O PALEOB Relationships of the Fossil and Recent Genera of Rabbitfishes (Acanthuroidei: Siganidae) Jp~" ' -v w-.< j S'. ■■ ■* , •u «£• 'is^w* ‘ ■ -dMk U wm FX?- P8SS A. - SERIES PUBLICATIONS OF THE SMITHSONIAN INSTITUTION Emphasis upon publication as a means of “diffusing knowledge” was expressed by the first Secretary of the Smithsonian. In his formal plan for the institution, Joseph Henry outlined a program that included the following statement: "It is proposed to publish a series of reports, giving an account of the new discoveries in science, and of the changes made from year to year in all branches of knowledge." This theme of basic research has been adhered to through the years by thousands of titles issued in series publications under the Smithsonian imprint, commencing with Smithsonian Contributions to Knowledge in 1848 and continuing with the following active series: Smithsonian Contributions to Anthropology Smithsonian Contributions to Botany Smithsonian Contributions to the Earth Sciences Smithsonian Contributions to the Marine Sciences Smithsonian Contributions to Paleobiology Smithsonian Contributions to Zoology Smithsonian Folklife Studies Smithsonian Studies in Air and Space Smithsonian Studies in History and Technology In these series, the Institution publishes small papers and full-scale monographs that report the research and collections of its various museums and bureaux or of professional colleagues in the world of science and scholarship. The publications are distributed by mailing lists to libraries, universities, and similar institutions throughout the world. Papers or monographs submitted for series publication are received by the Smithsonian Institution Press, subject to its own review for format and style, only through departments of the various Smithsonian museums or bureaux, where the manuscripts are given substantive review. Press requirements for manuscript and art preparation are outlined on the inside back cover. I. Michael Heyman Secretary Smithsonian Institution SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY NUMBER 84 Relationships of the Fossil and Recent Genera of Rabbitfishes (Acanthuroidei: Siganidae) James C. Tyler and Alexandre F. Bannikov SMITHSONIAN INSTITUTION PRESS Washington, D.C. 1997 ABSTRACT Tyler, James C., and Alexandre F. Bannikov. Relationships of the Fossil and Recent Genera of Rabbitfishes (Acanthuroidei: Siganidae). Smithsonian Contributions to Paleobiology, number 84, 35 pages, 21 figures, 1 table, 1997.—Four genera of fossil siganid fishes of early Eocene to early Oligocene age are recognized in addition to the single Recent genus. The osteological features of these five genera are described and illustrated. A phylogenetic analysis utilizing PAUP indicates that the genera have the following phyletic sequence convention: Ruffoichthys Sorbini (two species from the middle Eocene of Italy)— Eosiganus, new genus (one new species from the middle Eocene of Russia)— Siganopygaeus Danilchenko (one species from the early Eocene of Turkmenistan)— Protosiganus Whitley (one species from the early Oligocene of Switzerland)— Siganus Forsskal (27 Recent species in the Indo-Pacific). Official publication date is handstamped in a limited number of initial copies and is recorded in the Institution’s annual report, Annals of the Smithsonian Institution. SERIES COVER DESIGN: The trilobite Phacops rana Green. Library of Congress Cataloging-in-Publication Data Relationships of the fossil and recent genera of rabbitfishes (Acanthuroidei: Siganidae) / James C. Tyler and Alexandre F. Bannikov. p. cm.—(Smithsonian contributions to paleobiology ; no. 84) Includes bibliographical references. 1. Siganidae, Fossil—Europe. 2. Paleontology—Eocene. 3. Paleontology—Oligocene. 4. Animals, Fossil— Europe. 5. Siganidae. I. Bannikov, A. F. (Aleksandr Fedorovich) II. Title. III. Series. QE701.S56 no. 84 [QE852.P4] 560 s—dc21 [567'.7] 97-21827 The paper used in this publication meets the minimum requirements of the American National Standard for Permanence of Paper for Printed Library Materials Z39.48—1984. Contents Page Introduction. 1 Methods. 1 Acknowledgments. 2 Systematic Descriptions of the Genera of Siganidae. 2 t Ruffoichthys Sorbini, 1983 . 3 Ruffoichthys bannikovi Tyler and Sorbini, 1991, and Ruffoichthys spinosus Sorbini, 1983 . 3 f Eosiganus, new genus. 8 Eosiganus kumaensis, new species. 8 tSiganopygaeus Danilchenko, 1968 . 13 Siganopygaeus rarus Danilchenko, 1968 . 13 f Protosiganus Whitley, 1935 . 16 Protosiganus glaronensis (Wettstein, 1886). 16 Siganus Forssk&l, 1775 . 20 Siganus species.20 Analysis of Characters.25 Analytical Protocols.25 Characters Used in Phylogenetic Analysis.28 1. Number of Pelvic-Fin Spines.28 2. Procumbent Spine on First Pterygiophore of Spiny Dorsal Fin.29 3. Number of Anal-Fin Spines.29 4. Number of Supernumerary Dorsal-Fin Spines.29 5. Length of First Supernumerary Dorsal-Fin Spine.29 6. Association of Postcleithrum with First Anal-Fin Pterygiophore.30 7. Number of Dorsal-Fin Spines. 30 8. Number of Dorsal-Fin Rays.30 9. Number of Anal-Fin Rays . 30 10. Number of Supraneurals. 31 11. Shape of Teeth. 31 12. Ornamentation of Procumbent Spine .32 Other Characters. 32 Vacant Intemeural Space.32 Size of Uroneural.32 Subocular Shelf.32 Longest Dorsal-Fin Spine.33 Conclusion. 33 Literature Cited.34 Relationships of the Fossil and Recent Genera of Rabbitfishes (Acanthuroidei: Siganidae) James C. Tyler and Alexandre F. Bannikov Introduction Siganid rabbitfishes are one of the more prominent herbivo¬ rous components of the modern-day Indo-Pacific ichthyofauna. They are associated with coral reefs, the surrounding grass flats, and other algae-rich environments, such as mangroves and rocky shores. There is a single extant genus ( Siganus ), but it is exceptionally speciose (27 species that differ mostly in coloration and proportions). This genus is unique among Recent teleosts in having the palatine divided into two separate ossifications (palatine and prepalatine) and in having each pelvic fin with two spines (outer and inner, between which are three rays). Along with the description of a new fossil species in the morphologically primitive Eocene genus Ruffoichthys (palatine probably composed of two separate ossifications but no inner pelvic-fin spine), Tyler and Sorbini (1991) compared that genus with the two other fossil genera then known (the Eocene Siganopygaeus and Oligocene Protosiganus), but they did not analyze their phylogenetic relationships. We herein describe a new genus of Eocene siganid ( Eosiganus ), rede- scribe all of the other fossil taxa on the basis of our examination of all of the type materials, and utilize PAUP (Phylogenetic Analysis Using Parsimony, Version 2.4.1, written by D.L. Swofford, then of the Illinois Natural History Survey, now of the Smithsonian Institution) to analyze the relationships of the James C. Tyler, National Museum of Natural History, Smithsonian Institution (MRC-106), Washington, D C. 20560. Alexandre F. Bannikov, Paleontological Institute, Russian Academy of Sciences, Profsoyuznaya 123, 117647 Moscow, Russia. Reviewers: Carole Baldwin, Department of Vertebrate Zoology, National Museum of Natural History, Smithsonian Institution, Washington, D.C. 20560. Antony S. Harold, Department of Ichthyol¬ ogy and Herpetology, Royal Ontario Museum, Toronto, Ontario, Canada M5S 2C6. Robert L. Shipp, Department of Marine Sciences, University of South Alabama, Mobile, Alabama 36688. five genera based on the 12 informative characters that can be determined in the Recent and most of the fossil taxa. Methods. —The methodology used for the phylogenetic hypotheses is given under “Analytical Protocols” at the beginning of the “Analysis of Characters” section. Abbreviations for the repositories of materials are as follows: ANSP, Academy of Natural Sciences of Philadelphia; MCSNM, Museo Civico di Storia Naturale di Milano; MCSNV, Museo Civico di Storia Naturale di Verona; MCZ, Museum of Comparative Zoology, Harvard University; NMB, Naturhistorisches Museum Basel; PIN, Paleontological Insti¬ tute, Moscow; USNM, National Museum of Natural History, Smithsonian Institution (collections of the former United States National Museum). Specimen length is always standard length (SL). Vacant intemeural spaces are given the number of the preceding neural spine (e.g., the first space is between the first and second neurals, as in Baldwin and Johnson, 1993), and the terms supraneural (as in Mabee, 1988) and epineural (as in Patterson and Johnson, 1995) are used rather than predorsal bone and epipleural intermuscular bone; all three of these terms are changes from those used in Tyler et al. (1989) for siganids and other acanthuroids. For ease of comparison, descriptive features of the species are given in the same order, beginning with the few that are unique to either all or some siganids (palatine composed of two separate ossifications, two spines in each pelvic fin, deep notches on sides of teeth). All five genera of siganids have 10 abdominal and 13 caudal vertebrae, and these data are not repeated. Most other families of acanthuroids (luvarids, zanclids, acanthurids) have a more specialized condition of one fewer abdominal vertebra, 9+13=22, but the fossil kushlukiids (sister group of luvarids) have a secondarily increased number of vertebrae 10+-19- 1 2 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY 20=29-30. Scatophagids are like siganids in having 10+13=23, itself a specialized reduction of one caudal vertebra from the 10+14=24 of ephippidids and other, lower, squamipinnes. Other familial-level features not repeated herein that are common to all siganids are (1) the spines of the dorsal and anal fins are heteracanth; (2) some of the superficial skull bones, especially the frontal, are sculptured and cancellous (unknown in Siganopygaeus ), more so in Ruffoichthys, Eosiganus, and Protosiganus than in Siganus; (3) the ethmoid is block-like, the supraoccipital has only a very low crest, and the maxilla and premaxilla appear to be immovably articulated and nonprotru- sile, rotating around the ethmoid and flexible palatine (none of these features known in Siganopygaeus ); (4) the distal ends of the pterygiophores of the dorsal and anal fins are laterally expanded; (5) the first two anal-fin spines are in supernumerary association with the first pterygiophore; (6) there are 17 principal caudal-fin rays (only probable in Siganopygaeus ), nine in the upper lobe and eight in the lower, but we comment on the caudal fin because of variability in both its degree of preservation and the number of procurrent rays. In all five genera, the ventral shaft of the first pterygiophore of the dorsal fin is placed in the first intemeural space, but the differences in the location of the ventral end of the shaft relative to the first neural spine are described. Although they sometimes have been thought to be related to siganids or acanthurids, the Eocene fishes of the genera Pygaeus, Acanthopygaeus, Malacopygaeus, and Parapygaeus are chaetodontoids (Woodward, 1901; Eastman, 1904; Patter¬ son, 1993). The enigmatic Gazolaichthys vestenanovae Blot and Tyler (1991) was described from the Eocene of Monte Bolca, Italy, as an incertae sedis acanthuroid with characteristics suggestive of an intermediate position between acanthurids and siganids. However, we believe that it is more closely related to higher acanthuroids (the zanclid+acanthurid clade) than to siganids for the following reasons. Gazolaichthys shares several derived features with zanclids and acanthurids (e.g., vertebrae reduced to 9+13; first pterygio¬ phore of dorsal fin situated in front of neural spine of first vertebra; principal caudal-fin rays reduced to 16). By contrast, none of the similarities between Gazolaichthys and siganids are derived, and Gazolaichthys lacks such derived features of all or most siganids as increased numbers of dorsal- and anal-fin spines, reduced numbers of dorsal-, anal-, and pelvic-fin rays, pleural ribs on the second abdominal vertebra, and procumbent spine on first pterygiophore on the dorsal fin. Gazolaichthys differs from both siganids and the zan¬ clid+acanthurid clade by having the fourth intemeural space vacant and having two well-developed uroneurals. Gazolaich¬ thys differs most notably from acanthurids by lacking the highly specialized type of dorsal- and anal-fin spine locking mechanism by which the cap-like base of the first spine rotates into a deep indentation in front of the median flange at the distal end of the first basal pterygiophore of these fins. The relationships of Gazolaichthys as a basal member of the zanclid+acanthurid clade will be discussed in a work in progress by one of us (JCT) redescribing the genus on the basis of newly examined Monte Bolca specimens. Acknowledgments. —We thank Lorenzo Sorbini, Museo Civico di Storia Naturale, Verona, Italy, and Daniel Goujet and Paulo Brito, Laboratoire de Paleontologie, Museum National d’Histoire Naturelle, Paris, for facilitating our work with specimens in their care. Victor Springer, Division of Fishes, National Museum of Natural History, queried his ichthyophilatelic data base and made us aware of the existence of a stamp bearing a siganid and of the changed generic name for Protosiganus glaronensis. We were further aided in obtaining details about the issuance of this Swiss stamp by Joseph Geraci of the Smithsonian’s National Philatelic Museum and by its librarian, Timothy Carr, and then by receiving copies of the issuance descriptive materials from Peter Meier of the Philatelic Office in Bern. At the Naturhistorisches Museum Basel we are especially indebted to Peter Jung, Burkart Engesser, and Daniel Oppliger for their arduous and successful efforts to locate the long- misplaced holotype of Protosiganus glaronensis. At the Naturwissenschaftliche Sammlungen des Kantons Glarus we received much help from Hanspeter and Edith Schlielly in our unsuccessful search for additional specimens of P. glaronensis. We hoped to find other specimens among their rich holdings of Oligocene fishes from the black schists of Glarus, many of them dating to the time of the study of these collections by Louis Agassiz (1833-1843) for his “Recherches sur les Poissons Fossiles.” We thank the National Geographic Society, Washington, D.C., for grant no. 5037-93 that allowed one of us (AFB) to make excavations in the North Caucasus in 1993, from which two paratypes of Eosiganus kumaensis were obtained. Carole Baldwin, Division of Fishes, National Museum of Natural History, introduced us to the use of PAUP and helped answer many questions of phylogenetic interpretation; her help in improving our analysis and in reviewing the manuscript is greatly appreciated. The manuscript was further improved by the preacceptance reviews of Anthony Herald, Royal Ontario Museum, and Robert Shipp, University of South Alabama, and by discussions of acanthuroid characters with Richard Winter- bottom, Royal Ontario Museum. We appreciate the copy editing of Craig Warren at the Smithsonian Institution Press. Systematic Descriptions of the Genera of Siganidae Diagnosis for Both Fossil and Extant Taxa.— Acanthuroid fishes with the number of dorsal-fin spines increased to 11-14 (versus 3-9 in other acanthuroids), the number of dorsal-fin rays reduced to 9-11 (versus 20-42), the number of anal-fin spines increased to 4-8 (versus 0-3), the number of anal-fin rays reduced to 7-10 (versus 17-35), the NUMBER 84 3 scales small and cycloid (versus larger and spinulose or ctenoid), the first pleural rib inserted on the second vertebra (only probably so in one of the fossil species) (versus inserted on the third vertebra), and the pelvic fin usually with only 3 rays (occasionally 5 rather than 3 in one of the fossil species) (versus always 5 rays). The many other derived features known for the numerous species of the single Recent genus ( Siganus ) are either unknown in the fossil taxa (many features of the soft anatomy or those of regions not exposed in the fossil materials) or are present in some of the fossil taxa but not in others (e.g., inner pelvic-fin spine, palatine composed of two separate ossifica¬ tions). f Ruffoichthys Sorbini, 1983 TYPE Species. — Ruffoichthys spinosus Sorbini, 1983, by monotypy. Other species: Ruffoichthys bannikovi Tyler and Sorbini, 1991. Diagnosis. —Differs from all other siganids by having a 1,3 or, rarely, a 1,5 pelvic fin (versus 1,3,1), no procumbent spine anterodorsally on the first dorsal-fin pterygiophore (versus prominent procumbent spine present), 4 anal-fin spines (versus 6-8), and sixth intemeural space vacant (versus fifth). Ruffoichthys bannikovi Tyler and Sorbini, 1991 and Ruffoichthys spinosus Sorbini, 1983 Figures 1-4 Diagnosis of the Two Species. —The two species of Ruffoichthys differ as follows: there are 10 dorsal- and anal-fin rays in R. bannikovi versus 9 in R. spinosus-, there are 7-8 teeth on each side of the upper and lower jaws in R. bannikovi versus 5-6 in R. spinosus-, the greatest body depth is 41%-43% SL in R. bannikovi versus 23%-34% in R. spinosus, with the neural and haemal spines in R. bannikovi correspondingly longer; the soft portions of the dorsal and anal fins are higher in R. bannikovi (height about 7-8 times in SL) than in R. spinosus (height about 10-11 times in SL). Description. —Two additional specimens of R. spinosus (from Milan) have become available since the genus was redefined on the basis of 16 specimens (about 10-55.1 mm SL) of R. spinosus and two specimens (47.1-50.0 mm SL) of R. bannikovi (Tyler and Sorbini, 1991). We provide additional osteological information about these two species in the following description. The new information is given for both species together so that they can be compared more easily to the new genus and species described herein ( Eosiganus kumaensis ) and to the new data based on our examination of the holotypes of the other two species of fossil siganids ( Siganopygaeus rams and Protosiganus glaronensis). The pelvic fin in each of the two specimens of R. bannikovi has a single spine followed by three rays, whereas in R. spinosus the single spine is followed by either three or five rays. In neither species is there an inner spine that otherwise is typical of siganids. Of the eight specimens of R. spinosus in which the pelvic-fin counts could be obtained with some assurance, seven (including the holotype and one of the two newly examined Milan specimens in which the pelvic fin is clearly preserved) have three rays and one (MCSNV Tomelleri 53) has five rays. Tyler and Sorbini (1991) presumed that there was intraspecific variability in the number of pelvic-fin rays in this species, but another possibility is that the specimen with five rays represents a third species (or subspecies) of the genus. Many of the species of Recent Siganus differ from one another mainly in color pattern, with only subtle or no differences in the osteological features that are preserved in most fossils. Thus, if a fossil genus like Ruffoichthys was as speciose as Siganus and had as relatively minor differences between many of its species as is common in Siganus, the skeletons of several species could easily masquerade among the relatively numerous materials assigned here to R. spinosus. One of the specimens of R. spinosus (MCSNV Mantovani, 32.3 mm SL) has evidence that the palatine was composed of two separate bony elements, as is the case in all of the Recent species of siganids. The teeth are well exposed in several specimens of R. spinosus, including the holotype; they are distinctly notched on both the medial and lateral edges (Figure 19), and the teeth appear to have been fixed. In the upper jaw the medial notch is more distal and slightly less deep than the lateral notch, and the middle cusp is the largest. The teeth in the lower jaws of these specimens are not as completely preserved, but they seem to have the opposite pattern of notching, as is the case in the Recent Siganus. The teeth are less well exposed in R. bannikovi but seem to be somewhat less deeply notched than in R. spinosus, and there may be differences between the two species not only in the number of teeth (see diagnosis above) but in their shape as well; however, this cannot be determined until additional specimens of R. bannikovi with better preserved dentition become available. The dorsal fin has 11 spines, with the first two in supernumerary association with the first pterygiophore; the first dorsal spine is between two-thirds and three-fourths the length of the second spine, and the second to the fourth or fifth spines are of similar length. The number of dorsal-fin rays differs between the two species (see diagnosis above). The anterodorsal end of the first pterygiophore of the spiny dorsal fin is not prolonged as a procumbent spine. The anal fin has four spines. The number of anal-fin rays differs between the two species (see diagnosis above). The first anal-fin pterygiophore has a relatively vertical orientation, and the anteroventral process is short and well separated from the ventral end of the postcleithrum. Figure 1 .—Reconstruction of the skeleton of Ruffoichthys bannikovi Tyler and Sorbini, middle Eocene of Monte Bolca, Italy; based mostly on the holotype, MCSNV IG132595-132596,47.1 mm SL, but with many details from the acid-prepared paratype (for which see Figure 2). xufjoi NUMBER 84 7 FIGURE 4.—Photograph of the holotype of Ruffoichthys spinosus Sorbini (see Figure 3 for data and reconstruction). 8 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY There are about 14-15 pectoral-fin rays. In a few specimens it is clear that the postcleithrum is formed of separate upper and lower elements. The caudal fin has 17 principal rays and seven or eight procurrent rays above and below. There are five hypurals, three epurals, a large uroneural, and a free parhypural, with the haemal spines of PU2 and PU3 autogenous. There is no supraneural. The ventral shaft of the first dorsal-fin pterygiophore is placed between the short neural spine of the first vertebra and the longer one of the second vertebra. The sixth intemeural space is vacant. There are pleural ribs from the second to last (tenth) abdominal vertebrae, most, if not all, of which bear epineural intermuscular bones. The infraorbital series is well developed, with a large lachrymal and an unknown number of more-elongate elements in an arch to the middle of the rear of the orbit. It cannot be determined whether the second infraorbital is loosely articu¬ lated with the lachrymal and whether the main body of the lachrymal lies above the projected course of the infraorbital ring, both of which are synapomorphies of acanthuroids (Tyler et al., 1989). There is a well-developed subocular shelf, probably formed mostly from the third infraorbital. The posteroventral edge of the preopercle is serrate. There are 1+4=5 branchiostegal rays. The scales are small and cycloid. The illustrations of the two species of Ruffoichthys given in Tyler and Sorbini (1991) do not show surface sculpturing on the head bones. We have re-examined the specimens at the Museo Civico di Storia Naturale di Verona to confirm that they have the highly sculptured and cancellous condition typical of Recent siganids and higher squamipinnes, and these features are added to the lateral views of the skeletons of both species reproduced here (Figures 1, 3). We note that the descriptions of the teeth in both species of Ruffoichthys given by Tyler and Sorbini (1991) mention the notches that are present on both the medial and lateral edges but that the illustrations show only the deeper notch and larger cusp. This is rectified in the lateral views of the skeletons and is presented more clearly in the detailed drawing of a typical upper-jaw tooth of R. spinosus (Figure 19). AGE and LOCALITY. —Both species of Ruffoichthys are from the lower part of the middle Eocene (Lutetian; NP 14, Discoaster sublodoensis Zone) of Monte Bolca, Italy. Holotypes and Other Specimens. —Ruffoichthys ban- nikovi. Holotype: MCSNV IG132595 (head to right) and IG132596, in counterpart plates, 47.1 mm SL. Para- type: MCSNV IIB65, single plate, acid prepared, 50.0 mm SL. Ruffoichthys spinosus. Holotype: MCSNV T920, single plate, 55.1 mm SL. Additional Specimens: MCSNV IG43394, single plate, 34.9 mm SL; MCSNV IG43360, single plate, 28.6 mm SL; MCSNV IG186668, single plate, 38.2 mm SL; MCSNV Mercoledt 17/8 84 Lina, in counterpart plates, 34.0 mm SL; MCSNV Mantovani, single plate, 32.3 mm SL; MCSNV Mantovani, single plate, 30.4 mm SL; MCSNV Mantovani, single plate, 26.0 mm SL; MCSNV Mantovani, in counterpart plates, 22.3 mm SL; MCSNV Mantovani, single plate, ~25.5 mm SL; MCSNV Mantovani, single plate, ~21.0 mm SL; MCSNV Mantovani, in counterpart plates, ~10.0 mm SL; MCSNV Tomelleri 41, single plate, 17.7 mm SL; MCSNV Tomelleri 53, single plate, ~32.2 mm SL; MCSNV Tomelleri 40, single plate, ~28 mm SL; MCSNM MMV1194, single plate, 41.3 mm SL; MCSNM MMV148, single plate, 31.8 mm SL; BM(NH) P20931, single plate, 37.4 mm SL. t Eosiganus, new genus Type Species. — Eosiganus kumaensis, new species, by monotypy. DIAGNOSIS. —Differs from all other siganids by having a single supernumerary dorsal-fin spine (versus 2), this first dorsal-fin spine slightly to distinctly longer than the others (versus first spine shorter than the others), and by having about 7 anal-fin rays (versus 9-10). ETYMOLOGY. — Eo, for the Eocene age, and siganus for the extant genus of the family. Eosiganus kumaensis , new species Figures 5-9 This species is listed as “Siganidae, Gen. et sp. nov.” in a preliminary list of the Kuma ichthyofauna (Bannikov, 1993). Diagnosis. —That of the genus, of which it is the only known representative. Description. —Each pelvic fin has an outer and inner spine and what we interpret as three rays, but the rays are compressed between the spines, and the fins from both sides are superimposed, so we cannot be as sure of the number of rays as we are of the spines. The outer pelvic-fin spine is smooth in the 34.2 mm SL holotype, but it has prominent serrations in the ~18 mm SL paratype and perhaps slight serrations in the 19.6 mm SL paratype. We presume the presence of these serrations indicates that at least the ~18 mm SL paratype is a pelagic late larval stage or a juvenile, which has just settled into a benthic habitat, as this stage is also indicated by the serrations on its first dorsal spine (see below). Most Recent species of siganids metamorphose from pelagic larvae to benthic juveniles at about 20-30 mm SL (Leis and Rennis, 1983; Woodland, 1990, and contained references). The condition of the palatine is unknown. The teeth are well preserved. They are strong and bluntly conical, with smooth edges and no notches (Figure 19), and appear to have been fixed. There are about eight teeth to each side of both the upper and lower jaws. The dorsal fin has 11 spines, with a single spine in supernumerary association with the first pterygiophore; the Figure 5. Reconstruction of the skeleton of Eosiganus kumaensis, new genus and species, middle Eocene of North Caucasus, Russia; based on the holotype, PIN 4425-17A, 34.2 mm SL; because the paratypes are substantially smaller than the holotype, they are not used to render parts missing in the holotype. SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY ta ar NUMBER 84 11 FIGURE 7.—Photograph of the smallest paratype of Eosiganus kumaensis, new genus and species; PIN 4425-21 A, ~18 mm SL, with larval serrations on the first dorsal- and pelvic-fin spines and on the nasal (see Figure 8 for details). supernumerary spine is slightly longer than the succeeding spines in the holotype but is much longer than the second spine in the two much smaller paratypes. The first dorsal spine is smooth in the holotype and the 19.6 mm SL paratype, but it is serrate in the ~18 mm SL paratype. There are 11 dorsal-fin rays based on a combination of those that can be counted (in the paratypes) and the number of pterygiophores in places where the rays are not preserved. The anterodorsal end of the first pterygiophore of the spiny dorsal fin is prolonged as a prominent procumbent spine, with longitudinal grooves appar¬ ent in one of the paratypes (Figure 9). The anal fin has seven spines. The anal-fin rays are poorly preserved, but it is clear on the basis of the number of pterygiophores that there were no more than about seven rays. The first anal-fin pterygiophore has a relatively vertical orientation, and the anteroventral process is of moderate length but well separated from the ventral end of the postcleithrum. The total number of pectoral-fin rays cannot be determined. The postcleithrum is formed from separate upper and lower pieces. The caudal fin is well preserved only in one of the paratypes (PIN 4425-20A), but it is somewhat distorted. Our interpreta¬ tion is that the total number of principal rays is 17, with up to eight procurrent rays both above and below. There are five hypurals, three epurals, and a free parhypural; there is an element that we interpret as an uroneural, but it is incomplete 12 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY Figure 8.—Camera lucida drawings of parts of the smallest paratype of Eosiganus kumaensis, new genus and species; PIN 4425-21 A, ~18 mm SL, with some larval serrations present: A, ascending process of the premaxilla and serrate nasal bone, with part of the somewhat upwardly displaced parasphenoid; B, part of the lower region of the serrate first dorsal-fin spine; C, lower region of the serrate outer pelvic-fin spine. Anterior to left; scale applies to all three illustrations. and we cannot determine its full size. Because the haemal spine on PU3 appears to be autogenous, we presume PU2, the base of which is less clear, is also autogenous. There is no supraneural. The ventral shaft of the first dorsal-fin pterygiophore is placed between the poorly pre¬ served neural spines of the first and second vertebrae. The fifth intemeural space is vacant. Pleural ribs are present on most of the abdominal vertebrae and, from its position, it appears that the first rib was attached to the second vertebra, but we cannot be absolutely sure of this. There is evidence of only a few epineural intermuscular bones, and these are incomplete and poorly preserved; they may have been especially slender. The infraorbital series is not well-enough preserved to describe. The number of branchiostegals and the condition of the posteroventral edge of the preopercle cannot be determined. Scales are not preserved. Age, Locality, and Ichthyological Associations.— The type materials are all from the upper part of the middle Eocene (Kuma Horizon, also spelled Kumsky, which correlates with the Bartonian) of the North Caucasus in southwest Russia, on the Pshekha River, Apsheronsk District, about 0.5 km from the Gomy Luch farmstead. The Kuma Horizon is characterized by several late Eocene foraminiferans, and the rich fish fauna has a preponderance of oceanic pelagic species (e.g., represen¬ tatives of stomiiforms, trichiurids, acronurus-stage acanthurids, Bregmaceros cf. filamentosus, Thunnus abchasicus, Palaeorhynchus parini, Palimphyes pshekhaensis, Eomola bimaxillaria ) (Tyler and Bannikov, 1992a; Bannikov, 1993). 1 MM FIGURE 9.—Camera lucida drawing of the first two dorsal-fin spines and pterygiophores of a paratype of Eosiganus kumaensis, new genus and species, PIN 4425-20A, 19.6 mm SL; larval serrations absent. NUMBER 84 13 However, there are also some species typical of inshore benthic habitats, such as priacanthids and a syngnathiform ( Paraeolis - cus bannikovi Parin, 1992). We presume that the holotype and the 19.6 mm SL paratype were in a benthic reef or grass-flat habitat, whereas the ~18 mm SL paratype, with larval serrations, was either pelagic or had just recently settled to the bottom. The Kuma Horizon has been assigned previously to the upper Eocene (Tyler and Bannikov, 1992a; Bannikov, 1993) but according to Cavelier and Pomerol (1986) only the Priabonian (but not the Bartonian) should be included in the upper Eocene. Type Specimens. — Holotype: PIN 4425-17A (head to right) and 17B, in counterpart plates, 34.2 mm SL. Para- types: PIN 4425-20A (head to left) and 20B, in counterpart plates, 19.6 mm SL; PIN 4425-21A (head to left) and 2IB, in counterpart plates, ~18 mm SL. Etymology. —The specific name is for the Kuma Horizon that contains a wonderfully rich assemblage of middle Eocene marine fishes. ^Siganopygaeus Danilchenko, 1968 TYPE Species. — Siganopygaeus rams Danilchenko, 1968, by monotypy and original description. Diagnosis. —Differs from all other siganids by having 14 dorsal-fin spines (versus 11-13), 8 anal-fin spines (versus 4-7), one supraneural (versus no supraneural), the first of two supernumerary dorsal-fin spines very short (versus long when present), and the third dorsal-fin spine by far the longest (versus the third spine of similar length to the immediately preceding and succeeding spines). Siganopygaeus rarus Danilchenko, 1968 Figures 10, li This species originally was described on the basis of two specimens at PIN, but only the single plate of the holotype can now be located. DIAGNOSIS. —That of the genus, of which it is the only known representative. Description. —Most of the head is missing, with the exception of the opercular bones; the opercle and subopercle are well preserved and in place, whereas the preopercle is partially preserved and is displaced in front of the first dorsal-fin pterygiophore. Two other bones are displaced in the region of the mouth; we interpret these as the supracleithrum partially overlying a bone we cannot identify. The pelvic fin is well preserved and clearly has an outer spine followed by three rays and an inner spine (one other spine from the opposite pelvic fin also is exposed). The first ray is slender, the second ray slightly thicker, and the third ray substantially thicker than the other two, about as thick as the inner spine. The distal portions of the rays are incomplete, and, although there is evidence of them being composed of paired halves, no cross-striations are apparent. The palatine and teeth are not preserved. The dorsal fin has 14 spines, with the first two in supernumerary association with the first pterygiophore; the first dorsal spine is very short (2.2% SL) and was not mentioned by Danilchenko (1968). The second spine has only the base preserved but because this is somewhat less thick than the base of the third spine, we estimate that the second spine, although many times longer than the first spine, was shorter than the long third spine (28.8% SL for the third spine). The fourth and succeeding spines are all significantly shorter (19.2% SL for the fourth spine) than the third spine. The dorsal-fin rays are only partially preserved, mostly anteriorly, and the complete number cannot be determined; however, in most siganids ( Eosiganus being exceptional) the dorsal fin has about the same number of rays as the anal fin, and because there are 10 anal-fin rays, we presume that there are about 10 dorsal-fin rays. The anterodorsal end of the first pterygiophore of the spiny dorsal fin is prolonged as a prominent procumbent spine, without evidence of grooves. The anal fin is well preserved and relatively complete. It has eight spines and 10 rays. The first anal-fin pterygiophore has a relatively vertical orientation, and the anteroventral process is moderately long but well separated from the ventral end of the postcleithrum. The pectoral fin is poorly preserved, but there appears to be a minimum of 10 rays; Danilchenko (1968) gave the count as 12-14, perhaps based on the missing paratype. The postclei¬ thrum is formed of separate dorsal and ventral pieces. The caudal fin is essentially absent except for the impres¬ sions of rays from the upper lobe, in which there appear to have been about nine principal rays and an undetermined number of procurrent rays (there are only the impressions of the posterior three). The impression of the caudal skeleton is too vague to interpret how many separate elements were present. A well-developed supraneural is present, with a prominent anterior process from its dorsal end. The ventral shaft of the first dorsal-fin pterygiophore is placed between the neural spines of the first and second vertebrae. The fifth intemeural space is vacant. Pleural ribs are present on the second to ninth abdominal vertebrae. There is no evidence of epineural intermuscular bones, but this could as well be because of poor preservation or exposure of the vertebral centra and upper regions of the pleural ribs as from the true absence of epineurals. The infraorbitals and branchiostegals are not preserved. The displaced preopercle has at least a few large serrations along its posteroventral edge. Scales are not preserved in the holotype, but Danilchenko (1968) said that they are small and cycloid, with about 60-80 NUMBER 84 15 Figure 11. —Photograph of the holotype of Siganopygaeus rarus Danilchenko (see Figure 10 for data and reconstruction). 16 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY rows from just behind the opercle to the caudal-fin base. This information must have been obtained from the missing paratype. AGE AND LOCALITY. —Both specimens (holotype and miss¬ ing paratype) are from the lower Eocene Danatinian (Danata) Formation of Uylya-Kushlyuk, southwestern Turkmenistan (see Tyler and Bannikov, 1992b, and Bannikov, 1993, for the geology and ichthyological associations of the Danatinian Formation); based on nannoplankton, it is possible that the thin fish-bearing layer of the Danatinian Formation belongs to the upper Paleocene rather than to the lower Eocene (N. Muzylyov, Jan., 1995, pers. comm.). Type Specimen.— Holotype: PIN 2179-100,31.2 mm SL, single plate; the missing paratype was never catalogued. f Protosiganus Whitley, 1935 Archaeoteuthis Wettstein (1886), preoccupied by Ar- chaeoteuthis Roemer (1855), a cephalopod; Whitley (1935) gave the replacement name Protosiganus, which therefore takes the same type species. TYPE Species. —Archaeoteuthis glaronensis Wettstein, 1886, by monotypy and original designation. Diagnosis. —Differs from all other siganids by having 6 anal-fin spines (versus 4 or 7-8). Protosiganus glaronensis (Wettstein, 1886) Figures 12-14 Archaeoteuthis glaronensis Wettstein (1886):67—68. This species is known on the basis of the single plate of the holotype. At the time of its original description, the plate of black schist had not been prepared in any way, and the photograph of the specimen by Wettstein (1886, pi. 8: fig. 11) indicates that there was much matrix superimposed on the imprint. Sometime more than 30 years ago the holotypic plate was prepared at the Naturhistorisches Museum Basel using the air-brush abrasive technique to remove the excess matrix and enhance the exhibition qualities of the specimen, with outstanding results. The far greater detail of the skeleton that can now be seen is described below. We know that this preparation was done prior to 1961 because a postage stamp issued by Switzerland in 1961 as the 20 Rp denomination in the Pro Patria series has a representation of this species clearly based on the prepared specimen. The stamp does not bear the name of the fish, and the descriptions of this stamp in catalogs (e.g., Zumstein) simply state that it is a fossil fish (in German, Versteinerter Fisch). However, the publicity material issued from Bern by the Swiss Philatelic Office in 1961 relative to this fossil fish stamp states (in German) that it is “Scorpaena porcus Linne (Familie Scorpaenoidei)” from the early Oligocene of Canton Glarus, an inexplicable error because the specimen was properly identified when on exhibit at the Basel museum and the holotypic plate bears a label on the back side identifying it as the holotype of Archaeoteuthis glaronensis Wettstein. The erroneous identifi¬ cation associated with the stamp was corrected by Bearse (1976), Bearse et al. (1977), and Rice (1976), all of whom used the name Protosiganus glaronensis for it. Diagnosis. —That of the genus, of which it is the only known representative. Description. —The pelvic fin is well preserved and clearly has an outer spine followed by three rays and an inner spine (one other spine from the opposite pelvic fin also is exposed). The condition of the palatine is unknown. The teeth are relatively well preserved. They are long and slender (up to about 1.4 mm in length or 1.2% SL), with at least 14 exposed in the upper jaw and 20 in the lower jaw as based on actual teeth and the gaps in the series (Figure 19). The teeth have smooth anterior edges, but the distal half of the posterior edge has faint indications of what could be low lobations or irregularities. Because the teeth are relatively long and slender, we presume that they were probably moveable, as is typical of setiform teeth when present in other acanthuroids and higher squamipinnes. The dorsal fin has 13 spines, with the first two in supernumerary association with the first pterygiophore; the first dorsal spine is about three-fourths the length of the second spine, and the third spine (22% SL) is slightly longer than the others. There are 10 dorsal-fin rays. The anterodorsal end of the first pterygiophore of the spiny dorsal fin is prolonged as a prominent procumbent spine, with delicate longitudinal grooves. The anal fin has six spines and 10 rays. The first anal-fin pterygiophore has an oblique orientation and a long anteroven- tral process that closely approaches and probably directly contacts the ventral end of the postcleithrum (the actual point of probable contact is overlain by a narrow layer of matrix). Although some small portion of the oblique orientation of the anal pterygiophore may be attributable to distortion of the matrix, its close relationship with the postcleithrum is entirely natural. The total number of pectoral-fin rays cannot be determined. The postcleithrum is formed of separate upper and lower pieces. The caudal fin is well preserved. It has 17 principal rays, with nine procurrent rays above and seven or eight below. There are five hypurals, three epurals, and a large uroneural, but the region of the parhypural is covered with matrix. The haemal spine of PU2 is probably autogenous but that of PU3 appears to be fused with the centrum. There is no supraneural. Although matrix obscures the details, the ventral shaft of the first dorsal-fin pterygiophore seems to be placed directly over the short neural spine of the first vertebra and just in front of the longer neural spine of the second vertebra. The fifth intemeural space is vacant. Pleural ribs are present on the second to ninth abdominal vertebrae; a well-preserved epineural intermuscular bone is SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY rv-;> NUMBER 84 19 Figure 14. —Photograph of the holotype of Protosiganus glaronensis (Wettstein) (see Figure 12 for data and reconstruction), as described herein after air-brush preparation at the Naturhistorisches Museum Basel. 20 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY present on the pleural rib attached to the eighth vertebra, but there are only faint traces of other epineurals. The limits of the individual elements in the infraorbital series are not well-enough differentiated to describe. The posteroven- tral edge of the preopercle is serrate. Most of the branchios- tegals are preserved, but those from both sides are not well separated and it is impossible to determine their total number and grouping. Scales are not preserved. Age and Locality. —The holotype is from the early Oligocene (Rupelian) of Canton Glarus, Switzerland. Type Specimen. — Holotype: NMB E177, 113.6 mm SL, single plate. Siganus Forsskal, 1775 Type Species. — Scarus rivulatus Forsskal, 1775, by mono- typy; for the history of the nomenclatural confusion over the names Siganus = Teuthis and the respective type species, see Gill (1884), Woodland (1990), and Eschmeyer (1990). Other species: see Siganus species, below. Diagnosis. —Differs from all other siganids by the presence of a posterior barb on the procumbent spine of the first pterygiophore of the dorsal fin (versus no posterior barb). Three synapomorphies (long first dorsal spine, postcleithral- pterygiophore contact, 9-10 dorsal rays) indicate that Siganus is most closely related to the Oligocene Protosiganus, and both have 13 dorsal-fin spines (versus 11 or 14 in other genera). Siganus differs from Protosiganus by having the teeth wide and deeply notched (versus slender and with probable low lobations on one side in Protosiganus, but deeply notched teeth are also independently present in Ruffoichthys ), seven anal-fin spines (versus six in Protosiganus, but seven spines are also present in Eosiganus), and shorter dorsal-fin spines (longest spines ranging between 11%—18% SL versus 22% SL in Protosiganus ). Siganus species Figures 15-18 The species differ mainly in coloration, body proportions (including length and stoutness of rays and spines), numbers of scale rows, and, for the two subgenera, length of snout. Woodland (1990) has thoroughly treated the taxonomy and distribution of the 27 species of Siganus (22 in the subgenus Siganus and 5 in the subgenus Lo), all of which are Recent in the tropical and subtropical Indo-Pacific. We therefore limit ourselves to the osteology of the genus and its phylogenetic relationships within the family, topics not covered by Wood¬ land. Materials. —Our description of the osteology of Siganus is based on the following cleared and stained, radiographed, and alcohol preserved specimens. Cleared and Stained: Siganus (S.) canaliculatus (Park), ANSP 77804, 1, 55.9 mm SL. Siganus (S.) corallinus (Valenciennes), ANSP 49224, 1, 139 mm SL. Siganus (S.) luridus (Riippell), USNM 218868, 2, 41.2-61.0 mm SL. Siganus (S.) sp., USNM 109355, 3, 16.0-31.5 mm SL; MCZ 63119, 2, 7.7-9.7 mm SL. Siganus (Lo) vulpinus (Schlegel and Muller), USNM 270217, 1, 91.6 mm SL; USNM 325277, 2, 69.5-80.4 mm SL. Radiographed: Siganus (5.) argenteus (Quoy and Gai- mard), USNM 336445, 8, 96.4-124 mm SL. Siganus (S.) fuscescens (Houttuyn), USNM 336444, 11, 69.8-94.2 mm SL. Siganus (5.) luridus (Riippell), USNM 235620, 3, 105-156 mm SL. Siganus (S.) rivulatus Forsskal, USNM 336446, 3, 134-151 mm SL. Siganus (S'.) virgatus (Valenciennes), USNM 173031, 7, 81.7-95.4 mm SL. Siganus (Lo) vulpinus (Schlegel and Muller), USNM 182882-182883, 72.6-96.5 mm SL. Alcohol Preserved: Siganus (S.) fuscescens (Houttuyn), USNM 235569, 1, 56.7 mm SL. Siganus (S'.) spinus (Lin¬ naeus), USNM 273859, 1,61.1 mm SL. Siganus (S.) argenteus (Quoy and Gaimard; prior to Woodland, 1990, better known as rostratus Valenciennes), USNM 32508, 1, 106 mm SL. Siganus (S.) rivulatus Forsskal, USNM 235326, 1,112 mm SL. Siganus (S.) doliatus Cuvier, USNM 324412, 1, 121 mm SL. Siganus (S.) stellatus Forssk&l, USNM 235297, 1, 204 mm SL. Siganus (Lo) vulpinus (Schlegel and Muller), USNM 182883, 1, 96.5 mm SL. Siganus (L.) uspi Gawel and Woodland, USNM 243961, 1, 158 mm SL. (Compare Figure 18 of Recent species to Figure 19 of fossil species.) Description. —The pelvic fin has outer and inner spines and three rays. The palatine is composed of two separate ossifications, a character unique to siganids among teleosts, which was first described by Starks (1907) for Siganus fuscescens', the palatine and its upper-jaw articulation is well illustrated by Rosen (1984). The teeth are always deeply notched (Figure 18), with those of the upper jaw having a major notch medially (and most species with a lesser or more distally placed notch laterally) and those of the lower jaw with an opposite pattern, the major notch being lateral and the lesser notch, if present, medial. The teeth are fixed, with usually about 10-14 to each side of both the upper and lower jaws. Two dental autapomorphies of Recent siganids are that the replacement teeth lie free in connective tissue on the inner surface of the jaws rather than being enclosed by bone (Tyler et al., 1989) and that the tooth rows on the fifth ceratobranchial are preceded by deep, transverse, tooth-replacement trenches (Guiasu and Winterbottom, 1993); however we cannot determine these features in any of the fossil taxa. The dorsal fin has 13 spines, with the first two in supernumerary association with the first pterygiophore; the first dorsal spine is about two-thirds to three-fourths the length of the second spine, and the slightly longest spines usually are the third to the eighth (see Woodland, 1990). There is 22 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY 1st 2nd dorsal spines* supraoccipifal basal pterygiophore premaxillary mm . lst ’ -abdominal vertebrae supracleithrum -cleithrum dorsal postdeithrum scapula FIGURE 16.—Head bone configuration in species representative of the two subgenera of Recent Siganus: A, Siganus (Siganus) canaliculatus (Park), ANSP 77804, 55.9 mm SL, infraorbitals removed; B, Siganus ( Lo ) vulpinus (Schlegel and Muller), USNM 325277, 80.4 mm SL, infraorbitals in place, with the parts of the ectopterygoid and the two separate ossifications of the palatine that are obscured from view by the lachrymal indicated by dashed lines and dark stipple. The differences in snout length are not so extreme between some of the other species of the two subgenera. NUMBER 84 23 A B FIGURE 17.—Procumbent spine and posterior barb of the first dorsal-fin pterygiophore in Siganus (S.) luridus (Ruppell), USNM 218868, 61.0 mm SL: A, in dorsal view, with the first dorsal-fin spine removed, and B, in lateral view, with both supernumerary dorsal-fin spines in place. The edges of the median articular flanges of the pterygiophore that are clasped by the spines (chain-link type for the second spine) are shown by dashed lines in B. Distal radial removed in both figures. Abbreviations: art. sur. d.s.l, articular surface for the removed first dorsal-fin spine; b, barb on the procumbent spine; d.s.l and 2, first and second dorsal-fin spines; l.f., lateral flanges of the pterygiophore below the first and second dorsal-fin spines; p.s., procumbent spine of the first pterygiophore. extremely little variation in the norm of 10 dorsal-fin rays in all species of Siganus. The anterodorsal end of the first pterygio¬ phore of the spiny dorsal fin is prolonged as a procumbent spine, which bears a posteriorly directed barb from about the middle of its lateral surface. Whereas the dorsal-, anal-, and pelvic-fin spines have poison glands associated with them, neither the procumbent spine nor its barb is poisonous, but these are still formidable armature (Halstead et al., 1971). The anal fin has seven spines and nine rays, with just as little variation as is the case of the elements in the dorsal fin. The first anal-fin pterygiophore has an oblique orientation and a long anteroventral process that contacts the ventral end of the postcleithrum, to which it is firmly held by connective tissue. The pectoral fin most commonly has either 16 or 17 rays, but sometimes 15 or 18 (Woodland, 1990). The postcleithrum is formed of separate upper and lower pieces. The caudal fin has 17 principal rays and 7 to 11 (usually 8 to 10) procurrent rays above and below. There are five hypurals, three epurals, a small uroneural, and a free parhypural, with the haemal spines of PU2 and PU3 autogenous. There is no supraneural. The ventral shaft of the first dorsal-fin pterygiophore is placed directly over the short and 24 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY FIGURE 18.—Teeth in a representative selection of species of the two subgenera ( Siganus and Lo ) of the Recent Siganus ; numbers of teeth are for those to each side of the midline, i.e., on each premaxilla and each dentary. Camera lucida drawings of adjacent teeth from the middle of the left side of the upper (above) and lower jaws; anterior to left; in: A, Siganus (5.) fuscescens (Houttuyn), 14 teeth in both jaws; B, S. (S.) spinus (Linnaeus), 10 teeth in both jaws, with teeth relatively wider and fewer in number than in the other illustrated species; C, S. (5.) argenteus (Quoy and Gaimard; prior to Woodland, 1990, better known as rostratus Valenciennes), 13 teeth in upper jaw and 12 in lower jaw; D, S. (S.) rivulatus Forssk&l, 13 teeth in upper jaw and 14 below; E, S. (5.) doliatus Cuvier, 13 teeth in upper jaw and 14 below; F, 5. (5.) stellatus Forssk&l, 14 teeth in both jaws; G, 5. ( L.) vulpinus (Schlegel and Muller), 12 teeth in upper jaw and 13 below; H, S. ( L.) uspi Gawel and Woodland, 14 teeth in both jaws. The basal regions of the teeth are illustrated here as exposed distal to the edges of the sockets in the jaws bones and do not include the more posteriorly expanded anchoring regions at the base of the teeth within the bony sockets. Scale lines are all 0.5 mm. Nomenclature follows that of the systematic revision of Siganus by Woodland (1990). NUMBER 84 FIGURE 19.—Teeth in the fossil species of Siganidae. Camera lucida drawings of the teeth in the left side of the upper jaw from wherever they are best preserved (usually toward the front); anterior to left; teeth not consistently well-enough preserved in comparable places in both the upper and lower jaws to present drawings of opposing teeth like those given in Figure 18 for Recent species: A, two teeth from Protosiganus glaronensis (Wettstein), NMB E177, 113.6 mm SL, holotype, with the possibility that there are slight irregularities or low lobations along the distal half of the posterolateral edge; B, Eosiganus kumaensis, new genus and species, PIN 4425-17A-17B, 34.2 mm SL, holotype, simple, conical teeth; C, Ruffoichthys spinosus Sorbini, MCSNV T920, 55.1 mm SL, holotype, teeth with notches on anteromedial and posterolateral edges (the teeth are probably less deeply notched in the other species of the genus, R. bannikovi). Scale lines are all 0.5 mm. The teeth are unknown in Siganopygaeus rams Danilchenko, representative of the only other fossil siganid genus. open neural spine of the first vertebra and just in front of the longer neural spine of the second vertebra (with the neural canal effectively roofed over by the connective tissue binding the proximal tip of the pterygiophore to the side walls of the neural arch of the first vertebra). The fifth intemeural space is vacant (rarely the sixth as an intraspecific variation). Pleural ribs are present from the second to ninth abdominal vertebrae, with epineural intermuscular bones present from the first vertebra posteriorly, usually until the first caudal vertebra. The infraorbital series has a large lachrymal and usually four more-tubular elements with pores or broad openings, with the dermosphenotic firmly attached or fused to the sphenotic. At least in the species of the subgenus Lo with especially long snouts, the lachrymal is likewise especially elongate and the second infraorbital element is positioned entirely below the posteroventral edge of the lachrymal, whereas in most species of the subgenus Siganus the lachrymal is shorter and the second element has most of its length projecting behind the posterior edge of the lachrymal. In smaller specimens of species of the subgenus Lo, the second infraorbital has the same specialized loose attachment to the lachrymal as in most other acanthuroids (Tyler et al., 1989:58), although in larger specimens it is more firmly held to the lachrymal by fibrous tissue, but it is not in direct contact with it. There is no subocular shelf. The postero ventral edge of the preopercle is serrate. There are 1+4=5 branchiostegal rays. The scales are small and cycloid. 25 Analysis of Characters Analytical Protocols Highly corroborated hypotheses based on osteological and myological evidence (Tyler et al., 1989; Winterbottom, 1993; Winterbottom and McLennan, 1993; Guiasu and Winterbot¬ tom, 1993) indicate that the families of acanthuroid fishes (sensu stricto, as in Tyler et al., 1989) have the phyletic sequencing convention of Siganidae—Luvaridae—Zancli- dae—Acanthuridae (Nasinae—Acanthurinae), and that the first and second outgroups are, respectively, the Scatophagidae and Ephippididae (with Drepane of uncertain relationship to the ephippidids). The monophyly of the Siganidae is supported by a total of 18 autapomorphies (references above), but this is based only on the Recent species (i.e., those of Siganus). Three of the siganid autapomorphies are myological, and many others are cartilagi¬ nous features or those of the pharyngeals or other regions in which characters cannot be determined in the available fossil materials. However, the fossil siganids do have some of these 18 autapomorphies exposed: all four fossil genera have reduced numbers of dorsal- and anal-fin rays relative to all other acanthuroids; all have the first pleural rib attached to the second vertebra (only probably so in Eosiganus)', the scales are small and cycloid when known (Ruffoichthys and Siganopygaeus)', and the palatine probably is composed of two separate ossifications in the one genus (Ruffoichthys) in which at least one specimen has this bone exposed. Three of the fossil genera have an inner pelvic-fin spine, although this is absent in Ruffoichthys, and two of them (Ruffoichthys and Protosiganus) are known to have larger uroneurals than the specialized small size found in the Recent Siganus. The exposure of at least several of these siganid autapomorphies in each of the fossil taxa gives assurance that they are siganids. Winterbottom and McLennan (1993) believe that scatoph- agids and ephippidids should be included in an expanded Acanthuroidei, but in this paper we follow the more traditional usage of four families of acanthuroids, with scatophagids and ephippidids considered as higher squamipinnes. In the present analysis, the Luvaridae (and their fossil sister group, the Kushlukiidae), Zanclidae, and Acanthuridae are used as the sequential convention members of the first outgroup (the sister group) of the Siganidae, with the Scatophagidae and Ephippididae as other sequential outgroups, in that order. When appropriate, conditions in Drepane and chaetodontoids are mentioned. Statements about osteological conditions in all of these acanthuroid and higher squamipinne outgroups are based on the cleared and stained specimens listed in Tyler et al. (1989). These statements include the conditions, when known, for at least the more fully preserved and described of the fossil representatives of these groups, all of which are from the Eocene: for the Ephippididae, Archaephippus asper (Volta) and Eoplatax papilio (Volta); for the Scatophagidae, Scatopha- 26 SMITHSONIAN CONTRIBUTIONS TO PALEOBIOLOGY gus frontalis Agassiz (for the three preceding species see Blot, 1969); for the Zanclidae, Eozanclus brevirostris (Agassiz) (see Blot and Voruz, 1970, 1974); for the Luvaridae and Kush- lukiidae, the several species of the genera Avitoluvarus, Luvarus, and Kushlukia described by Bannikov and Tyler (1995); and for the Acanthuridae, the numerous genera and species of acanthurids described by Blot and Tyler (1991). Also included as appropriate are data on the poorly known acanthurids from the Oligocene and Miocene that one of us (JCT) is in the process of redescribing. Character polarity is hypothesized using the outgroup comparison method of Maddison et al. (1984). The ancestral states in Table 1 represent those hypothesized at the outgroup node (i.e., the ancestral acanthuroid states) based on conditions in the siganid sister group and outgroups. The matrix for the 12 polarizable osteological features (Table 1) present in both the Recent genus and in at least most of the fossil genera was analyzed using the “Branch and Bound” option of the software package PAUP, an approach that is designed to find all of the most-parsimonious trees. Characters were optimized using both ACCTRAN (acceler¬ ated transformation; favoring reversal over independent acqui¬ sition) and DELTRAN (delayed transformation; favoring independent acquisition over reversal). No differences were found in tree structure with these two methods of optimization. In the phylogeny given in Figure 20, we have chosen to show the DELTRAN optimization; other equally parsimonious distributions of states are discussed in the analysis of each character as appropriate. Four of the characters included in the matrix, namely numbers of dorsal- and anal-fin spines and soft rays, have multiple states (characters 3, 7, 8, and 9 in the “Analysis of Characters” and in Table 1 and Figure 20). The numbers of dorsal- and anal-fin elements are conservative within siganids, ranging for all taxa from 11-14 dorsal spines, 9-11 dorsal rays, 4-8 anal spines, and 7-10 anal rays, and are relatively invariable within taxa. We presume that our sampling of fossil siganid species is highly incomplete. We also have very few specimens of most of the fossil species, all except Ruffoichthys spinosus being known on the basis of only one to three specimens. There is no variation in the dorsal- and anal-fin counts in the 16 specimens of Ruffoichthys spinosus, and it differs in fin meristics from the other known species of the genus only by having one fewer dorsal- and anal-fin ray. The 27 species of Recent Siganus have a remarkable constancy in the numbers of dorsal- and anal-fin spines and rays, almost invariably D.XIII.IO and A.VI 1,9. Relative to the outgroups, the numbers of dorsal- and anal-fin rays in siganids are especially constant (e.g., dorsal rays 20-30 in luvarids+kushlukiids, 20-33 in acanthurids, and 15-53 in ephippidids; anal rays 17-26 in luva¬ rids+kushlukiids, 19-33 in acanthurids, and 14-46 in ephip¬ pidids). The evidence above suggests that the numbers of dorsal- and anal-fin spines and rays evolved conservatively in siganids. Therefore, we believe it justified to treat the four multi-state characters in the analysis as an ordered transformation series. However, because such ordering of data has been criticized (Michevich, 1982), we have analyzed the multi-state characters by treating them as both unordered (Swofford, 1985) and ordered. TABLE 1.—Numerical character matrix: 0 = primitive; 1 = derived, or 1, 2, 3 progressive degrees of specialization; 9 = either unknown because of incompleteness of fossil material (i.e., teeth and dorsal rays in Siganopygaeus) or character undeterminable (i.e., length of absent first supernumerary dorsal spine in Eosiganus, and lack of procumbent pterygial spine and, therefore, of a possible barb in Ruffoichthys). Ancestral states represent those hypothesized at the outgroup node (i.e., the ancestral acanthuroid states) based on conditions in the siganid sister group and outgroups. Genera ■i t g i o / g $ Jj ■i £ £ / t 5 i A / .*: >5fe*jr • ; fe* •»• ' v V^: V,| ifi m’AfK '’ 1*,: ?>-# '■ mm^i m w&\ \ .xHSR |Hj| f , ;i, Q.. tj R »SJ >£~ F \y Jk?- - ^