial length— midline measurement from snout-tip to a transverse line parallel to anterior border of nostrils Prebranchial length— midline measurement from snout-tip to a transverse line parallel to opening of first gill slits Head length — midline measurement from snout-tip to a transverse line parallel to opening of fifth gill slits Pectoral fin inner margin— from posterior in- sertion to posteriormost extension of pectoral fin Pelvic fin span— distance between apices of pelvic fins when maximally extended Upper and lower tooth rows — maximum number of tooth rows across upper and lower jaws Vertebral counts in stingrays are complicated by the extraordinary specialization of the ver- tebral column as a support for the pectoral fins, and by its termination in an elongate tail, which is frequently damaged. Anteriorly the column is fused into two synarcuals incorporating a vari- able number of centra. In Dasyatidae the ante- riormost 23-40 vertebrae are incorporated into the first synarcual. In most of these vertebrae the centra are completely fused, but their number can be determined by counting the spinal nerve foramina. The second synarcual frequently is separated from the first by a small number of intersynarcual vertebrae; in most of the Dasyatis herein reported, however, there is only a joint between the two synarcuals. In the second syn- arcual the centra, although fused, retain their form and are readily counted in radiographs. Some- times the posteriormost centrum in the second synarcual is sharply distinguished from the monospondylous trunk centra succeeding it. In specimens in which the end of the second syn- arcual cannot be determined, we give a combined count of second synarcual plus monospondylous trunk vertebrae. This is usual in late fetal or new- born specimens with poor calcification and in heavily denticulated specimens in which this portion of the vertebral column is obscured in radiographs (e.g., in D. ukpam). Posteriorly the vertebral column ends in a long series of diplos- pondylous tail centra followed by a highly flex- ible, unsegmented rod (apparently consisting of the notochord and a heavily calcified notochor- dal sheath). The monospondylous and diplos- pondylous sections of the vertebral column are usually sharply demarcated in radiographs. For terminology and illustrations of dasyatid clasper morphology see Compagno and Roberts (1982). Family DASYATIDAE Jordan, 1888 We follow Bigelow and Schroeder (1953) in restricting Dasyatidae to the whiptailed sting- rays, and tentatively recognize the following gen- era: Dasyatis, Himantura, Hypolophus, Taeni- ura, Urogymnus, and Urolophoides (see also Compagno and Roberts 1982). COMPAGNO AND ROBERTS: WEST AFRICAN STINGRAYS 285 FIGURE 1. Disc shape in West African Dasyatidae. (a) Dasyatis violacea (trapezoidal); (b) Dasyatis centroura (diamond- shaped); (c) Taeniura grabata (circular); (d) Urogymnus asperrimus (oval). KEY TO DASYATIDAE OF WEST AFRICA la. Disc oval (Fig. Id); tail without dermal folds; sting invariably absent Urogymnus africanus (Bloch and Schneider, 1801) Ib. Disc variable in shape, tail with dermal fold or folds, sting usually present (ab- sent in some Dasyatis ukparri) 2 2a. Ventral tail fold extending to end of tail; disc circular (Fig. Ic) Taeniura grabata (E. Geoffrey Saint-Hilaire, 1817) 2b. Ventral tail fold ending far anterior to end of tail (Dasyatis) 3 3a. Disc trapezoidal or diamond-shaped (Fig. la-b) 4 3b. Disc oval or circular 9 4a. Disc trapezoidal, anterior margin broad- ly rounded, snout not projecting as an angular lobe from disc (Fig. la); upper and lower surfaces of disc dark D. violacea (Bonaparte, 1832) 4b. Disc diamond-shaped, anterior margin angular, snout projecting as an angular lobe from disc (Fig. Ib); lower surface of disc light 5 5a. Upper surface of disc with a dark retic- ular pattern; ventral tailfold very short, about twice length of sting D. marmorata (Steindachner, 1892) 5b. Upper surface of disc plain; ventral tail- fold long, much more than twice sting length 6 6a. Entire dorsal surface of disc covered with small denticles; no middorsal row of en- larged denticles or thorns; adults with over 100 rows of teeth in each jaw; disc very broad, about 1.5 times as wide as long in adults D. rudis (Gunther, 1870) 6b. Dorsal surface of disc only partially cov- ered with small denticles, along middle of back, or naked except for a middorsal row of enlarged denticles or thorns; adults with much less than 100 rows of teeth in each jaw; disc narrower, 1.0-1.3 times as wide as long 7 7a. Anterior margin of disc anterior to spi- racles nearly straight behind snout-tip, with tip projecting; posterior parts of pel- vic fins projecting well rearward beyond rear tip of pectoral fins D. pastinaca (Linnaeus, 1758) 7b. Anterior margin of disc anterior to spi- racles slightly concave behind snout-tip, with tip not conspicuously projecting; posterior parts of pelvic fins extending slightly behind rear tips of pectoral fins 8 8a. Ventral tailfold high, about as deep as tail above it; a dorsal ridge present on tail behind sting; disc and tail in large juveniles and adults without enlarged, heavy, broad-based denticles, but with moderately enlarged middorsal and scapular denticles only D. ameri- cana (Hildebrand and Schroeder, 1928) 286 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 18 8b. Ventral tailfold low, half as deep as tail above it; no dorsal ridge on tail behind sting; disc and tail in large juveniles and adults with scattered enlarged, heavy, broad-based denticles, forming a mid- dorsal row at center of disc and tail D. centroura (Mitchill, 1815) 9a. Anterior margins of disc broadly convex anterior to spiracles, with tip of snout not projecting from them; dorsal disc surface in juveniles to subadults com- pletely covered with denticles, with flat- tened large denticles on midbelt, small pointed denticles laterally, and large, conical, erect, sharp-cusped denticles on thorns scattered on disc and tail base; sting reduced or absent; no dorsal keel on tail; base of tail circular in cross sec- tion; ventral surface of disc light with a broad dusky marginal band D. ukpam (Smith, 1863) 9b. Anterior margins of disc concave ante- rior to spiracles, with tip of snout con- spicuously projecting from them; dorsal disc surface partially naked, with a mid- belt of flattened denticles and often a midscapular pearl spine, or naked; no large conical thorns on disc and tail base; sting large; a low dorsal keel on tail be- hind sting; base of tail horizontally oval in cross section; ventral surface of disc light without a broad dusky marginal band 10 1 Oa. Back flattened, without an enlarged mid- scapular pearl spine (sometimes a row of up to 4 moderately enlarged flattened spines in its place); midbelt of flattened denticles reduced or absent, disc some- times entirely naked; snout long, preor- bital length 2.8-3.2 times interorbital space (down to 2.3 in late fetuses or new- born specimens) and 27-32% of disc width; disc very flat, thickness at scap- ulocoracoid 8-11% (usually less than 1 1%) of disc width; lateral prepelvic pro- cesses of pelvic girdle greatly expanded D. garouaensis (Stauch and Blanc, 1962) 1 Ob. Back somewhat arched, with an enlarged midscapular pearl spine; midbelt of flat- tened denticles well developed in large juveniles and adults; snout shorter, preorbital length 1.5-2.4 times interor- bital space and 1 9-26% of disc width; disc thicker, 1 1-15% of disc width over scapulocoracoid; lateral prepelvic pro- cesses slightly expanded 11 1 1 a. Upper jaw strongly undulated, with teeth greatly enlarged on prominent lateral projections; teeth less numerous, in 26- 29/31-34 rows; snout more broadly pointed; pearl spine usually larger and circular, length about 5-6 mm; pectoral radials 133-135; size larger, adults to 65 cm D. margarita (Gunther, 1870) lib. Upper jaw moderately undulated, with teeth moderately enlarged on low lateral projections; teeth more numerous, in 36- 42/38-50 rows; snout usually more acutely pointed; pearl spine usually smaller and often axially oval, length 2- 4 mm; pectoral radials 116-127; size smaller, adults to 26 cm D. margaritella new species Dasyatis Rafinesque, 1810 Dasyatis RAFINESQUE, 1810:16 (type-species Dasyatis ujo Ra- finesque, 1810 [=Rajapastinaca Linnaeus, 1 758], by mono- typy). For full generic synonymy of Dasyatis see Bigelow and Schroeder(1953). DIAGNOSIS. — Dasyatidae with disc circular, oval, trapezoidal, or diamond-shaped (Fig. 1), its dorsal surface smooth or variably covered with small, flat or prickle-like denticles; large, sharp, spine- or plate-like denticles present or absent on dorsal surface; snout angular or broadly rounded, its projecting tip variably developed; pectoral fins rounded or angular; pelvic bar mod- erately arched; tail long, slender, with dorsal and ventral folds or ventral folds only; ventral fold not reaching tip of tail; sting usually present (re- duced or absent in Dasyatis ukpam). Teeth small, rhomboidal, thin-crowned. Dasyatis as here recognized is a large, heter- ogeneous assemblage of about 33 species and may be polyphyletic. Dasyatis margarita, D. margaritella, and D. garouaensis are not far re- moved morphologically from the generic type- species D. pastinaca. Dasyatis ukpam, however, is distinct, approaching Urogymnus Muller and Henle, 1837 in general morphology, heavy den- COMPAGNO AND ROBERTS: WEST AFRICAN STINGRAYS 287 FIGURE 2. (a) Dasyatis margarita, lectotype, 200-mm immature female, West Africa (BMNH 1865.7.4:1); (b) Dasyatis margaritella, 226-mm mature male, Conakry, Guinea (ISH 183/63). ticulation, and sting reduction. Urogymnus species invariably lack the sting, while specimens of D. ukpam either lack the sting or have a very small one. Smith (1863) noted that D. ukpam seemed intermediate between Urogymnus and Trygon (=Dasyatis), at least in the nature of its sting, but included it in Hemitrygon Miiller and Henle, 1837 (=Dasyatis) because it has a short ventral tail fold and no dorsal tailfold. We retain D. ukpam in Dasyatis pending modification of the limits of Dasyatis and other dasyatid genera. The species is readily distinguished from known Urogymnus species in having a ventral tailfold, much longer tail (less than 1.5 times disc width in Urogymnus), a less thick, more circular disc, darker dorsal coloration (dorsal surface pale brown in all Urogymnus examined), a dark mar- ginal band on ventral surface of disc, smaller flat denticles on dorsal surface of disc, and in some specimens a small sting. Dasyatis margarita (Giinther, 1870) (Figure 2a) Trygon margarita GUNTHER, 1870:479 (type-locality West Af- rica). Dasyatis sp. BLACHE ET AL., 1970:53, fig. 117. MATERIAL EXAMINED. — BMNH 1865.7.4:1, 200-mm im- mature female, West Africa (formerly syntype ofD. margarita; designated lectotype below); USNM 222589, 1 30-mm late fetal or newborn male, Sierra Leone; BMNH 1930.3.24:3, 212-mm immature male, Accra, Ghana; BMNH 1936.8.20:2-3, 216- mm and 315-mm immature females, Lagos, Nigeria; BMNH 1899.2.20:35, 206-mm immature female, Banana, Congo Riv- er mouth, Zaire; AMNH 40408, 235-mm female, Angola. LECTOTYPE DESIGNATION.— In the original description of D. margarita Glinther ( 1 870:479) listed two specimens from West Africa without indicating either as holotype: a. Disk 8'/2 inches long, tail 19 inches. Purchased of Mr. J. Wood. b. Young. From the collection of the Zoological Society. These two specimens are therefore syntypes, but our studies indicate they are not conspecific. Specimen a is BMNH 1 865.7.4: 1 , a 200-mm immature female with a single, large, round pearl 288 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43, No. 18 «l f^OoONOinromOvNOrou-i — oo — ooooTr-ir-oOTfir'loO'OON — H §r«^>^>(Nr^(— OOi^loOOOmnror— "S 22^^^od 1O ^^ O "^ VN. § 8 S — — — (N(Nr4TtiOONOO — or^r^rji — fNfN — -^ — NO J^g-rn-* — C ^ Oiqr-oooNONor^cvioooN'nmTj-ONOo — ONONf—. — f^fnooooNoo — _^ V — O *^ ^i ^ fN "^ ON ^i O ^1 ^ °^ fN C^ l/~l ^^ OO ON f*^ O ON fN ^^ NO ^ r— ^ ^®. ^ f*"l ON fS Ci % ;i | »^^00^ — >0 — u^ — OOCj) — (N — JNo — COTT — pn V % ^ j^^u^r-m 001^1 OO0 4 i ^ ,4 rA ^ ^ ^ ^ n 4 ON ^ ?? i J*^ ri NO -I A 'C w c ^J^r^iriroONdr-' o^«X od od — (Nror-i g> z VI —i -~ —i — — — a S X Ci £p S Ou-NW-.'^OOOO'OO'^OOOO'nO^w-.^'nO -n 0-^1 0 ^ t/^r*^rO»^NI^'3 ^ S *^ *^ c •rt "O r^ C "T\ I '5'5 « "i'li "s 1 ill lift 1 4* MS, „« « lii1 g 'i 1 S 1 * l^c|>5^ g»fe.o-g^|^SS ^-2^2-g Cj- | 9 *3 " 2 JrtCd^SoOcOj^53oct- o^'>^ Si S* S fi S 1 9 4 5 s «c-«-'2.t;s^i:i-g2S)--.E^*'-re^ i -1 11 1 CC^WCL •*" 2^-r-i^"r;^/5tuZc^'z;— ucs*-* C CC o.edOO?^ 0^£&-ca-ei-S.3S^v'wca0ltCSc^£Z>S2 — f, — iKa,a,a,a.Xcu!>r!CucuCL ^ Q. «fcs 44, 49-50 peruviana 56 spedosa 52 stolH44, 51 frogort 53 Romalea psittacus 54 Romaleidae 43 Romaleinae 43, 47 136 Sagraea 274 Salangichthyinae 179, 181-182, 186, 189-190, 192, 194-195, 198, 202,204,211 Salangichthys 184, 186, 202-203, 211-213 ishikawae 179-181, 185-186, 199, 204, 209, 213 kishinouyei 2 1 3 microdon 179, 181-187, 201-202, 204, 209, 212- 215,218 328 PROCEEDINGS OF THE CALIFORNIA ACADEMY OF SCIENCES, Vol. 43 Salangidae 179, 181-183, 185, 194-195, 198-206,210- 211,213-218 Salanginae 179, 181-182, 185, 189-190, 192-195, 198, 204-206, 210, 218 Salangoidea 179-200 Salanx 179, 184, 201-204, 206-207, 210-211 acuticeps 206 argentea 211 ariakensis 181, 185-186, 206-208, 210, 213 brachyrostralis 210 chinensis 181, 187, 204-205, 210, 216 cuvieri 180-182, 186, 188, 192, 194, 197, 199, 202, 206-208, 210 (Hemisalanx) 186, 210 (Hemisalanx) prognathus 179, 210 hyalocranius 204-205, 208, 212 (Leucosomd) 186, 198, 210 (Leucosomd) reevesi 179, 198 (Leucosomd) reevesii 210 prognathus 181, 185-186, 189, 195, 208, 210 reevesi 179, 181, 186, 208, 210 reevesii 210 (Salanx) 186, 202, 207, 210 (Salanx) ariakensis 179, 206 (Salanx) cuvieri 179, 207 tSalanx argentea 212 brachyrostralis 210 cuvieri 206 Salmacina 244 Salmo 194 Salmonidae 200-201, 203, 214 Salmoniformes 179, 181, 195, 199-200, 214-215, 218 Salmonoidea 179, 203 Salmonoidei 203 Salmostoma 144, 151 Sardina pilchardus 200 Sardinops sagax 232 Schismorhynchus 60, 62 Schultzidia 60 Sciaena hamrur 302 Scorpaenichthys marmoratus 231-232, 234 Securicula 151 Seriola lalandi 301 Sideria chlevastes 17, 22 Siluriformes 201 Solanaceae 49 Solanum 49 argentinum 49 elaeagnifolium 49 verbascifolium 49 Somniosus microcephalus 89, 105 padficus 89, 105 (Somniosus) 105 Sparidarum rutoti 3 1 2 Sphecidae 27, 123 Sphex figulus 125-126 fuliginosa 126 fuliginosus 123, 126-127 Sphyrna mokarran 89 Spirinchus 194 Spirobis 244 Squalus 228 acanthias 231-232 Stereolepis gigas 301 Styela 246-247 uniplicata 247 Styelopsis 246 Sundasalangidae 179, 181-182, 184-185, 189-190, 192, 194-195, 199, 201-202, 204, 211, 213, 218 Sundasalanx 179, 181, 185, 192-193, 198, 201, 204, 213-214,218 microps 170-181, 186, 191-192, 196, 198, 200- 202,209,213-214 praecox 179, 181, 186, 193, 196, 209, 213-214 Synodontidae 201 Tachysphex 27-42 acanthophorus 27, 29, 39-41 alayoi 29 apricus 27. 29-31, 34 arizonac 27, 29-30, 32, 35 armatus27, 29,40-41 ashmeadii 3 1 belfragei 36, 41 bohartorum 27, 29-31 brevicornis 29 brulli group 28-29, 36, 39-40 crenulatus 35-36 fulvitarsis 29 glabriorll, 36 idiotrichus 27, 29, 31, 34, 36 irregularis 27, 29, 31-32, 34, 36 julliani group 28-29 krombeini 34 krombeiniellus 27, 29, 41 lamellatus 27, 29-30, 32, 35-37 maurus 4 1 menkei 27, 29, 41-42 mirandus 27, 29, 32-34 mundus29, 36,40-41 musciventris 27, 29, 33-35 occidentalis 27, 29, 34-35 papago 27, 29, 35 pechumani 30-3 1 pompiliformis group 28-29, 31,36 psilocerus 30, 35 semirufus 33-34 Solaris 27, 29, 35 sonorensis 32, 37 spatulifer 27, 29, 35-36 spinulosus 40 INDEX 329 tarsatus 34 terminatus group 28-29 texanus 30 undescribed species 33 verticalis 27, 29, 31,36-37 yolo 27, 29, 37-39 yuma 27, 29, 39 Taeniura 284 grabata 285 Tetraodon 2,15 modestus 2 Tetraodon (Arothrori) modestus 7-8 modestus 2 naritus 2 Tetraodontidae 1, 14-15 Tetraodontiformes 14 Thryssocypris 141-158 smaragdinus 141-158 tonlesapensis 141-158 Thunnus albacares 301 obesus 301 Thysanopoda pect inata 108 Topobea 269-270, 281-282 brenesii 270 calophylla 269, 280-282 durandiana 282 elliptica 269-270, 281 pittieri28l Trichoptera 14 Trygon 287, 295 margarita 287, 290 ukpam 295 Trypanorhynchida 110 Trypoxylon 125-126 apicale 126-127 apicalis 126 fieuzeti 127, 131 figulus 123-140 figulus barbarum 123, 127, 131 figulus koma 123, 132 figulus major 1 26 figulus forma major 126 figulus var. mo/or 1 36 figulus var. majus 1 26 figulus media 1 36 figulus forma media 136 figulus var. media 136 figulus medium 132, 136 figulus var. medium 1 36 figulus minor 1 32 figulus forma minor 132 figulus var. minor 132 figulus minus 132, 136 figulus var. minus 1 32 figulus minus var. rwZ?/ 1 36 figulus yezo 123, 127 WO/MS 123, 127 medium 123-140 mmus 123-140 rwfci 123, 139 Umbridae217 Urogymnus 283-284, 286-287, 295 africanus 285 asperrimus 285 Urolophoides 284 Uropterygius 22 alboguttatus 17, 22 kamar 22 marmoratus 17, 22-23 xanthopterus 17, 22 Verbesina encelioides 49 Xenopterus 1-3, 9 bellengeri 2 naritus 2-3 Xestotrachelus 43-58 hasemani 44, 56-57 , 48, 50, 55-57 Zalophus californianus 77-85, 229-230 Zoniopoda robusta 56