a) errey evress Tey heknnmeboeee peep eyeis Nr aeesesbiote Set saints are Roe. eer yern ds a pam r= Aaa hn tenain HARVARD UNIVERSITY e Library of the Museum of Comparative Zoology uh i , i un Gia ie rie ut i Uh ie pat i". rhe y i) area U il { AAU ale i if Pair nial at iu y f ih erg ante i \ - J ‘ LM \ PN Ne } ; Rn ' vai Rie \ ; : Wh ‘ ete MRT: F Ni he mh rd ¢ H eer as i ' , 2 eth) Vt Acta LAN HERR RAY nee Me Nog “Lethe ban tiat aA Youn: yea i. ey rf mise oe ati bisa ae WS i a. § " eA mie Yes i >) 2 ¥ oe 7 q ; , oe me \ 1 ; we a ae FG y a), TA Ay P , es . *y - a : Oe ae Fi } ‘vy Y ‘ \ fal in ; 4 in = if } | ri > ! é 1 ; . { 4 P } ra Ng iM id f } & ; \ ni i { Rulletin OF THE Museum of Comparative Loology Volume 139 1970 HARVARD UNIVERSITY CAMBRIDGE, MASSACHUSETTS 02138 U.S.A. vane int ave (venti No. No. No. No. CONTENTS The Argyrolagidae, Extinct South American Marsupials. By Ceorge Caylord simpson. Hebruany. lOO ie eee North American Fossil Anguid Lizards. By Charles A. M. Mes- PHONE Nig LS OB 1S 0 ete et ee NN ol Mare eo A Subfamilial Classification of Scincid Lizards. By Allen E. Greer. Kees SS (aac a RC 7 le aN New Species of Bottom-Living Calanoid Copepods Collected in Deepwater by the DSRV Alvin. By George D. Grice and Kuni Tauern amie, Aron lei) yer ses eee ae ae eae vt ee een The Proterosuchia and the Early Evolution of the Archosaurs; an Essay About the Origin of a Major Taxon. By Osvaldo A. Reig. /A\ ayes ss ea ee eee cee toe ORR Ueteel Cy Oe New Fossil Pelobatid Frogs and a Review of the Genus Eopelo- nates, [By euieloinel |dswess Web (0) ee See The Galaxiid Fishes of New Zealand. By R. M. McDowall. June, RS 0) ec he 2 elt ea RPS re Ie EO ae eae ee ee ree The Spider Genus Ariadna in the Americas (Araneae, Dysder- idae). By Joseph A. Beatty. June, 1970 _ 293 — 433 Bulletin « OF THE Museum. ° f : Comparative Toology The Argyrolagidae, Extinct South American Marsupials GEORGE GAYLORD SIMPSON HARVARD UNIVERSITY CAMBRIDGE, MASSACHUSETTS, U.S.A. VOLUME 139, NUMBER 1 FEBRUARY 19, 1970 PUBLICATIONS ISSUED OR DISTRIBUTED BY THE _ MUSEUM OF COMPARATIVE ZOOLOGY HARVARD UNIVERSITY BULLETIN 1863- Breviora 1952— Memorrs 1864-1935 Jounsonia, Department of Mollusks, 1941- OccasIonaL Parers oN Mo.uusks, 1945- Other Publications. Bigelow, H. B., and W. C. Schroeder, 1953. Fishes of the Gulf of Maine Reprint, $6. 50 cloth. Brues, C. T., A. L. Melander, and F. M. Carpenter, 1954. Classification of Ine sects. $9. 00 cloth. EK Creighton, W. S., 1950. The Ants of North America, Reprint, $10.00 ae - Lyman, C. P., and A. R. Dawe (eds.), 1960. eae cain on Natural Mam malian Hibernation. $3.00 paper, $4.50 cloth. © ly Peters’ Check-list of Birds of the World, vols. 2-7, 9, 10, 12, 15. Rain list on 8 request. ) a 3 My, Turner, R. D., 1966. A Survey and Illustrated Catalogue of the Teredinidae ( Mollusca: Bivalvia). $8.00 cloth. i Whittington, H. B., and W. D. I. Rolfe (eds.), 1963. Phylogeny Aa es) aa of Crustacea. $6.75 cloth. sat Proceedings of the New England Zoological Club 1899-1948. (Complete sets only. ) pas: oe Publications of the Boston Society of Natural History. Publications Office Museum of Comparative Zoology Harvard University Cambridge, Massachusetts 02138, U. S. A. \ © The President and Fellows of Harvard College 1970. THE ARGYROLAGIDAE, EXTINCT SOUTH AMERICAN MARSUPIALS GEORGE GAYLORD SIMPSON ABSTRACT 1. The known Argyrolagidae include two genera, for which the names Micro- tragulus and Argyrolagus are provisionally retained, with certainly four, possibly five or six valid species. M. reigi and A. scagliai are here described as new. The known range is mid or late Pliocene to early or mid Pleistocene in Argentina. 2-0-1-4 . 20.1.4 incisors en- larged, procumbent below; all teeth root- less. The skulls have tubular bony snouts in advance of the incisors, a large, covered masseteric origin in the anterior part of the orbit, no distinct temporal fossa, and globu- lar crania with somewhat inflated ear regions. Forelegs are reduced, hindlegs elongate; tibia and fibula are fused distally; metatarsals III and IV are appressed; there are only two digits in the pes; locomotion was bipedal ricochetal. The habitus is re- markably convergent toward some placen- tals, especially kangaroo rats and jerboas. 3. Argyrolagids are marsupials but show no clear affinity with any others known. They probably arose from didelphids in- dependently of other known families and are distinct at the superfamily level, at least. 4, Although early steps in argyrolagid ancestry and specialization are unknown, they probably became differentiated in South America, and there is no evidence or present reason to postulate that they have ever occurred elsewhere. They do not 2. The dentition is indicate direct or indirect connection with Australia or the presence of a Southern Hemisphere bridge or intervening land. 5. Argyrolagids represent a distinct eco- logical habitus also found among indepen- dently evolved placentals in North America, Africa, Asia, and Australia (there also placental, not marsupial). The living ani- mals of this habitus are characteristic of, although not confined to, deserts. The argyrolagids probably evolved also in adaptation to more or less local desert habitats, although the few specimens so far found were apparently not living under true desert conditions put marginally, in areas perhaps semiarid but not fully arid. 6. Argyrolagids (along with necrolestids and groeberiids) demonstrate that mar- supial radiation in South America was even wider than indicated by the four families, Didelphidae, Borhyaenidae, Caenolestidae, Polydolpidae, usually considered in this connection. Prior to Pliocene-Recent in- vasions, all South American carnivores were marsupials, all medium to large herbivores were placentals, and other ecological niches were divided between placentals and mar- supials, some of the latter, such as the argyrolagids, having extreme adaptive spe- cializations. Marsupial radiation was almost as broad and reached almost as great ex- tremes in South America as in Australia, the most important over-all difference being that in the latter continent the medium to large herbivores were mar- supials. Bull. Mus. Comp. Zool., 139(1): 1-86, February, 1970 I 2 Bulletin Museum of Comparative Zoology, Vol. 139, No. 1 CONTENTS Intro c tong sete te ees et ee ee 2 Noknowlecdienientsye see 2 Glassihicats ome eae ee re eee ee 3 BART ECO T yp aes ee a a pes 15 ANGERS OUU YES * Seto Te 2 ae eee ad ee a see 32 ASO ECO STAT iy agp ees ae a eee WN I 43 iolosay aml Ieolomy 49 Conspectus of South American Marsupials —__- 56 IniteratunemsGited tess tee Aol 62 Aka kevin kownay es ae A ee 65 INTRODUCTION The Argyrolagidae, also sometimes called the Microtragulidae, have been known after a fashion since 1904. After a fashion, only, for something has been lacking in knowledge of animals that have been con- sidered rodents, ruminant artiodactyls, lagomorphs, notoungulates, diprotodont marsupials, paucituberculate marsupials, and polyprotodont marsupials, each after sober consideration by a qualified verte- brate paleontologist. That confusion had several causes. First, until the 1950’s the animals were known only from isolated scraps. (The discoveries of the 1950's and later are here described for the first time. ) Second, dissociated fragments, mainly metatarsals and mandibles, violated the Cuvierian “law” of association; it was impossible for a rational student to predict one part from another. Third, each part was decidedly sui generis, unlike anything else known. And fourth, in spite of that uniqueness, each part had certain (con- vergent, as we now know) broader simi- larities to various unrelated groups of . mammals. It is now possible to bring much, al- though not yet quite complete, order out of that confusion. First, skulls, mandibles, and considerable parts of skeletons are now known. Second, indubitable associ- ation can now be established among the diverse anatomical parts. Third, the oddity of the group is enhanced rather than less- ened by these discoveries, but that makes its definition all the sharper. Fourth, it has thus become possible to identify most merely convergent resemblances as such. There thus comes clearly into view, after all those years, a fascinating and absolutely unique group of marsupials that has evolved in a direction unlike any other marsupials or indeed any other mammals in South America, and yet in doing so has occupied ecological niches resembling those of unrelated, or only distantly related, groups elsewhere in the world. On this and other evidence, it is also becoming increasingly evident that marsupial radi- ation in South America was considerably more complex that has been generally realized. That, in turn, raises interesting evolutionary and biogeographical prob- lems. All of the measurements, in text and tables, are given in millimeters. In the tables L=length and W=width. The fol- lowing abbreviations are used: MMMP, Museo Municipal . . . de Mar del Plata (full name noted below). MACN, Museo Argentino de Ciencias Naturales “Bernardino Rivadavia” (also known, e.g. in publication by L. Kraglie- vich, as the “Museo Nacional”), Buenos Aires. ACKNOWLEDGMENTS The discoveries that now add so enor- mously to our knowledge of this family were all made in the Chapadmalal-Mira- mar region, along the shore line exposures between Punta Mogotes and Punta Her- mengo (southwest of Mar del Plata) under the auspices of the Museo Municipal de Ciencias Naturales y Tradicional de Mar del Plata, Buenos Aires Province, Argen- tina. Most of the discoveries were made personally by Galileo Scaglia, Director of that Museum. Sr. Scaglia has made all the pertinent materials of that museum available for this study and has supplied excellent, detailed field data for each speci- men. The very possibility of this work, along with its details of field occurrence, is thus due to him. Osvaldo Reig worked for a time in collaboration with Sr. Scaglia and also found some of the specimens here described. More recently he had planned to describe these materials himself, and, although he had not yet compiled any notes or manuscript, a number of illustra- tions in various stages toward completion were made under his direction. When he left Argentina to go first to the United States and then to Venezuela, he found it impossible to continue that research. He then most generously turned it over to me, arranging for delivery to me of the speci- mens, and also placing his illustrations at my disposal. I urged Dr. Reig to let his name appear as co-author of this mono- graph, but he firmly declined on the grounds that he would be unable to do any of the actual research. It must never- theless be recorded that the study would not have been made by me or at this time if it had not been for Dr. Reig. Bryan Patterson had also long been interestd in these materials, and it was hoped for a time that he might undertake their study either alone or with me, but he waived his prior rights and insistently transferred the research to me. Sr. Carlos Rusconi kindly supplied in- formation on the type of Argyrolagus parodii Rusconi and gave me two unpub- lished photographs of that specimen. For functional comparisons data were provided by Dr. William D. Turnbull, Field Museum of Natural History, and Dr. Richard G. Van Gelder, American Museum of Natural History. The latter also provided a dipodid skeleton. Other specimens for comparison have been made available in the division Tue ARGYROLAGID MARSUPIALS °* Simpson 3 of mammals of the Museum of Compar- ative Zoology, Harvard University, and from the zoological collections of the Department of Biology, University of Ari- zona. Drawings are by RaVae Marsh. Most of this study was carried out under professorships, half-time each, in the Mu- seum of Comparative Zoology, Harvard University, and the Department of Geology, University of Arizona. This monograph is a joint contribution from these institutions. CLASSIFICATION Superfamily Argyrolagoidea The only known members of this taxon are the Argyrolagidae. It is sufficiently characterized by the diagnosis of that family and description of its members that follow. Justification for ranking as a super- family is given in a later section on affini- ties. Family Argyrolagidae Ameghino, 1904 Argyrolagidae Ameghino, 1904, vol. 58, p. 255. Microtragulidae Reig, 1955, p. 61. Type. Argyrolagus Ameghino, 1904. Referred genus. Microtragulus Ameghino, 1904. Known distribution. Late (possibly mid- dle) Pliocene to early (possibly middle ) Pleistocene, “Araucanian” to San Andrés Formation, Argentina. Diagnosis. Small marsupials with dental 2.0.1.4 formula ———. Teeth hypselodont, root- 2.0.1.4 less. Upper incisors recumbent and lower procumbent, forming a pinching apparatus. Presumed premolars small, nearly styliform. Upper molars simple, rounded lingually and flattened labially. Lower molars bi- lobed, anterior lobe larger, separation of lobes definite labially but may be obscure or absent lingually. Rostrum projecting well anterior to palate and incisors. Enor- mous palatal vacuities. Cranium inflated, with epitympanic and bullar cavities Mandible with small coronoid process 4 Bulletin Museum of Comparative Zoology, Vol. 139, No. 1 relatively low condyle, and inflected angle. Anterior limbs small. Three main segments of posterior limbs greatly elongated. Only two functional metatarsals and toes. Loco- motion probably ricochetal. Generic and family nomenclature. Ame- ghino proposed the then new generic and specific names Microtragulus argentinus and Argyrolagus palmeri in the same paper (Ameghino, 1904). That paper was dis- tributed as a unit with serial pagination, but its original publication was in six different installments extending over three volumes of a journal. The two genera and species were published in the same volume and year, but Microtragulus argentinus was in an earlier issue and therefore has definite temporal, and not only page, priority. In the original publication, Microtragulus was referred to the artiodactyl family Tragulidae, with which it has nothing to do as is now evident, although the misin- terpretation was almost inevitable at the time. The type and then only species was based on a supposed cannon bone, appar- ently: fused metatarsals, which are indeed similar to those of some advanced artio- dactyls. Argyrolagus, its type based on a mandible obviously not artiodactyl but somewhat rodentlike, was made the sole member of a then new proposed family Argyrolagidae. Until 1955 no other name for a family including either of these generic names was proposed. Rusconi (1936) suggested that Argyrolagus and Microtragulus are synonymous, but he did not note that in that case Microtragulus would have priority, and he continued to use the names Argyrolagus and Argyro- lagidae. Reig (1955, 1958) indicated Argyrolagus as a junior synonym of Micro- tragulus and used the family name Micro- tragulidae. No explanation was given, but it was evident (and has been confirmed in personal communication) that Micro- tragulus was considered as a_ senior synonym of Argyrolagus and therefore the valid name for what was believed to be the sole genus of the family. The family name was changed to accord with the only supposedly valid generic name. That com- mon sense procedure was then usual and was not contradicted by any rule or usage, although it has since been modified, con- trary to common sense or earlier usage, by the Code of Nomenclature later promul- gated (Stoll et al., 1961, revised 1964). There has been no first-hand study of this group since Rusconi (1933, 1936). Romer (1966, p. 379) listed it as “PMicro- tragulidae, Microtragulus [Argyrolagus]” (square brackets in the original). Rusconi (1967) continued to use the name Argyro- lagidae, appearing somewhat less positive that Argyrolagus and Microtragulus are synonymous and continuing to ignore the fact that Microtragulus has priority. He again tentatively puts forward that syn- onymy and does recognize that priority in later personal communication. The situation, never clear, is now further obscured by the fact that there are defi- nitely two genera and at least four species, possibly as many as six, among the known specimens. One cannot therefore simply take it that Microtragulus is a senior syn- onym of Argyrolagus and decide the family name on that basis, a decision that would be equivocal enough under the peculiar provisions of the current code. That two genera exist is established on the basis of mandibles and lower dentitions, none of them from the same locality or horizon as the types of M. argentinus and A. palmeri. One of these genera known from other materials does, in all probability, in- clude A. palmeri. The problem at the generic level is that it is unknown whether M. argentinus also belongs to that genus, in which case Argyrolagus is a synonym of Microtragulus, or whether it belongs to the other genus known from mandibles from other horizons and localities, in which case both Microtragulus and Argyrolagus are . valid names. A direct and positive solution to this problem will require finding metatarsals clearly referable to M. argentinus and a mandible of the same individual. The mandible presumably would then indicate whether the type of M. argentinus is or is not congeneric with the mandible type of A. palmeri and whether it does or does not belong to the second genus known from dentitions. One may hope for such a so- lution but cannot reasonably expect it in the near future, at least—well over sixty years have elapsed without the production of a single scrap of an argyrolagid, let alone associated skull and limb bones, from the type deposit of Microtragulus and Argyro- lagus. In the meantime, only quite indirect comparisons are possible. Metatarsals are known from only one individual of this | family aside from the type of M. argentinus. Fortunately in that one instance, MMMP No. 785-S, from the Chapadmalal for- mation, the bones are individually associ- ated with a mandible. The mandible is considered congeneric but not conspecific with the type of A. palmeri. The meta- tarsals of that individual are morpho- logically like the type M. argentinus but are 25 per cent larger, a difference not impossible but improbable within a single species. The second genus now known to belong to this family (whatever names may be given to the genera and the family) occurs at an “Araucanian” horizon prob- ‘ably earlier and almost certainly not later than Ameghino’s types from Monte Her- moso, and also in the Chapadmalalan to the San Andrés formations in the Chapad- malal-Miramar region, beds younger than Monte Hermoso. Hence that genus must also have existed in Monte Hermoso time. The known specimens of jaws and cheek teeth are all definitely smaller than those of A. palmeri and others considered con- generic with the latter, including MMMP No. 785-S. Hence there is at least a pos- sibility that the small metatarsals, type of M. argentinus, belong to the second genus (i.e. not to be the same genus as A. pal- THE ARGYROLAGID MARSUPIALS * Simpson 5 meri), for which Microtragulus would then be the valid generic name. In MMMP No. 785-S the ratio of the length of the metatarsals to the length of M,-4 is 4.19. If the ratio were the same in an individual represented by a_ lower dentition from the San Andrés Formation (MMMP No. 960-M) belonging to the second genus, its metatarsals would be 26.4 mm in length. The length of the type M. argentinus is 28.5, only 9 per cent longer. Such little cogency as this very incomplete, very indirect comparison has, is, however, still further reduced by the facts that the San Andrés and Monte Hermoso individuals are quite unlikely to be conspecific, if only because there is a considerable difference in age, that they might well have had different limb-tooth proportions, and that similarity in length of metatarsals is not in any case a con- vincing generic character. There is no possible objective solution to this problem. Any definite choice as to recognition and naming of the taxa at present must be purely arbitrary. Most clear-cut would be the tempting solution of having all the previous generic and family names officially rejected and then starting anew. In fact, the specific names, as will soon appear, are hardly in better shape and might be included in the holo- caust. However, so radical an action is not likely to be accepted by the Inter- national Commission, would require long and costly argument and action, and even if officially approved, would be personally condemned by many zoologists. I therefore propose action no _ less arbitrary but more conservative. I shall assume, until and unless contrary evidence is discovered, that the type of Microtra- gulus argentinus belongs to my “other genus” including “Argyrolagus” catamar- censis Kraglievich and MMMP No. 960-M (named Microtragulus reigi on a_ later page). On that assumption, Microtragulus is not synonymous with Argyrolagus. This 6 Bulletin Museum of Comparative Zoology, Vol. 139, No. 1 arbitrary assumption has these advantages: It validates both of Ameghino’s classic names for current continued usage, avoiding the necessity of coining any new generic names. It validates the prior (by 51 years) family name and avoids a choice be- tween family names that would be equi- vocal or insoluble under the current Code. Nothing is known in definite contra- diction of this usage. A final objective settlement, if one is ever achieved, is as likely to support this usage as to upset it. The chances are that any change that may later become necessary will be as slight under this usage as under any other. Specific nomenclature. The status of the various previously proposed and here new specific names will be more particularly discussed under the following generic and specific headings. Here it may be pointed out in a more general way that this also is a problem that cannot at present be satisfactorily solved and can only be treated in a somewhat arbitrary way. Four specific names have previously been proposed: Microtragulus argentinus Ameghino, 1904; Argyrolagus palmeri Ameghino, 1904; Argyrolagus catamarcen- sis Kraglievich, 1931; and Argyrolagus parodii Rusconi, 1933. The type of the first is not comparable with any of the others among these types. It is directly comparable with only one other known specimen, from which it differs specifically, at least. Indirect comparisons at the spe- cific level are practically worthless. The type of A. palmeri is not at hand, although it may still be in existence. Available figures and descriptions of it seem to be adequate for comparison. The type of the last named species, A. parodii, has been virtually destroyed and available data are inadequate. Only for “Argyrolagus” cata- marcensis, which proves to be a_ valid species but not to belong to Argyrolagus, is an adequate type actually in hand. In the collections here first described there are two clear-cut genera and species. The species are distinct from “A.” catamarcensis but one is considered congeneric with the latter. Comparisons of the species with M. argentinus, A. palmeri, and A. parodii are unsatisfactory and inconclusive because of the noted deficiencies of these types. I have more or less arbitrarily given new names to the fully definable and distinctive species in the new collection. As previously noted, I have with even greater arbitrari- ness here assigned Ameghino’s two generic names to the two genera recognized in the hitherto undescribed materials. The facts that the teeth are continuously growing and that the sequence of size is also a morphological sequence (see Figure 1) suggest the possibility that the apparent generic and specific distinctions are in fact merely functions of individual age in a smaller number of taxa, perhaps even in one species. The possibility cannot be absolutely ruled out, but it is quite im- probable. The larger species have more, and more distinct, vertical grooves (or flexids ) on the lower molars, whereas it is a rule with few, perhaps no, exceptions in mammals that these tend either to disap- pear or to become fossettids with increas- ing age (wear). There are certain other structural differences, such as the propor- tions of trigonid and talonid lengths, that are not likely to change individually in just this way. Size differences in lower jaws are not associated with other evidence of individual age differences. In all known specimens M} have erupted and are worn. Teeth measurable at the wear surface and the alveolar end are not distinctly larger at the latter end. Limb bones with fused epiphyses are of markedly different sizes, indicating that size differences are not in all, if in any, instances caused by growth. The apparently different species are in part from different geological horizons. Distribution: The known distribution of named taxa is as follows: Chapadmalal- Miramar area, Monte Hermoso, Buenos Aires Buenos Aires Catamarca Province Province Province San Andrés Formation: Microtragulus reigi Barranca de los Lobos For- mation: Microtragulus reigi Vorohué For- mation: Microtragulus reigi Chapadmalal Formation: Microtragulus reigi Argyrolagus scagliai Monte Hermoso Formation: Microtragulus argentinus Argyrolagus palmeri PAndalhuala member or formation (in the “Araucan- . > ian”): Microtragulus catamar- censis The exact level of PArgyrolagus parodii is unknown. It is probably from either the Chapadmalal or the Vorohué Formation and is surely from within the indicated Chapadmalal-Miramar sequence. The Chapadmalal Formation, as re- stricted by Kraglievich (1952), is probably basal Pleistocene in age, now that Blancan in North America and Villafranchian in Europe are generally considered Pleisto- THE ARGYROLAGID MARSUPIALS + Simpson 7 cene rather than Pliocene as sometimes earlier designated. The three overlying formations in which argyrolagids occur do not seem to cover any considerable span of time and are probably also early Pleisto- cene but could just possibly extend into the middle Pleistocene. Actual super- position of Chapadmalal on Monte Her- moso has not been demonstrated, but Monte Hermoso is generally considered somewhat older, and hence Upper Pliocene, on faunal grounds. “Araucanian” is an obsolescent and inappropriate name for a long sequence of mainly Pliocene beds. It seems to include strata of Monte Hermoso age, but also some distinctly older. The type of M. catamarcensis is labeled “Andal- huala,” presumably for the locality, and may well be from the beds so named with that as type locality. If so, the age of this type is probably pre-Monte Hermoso and approximately middle Pliocene. Genus Microtragulus Ameghino Microtragulus Ameghino, 1904, vol. 58, p. 191. Type. Microtragulus argentinus Ame- ghino. Referred species. M. catamarcensis Krag- lievich and M. reigi, new species. Known range. Upper (or possibly Mid- dle) Pliocene to Lower (or possibly Mid- dle) Pleistocene, Argentina. Diagnosis. (For reasons explained else- where, this diagnosis is arbitrarily based on the referred and not the type species. ) M,-3 with rounded lingual faces, internal groove absent or slight, external groove present, relatively posterior, partially dis- tinct second lobe short and wide. These teeth almost as wide as long. My, elongate, distinctly bilobed, second lobe narrow. Discussion. The above characters, shared by M. catamarcensis and M. reigi, sharply distinguish those species from Argyrolagus palmeri and A. scagliai. The teeth are more fully described and other morphological distinctions are mentioned in the anatomi- cal section of this study, 8 Bulletin Museum of Comparative Zoology, Vol. 139, No. 1 Microtragulus argentinus Ameghino Microtragulus argentinus Ameghino, 1904, vol. 58, p. 191; 1906, fig. 344. Type MACN No. 4743, metatarsals ITI-IV and associated tarsal bones. Hypodigm. Type only. Known range. Monte Hermoso Forma- tion, Monte Hermoso, Buenos Aires Prov- ince. Original diagnosis (of genus and _ spe- cies). “The smallest of known artiodactyls, since its size did not exceed that of a small rat. . . . The metatarsal or cannon bone, formed by the fusion of the two median metatarsals III and IV, is only 27 mm long and 2 mm wide in its narrowest middle part. . . . The two metatarsals in question . are fused for their whole length but between the two there remains a deep longitudinal groove in the anterior face and another, shallower, on the pos- terior face. The lateral metatarsals I and V are atrophied, represented only by their styliform proximal ends, this part of metatarsal V being fused with that of meta- tarsal IV, but that of metatarsal II remained separate. The cuboid, scaphoid [navicular], and cuneiforms are separate, but are con- structed, like the other bones, on the same type as that of the Tragulidae.”! (Parts of the diagnosis merely descriptive of 1“Fs el mas pequefio de los artiodactilos cono- cidos, pues su tamanfo no excedia al de una pequefa rata. El hueso metatarsiano 6 canon formado por la fusién de los dos meta- tarsianos 3 y 4, solo tiene 27 mm. de largo y 2 mm. de ancho en su parte media mas angosta. .. . Los dos metatarsianos en cuestidn . . estan soldados en todo su largo, pero:se conserva entra ambos un profundo surco longitudinal en su cara anterior y otro mas superficial en la cara posterior. Los metatarsianos laterales 2 y 5 son atrofiados, representados tan sdlo por sus extremidades proxi- males estiliformes, siendo esta parte del meta- tarsiano 5 soldada con la del metatarsiano 4, pero la del metatarsiano 2 se conservaba independiente. El cuboides, el escafoides y los cuneiformes se conservan independientes, pero construidos, como también los demas huesos, sobra el mismo tipo del de los Tragulidae.” Hypisodus and now known to be irrelevant are omitted. ) Discussion. The name is not known to be preoccupied, was the first ever applied to a member of this family, and was given a definition technically sufficient under the Code. It is therefore necessarily valid as a name. However, the diagnosis, which is relative to tragulids and to Hypisodus (a hypertragulid), placental artiodactyls, is simply irrelevant now that Microtragulus is known to be a marsupial. Direct com- parison is possible only with the type of Argyrolagus scagliai. The metatarsals (not in fact fused) agree except that those of the latter are 25 per cent longer. The rea- sons for arbitrarily placing them in differ- ent genera, as well as species, have been given above. It is probable that M. argen- tinus was of approximately the size of M. reigi, with which direct comparison is impossible at present. I find the dimensions of the type (appressed metatarsals) of M. argentinus to be slightly larger than those given by Ameghino: 28.6 in length and 2.4 in (transverse) width at the narrowest point. The minimum anteroposterior diam- eter is 1.5. Microtragulus catamarcensis (L. Kraglievich) Argyrolagus catamarcensis L. Kraglievich, 1931, reprinted in L. Kraglievich, 1940, p. 592. (Not previously figured. ) Type. MACN No. 5529. Parts of both rami of the mandible with left I, and P3-M, (poorly preserved), right I,, and M,-s, and other alveoli and fragments.” Hypodigm. Type only. Known range. Araucanian of Catamarca (“Ios yacimientos araucanenses de Cata- marca’). A label with the specimen says, “Andalhuala Catamarca F. Araucana.” Andalhuala is evidently named as at or 2In the same vial there is a fragment of bone with a lower molar tooth apparently of Carolo- ameghinia mater, a rare genus and species known only from the Casamayoran of Patagonia. There cannot be any connection between the two speci- mens. near the locality where the specimen was found. It is also the type locality for a ~ subdivision of the Araucanian beds, and there is some probability that the speci- men came from that stratigraphic sub- division. It is shown, for example, on the correlation chart, pl. IV, of J. L. Kraglie- vich (1952). Original diagnosis. No formal proposal or diagnosis was given, but, in the course of a discussion of Argyrolagus palmeri, L. Kraglievich gave this name as new,’ with enough description to validate the name under the then existing code of nomen- clature; from that discussion I have ab- stracted such comments as might have been considered distinctive of the species in comparison with A. palmeri. “In the Araucanian deposits of Catamarca, somewhat older than that of Monte Her- moso, I have established the presence of another argyrolagid, which I shall call Argyrolagus catamarcensis n. sp., much smaller than A. palmeri. The animal is represented by a large part of the mandible (No. 5529, paleontological collection of the National Museum), with the body of both rami, the median incisors and several cheek teeth of one side or the other, of really tiny size but of a structure similar to the genotype species in every respect.”* There follows a description not said to be and not in fact distinctive from A. palmeri. Then: “Perhaps the anterior accessory groove of the second cheek tooth [M;] (the first preserved) is a little weaker than in A. palmeri. The anteroposterior diameter ° This offhand presentation, buried in a_ text paragraph, doubtless explains why “A.” cata- marcensis does not figure in the relevant biblio- graphy, Camp and Vanderhoof (1940), which does cite the publication in which the name ap- peared. “Tn fact the structure of these teeth is strikingly different from that of Argyrolagus palmeri, as is now shown. That an observer of L. Kraglievich’s high caliber thought the structure the same is due to the fact that the outlines of the molars were obscured by matrix, which has subsequently been removed without damage to the specimen. THe ARGYROLAGID MARSUPIALS * Simpson 3) of the median incisor scarcely exceeds 1 mm. The maximum height of the rami below the cheek teeth does not reach 5 mm, and the three intermediate cheek teeth [i.e., the second to fourth or Mj,-3] occupy a space of only 4 mm.” Revised diagnosis. Smaller than H. reigi; M,-4 31 per cent longer in type of the latter than in type of H. catamarcensis. Lingual groove absent on M;,-3. Measurements in Table 1. Discussion. The brief new diagnosis suffices to distinguish this species from others in which the lower dentition is known. Further details are given in the discussion of anatomy. It is improbable that this name is synonymous with H. argentinus. The type metatarsals of the latter are probably too large for H. cata- marcensis, and there is considerable differ- ence in geological age. Microtragulus reigi®, new species Type. MMMP No. 960-M, part of right mandibular ramus with all teeth. Collected by G. Scaglia at Punta San Andrés, San Andrés Formation. Hypodigm. The type and the following: MMMP No. 714-S, part of left mandibular ramus with M»s4; collected by O. Reig in 5“ ~ . . en los yacimientos araucanenses de Catamarca, algo mas antiguos que el de Monte Hermoso, he comprobado la presencia de otro argirolagido, que denominaré Argyrolagus cata- marcensis n. sp., mucho mas pequefio que A. palmeri. El animal esta representado por una gran parte de la mandibula (N° 5529, colec. paleont. Mus. Nac.), con el cuerpo de ambas ramas, los incisivos medios y varios molares de uno y otro lado, de un tamafo verdaderamente diminuto, pero de una conformaciédn en todo similar a la especie genotipo. . . . Tal vez el surco accesorio anterior del m 2 (primero de los molares conservados) es un poco mas débil que en A. palmeri. El] didmetro anteroposterior del incisivo medio apenas pasa de 1 milimetro; la altura maxima de las ramas debajo de los molares no llega a 5 milimetros y los tres molares intermedios ocupan tan solo un espacio de 4 milimetros.” 6 For Dr. Osvaldo Reig whose essential contri- butions to this study are acknowledged above. 10 Bulletin Museum of Comparative Zoology, Vol. 139, No. 1 a disgorged food pellet in the Atlantic coastal cliff 300 meters south of the Arroyo Loberia, Vorohué Formation, bed III. MMMP No. 691-S, nearly complete skull, lacking snout; collected by G. Scaglia at foot of cliff 120 meters south of the Bajada de las Palomas, Chapadmalal Formation, probably bed 3 or 4. MMMP No. 661-S, right maxilla with P?-M?‘; collected by G. Scaglia 500 meters south of Punta Vorohueé, one meter above sea level, Barranca de los Lobos Formation. MMMP No. 395-M, frag- ments of maxilla and mandible, with limb bones and fragments of several (probably three) individuals, perhaps not all of this species; collected by G. Scaglia in the cliff 590 meters northeast of Arroyo Brusquitas, Barranca de los Lobos Formation, bed I. Some of the skeletal remains, not associated with teeth, listed and described under “anatomy” probably belong to this species, but only specimens with teeth are explic- itly placed in the hypodigm. Known range. Early Pleistocene (Cha- padmalal to San Andrés formations) of the Chapadmalal-Miramar region, Buenos Aires Province. More precise localities and hori- zons of the known specimens given above. Diagnosis. Larger than M. catamarcensis. M,-s with shallow but definite lingual grooves. Measurements in Tables 1 and 2. Discussion. No metatarsals that could be referred to this species are known from the beds in which the teeth of the hypodigm were found. Comparison with the type of M. argentinus is therefore im- possible. The two nominal species are of about the same size, although a metatarsal somewhat smaller than the type of M. argentinus might be expected in M. reigi. The Monte Hermoso and Chapadmalal, sensu lato, faunas are largely different. Virtually no species are recorded as com- mon to both, and L. Kraglievich’s enumer- ation (1934) shows only 24.3 per cent of total then known and well-identified genera as present in both, although he indicated that 54.5 per cent of the well-identified genera of the smaller Monte Hermoso fauna are present in the (unrestricted ) Chapadmalal. The specific distinction may well be exaggerated by the tendency to define nominal species as distinct just be- cause the specimens in question are from different beds. Nevertheless, the Monte Hermoso fauna clearly is largely different from that of the Chapadmalal or any known later fauna. Reference of the Chapadmalal-San Andrés specimens to the Monte Hermoso species would go against some probability. A name is needed for these excellent and important specimens, and dubious reference to the earlier species would now be more misleading than refer- ence to a new species, even though the name of the latter could conceivably later prove to be a synonym. Genus Argyrolagus Ameghino Argyrolagus Ameghino, 1904, vol. 58, p. 255. Type. Argyrolagus palmeri Ameghino. Referred species. A. scagliai, new species, and doubtfully ?A. parodii Rusconi. Known range. Late Pliocene (Monte Hermoso) to early Pleistocene (Cha- padmalal), Argentina. Diagnosis. (Differential from the only other genus now recognized in the family, called Microtragulus by the arbitrary usage previously explained.) My,-4 strongly and definitely bilobed, with opposite labial and lingual vertical grooves of approximately equal strength; definitely longer than wide; second lobe relatively longer than in Microtragulus. M4, not so markedly unlike Ms. Discussion. The characters noted in the diagnosis sharply distinguish these species from those here designated as Micro- tragulus catamarcensis and M. reigi. Com- parison with the genotype of Microtragulus, M. argentinus, is possible for A. scagliai, but indicates only that the metatarsals of the latter species are longer and_ stouter than those of the former. As previously explained, it is possible that the type of M. argentinus does belong to Argyrolagus, in which case Argyrolagus is a synonym of THe ARGyROLAGID MARSUPIALS * Simpson Mt Microtragulus, and the species here called M. catamarcensis and M. reigi do not be- long to that genus. It is, however, at least equally plausible that a real generic dis- tinction simply is not evident in the meta- tarsals. There are many examples among mammalian genera closely related but generally accepted as distinct in which the generic distinction is not evident in meta- tarsals or other limb segments. The re- semblance does indicate close relationship and shows beyond serious doubt that Microtragulus and Argyrolagus, whether truly distinct genera or not, do belong in the same family. Other characteristics of Argyrolagus, as here restricted, are given in the section on anatomy. Argyrolagus palmeri’ Ameghino Argyrolagus palmeri Ameghino, 1904, vol. 58, p. 255; Ameghino, 1906, fig. 221; L. Kraglievich, 1931, fig. 2. Type. Ameghino Collection, presumably in MACN but not seen, part of a left mandibular ramus with I,, M,_4, and alveoli of I, and Joa Hypodigm. Type only. Known range. Monte Hermoso For- mation at Monte Hermoso, Buenos Aires Province. Original diagnosis (of genus and _spe- cies). “Medial incisor narrow, flat on the internal and convex on the external side, as in Prolagus; the root of this incisor reaches only as far as below the fifth cheek tooth. The second incisor smaller, elliptical, located posterior to the medial incisor and separated from the following cheek tooth *The name was given in honor of the North American mammalogist T. S. Palmer, author of the Index Generum Mammalium, a work now sometimes maligned but still extremely useful and irreplaceable; indeed even now, as Ameghino wrote in 1904, “The most complete and perfect compilation of its sort ever written.” It had just been issued when Ameghino wrote those words in a footnote to his description of this genus and species. by a short diastema. The five cheek teeth in continuous series, the first elliptical and the following four composed of two prisms, all very long and with open roots. Hori- zontal ramus with a very convex ventral border. Length from the anterior part of the medial incisor to the posterior edge of the last cheek tooth 14.5 mm. Length of the space occupied by the five cheek teeth 9 mm.” Revised diagnosis. About the size of A. scagliai or slightly smaller. Anterolabial projection of M, less pronounced. My-s narrower relative to length. Talonid of My without posterior projection. Measurements derived from illustrations in Ameghino (1906) and L. Kraglievich (1931) are given in Table 1. Discussion. Ameghino’s original diag- nosis or, rather, description was not dif- ferential, as there was then nothing to compare with. Even the supposedly re- lated lagomorphs are all so obviously dif- ferent that a diagnosis against them was unnecessary. Of course it has long since been recognized that this was because the groups are not, in fact, related. Although the specimens here grouped in Micro- tragulus are indeed related to A. palmeri, those with known lower dentitions (M. catamarcensis and M. reigi) are quite dis- tinct, as indicated here by their generic 8“TIncisivo interno angosto, plano sobre lado interior y convexo sobre el externo, igual al de Prolagus; la base de este incisivo sdlo Mega hasta debajo de la muela 5. Incisivo segundo mas pequeno, eliptico, colocado detras del incisivo interno y separado de la muela que sigue por una barra corta. Las cinco muelas en serie continua, la primera eliptica y las cuatro siguientes com- puestas de dos primas [sic!], todas muy largas y de base abierta. Rama horizontal de borde in- ferior muy convexo. Longitud de la parte anterior del incisivo interno al borde posterior de la ultima muela, 14.5mm. Longitud del espacio ocupado por las 5 muelas, 9mm.” Ameghino designated all permanent postcanine teeth in mammals, premolars and molars of other authors, as molars; I therefore translate his “muela” as “cheek tooth.” “Primas” is an obvious misprint for “prismas.” 2 Bulletin Museum of Comparative Zoology, Vol. 139, No. 1 separation. A. scagliai, new here, is much closer to A. palmeri, but the characters in- dicated in the diagnosis adequately indicate specific distinction. The two are of defi- nitely, although not greatly, different ages. I have not seen the actual specimen, but the descriptions by Ameghino and _ espe- cially by L. Kraglievich are detailed, and they agree well, as do their figures, three of which are given by Ameghino and two by L. Kraglievich. As shown in Table 1, ‘measurements made on these figures (all of which are X2), although differing by as much as 0.4 mm in extreme cases, are in sufficiently close agreement to be trusted as approximate, at least, when averaged. Argyrolagus scagliai,’? new species Type. MMMP No. 785-S, nearly com- plete skull, left ramus of mandible, pelvis and sacrum, right and left femora, right and left tibiae and fibulae, right and left metatarsals, partial right and left tarsi, part of scapula, partial right and left humeri, vertebrae, and various fragments; collected by G. Scaglia, 200 meters north of the Bajada de los Lobos, Chapadmalal For- mation, bed 9. Hypodigm. Type and the following: MMMP No. 741-M, part of right mandib- ular ramus with all teeth; from the Bajada las Palomas, Chapadmalal Formation, bed 9. MMMP No. 802-M, most of skull, lack- ing snout; from Punta Plataforma, Cha- padmalal Formation. MMMP No. 281-S, partial left side of skull; collected by G. Scaglia 100 meters south of the Bajada de la Barranca de los Lobos, Chapadmalal Formation, bed 9. MMMP No. 973-M, most of palatal and adjacent facial parts of skull with all teeth except right I’; collected by G. Scaglia at Vivero, Arroyo Loberia, Chapadmalal Formation, bed 8. MMMP No. 974-M, part of left ramus with P3-Ms; collected by G. Scaglia on the south side ®For Galileo Scaglia, who collected many of the specimens here described, who made them all available, and who supplied the data on lo- calities and levels. of Arroyo Brusquitas, Chapadmalal For- mation, bed 9. Some of the skeletal parts mentioned in the section on anatomy may also belong to this species, but they are not formally included in the hypodigm. Known range. Chapadmalal Formation, early Pleistocene, of the Chapadmalal- Miramar region, Buenos Aires Province. Details given above. (By what is probably coincidence, all the identified specimens of exactly known level are from beds 8 and 9 of the Chapadmalal, relatively high levels in that formation. ) Diagnosis. About the size of A. palmeri or slightly larger. Pronounced anterolabial projection on M;. My,-3 relatively wide (more than in A. palmeri, less than in Microtragulus ). M, talonid relatively com- plex, with posterior projection. Discussion. This, now much the best- known species of the family, is described in detail in the section on anatomy. ?Argyrolagus parodii Rusconi Argyrolagus Parodii Rusconi, 1933, p. 245, figs. 1 and 10; 1936, figs. 6b, 9, 10, and 12. Type. Part of a left mandibular ramus with M34. This was collected by Lorenzo W. Parodi, apparently on his own and not for a museum or other institution. He turned the specimen over to Carlos Rus- coni, who has informed me (letter of 24 October 1967) that the specimen “is in my possession (in my house), but unfortu- nately someone has broken it and it is in small bits. I do not know whether it can be reconstructed.”!° Evidently comparisons are now impossible, and Sr. Rusconi did not think it worth while to forward the remaining fragment or fragments for com- parison. io“ | |. se halla en mi poder (in my house) [parenthetical expression English in the original]. Pero, desgraciadamente alguna persona me la ha roto y se encuentra en pequefios trozos. Ignoro si podria ser reconstruida.” Rusconi then adds in English, “(This mandibular fragment in [is] broken or destroyed but [I] preserve some frag- ment. )” THE ARGYROLAGID MARSUPIALS * Simpson 13 TABLE 1. MEASUREMENTS OF LOWER TEETH OF ARGYROLAGIDAE M, M, M, M, LM, , LM,/LM, LM,/WM, Ey Wi Wi ee Ni Ee ew. Microtragulus M. catamarcensis, type 12 0.8 Me el 2 aleO eG 4.8 1.09 1.20 M. reigi, type Me 1:3 1.6 1.4 elie 16 Is) Frontoparietal scales separated by interparietals ae = ? p+++ ++++ + + - 4) pear ear | Paar [AP or ae 5. Lack of anterior expansion of the frontal, anterior to the orbital emargination. 6. Indications of a longitudinal crest along the temporal arch. Each of these points will be considered below: 1. The exoccipital and prootic region of the holotype is damaged; the smooth surface bone layer is nowhere preserved in this region. It is only by comparison and by knowing the region where the prootic occurs that one can recognize this bone. The element referred to by Camp as the paroccipital appears as a separate piece, but may very well be a flake from the prootic region. Similar conclusions were reached by Hecht (1959). This small ele- ment bears little resemblance to the so- called “epiphysis” of the paroccipital of Xenosaurus. Estes (personal communica- tion) has observed similar ossicles to those of Xenosaurus in some recent anguids (e.¢. Gerrhonotus multicarinatus, MCZ 32250). 2. The holotype of Melanosaurus maxi- mus shows a very large inner wing on the quadrate, as indicated by McDowell and Bogart, but the three Xenosaurus grandis skeletons in my possession show practically no development of this inner wing. How- ever, the anguid Ophisaurus apodus has a well-developed internal wing of the quad- rate, although it is smaller than that of Melanosaurus. 3. On the jugal of Melanosaurus maxi- mus the postorbital and maxillary processes meet at a right angle, while the postorbital process of Xenosaurus grandis is strongly tilted caudad. The striations of the jugal are confined to the posterior and ventral border; similar striations also occur in some glyptosaurs, to which the jugals of Melanosaurus bear the closest resemblance. There is little resemblance between the jugal of Melanosaurus and that of Xeno- saurus. 4. The coossification of cranial osteo- derms with the underlying bones is an anguid character as well as a xenosaurid one. However, frontals of Xenosaurus grandis are covered with large raised osteo- scutes that are coossified, and lines indi- cating the position of the epidermal scales NortH AMERICAN Foss. ANGUIDAE + Meszocly are absent. The skull roofing bones of Melanosaurus are covered by osteoscutes bearing a sculpture similar to that of Xestops or Peltosaurus granulosus, and the parietal osteoscutes clearly show the out- line of interparietal and occipital epidermal scales characteristic for nearly all anguids, excepting Glyptosaurus. 5. The outline of the frontals (especi- ally that of the paratype) bears an ex- tremely close resemblance to that of some glyptosaurs (e.g. G. rugosus) and shows little resemblance to the emarginated, nar- row frontals of Xenosaurus grandis. The frontal outline of the latter bears the closest superficial resemblance among Recent an- guids to that of Gerrhonotus liocephalus. 6. It is difficult to comment on this statement, since it may refer to the supra- temporal processes of the parietal or to the squamosal. It probably refers to the latter bone. The squamosal in Melanosaurus is very long and slender, whereas in Xeno- saurus it is short and thick, with a posterio- mesial weblike process that is in contact with the parietal, roofing over most of the upper temporal opening. The supratempo- ral processes of the parietal are also elongate, whereas in Xenosaurus these two are very short and stubby. Gilmore (1928, p. 138) suggested that Melanosaurus maximus “has its nearest affinities with the genus Peltosaurus.” There is a resemblance between Peltosau- rus granulosus and Melanosaurus in that they both have fused frontals with concave orbital borders. The outline of the frontals of the holotype also resembles that of P. granulosus, while that of the larger para- type is closer to Arpadosaurus (see below ) and Glyptosaurus. Sculpture type of the cranial osteoderms is essentially the same as in P. granulosus, but the same sculpture type also occurs in Xestops, Arpado- saurus, and Glyptosaurus. There is a simi- larity between the pattern of labial articu- lation between dentary and_postdentary bones of Melanosaurus and P. granulosus; 135 but the similarity is even greater between Melanosaurus and Glyptosaurus. The heavy, sutured osteoscutes have a deep groove between the gliding surface and the sculp- tured surface and are very similar to those of Glyptosaurus; they resemble less the thin, essentially beveled scutes of Pelto- saurus. granulosus, which also lack the groove between gliding and_ sculptured surfaces. Also, with respect to size, Melano- saurus (especially the paratype) is in the range of the genus Glyptosaurus. Melano- saurus is more primitive than P. granulosus in having toothed vomerine bones and in having separate postorbital and postfrontal bones. Thus, Melanosaurus shares a number of characters with Peltosaurus granulosus, but the majority are not unique to the two genera, being found also in other Eocene and Oligocene forms and thus may be regarded as specialized features character- izing a side line of anguid evolution. How- ever, M. maximus, with regard to body scutes as well as size, shows a specialization toward the Glyptosaurus line. This special- ization is not present in P. granulosus, whose geological occurrence is later ( Oligo- cene) than that of the Eocene M. maximus. The affinities of Melanosaurus and Glypto- saurus are further discussed in the section dealing with Arpadosaurus. Camp (1923) figured the reconstructed skull of Melanosaurus (“Xestops”) maxi- mus based on the holotype, and Gilmore (1928, p. 140) refigured Camp’s recon- struction. In this figure, the postorbital is shown to exceed the postfrontal an- teriorly, but examination of the holotype indicates a postorbital incision on the pos- terior border of the postfrontal bone, sug- gesting that the postorbital was excluded from the orbit, as in the glyptosaurs. It may be also pointed out that Gilmore’s photographs of the vertebrae (pl. 23) and the jugal (pl. 24) do not represent the holotype (as indicated in the plate legend ), but are skeletal elements of the paratype AMNH 517s. 136 Arpadosaurus gazinorum n. gen., n. sp. Holotype. USNM 25826. The type mate- rial consists of disarticulated skull elements, osteoscutes, and fragments of vertebrae. The following skull elements have been identified: frontals, parietal, left and right dentary fragments, tip of the right maxilla, fragments of left and right palatines, right articular, and occipital condyle. Horizon. Late early Eocene. Locality. Twelve miles north of Big Pi- ney, Wyoming; Wasatch Formation. Diagnosis. A large anguid structurally intermediate between Melanosaurus and Glyptosaurus in scalation. Arpadosaurus gazinorum differs from the former in hav- ing an unusual epidermal scalation, as indicated by grooves on the frontal bone, and a less extensive patch of palatine teeth; from the latter it differs in that the head osteoscutes are not broken up into poly- gonal plates. Etymology. Arpad—name of a 10th Century Hungarian leader who was lifted on shields into the air by his peers when they elected him as head of the seven tribes then inhabiting Hungary. The name refers to the shield-shaped fused frontal bones of the lizard. Greek, sauros—lizard. The specific name honors C. L. and Chester Gazin, who collected the specimen. Description of skeletal elements. The frontals (Fig. 14) bear a dorsal metopic suture on the osteoscutal crust but are indistinguishably fused ventrally. They are roughly triangular in outline and short in relation to their width. Their maximum length is 32.8 mm, and they are 28.6 mm along the frontoparietal suture. The dorsal surface of the bone is covered by fused osteoscutes, which have a sculpture of small tubercular mounds. The frontal osteoscutal crust is traversed by numerous grooves, indicating the outlines of epidermal scales. A prominent groove traverses the frontal transversely about one-third the distance of total length of this bone from the fronto- parietal suture. Posterior to the above Bulletin Museum of Comparative Zoology, Vol. 139, No. 2 Figure 14. Arpadosaurus gazinorum, n. gen., n. sp., USNM 25826, holotype. Early Eocene, Wyoming. Frontals and parietals, dorsal view, X 2. groove the dorsal surface of the frontal is further subdivided by several grooves. Two grooves, corresponding to the outline of frontoparietal scutes in other anguids, converge on one another toward the pos- terior midline, but are separated from one another by a small wedgelike area devoid of osteoscutes, presumably representing the interparietal epidermal scute area. The frontoparietal areas are again subdivided into unequal halves by a groove parallel to the metopic suture. On the lateral sur- face the prefrontal and postfrontal incisions are well defined. The prefrontal incision is the larger, occupying about two-thirds of the lateral surface of the frontal and separated from the postfrontal by only a 3mm gap. The anterior extremities of the NortH AMERICAN Fossr. ANGUIDAE + Meszoely 137 Taste VI. New Wortp Fossr. Record OF THE ANGUIDAE, Gerrhonotinae Anguinae Glyptosaurinae Diploglossinae 2 Diploglossus spp. 3 from Jamaica and S Ophisaurus ventralis the Dominican n . ° 7 O. compressus Republic (Etheridge, — aa 1964, 1965) o iB Ophisaurus é attenuatus a Paragerrhonotus Gerrhonotus Osteoscutes g noted by Estes 3 and Tihen, = 1964 = @ Peltosaurus abbottii 2 Pancelosaurus P. granulosus 5p pawneensis Glyptosaurus ro) (3 species ) Glyptosaurus © (8 species ) o Arpadosaurus gazinorum S Melanosaurus maximus cf. Gerrhonotus Xestops vagans 2 es Pancelosaurus jo} A o piger is] ay n 5 0 & = 2 cf. Gerrhonotus Pancelosaurus 4 o piger Oo olfactory processes are broken off. Posterior to these processes wedge-shaped impres- sions are present close to the posterior lateral extremity of the frontal. The parietal (Figs. 14, 16c) is a qua- drangular bone; most of its upper right quadrant is missing. The supratemporal processes are not preserved. Most of the dorsal surface of the parietal, except a small posterior area, is covered by coossified osteoscutes, which are covered with the same tubercular mounds as the frontals. Grooves indicating the outline of inter- parietal and occipital scutes are present. The interparietal area is narrow and tri- angular, its apex directed posteriad and en- closing the parietal foramen. The occipital scale impression is subequal to the inter- parietal, but is subrectangular rather than triangular in shape. Large paired parietal scutes flank the interparietal area, and two similar ones flank the occipital. The occipital area mesial to the two large parietal scutes is also flanked by numerous elongate wedge-shaped irregular scutes. On the ventral surface two anterior ridges converge posteriad to merge and continue as a single median ridge. The 138 Bulletin Museum of Comparative Zoology, Vol. 139, No. 2 F Figure 15. Arpadosaurus gazinorum, n. gen., n. sp., USNM 25826, holotype. A, left palatine, dorsal view; B, the same, ventral view; C, medial body osteoscute; D, lateral body osteoscute; E, right dentary fragment, lingual view; F, the same, dorsal view. All & 4. Early Eocene, Wyoming. NortH AMERICAN Fossin ANGUIDAE + Meszocly area surrounded by the anterior ridges is triangular and contains the parietal fossa close to its posterior apex. The parietal fossa is a vertical pit. The posterior ridges are less prominent than the anterior ones, and merge with the latter at a distance about one-half the total length of the parietal. The posterior portions of both dentaries are preserved. The right fragment (Fig. 15e-f) contains five teeth, the left only four. The teeth increase in size anteriorly; the most anterior tooth is the largest on both fragments. The teeth are robust crushing teeth with blunt, rounded crowns. In dorsal view the crowns are transversely widened, and have weak, longitudinal cut- ting edges, from which fine ridges extend at a right angle. The bases of the teeth are expanded and weakly striated. The first, second, and fourth teeth of the right den- tary fragment have a basal foramen, and the bases of the latter two are excavated. In the right fragment, the anterior labial coronoid articulation surface ends under the first tooth from the rear; both fragments indicate that on the lingual side the coro- noid reached the fourth tooth from the rear. A small portion of the lower jaw in the articular region includes the jaw articulation surface, which is roughly saddle-shaped and raised anteriorly and posteriorly. Close to the articular facet two foramina occur, one posteromesial and the other anterolateral. The former corresponds to the foramen for the chorda tympani, the latter is the anterior articular foramen. The greater portion of the right palatine is preserved, with a well-defined patch of tubercular teeth on its ventral surface (Fig. l5a-b). The teeth approach closely, but do not extend as far as, the choana. On the dorsal surface, posterior to the maxil- lary process, a large, well-defined infra- orbital foramen is present. The vomerine process is broken near its base. The occipital condyle of the occipital region is robust and semi-circular, kidney- 139 shaped in outline. The basi- and exoccipital] contributions to the condyle are indis- tinguishable, as a result of coossification of these bones. The body osteoscutes (Fig. 15c-d) are longer than wide and covered with the same tubercular mounds as those fused to the frontal and parietal bones. The sculp- tured surface is separated from the smooth anterior gliding surface by a deep groove. The gliding surface occupies one-third to one-fourth the entire length of the osteo- scute. Most of the osteoscutes are rec- tangular, some with an asymmetrical keel. Other osteoscutes are wedge-shaped, the apex directed posteriad; all of these bear keels. The lateral edges are beveled, and at the same time the irregular surfaces present in this region indicate suturing between adjacent osteoscutes. Discussion. Arpadosaurus gazinorum is known only by a single specimen from what Gazin (1962) refers to as the La Barge fauna. It is comparable in size to the paratype of Melanosaurus maximus and to Glyptosaurus rugosus. In the general shape of the frontals, and in having rec- tangular body osteoscutes that are sutured to one another laterally, A. gazinorum also resembles the above genera. However, in a number of characters A. gazinorum is intermediate between Melanosaurus maxi- mus and Glyptosaurus rugosus. On the ventral surface of the parietal (Fig. 16) the triangular area enclosed by the anterior ridges is largest in Melanosau- rus, smallest in G. rugosus, and interme- diate in A. gazinorum. These same ridges converge in G. rugosus to form a narrow but high mesial ridge. This same mesial ridge is broad and flat in Melanosaurus; the anterior ridges continue side by side instead of forming a strong single ridge, as in G. rugosus. The situation in A. gazino- rum is once again intermediate. Arpado- saurus resembles Glyptosaurus in that the parietal fossa is a vertical pit. Melanosaurus has a slanted parietal fossa. The general proportions of the parietal table of the 140 Bulletin Museum of Comparative Zoology, Vol. 139, No. 2 Parietals of fossil anguids in ventral view. A, Melanosaurus maximus, AMNH 5175, paratype, X 2; B, Glypto- Figure 16. saurus rugosus, AMNH 6055, * 2; C, Arpadosaurus gazinorum, n. gen., Nn. sp., holotype, 2; D, Pelfosaurus granu- losus, KU 620, * 5; E, Pancelosaurus piger, n. gen., MCZ 3498, & 6 (Hell Creek Formation, Montana). Nort AMERICAN Foss. ANGUIDAE + Meszoely 14] former two genera are also similar. This bone is essentially quadrangular in both of the above, while in Melanosaurus the same bone is more elongated. The scutellation on the dorsal surface of the parietal of Arpadosaurus resembles much more that of Melanosaurus than of Glyptosaurus. In the first two genera, interparietal and occipital impressions are discernable; in Glyptosaurus the dorsal sur- face is covered by subequal polygonal plates. The osteoscutal crust on the dorsal sur- face of the frontal bones of A. gazinorum is subdivided into unequal epidermal scute areas by numerous grooves; this surface is also irregular, featuring large depressed areas as well as elevated mounds. It is conceivable that further subdivisions of the osteoscutal crust could have led to the situation encountered in Glyptosaurus, in which numerous polygonal osteoscutal plates cover the frontal and other cranial bones. Arpadosaurus gazinorum displays a num- ber of features intermediate between Melanosaurus and Glyptosaurus, but since it occurs in strata of the same age as Melanosaurus and some glyptosaurs, A. gazinorum cannot at this time be regarded as anything more than a structural inter- mediate between the two genera. The origin of glyptosaurs perhaps lay in the Paleocene or very early Eocene. Glyptosaurus Marsh 1871 Helodermoides Douglass 1903 Type species of the genus. Glyptosaurus sylvestris Marsh 1871. Referred species. Glyptosaurus nodosus, G. rugosus, G. brevidens, G.? sphenodon, G. princeps, G. hillsi, G. obtusidens, and G. donohoei from the Eocene of North America; G. montanus, G. giganteus, and G. tuberculatus from the Oligocene of North America; Glyptosaurus near nodusus was recognized by Gilmore (1943) from the Eocene of Mongolia. Geological range. Early Eocene to Oligo- cene of North America, Eocene of Mon- golia, and questionably Paleocene to Eocene of Europe (see below). Synopsis of known characters of the genus The frontals, parietals, and the cheek region are covered by numerous polygonal osteoscutal plates. These cranial osteo- scutes, as well as those of the body, are covered with raised tubercular mounds, which are often arranged in concentric patterns. The frontals are distinct or fused; in the latter case the point of fusion is generally marked by a raised ventral ridge. The palatines and pterygoids bear teeth. The postfrontals and prefrontals are nar- rowly separated above the orbit. The parietal foramen is present, and the post- orbital is excluded from orbit formation. The body osteoscutes are rectangular, have a uniform width, and are much longer than wide. They are covered with tubercular mounds that are arranged in a concentric pattern. A deep groove is present between the smooth anterior gliding surface and the sculptured area. The gliding surface is an anterior transverse band and comprises about one-quarter of the total length of the osteoscutes. In every species some of the osteoscutes have feeble keels. Strongly jagged lateral edges indicate suturing be- tween adjacent osteoscutes. On the mandible the anterior extremities of the coronoid and surangular are in line on a vertical plane on the labial surface. The dentary reaches posteriad between the above two bones to the anterior supra- angular foramen (Fig. 17b). A survey of the glyptosaurs Numerous polygonal plates sculptured with tubercular mounds cover the cranial elements as well as the cheek region in all the species of Glyptosaurus. These individ- ual cranial plates may be rather flat, as in G. hillsi, G. princeps, and G. sylvestris, or strongly raised, forming a highly irregular 142 surface, as in G. donohoei, G. giganteus, G. nodosus, G. rugosus, and G. tuberculatus. These cranial osteoscutes tend to become larger in an anteroposterior direction (e.g. plates covering the parietal are larger than the ones on the frontal, and those covering the cheek region are even larger). The frontals are distinct in G. sylvestris, G. nodosus, G. montanus, and G. tuberculatus. These bones are fused but with a promi- nent suture line or a raised ridge marking the point of fusion, in G. rugosus, G. hillsi, G. princeps, and G. giganteus. In the other species the condition of the frontals is not determinable. The parietals are much shortened in G. giganteus and G. montanus relative to other species where this bone is known. In these same two species the frontals are also foreshortened, and_ post- frontals and prefrontals closely approach one another over the orbit. In G. giganteus the latter two bones are almost touching. In other species of Glyptosaurus prefrontals and postfrontals are moderately separated from one another. Teeth in this genus range from the robust bulbous crushing teeth of Glyptosaurus hillsi (similar to the ones described and figured for Arpadosaurus gazinorum) to the recurved teeth with pointed unstriated apices of G. tuberculatus (similar to those of Pancelosaurus pawneensis, Pl. 1). G.? sphenodon has very slender teeth, and is much smaller than any of the other glypto- saurs; it is probably not a Glyptosaurus. The teeth in general are moderately heavy, with obtuse crowns that bear striations and an anteroposteriorly directed cutting edge. The largest members of this genus are G. hillsi and G. giganteus. The former prob- ably had a slender and elongated head; in the latter the head was broad and relatively short. Both forms exceed by far the maxi- mum size of any other known anguid. G. nodosus is the smallest glyptosaur. All species in which the mandible is known display the characteristic labial suturing between dentarv and postdentary bones, in which the anterior extremities of the Bulletin Museum of Comparative Zoology, Vol. 139, No. 2 coronoid and surangular are on a vertical line (Fig. 17b), and the anterior inferior alveolar foramen is between the dentary and the splenial. Discussion. The glyptosaurs are large lizards, some comparable in size to Arpa- dosaurus and Melanosaurus, while some (e.g. G. giganteus and G. hillsi) are much larger. The glyptosaurs are unique among anguids in that the head osteoscutes are broken up into numerous polygonal plates, resembling superficially in this respect the living Heloderma. These small bony osteo- scutes are more loosely associated with the skull bones that they cover than is the case in many other anguids and consequently some are often missing. The polygonal plates, as well as the body osteoscutes, are covered with the same tubercular mounds as skull and body osteoscutes of Melano- saurus, Arpadosaurus, Peltosaurus granu- losus, and Xestops vagans. This type of sculpture is not found in other fossil or Recent members of the Anguidae. The labial contact between the dentary and postdentary bones on the mandible is also nearly identical, in the glyptosaurs, to the latter four fossil genera further indicating that the affinities of glyptosaurs are with these forms. Osteoscutes of glyptosaurs are very similar to those of Melanosaurus and Arpadosaurus in respect to sculpture. general shape, strong lateral suturing, and in having a deep groove between the glid- ing surface and the sculptured area. As indicated earlier (p. 141), Melanosaurus, Arpadosaurus, and Glyptosaurus, respec- tively, can be placed in a structural series that may represent steps in the evolution of glyptosaurs from a more primitive Melanosaurus-like ancestor. Several authors have suggested that the European Placosaurus is very close to Glyptosaurus, if not congeneric with it, and Romer (1967) lists Glyptosaurus as one of the synonyms for Placosaurus. Much of what is in the literature at present concern- ing this question is repetition of assump- tions made by early workers. The generic NortH AMERICAN Fossu. ANGUIDAE * Meszoely 143 Figure 17. A, left mandible of Peltosaurus granulosus, labial view, FAM 42913; Note coronoid in advance of surangular in A, the two bones essentially Glyptosaurus, cf. G. obtusidens, AMNH 5176. in line in B. A X 3.3, B X& 2.7. name Glyptosaurus is retained here for the North American forms until actual study and comparison of the North American and European fossil forms is made. THE SUBFAMILIES OF ANGUIDAE The preceding discussion of fossil an- guids gives a basis for evaluating the utility of anguid subfamilies based on modern forms. When the fossil record of the group is considered, it is difficult to list a series of characters that characterize the sub- families, as will be seen below and as has been intimated frequently in the above review. B, fragmentary right mandible of Anguinae. This group includes Pancelo- saurus, the most primitive of known fossil anguids, and its limbless modern relatives Anguis and Ophisaurus. Because of the adaptive differences between the fossil and Recent forms, this subfamily is especially difficult to characterize, but all included forms share unfused frontals with well- separated frontoparietal epidermal scales, and a similar type of scutellation. Anguis has somewhat modified its scutes from the Ophisaurus condition (Fig. 6); the latter more closely resembles Pancelosaurus, ger- rhonotines, and glyptosaurines in scute morphology. In the mandible, the sur- 144 angular and angular extend far forward of the coronoid in the Recent forms, slightly less so in Pancelosaurus. A premaxillary foramen is present in the Recent forms and probably, but not certainly, also occurred in Pancelosaurus. Vomers, palatines, and pterygoids are toothed in Ophisaurus, and at least the latter two bones were toothed in Pancelosaurus as well, palatal teeth are absent in Anguis. Pancelosaurus was a limbed form, and was probably ancestral to the glyptosau- rines; it also shows strong similarities to diploglossines and to gerrhonotines. It is close to the basal stock of known Anguidae. Gerrhonotinae. Only. the Recent genera Gerrhonotus, Abronia, and, perhaps, Colop- tychon are included. This specialized group has fused, hourglass-shaped frontals, the frontoparietal scales almost or barely in contact on the midline, and lacks the premaxillary foramen of anguines and diploglossines. The forward extension of surangular and angular, and the type of scutellation, suggest that the group is de- rived from primitive limbed Anguinae. Glyptosaurinae. This group includes only the extinct genera Xestops, Peltosaurus, Melanosaurus, Arpadosaurus, and Glypto- saurus (and their European relatives, for which synonymy is not yet clear). This subfamily possesses tuberculated osteo- scutes, and the latter three genera show an increasing tendency to fragment the cephalic scutellation. The palate is toothed on palatines, and pterygoids in forms in which these bones are known. The premaxillary foramen is absent in Pelto- saurus, but the condition in the other genera is not known; this character thus cannot be used to link Peltosaurus with the gerrhonotines. On the mandible, labial extensions of coronoid and surangular an- teriorly are in a vertical plane except in Peltosaurus, which is anomalous in having the coronoid exceed the surangular. This group is probably derived from Xestops, which is differentiated from the anguine Pancelosaurus mainly by its tuber- Bulletin Museum of Comparative Zoology, Vol. 139, No. 2 culated osteoscutes. Xestops gave rise, probably, to the Oligocene Peltosaurus, in which the frontoparietal scales meet broadly on the midline. Glyptosaurus, the most extreme of the group in fragmentation of cephalic scutellation, was probably de- rived from Xestops through Melanosaurus- like and Arpadosaurus-like ancestors, and it lived through Oligocene time at least. Diploglossinae. Only the Recent genera Diploglossus (including Celestus and Sau- resia), Ophiodes, and Wetmorena are in- cluded here. This primitive group is distinctive, and has no fossil record outside of its present distribution in the West Indies. Cycloid scales with a peaked glid- ing surface distinguish them from all other anguids, although they are linked to the Anguinae by their separate frontals with well-separated frontoparietal scale impres- sions. As might be expected, the fossil record thus shows intergradation between the rather well-defined Recent subfamilial groups. The resemblances of the fossil Pancelosaurus are primarily to the An- guinae; nevertheless, strong resemblances to the other subfamilies also occur. Regardless of the intergradation shown above, the subfamily categories based on Recent forms may be maintained because they appear to represent actual lineages long established in the fossil record. The addition of the Glyptosaurinae is necessary, in addition, to encompass the Eocene- Oligocene radiation of forms with tuber- culated osteoscutes. SUMMARY AND CONCLUSIONS The foregoing survey of the osteology and epidermal scalation of Recent Anguidae indicates that all extant forms fall into three groups worthy of subfamily status: (1) Anguinae (including Ophisaurus and Anguis), (2) Gerrhonotinae (including Abronia, Gerrhonotus, and possibly Col- optychon), (3) Diploglossinae (including Diploglossus, Wetmorena, and Ophiodes). All these extant forms, excepting the ger- NortuH AMERICAN Fosst ANGUIDAE ° rhonotines (Abronia and Gerrhonotus), have paired frontal bones and a premaxil- lary foramen (see Fig. 2). This foramen is restricted, among Recent lizards, to the genera Ophisaurus, Anguis, Diploglossus, Wetmorena, and Ophiodes. It may possibly have been present in the fossil genus Pancelosaurus, but its presence in other fossil forms cannot be determined except in Peltosaurus, in which it is absent. These characters clearly set the Gerrhonotinae apart from the other Recent anguids, but gerrhonotines do share a_ similar body scutellation and a lateral fold with the ophisaurs. In this respect the Diploglos- sinae, with their unique (in anguids) cycloid body osteoscutes, stand apart from the rest of the Recent forms. The anguines, sharing some characters with both, appear to be structurally inter- mediate between the Gerrhonotinae and the Diploglossinae. They are primitive (except for Anguis) in regard to their toothed pterygoids and palatines (and vomers in some), although Recent forms are limbless. Anguis shares a great number of osteo- logical as well as scalation characters with Ophisaurus (see page 101), differing from this genus only in characters judged here to be degenerate: (1) reduced interclavi- cle; (2) complete absence of palatal teeth; (3) thin, feebly-developed osteoscutes with the anterior gliding surface reduced so as to become confluent with the lateral bevel, (4) absence of a lateral fold, a feature apparently correlated with the reduction of the body armor; (5) absence of an external ear opening (a condition also found in some Ophisaurus). Anguis ap- pears to be a degenerate ophisaur derived from a limbless Ophisaurus ancestor. In the North American fossil record, fossil remains unquestionably those of lizards of the family Anguidae first occur in late Cretaceous sediments of Wyoming and Montana. This anguid, formerly known as Peltosaurus piger, was first described by Gilmore (1928) on the basis of two jaw Meszoely 145 elements. The generic assignment was based on a tooth structure similar to that of the Oligocene Peltosaurus granulosus. This early anguid is here placed in a new genus, Pancelosaurus, based on study of a large number of previously unknown cranial elements recovered through washing and screening methods. The cranial elements of Pancelosaurus piger, in their close resemblance to those of some Recent genera (Ophisaurus, Ger- rhonotus, Diploglossus), point to the fact that most of the Eocene and Oligocene fossil anguids represent a side line of anguid evolution rather than being an- cestors of the Recent forms (see below). Pancelosaurus piger exhibits a mosaic of characters in its skeleton, showing some resemblance to each of the above Recent genera, especially Ophisaurus. Since its body osteoscutes most closely resemble those of Gerrhonotus and Ophisaurus, per- haps this Cretaceous form was already specialized toward the line leading to these two Recent genera. It exhibits many of the same primitive and intermediate characters between the diploglossines and the gerrhonotines as do the ophisaurs and is regarded here as a primitive limbed member of the Anguinae. Pancelosaurus piger is known from late Cretaceous through late Paleocene deposits. The genus extends into the middle Oligo- cene as P. pawneensis (formerly Xestops pawneensis ). Pancelosaurus is the only fossil anguid in late Cretaceous and Paleo- cene deposits of North America that is known to such an extent as to allow for generic diagnosis. Peltosaurus jepseni, de- scribed by Gilmore, is a synonym of P. piger. The large Eocene and Oligocene fossil forms such as Xestops, Peltosaurus, Mel- anosaurus, Arpadosaurus, and Glyptosaurus may also have been derived from Pancelo- saurus or its relatives, as is indicated by similar body osteoscutes. These are more robust than those of P. piger and contrast with those of the latter in being covered 146 with tubercular mounds. This sculpture type is found only in the above Eocene and Oligocene forms, and combined with a characteristic pattern of labial suturing between dentary and postdentary bones, indicates their distinctness from other anguids. They are placed here in the extinct subfamily Glyptosaurinae. In this concept of the glyptosaurines, the Eocene species Xestops vagans appears to be the most primitive form, sharing some primitive characters of the frontal bones with P. piger, while exhibiting the above-men- tioned glyptosaurine characters. Dimeto- pisaurus twyomingensis is a synonym of Xestops vagans. All the known North American genera referable to the Glyptosaurinae occur in deposits only of Eocene (Arpadosaurus, Glyptosaurus, Melanosaurus, and Xes- tops) or Oligocene (Glyptosaurus and Peltosaurus) age. However, three large, keeled body osteoscutes bearing the diag- nostic tubercular sculpture of the Glyp- tosaurinae have been reported by Estes and Tihen (1964) from the Mio-Pliocene Valentine Formation of Nebraska. Unless they are reworked, these scutes indicate that some unknown glyptosaurine survived until the late Miocene or early Pliocene. In the Glyptosaurinae, the new genus Arpadosaurus appears to be structurally intermediate between Melanosaurus and Glyptosaurus, and it is postulated that the glyptosaurs may have originated from a form like Melanosaurus; more evidence is needed to confirm this hypothesis, how- ever. It is postulated above that Pancelosaurus piger or a closely related form may have given rise to both the extant ophisaurs and gerrhonotines or only to the former. There is indeed nothing in the known anatomy of P. piger that would contradict such an assumption, but in the absence of a more complete fossil record, the origin of these Recent forms is still hypothetical. The only fossil form referred to the Gerrhonotinae and consisting of more than Bulletin Museum of Comparative Zoology, Vol. 139, No. jaw elements is Paragerrhonotus ricardensis (Estes, 1964) from the early Pliocene of California. This fossil form poses more problems than it solves. Although frontal bones are fused and display strongly emarginate concave borders as in the Re- cent Gerrhonotus, the osteoscutal crust on the frontal is broken up into several facets instead of exhibiting the usual epidermal — imprints of Recent aneuids! This is a trend | often seen in anguimorphs (cf. Parasaniwa, Glyptosaurus, Exostinus, Heloderma). It is probably a specialized sideline, derived from Gerrhonotus. Estes (1964) regarded some anguid dentaries from the Lance Formation as near Gerrhonotus. These jaw fragments with Gerrhonotus-like teeth occur in the same deposits as Pancelosaurus piger. No other cranial elements have been recovered. Without the presence of the specialized frontals of Gerrhonotus (or a closely allied form) in the Cretaceous, its presence can- not be confirmed, and it is also possible that these jaw fragments may represent a smaller species of Pancelosaurus. Gerrhonotus-like jaws were also noted by McKenna (1960) from the Eocene (Wasatchian) Four Mile local fauna of Colorado, and the Mio-Pliocene Valentine Formation of Nebraska (Estes and Tihen, 1964). These identifications, like those from the Lance Formation, are not sup- ported so far by any other cranial elements. Remains of ophisaurs referable to or close to Recent species have been described both from the Pliocene of Europe and from the Pleistocene and Pliocene of North America, and Mtynarski (1960, 1964) de- scribed the remains of Ophisaurus pannoni- cus from the early Pliocene of Poland. He regarded this form as very close to if not conspecific with Ophisaurus apodus. From these same deposits Mtynarski (1964) de- scribed the remains of Anguis cf. fragilis. Etheridge (1960) recognized the extant O. attenuatus as early as the late Pliocene of Kansas. Auffenberg (1955) recognized both O. ventralis and O. compressus from Nort AMERICAN Foss, ANGUIDAE Meszoely the Pleistocene of Florida. Thus it appears that establishment of the Recent species of Ophisaurus had begun at least by the beginning of the Pliocene, and that Anguis was distinct from the ophisaurs by that time. The middle Eocene European Geiseltal deposits contain numerous anguid fossils; some were said by Kuhn (1940) to show resemblance to the ophisaurs. I have not seen this material, and Kuhn’s figures are poor, but study of this material will shed some more light on the phylogeny of the Anguinae as here constituted, as well as on that of the Anguidae in general. The diploglossines have no fossil record beyond Pleistocene cave finds (Etheridge, 1964, 1965). The preceding account illustrates the in- complete state of the fossil record of the Anguidae, especially that of the Recent forms, and also the need for new discoy- ries, as well as reworking of existing fossil material. Some of the main conclusions reached by this investigation may be summarized as follows: (1) The Recent anguids and the North American fossil forms fall into four groups worthy of subfamily status: (1) Anguinae, (2) Diploglossinae, (3) Gerrhonotinae, (4) Glyptosaurinae. The Diploglossinae, in many ways primitive, has essentially no fossil record; the Glyptosaurinae is extinct. (2) The Glyptosaurinae (s.1.) contains most of the well-known fossil anguids (Glyptosaurus, Melanosaurus, Peltosaurus, and Xestops) and a new genus, Arpado- SaUTUS. (3) The glyptosaurines appear in the early Eocene of North America and dis- appear from the fossil record by or just before the beginning of the Pliocene. (4) Peltosaurus, long considered a close relative of Gerrhonotus, is shown to have a greater resemblance to the glyptosaurines (especially Xestops). (5) The earliest known anguid is Pan- 147 celosaurus piger from the late Cretaceous of Wyoming and Montana. This is the only anguid fossil in Cretaceous and Paleo- cene deposits of North America that is known to such an extent as to allow generic diagnosis. A related species, P. pawneensis, occurs in the Oligocene of Wyoming. (6) Pancelosaurus piger displays a body scutellation similar to that of Gerrhonotus and Ophisaurus, and this early fossil ap- pears to be already a member of the An- guinae, but the imperfect fossil record prevents more positive statements concern- ing the phylogeny of the above Recent forms. Abbreviations a = angular aaf = anterior inferior alveolar foramen ain = anterior internasal epidermal scale amf = anterior mylohyoid foramen asf = anterior supra-angular foramen c = coronoid d = dentary e = ectopterygoid fn = frontonasal epidermal scale fp = frontoparietal epidermal scale fr = frontal epidermal scale ip = interparietal epidermal scale m = maxilla oc = occipital epidermal scale pa = parietal epidermal scale part = fused articular and prearticular pf = premaxillary foramen pfr = prefrontal epidermal scale pin = posterior internasal scale pl = palatine pm = premaxilla pt = pterygoid r = rostral epidermal scale sa = surangular so = supra-orbital epidermal scale sp = splenial Vv = vomer LITERATURE CITED AUFFENBERG, W. 1955. Glass lizards (Ophisau- rus) in the Pleistocene and Pliocene of Florida. Herpetologica 2 (2): 133-136. Barrows, S., AnD H. M. Smiru. 1947. The skele- ton of the lizard Xenosaurus grandis (Gray). Univ. Kansas Sci. Bull. 31: 227-281. Bocert, C. M., anp A. P. Parker. 1967. A new species of Abronia (Sauria, Anguidae) from 148 the Sierra Madre del Sur of Oaxaca, Mexico. Amer. Mus. Novit. No. 2279: 1-21. BouLENcER, G. A. 1885. Catalogue of the Lizards in the British Museum (Natural History). Vol. II, 2nd ed., London, xii + 497 pp. 1918. Les lézards hélodermatides de l'Eocéne supérieur de la France. C. R. Acad. Sci. Paris. 166: 1-4. Burt, C. E., anp M. C. Burr. 1991. South Nearer lizards in the collection of the American Museum of Natural History. Bull. Amer. Mus. Nat. Hist. 61: 227-395. Camp, C. L. 1923. A classification of the lizards. Bull. Amer. Mus. Nat. Hist. 48: 289-481. Corr, E. D. 1864. On the characters of the higher groups of Reptilia, Squamata and especially of the Diploglossa. Proc. Acad. Nat. Sci. Philadelphia 1864: 224-231. 1873. Second notice of extinct Vertebrata from the Tertiary of the Plains. Paleont. Bull. 15: 5-6. 1884. The Vertebrata of the Tertiary formations of the West. Rep. U.S. Geol. Surv. Terr. (Hayden) 3: 770-781. 1900. The crocodilians, lizards and snakes of North America. Rep. U.S. Nat. Mus. 1898: 153-1294. Critey, B. B. 1968. The cranial osteology of gerrhonotiform lizards. Amer. Midl. Nat. 80: 199-219, DariNncTon, P. J., JR. 1957. Zoogeography: the Geographical Distribution of Animals. New York, Wiley and Sons, xi + 675 pp. Douc.ass, E. 1903. New vertebrates from the Montana Tertiary. Ann. Carnegie Mus. 2: 145-199. 1908. Some Oligocene lizards. negie Mus. 4: 278-281. Estes, R. 1963a. Early Miocene salamanders and lizards from Florida. Quart. J. Florida Acad. Sci. 26: 234-256. 1963b. A new gerrhonotine lizard from the Pliocene of California. Copeia 1963 (4): 676-680. 1964. Fossil vertebrates from the late Cretaceous Lance Formation, Eastern Wyo- ming. Univ. Calif. Publ. Geol. Sci. 49: 1-180. Estes, R., P. Berberian, and C. A. M. Mezoely. Ann. Car- 1969. Lower vertebrates from the late Cre- taceous. Hell Creek Formation, _McCone County, Montana. Breviora, Mus. Comp. Zool., 337: 1-33. Estes, R., AND J. A. TrHeEN. 1964. Lower verte- brates from the Valentine Formation of Ne- braska. Amer. Midl. Nat. 75: 495-515. Ernermcr, R. 1960. The slender glass lizard, Ophisaurus attenuatus, from the Pleistocene (Ilinoian Glacial) of Oklahoma. Copeia 1960 (1): 46-47. Bulletin Museum of Comparative Zoology, Vol. 139, No. 2 1961. Late Cenozoic glass lizards (Ophi- | saurus) from, the southern Great Plains. | Herpetologica 17: 179-186. 1964. Late Pleistocene lizards from Bar- | buda, British West Indies. Bull. Florida State Mus. 9: 43-75. | 1965. Fossil earls from the Dominican | Republic. Quart. J. Florida Acad. Sci. 28: 83-105. ! Gazin, C. L. 1956. Paleocene mammalian faunas | of the Bison Basin in south-central Wyoming. | Smithsonian Misc. Coll. 131 (6): iv +957 | pp. i 1962. A further study of the Lower | Eocene mammalian faunas of southwestern | Wyoming. Smithsonian Misc. Coll. 144: 1-98. | Gervais, P. 1859. Zoologie et Paléontologie 2nd Ed. Paris, 543 pp. 1928. Fossil lizards of North Nation. Acad. Sci. 22: Francaises. GILMorE, C. W. America. Mem. ix + 201 pp. 1938. Description of new and _little- | known fossil lizards from North America. } Proc. U.S. Nat. Mus. 86: 11-26. y 1942. Paleocene faunas of the Polecat | Bench Formation, Park County, Wyoming. Part Il: Lizards. Proc. Amer. Philos. Soc. 85: 159-167. 1943. Fossil lizards of Mongolia. Amer. Mus. Nat. Hist. 81: 361-384. Hecut, M. K. 1959. Amphibians and reptiles. In: McGrew, P. O., The geology and paleon- | tology of the Elk Mountain and Tabernacle | Butte area, Wyoming. Bull. Amer. Mus. Nat. | Hist. 117: 130-146. HorrstettTer, R. 1962a. Observations sur les | osteodermes et la classification des anguides | actuels et fossiles (Reptiles, Sauriens). Bull. | Mus. Nat. Hist. Nat. 34: 149-157. 1962b. Additions a la faune reptilienne de | VEocene supérieur de Mormont-Saint- Loup | (Suisse). Bull. Soc. Géol. France (7) te 109-117. | Horan, J. A. 1958. The Pleistocene herpeto- fauna of Saber-tooth Cave, Citrus County, | Florida. Copeia 1958 (4): 276-280. 1965. A new glass lizard from Veracruz, Mexico. Quart. J. Florida Acad. Sci. 27 (oie 311-315. Kunn, O. 1940. Die Placosauriden und Anguidem aus dem Mittleren Eozin des Geiseltales. | Bull. Nova Acta Acad. Leop.-Carol. 53 (8): 461-486. Marsu, O. C. 1871. Notice of some new fossil reptiles from Cretaceous and Tertiary for- mations. Amer. J. Sci. 3 (1): 447-459. — 1872. Preliminary description of new Tertiary reptiles, part 1. Amer. J. Sci. 4: 298-309. NortH AMERICAN Fossi ANGUIDAE ° McConkey, E. H. 1954. A systematic study of the North American lizards of the genus Ophisaurus. Amer. Mid]. Nat. 51 (1): 133-174. 1955. A new lizard of the genus Ophisau- rus from Mexico. Nat. Hist. Misc., Chicago Acad. Sci. 145: 1-2. McDoweEL., S. B., anp C. M. Bocert.~ 1954. The systematic position of Lanthanotus and the affinities of the anguinomorphan lizards. Bull. Am. Mus. Nat. Hist. 105 (1): 1-145. McKenna, M. C. 1960. Fossil Mammalia from the early Wasatchian Four Mile fauna, Eocene of northwest Colorado. Univ. Calif. Publ. Geol. Sci. 37: 1-130. 1962. Collecting small fossils by washing and screening. Curator 5: 221-235. Mrynarski, M. 1960. Pliocene amphibians and reptiles from Rebielice Krolewskie (Poland). Acta Zool. Cracov. 5 (4): 131-150. 1964. Die jungpliozane Reptilien-fauna von Rebielice Krolewskie, Polen. Senck. Biol. Frankfurt a.M. 45: 325-347. Romer, A. S. 1956. Osteology of the Reptiles. Chicago, Univ. Chicago Press, xxi + 772 pp. 1967. Vertebrate Paleontology. 3rd Ed., Chicago, Univ. Chicago Press, viii + 468 pp. SLOAN, R., AND L. VAN VALEN. 1965. Cretaceous Meszoely 149 mammals from Montana. Science 148: 220- DON. STEBBINS, R. C. 1958. A new alligator lizard from the Panamint Mountains, Inyo County, California. Amer. Mus. Novit. No. 1883: ile TinEN, J. A. 1949. The genera of gerrhonotine lizards. Amer. Mid]. Nat. 41 (3): 580-601. 1949. A review of the lizard genus Barisia. Univ. Kansas Sci. Bull. 33 (3): 217-256. 1954. Gerrhonotine lizards recently added to the American Museum collection with further revisions of the genus Abronia. Amer. Mus. Novit. No. 1687 (4): 1-26. Unverwoop, G. 1959. A new Jamaican galliwasp (Sauria, Anguidae). Breviora, Mus. Comp. Zool., No. 102: 1-13. 1964. An anguid lizard from the Leeward Islands. Breviora, Mus. Comp. Zool., No. 200: 1-10. VANZOLINI, P. E. 1952. Fossil snakes and lizards from the Lower Miocene of Florida. J. Paleon. 26: 452-457. Wuite, T. E. 1952. Preliminary analysis of the vertebrate fossil fauna of the Boysen Reser- voir area. Proc. U.S. Nat. Mus. 102: 185- 207. (Received 4 April, 1968.) Pe | 4 A Subfamilial Classification of Scincid } ! Lizards ALLEN E. GREER HARVARD UNIVERSITY VOLUME 139, NUMBER 3 CAMBRIDGE, MASSACHUSETTS, U.S.A. MARCH 27, 1970 PUBLICATIONS ISSUED OR DISTRIBUTED BY THE — MUSEUM OF COMPARATIVE ZOOLOGY HARVARD UNIVERSITY BuLLetin 1863- BREVIORA 1952— Memorrs 1864-1938 JounsoniA, Department of Mollusks, 1941- OccASsIONAL PAPERS ON Mo.uusks, 1945— Other Publications. Bigelow, H. B., and W. C. Schroeder, 1953. Fishes of the Gulf of Maine. Reprint, $6.50 cloth. Brues, C. T., A. L. Melander, and F. M. Carpenter, 1954. Classification of In- sects. $9.00 cloth. Creighton, W. S., 1950. The Ants of North America. Reprint, $10.00 cloth. Lyman, C. P., and A. R. Dawe (eds.), 1960. Symposium on Natural Mam- malian Hibernation. $3.00 paper, $4.50 cloth. Peters’ Check-list of Birds of the World, vols. 2-7, 9, 10, 12, 15. (Price list on request. ) Turner, R. D., 1966. A Survey and Illustrated Catalogue of the Teredinidae (Mollusca: Bivalvia). $8.00 cloth. Whittington, H. B., and W. D. I. Rolfe (eds.), 1963. Phylogeny and Evolution of Crustacea. $6.75 cloth. Proceedings of the New England Zoological Club 1899-1948. (Complete sets only. ) Publications of the Boston Society of Natural History. Publications Office Museum of Comparative Zoology Harvard University Cambridge, Massachusetts 02138, U. S. A. © The President and Fellows of Harvard College 1970. A SUBFAMILIAL CLASSIFICATION ALLEN E. GREER ABSTRACT The subfamilial classification of skinks which is proposed is based primarily on the osteology of the skull, particularly on the relationships of the bones of the second- ary palate and the frontal bones, and on external morphology. These, plus other characters important in understanding the evolution and classification of the higher taxa of skinks, are discussed in some detail. Four subfamilies are recognized. The Scincinae (approximately 28+ genera and 182 species) are considered to be the most primitive subfamily of skinks and are ap- parently independently ancestral to the other three subfamilies. The scincines occur in Asia and Africa, and in the New World north of Costa Rica (Eumeces and Neoseps), but are conspicuously absent from the Australian Region. The center of abundance and diversity of scincines today is in subsaharan Africa and the islands of the western Indian Ocean. The Feylininae (2 genera and 4 species) and Acontinae (3 genera and 15 species) are specialized burrowing taxa which almost surely evolved independently from the scincines of sub- saharan Africa where both subfamilies are confined today. The fourth subfamily, the Lygosominae (appoximately 40+ genera and 600+ species), is the most numerous and diverse subfamily of skinks. They appear to have arisen from a scincine an- cestry and have radiated spectacularly in the Australian Region and southeast Asia. From this area, they have spread west into Bull. Mus. Comp. Zool., 139(3): 151-184, March, 1970 OF SCINCID LIZARDS Africa and across the Atlantic into the New World (Mabuya), and north and east across a Bering Straits land bridge into North and Central America ( Leiolopisma). It is suggested that the radiation and ex- pansion of the lygosomines is responsible in part for the apparent decline of the scincines in certain areas such as Asia. INTRODUCTION The only attempt at a suprageneric classification of skinks was provided by Mittleman (1952) as a kind of preface to his synopsis of the genera that are related to or often grouped under (as subgenera ) the catch-all genus Lygosoma. The four subfamilies recognized were diagnosed by means of a key, and the general distribution of each subfamily was given. Only the genera of the subfamily Lygosominae, however, received further attention. Mittleman’s (1952) diagnostic key to the four subfamilies of skinks is as follows: A. Palatine bones in contact on median line of palate. 1. Pterygoid bones separated on the median line of palate; palatal notch extending anteriorly to level of centers of eyes 2. Pterygoid bones in contact ante- riorly; palatal notch not extending anteriorly to level of centers of eyes LYGOSOMINAE. B. Palatine bones separated on median line of palate. 151 152 1. Nostril pierced in nasal, or be- tween two adjacent plates, but never touching rostral SCINCINAE. Nostril pierced between rostral and adjacent plate, thereby con- tacting rostral, or else within ros- traliitsel fea CHALCIDINAE. bo In the course of my research on the supraspecific relationships of skinks, I have attempted to correlate skull osteology with external morphology in delimiting taxa. This study, which is based on data from the complete skulls of over 350 species of skinks, has suggested to me a subfamilial classification that has a sounder basis than that of Mittleman. Mittleman’s (1952) diagnoses of the three subfamilies Mabuyinae, Scincinae, and Chalcidinae are accurate descriptions of three possible assemblages of skinks, but none of these assemblages can be defended as a monophyletic unit. This should be- come evident in the discussion of the new classification. Mittleman’s diagnosis of the Lygosominae, on the other hand, con- stitutes an inaccurate description of many of the genera which he included in the group. but his generic list for the subfamily includes most of the genera that I believe should constitute a subfamily Lygosominae. The Lygosominae of Mittleman is, in other words, an inaccurately diagnosed but well conceived taxonomic group. In his characterization of the Lygosomi- nae, Mittleman (1952) fell into the same trap as did Boulenger (1887) and M. A. Smith (1935) in their skink classifications. All three authors attempted to interpret the important relationships of the bones of the palate without removing the over- lying buccal mucosa. In several lygosomine genera the pterygoids (ie., their palatal Bulletin Museum of Comparative Zoology, Vol. 139, No. 3 rami) are completely separated along the midline by the interpterygoid vacuity or by processes that project posteriorly from the palatines. Thus not only does the type species of the type genus of the sub- family Lygosominae, Lygosoma quadrupes, disagree with Mittleman’s diagnosis of the subfamily, but the following genera do as well: Ablepharus, Cryptoblepharus, Emoia, Eumecia, Eugongylus, Leiolopisma (part), Leptosiaphos, Riopa (part), Sia- phos (part), and Sphenomorphus (part). The four subfamilies in the classification proposed below are each based on the correlation of several skull and external characters, rather than on a single skull or external character, as were Mittleman’s (1952) subfamilies, and the resulting distribution of the subfamilies is more meaningful zoogeographically than was Mittleman’s arrangement. CHARACTERS UTILIZED Before discussing the four subfamilies of skinks, it will be worthwhile to review briefly some of the characters that have been most useful in diagnosing the higher taxa of skinks. This discussion will be limited to the taxonomic use of these char- acters, as the phylogenetic significance of these and other characters will be con- sidered in a later section of the paper. Secondary palate. Apart from Dibamus and Anelytropsis, which appear to be re- lated to one another but whose relation- ships with other lizards are obscure ( Miller, 1966b), skinks are the only family of lizards with a bony secondary palate. The secondary palate may be complete or in- complete depending on the degree of apposition (meeting along the midline or not, respectively ) of the horizontal lamellae of the palatine bones. As a further advance- => Feylininae: Feylinia polylepis Acontinae: Acontias breviceps (MCZ 38559); Lygosominae: Figure 1. Dorsal view of the skulls of representatives of the four subfamilies of skinks. (MCZ 61215); Scincinae: Proscelotes arnoldi (MCZ 55145); Sphenomorphus jobiensis (BM 1935.5.10.108). Drawn to scale. FEY LININAE , ACONTINAE SUBFAMILIES OF SKINKS © Greer LYGOSOMINAE 154 ment on the complete secondary palate, the palatal rami of the pterygoids may also meet along the midline with the palatines to make an even more extensive secondary palate. The secondary palate is a diagnostic feature of skinks as a family, and the com- plex relationships, as well as shapes, of the bones forming the palate (and those bordering it) are useful in recognizing sub- families and taxa of lower rank (Greer, 1967a and b; Greer and Parker, 1968). Osteoderms. The second partially diag- nostic feature of skinks as a family is the characteristic arrangement of the tubules in the osteoderms, i.e., an approximately transverse canal with anteriorly and poste- riorly projecting longitudinal canals (see Gosse, 1848; Duméril and Bocourt, 1881; Otto, 1908; Hewitt, 1929; Smith, 1935; Sibtain, 1938; Ali, 1947; Oliver, 1951; Fitch, 1954; Ganapati and Rajyalakshmi, 1958; Deraniyagala, 1960; Tilak and Rastogi, 1964, and Rathore, 1967 for figures of skink osteoderms ). A similar pattern of tubules is found in some gerrhosaurine osteoderms and serves to align this subfamily of cordylids with skinks. Hewitt (1929) has sought to use the number of “cells” created by the radiating osteoderm tubules as a means of working out the relationships of major groups of skinks. No one has followed Hewitt’s lead, but it might be profitable to do so in the future. Frontal bones. The separation or fusion of the frontal bones correlates well with certain relationships of the bones in the secondary palate and is important in diag- nosing the four subfamilies of skinks. The condition of the frontal is, of course, also an important character in diagnosing major taxa in other lizard families. Nasal bones. These bones are fused in one subfamily (Feylininae) and distinct in the other three. Given the great number and diversity of species in these latter three Bulletin Museum of Comparative Zoology, Vol. 139, No. 3 subfamilies, the fusion of the nasals in one group of skinks is important. Ectopterygoid. This bone can show a good deal of variation in its relationships — with the bones (palatine and pterygoid) of © the secondary palate. These relationships — are important at the subfamily level and below. Jugal. This bone is lacking in one small — subfamily (Feylininae ), and, as it is lacking in only two other very closely related genera of skinks, its loss is an important feature. Teeth. The presence or absence of — pterygoid teeth and the number of pre-— maxillary teeth seem to correlate well with other characters of taxa at the level of genus or species group. In general, these | two characters, especially the number of premaxillary teeth, seem to be of greater taxonomic significance in skinks than in most other lizard families. Supratemporal arch. Skinks are often diagnosed as displaying a complete supra- — temporal arch, that is, the postfrontal and squamosal bones articulate with one an- other either directly or through a_post- orbital bone. This is true in three of the subfamilies of skinks, the Feylininae, Scincinae and Lygosominae, but it is not true in the fourth subfamily, the Acontinae (with the exception of two species). In this last group the postfrontal and squa- mosal are generally small (a postorbital is lacking ) and do not form a complete arch. Meckel’s groove. Meckel’s groove may either be open anterior to the splenial or be closed by the overlapping and fusion of the dentary. There are only a few species with an intermediate condition (dentary overlapping but not fused along the resulting suture), and either one con- | dition or the other seems to be character- istic of major groups of skinks. External naris. In two of the four sub- families (Feylininae and Acontinae) the external naris is situated in a large, poste- riorly expanded rostral, and in the lygo- somines the naris is in a discrete nasal scale. The Scincinae, however, show a variety of relationships between the ex- ternal naris and the surrounding scales; these relationships are of some value in recognizing the taxa within this group. Preanal scales. The single, transverse preanal scale in one subfamily (Acontinae ) is unique among skinks, with the exception of a few species of Tropidophorus, and is unusual in lizards. The size of the preanal scales relative to one another and to the posterior ventral scales is a useful char- acter for aligning major groups of lygo- somine skinks. Appendages. The relative frequency of the species that have completely lost the external appendages in the four subfamilies reflects, to some extent, the degree to which the subfamilies have “gone under- ground, that is, have become burrowers. Length of tail. The relatively short tails (less than one-third of the total length) of two of the subfamilies (Feylininae and Acontinae ) might be thought of simply as an adaptation to a burrowing way of life, but the burrowers of the other two. sub- families have the relatively longer tails characteristic of their groups. Mode of reproduction. The two smallest subfamilies, the Feylininae and the Acon- tinae, seem to be live-bearing, whereas the two larger subfamilies, the Scincinae and Lygosominae, are both egg-laying and live- bearing. Since egg-laying habits are un- doubtedly ancestral to live-bearing habits, this character helps to establish the possible phylogeny of the four subfamilies. THE SUBFAMILIES OF SKINKS The four subfamilies discussed below are not arranged in any phylogenetic order, as is often the case in papers of this nature. Instead, the two small (in terms of number of species) and highly specialized sub- families, the Feylininae and Acontinae, are discussed first; the Scincinae, which are considered to be independently ancestral to all three other subfamilies, come next Ol Ol SUBFAMILIES OF SKINKS * Greer L and are followed by the Lygosominae, the most numerous, diverse, and advanced group of skinks. In the description of the skull features of the four subfamilies, only the taxo- nomically important characters will be considered. The bones of the secondary palate are described in detail first, as they offer the most diagnostic characters for recognizing the subfamilies. The remainder of the bones of the skull are then described in a generally anterior-posterior order. The teeth and mandible are described last. Feylininae Diagnosis. Frontal bones separate (Fig. 1); premaxillae and nasal bones fused. Horizontal laminae from lateral sides of palatines approaching but not touching on ventral midline. An anteriorly projecting process from palatal ramus of pterygoid articulates with maxilla to exclude palatine from position on medial edge of infra- orbital vacuity (Fig. 2). Post-temporal fenestra reduced in size; supratemporal arch complete, i.e., post- frontal articulates with squamosal, which is closely applied to parietal. Postorbital and jugal bones lacking. Lateral descend- ing processes from frontals and parietal fingerlike, i.e., not expanded. Bony shaft of stapes abutting directly against quadrate. Seven teeth on premaxillae; 13-14 teeth on maxilla. Meckel’s groove open anterior to splenial. Rostral and mental scales slightly en- larged. External naris connected with posterior edge of rostral by short suture. Limbs totally lacking. Preanal scales not enlarged, i.e., approximately same size as other ventral, posterior body scales. Description of skull. Cope (1892) has figured and described certain features of the skull of Feylinia currori. The following account is based on the skulls of two species, Feylinia polylepis and F. currori. The skulls of both species are very similar. 156 Bulletin Museum of Comparative Zoology, Vol. 139, No. 3 SC INC INAE fr FEYLININAE ACONTINAE Figure 2. Ventral view of the skulls of representatives of the four subfamilies of skinks. Same species as represented in Figure 1 except for the Acontinae, which are represented by Acontias meleagris (MCZ 11934). Drawn to scale. The skull as a whole is somewhat de- pressed for its length, and the postorbital region is not as elongate as might be ex- pected in a lizard so obviously adapted to a burrowing existence. The palatines are scroll-like, with the two free edges of each “scroll” just failing to meet medioventrally. Each palatine thus forms a separate air passage, with the ventral surfaces of the palatines acting as a rudimentary secondary palate to par- tially separate the food and air passages. The palatal rami of the pterygoids are separated medially and therefore do not participate in the formation of the second- ary palate. The palatal ramus of the pterygoid articulates with the ectopterygoid along the posterior edge of the _ infraorbital vacuity and sends an anterior process for- ward to the maxilla to exclude the palatine from a position on the medial edge of the infraorbital vacuity (Fig. 2). The premaxillae, vomers, and nasals are each fused to form single elements, but the frontal is paired. The parietal is single and anteriorly bears a foramen, which, however, is overgrown by a bony boss on the dorsal surface. The prefrontal is very large and occupies most of the anteromedial side of the orbital area. On the dorsal surface of the skull, the prefrontal articulates with the nasal. The suborbital bar is composed entirely of the ectopterygoid and an articulating proc- ess from the palatal ramus of the ptery- goid. A jugal (postorbital bar) is lacking. The postfrontal bone is small and compact. There is a single pair of thin, fingerlike. lateral processes descending from both the frontals and parietal. Those from the frontal are closely applied to the prefrontal and curve inward toward the midline but do not meet to encircle the forebrain. The processes from the parietal hang free and touch only the dorsally projecting epi- pterygoid. These parietal processes are similar to those of most non-burrowing lygosomine skinks and are not in the least Ol ~l SUBFAMILIES OF SKINKS © Greer [I expanded into long processes such as those which enclose most of the hindbrain of such typical burrowers as the Acontinae. There is a small post-temporal fenestra, but although the supratemporal arch is complete, the supratemporal fenestra is obliterated by the close apposition of the squamosal to the parietal. A postorbital bone is absent. The quadrate is short and stout, with a vertical ridge on its anterior surface. A horizontal, posteriorly projecting process with a ventral, terminal inflection arises from the posterodorsal surface of the quad- rate. The footplate of the stapes is large, and the bony shaft abuts against the inner side of the ventral inflection of the poste- rior process of the quadrate. Posteriorly curved, almost fanglike teeth are present on the fused premaxillae, maxillae, and dentaries. There are no teeth on any of the other bones of the skull or jaw. Both species of Feylinia examined possess seven teeth on the fused pre- maxillae and 13-14 teeth on the maxilla. The skull and mandible lack pigment. In the lower jaw, the articular, pre- articular, and surangular are fused. The angular is reduced in size. The splenial extends posteriorly to occupy much of the position held by the angular in other skinks. The coronoid process is low, and Meckel’s groove is present. Description of external characters. The single rostral and mental scales are slightly enlarged; the external naris lies within the rostral and is connected with the posterior edge of the rostral through a short hori- zontal or curved suture. The middorsal head scales consist of a pair of postrostral scales (in Feylinia) or a single postrostral scale (in Chabanaudia ), and following this, three single, large, median scales. An external ear opening is lacking. The body scales are smooth and disposed in 16-30 longitudinal rows at midbody. The preanal scales are subequal with the other ventral, posterior body scales. Limbs are absent, although rudimentary 158 pectoral and pelvic girdles are present (Essex, 1928). The tail is relatively short, comprising approximately one-third of the total length. Mode of reproduction. The only infor- mation available on this topic is a note by de Witte (1953) on two gravid Feylinia currori, which contained two and_ three “embryons. This meager evidence sug- gests that F. currori is probably live- bearing. Distribution. Central and west Africa and Principe Island, primarily in lowland, evergreen forest (Fig. 3). Genera. Two genera and four species are currently recognized in the subfamily Feylininae: Feylinia Gray, 1845; 3 species; central and west Africa and Principe Island. Chabanaudia de Witte and Laurent, 1943; 1 species; Gabon. Discussion. Chabanaudia has been sepa- rated from the genus Feylinia by de Witte and Laurent (1943) on the basis of its single rather than double postrostral scale. As I have not examined a skull of the single species of Chabanaudia (boulen- geri), I can add nothing to our knowledge of its generic characters or relationships. Feylinia is, on the basis of osteoderms and the secondary palate, clearly a skink and, according to Miller (1966a), the cochlear duct of Feylinia is so “similar in all details to the scincid duct that it may be included in that general group.” Boulenger (1887) distinguished Feylinia, Typhlosaurus, and Anelytropsis as a sepa- rate family (Anelytropidae) and regarded it as a “degraded type of the Scincidae (italics his), with which they are closely connected through the genus Acontias.” As will be shown below, Typhlosaurus is indeed very closely related to Acontias, comprising, with this genus and the mono- typic Acontophiops, a separate subfamily of skinks. The affinities of the rare monotypic Mexican genus Anelytropsis are not so clear, however. Recent studies of the Bulletin Museum of Comparative Zoology, Vol. 139, No. 3 cochlear duct (Miller, 1966b), skull (Mc- Dowell, personal communication), and vertebral morphology (Etheridge, 1967) indicate that this genus is perhaps related to the Southeast Asian-New Guinean Di- bamus, but the broader relationships of these two genera are obscure. Specimens examined.' I have examined the skulls of the following species of fey- linines: Feylinia currori elegans (MCZ 49886), F. currori (MCZ 106990), and F. polylepis (MCZ 61215). Acontinae Diagnosis. Frontal bone divided (Fig. 1); palatine bones just separated ventrally along midline of secondary palate; palatine in broad contact with ectopterygoid along posterior edge of infraorbital vacuity, thereby usurping extensive contact of pterygoid with ectopterygoid and exclud- ing palatal ramus of pterygoid from a position on infraorbital vacuity (Fig. 2). Supratemporal arch and_ post-temporal fenestra usually completely lacking (except in Acontias plumbeus and Typhlosaurus lineatus, see below). Prefrontal and squa- mosal bones reduced in size (except in Acontias plumbeus and Typhlosaurus line- atus), the squamosal especially so, being much smaller than the supratemporal bone directly posterior to it. Maxilla borders orbit ventrally, the jugal being reduced to a small, vertical element suspended between postfrontal and maxilla. Four to six teeth on premaxillae and three to ten teeth on maxilla. Meckel’s groove closed and fused. Rostral and mental scales greatly en- larged, the external naris being situated well forward in rostral and connected to its posterior edge by a horizontal suture. No external trace of limbs. A single, trans- versely enlarged preanal scale. Tail less than 22 per cent of total length. Description of skull. Detailed descrip- ! Abbreviations used in this section and similar sections to follow will be found on pp. 180 and 181. tions of the skull of Acontias meleagris have been provided by de Villiers (1939), Brock (1941), and van de Merwe (1944), and figures of the skulls of Acontias plumbeus and Typhlosaurus aurantiacus in Peters (1882). As practically all the important skull features of the subfamily can be seen in the well-described A. meleagris, no more than a brief description of the skull mor- phology characteristic for the group will be given here. As is generally true in other burrowing lizards, the postorbital region of the skull has become elongated, a feature which, along with the blunt rounded snout, gives the whole skull a bullet-shaped appear- ance. The palatines are two long, almost com- plete scroll-like tubes whose ventral sides approach closely, but do not meet, along the midline of the palate. The medial sides of the palatines do touch, however, and articulate with medial posterior projections of the vomers to separate partially two tubular air passages. The palatine is in broad contact with the ectopterygoid along the posterior edge of the infraorbital vacuity. The palatal ramus of the pterygoid thus lacks the broad con- tact with the ectopterygoid seen in all other skinks and is completely excluded from the edge of the infraorbital vacuity. The pterygoids are also widely separated from one another along the midline of the palate G@isig. 2). The premaxillae, vomers, nasals, and frontals are divided by a median suture. Closely apposed medial processes from the vomers project posteriorly for about half the length of the palatines and articulate with the closely apposed medial sides of the palatines to separate partially the two air passages formed by the scroll-like pala- tines. The parietal bone is single; there is a parietal foramen in the anterior part of the parietal, although in some specimens it tends to be covered dorsally with a bony boss. SUBFAMILIES OF SKINKS *« Greer 159 A long, thin, anteriorly projecting process from each frontal bone wedges part way between the nasal and maxilla to separate the reduced prefrontal from the nasal. The prefrontal is a very small bone on the dorsal edge of the orbit, which articulates with the postfrontal to exclude the frontal] from the orbit. Lateral descending processes from each frontal approach closely or meet below the forebrain. The lateral descending parietal processes are expanded longitudinally to varying degrees, thereby enclosing the hindbrain to varying degrees. The supratemporal arch and the post- temporal fenestra are lacking, except in Acontias plumbeus, which has retained both the arch and the fenestra, and Typhlosaurus lineatus, which has retained the supratemporal arch but has lost the post-temporal fenestra. The postfrontal and squamosal are reduced in size, the squa- mosal especially so, being much smaller than the supratemporal bone directly pos- terior to it. In A. plumbeus and T. lineatus the squamosal and postfrontal are well developed and form a supratemporal arch. There is also a clear post-temporal fenestra in A. plumbeus, but not in T. lineatus. All species in the subfamily lack the post- orbital bone. The jugal does not take part with the maxilla in forming the ventral border of the orbit as in most skinks, but is reduced to a small vertical element hanging be- tween the postfrontal and maxilla. An epipterygoid is present. The quadrate is short, stout, and slightly concave posteriorly. The end of the bony shaft of the stapes never articulates directly with the quadrate. In some species (e.g., Typhlosaurus caecus and T. vermis), how- ever, the quadrate is very compressed, and the shaft of the stapes projects anteriorly, oblique to the lateral edge of the quadrate. Teeth are present only on the pre- maxillae, maxillae, and dentaries. The number of teeth ranges from four to six on the premaxillae and from three to ten on 160 the maxilla. The maxillary teeth vary from the short, blunt crushing teeth of Acontias plumbeus to the pointed, slightly curved teeth of Typhlosaurus vermis. The skull and mandible lack pigment. In the lower jaw, the articular, pre- articular, and surangular are usually fused, although the labial suture between the sur- angular and articular may be evident. The splenial is usually reduced in size, but the angular is well developed. Meckel’s groove is obliterated by the overlapping and fusion of the dentary. Description of external characters. The rostral and mental scales are greatly en- larged. The external naris is situated well forward in the large rostral and is con- nected with its posterior suture through a horizontal suture. The middorsal head scales consist of one to three single, large, median scales between the posterior edge of the enlarged rostral and a pair of parietals. The external ear opening is completely covered by scaly epidermis. The body scales are smooth and disposed in 12 to 20 longitudinal rows at midbody. There is a single, transversely enlarged preanal scale. All external traces of limbs are lacking, although there are rudimentary pectoral and pelvic girdles (Essex, 1928). The tail is very short, comprising less than 22 per cent of the total length. Mode of reproduction. The three species of acontines for which the mode of re- production is known (Acontias meleagris, Typhlosaurus bicolor, and T. lineatus) are live-bearing and produce one to four young in a clutch. Distribution. Southern Africa with an isolated population in extreme southeastern Kenya (Fig. 3). Genera. Only three genera, encompass- ing 15 species, are included in the sub- family: Acontias Cuvier, 1817; 6 species; south- ern Africa, with an isolated population in extreme southeastern Kenya. Bulletin Museum of Comparative Zoology, Vol. 139, No. 3 Acontophiops Sternfeld, 1911; 1 species; northern Transvaal of South Africa. Typhlosaurus Wiegmann, 1834; 8 species; southern Africa. Discussion. An important problem is the status of the Malagasy Acontias. Boulenger (1887) included Cingalese and Malagasy species as well as South African species in his genus Acontias. Hewitt (1929) pointed out certain differences in the head scales and tubular system of the osteoderms among the Acontias of Ceylon, Madagas- car, and Africa. He thereupon referred the Cingalese skinks to their (original) genus Nessia Gray, 1839 (type species: burtoni), and proposed the generic name Pseuda- contias for the two Malagasy species (type species: holomelas), leaving the name Acontias Cuvier, 1817 (type _ species: meleagris ), for the mainland African forms. M. A. Smith (1935) noticed the great differences in the relationships of the bones of the palate between Nessia and Acontias (outlined here in the diagnoses of the respective subfamilies, the Scincinae and Acontinae) and supported Hewitt’s taxonomic decisions for these two groups. Angel (1942) noted that the generic name Pseudacontias Hewitt, 1929, was pre- occupied by Pseudacontias Bocage, 1889, another genus of Malagasy skinks, and, minimizing the differences pointed out by Hewitt (1929), put the two disputed Malagasy skinks back in Acontias. In addition to the differences in the tubular pattern of the osteoderms and the relationships of certain head scales, Hewitt (1929) had noted that the Acontias of Africa differed from those of Madagascar in having very much shorter tails and fewer scales around midbody. These two differ- ences distinguish the subfamily Acontinae and the subfamily next discussed, the Scin- cinae, to which in fact the Malagasy (and the Cingalese Nessia) belong. The skull differences between South African Acontias and Malagasy Acontias are also those of the two subfamilies. I have examined only the secondary SUBFAMILIES OF SKINKS ° Greer 16] @ = FEYLININAE Y ACONTINAE Figure 3. Distribution of the Feylininae and Acontinae, two subfamilies which have apparently evolved independently from the scincines in Africa. 162 Bulletin Museum of Comparative Zoology, Vol. 139, No. 3 palate in the skull of the Malagasy Acontias, and both species differ from Nessia (layardi) in having the postero- medial edges of the palatal rami of the pterygoids smoothly diverging, instead of deeply emarginated as in Nessia. Such palatal differences are indicative of generic separation. I therefore suggest that the two species of Malagasy “Acontias” (holo- melas and hildebrandti) be placed in a distinct genus which may be known as Malacontias' new genus The type species, herewith designated, is Acontias holomelas Giinther, 1877. Specimens examined. The skulls of the following acontine species have been ex- amined: ACONTIAS: breviceps (MCZ 38559), g. gracilicauda (MCZ 100905), g. occidentalis (MCZ 67859, 67861), g. tasmani (MCZ 96905), lineatus (MCZ 21416, 21659), meleagris (MCZ 11934, FMNH 84189), plumbeus (MCZ 14233), percivali (MCZ 40180). TYPHLOSAURUS: caecus (AMNH 50669), cregoi (MCZ 41935), lineatus (FMNH 142754), vermis (MCZ 41938). Scincinae Diagnosis. Frontal bone divided (Fig. 1). Palatines almost always separated medially except in some Scelotes, Pro- scelotes and Gongylomorphus? — bojeri. Palatal rami of pterygoids almost always separated medially except in Gongylo- morphus bojeri and the three endemic “Scelotes’ of the Seychelles (gardinieri, braueri and veseyfitzgeraldi). Palatine 1The generic name Malacontias derives from the first syllable of the word “Malagasy’—an inhabitant of Madagascar—and the previous generic name (Acontias) for the species now placed in the new genus. 2 Loveridge (1957) has shown that the generic name Thyrus Gray, 1845, for the endemic Mau- ritian scincine is antedated by the more unwieldy name Gongylomorphus Fitzinger, 1843. bones widely separated from ectopterygoid along posterior edge of infraorbital vacuity in most genera and species, i.e., palatal ramus of pterygoid borders infraorbital vacuity and articulates with ectopterygoid along posterior edge of this vacuity (Fig. 2). In a few species, ectopterygoid con- tacts palatine along posterior edge of infra- orbital vacuity by anteriorly projecting process that excludes palatal ramus of pterygoid from infraorbital vacuity. Supratemporal arch complete, i.e., squa- mosal and postfrontal bones always in contact directly or by way of postorbital bone. Lateral descending processes from parietal to epipterygoid sometimes ex- panded longitudinally, but more frequently simply fingerlike projections. Nostril usually pierced in rostral, or be- tween rostral and various other small head scales, or between two or more small head scales, rarely in large, discrete nasal scale. Limbs present in most species. At least one pair of enlarged preanal scales; tail more than 30 per cent of the total length. Description of skull. The skulls of the following scincines have been figured and discussed in the literature: Barkudia insu- laris (Ganapati and Rajyalakshmi, 1958); Chalcides guentheri (Haas, 1936); Chalci- des ocellatus (Kamel, 1965); Chalcides sp. (Romer, 1956); Eumeces schneideri (Duméril and Bocourt, 1881); Eumeces quinquelineatus (Rice, 1920); Eumeces spp. (Kingman, 1932); Nessia smithi (De- raniyagala, 1953); Scincus scincus (EI- Toubi, 1938); Voeltzkowia mira (Rabanus, 1911). The palatine bones are apposed to vary- ing degrees, but do not actually meet along the ventral midline except in some Scelotes, Proscelotes, and Gongylomorphus, where the palatines meet along their medial edges to various degrees. Dorsally the palatines meet above the air passage. The pterygoids (palatal rami) are al- ways separated medially except in Gongylo- morphus bojeri from Mauritius and_ the three endemic “Scelotes” on the Seychelles. In these species the palatines and ptery- goids form as complete a secondary palate as that seen in any lygosomine. The palatine is usually separated from the ectopterygoid by the palatal ramus of the pterygoid along the posterior edge of the infraorbital vacuity, but in the genus Scincus and in a few species or even in- dividuals of one species of some genera (e.g., Chalcides ocellatus and “Scelotes” astrolabi), the ectopterygoid may make contact with the palatine by an anteriorly projecting process that excludes the palatal ramus of the pterygoid from the infra- orbital vacuity. At the anterior edge of the infraorbital vacuity, the ectopterygoid may extend along the bordering edge of the maxilla to varying degrees and in some species may actually articulate with the palatines to exclude completely the maxilla from the infraorbital vacuity. The premaxillae and vomers may be paired, partially fused, or completely fused. The nasals and frontal are always divided. The parietal is single and bears a parietal foramen. The frontal may form a surface suture with the maxilla to separate the nasal and prefrontal, or the nasal may articulate with the prefrontal to separate the frontal and maxilla, or all four bones may meet at a point. Lateral descending processes from the frontal may be present or absent. When present, they may be long and deep, vir- tually meeting below the forebrain. I.ateral parietal processes are always present and are usually fingerlike projections to the epipterygoid. In some species, however, the parietal processes become somewhat expanded longitudinally, enclosing part of the hindbrain. This is especially true of species adapted to a burrowing existence. The post-temporal fenestra is often re- duced or obliterated in burrowing species, but otherwise the arch is usually present. The postfrontal and squamosal bones are always present and in contact with one SUBFAMILIES OF SKINKS * Greer 163 another directly or through a separate post- orbital bone. An epipterygoid is always present, as is the jugal in all species examined except in the closely related Typhlacontias gracilis, T. rohani, and Fitz- simonsia brevipes. The quadrate is usually concave poste- riorly and convex anteriorly, although in some species this bone becomes very stout and rodlike. The bony shaft of the stapes articulates directly with the quadrate in some genera (Fitzsimonsia, Melanoseps, Ophiomorus, Scolecoseps, Typhlacontias, and Brachymeles vermis, although in no other species of Brachymeles examined). In these scincines, as in the feylinines, the distal end of the stapes abuts against a ventral inflection of a posteriorly project- ing nub of the quadrate. Teeth are always present on the pre- maxillae, maxillae, and dentaries. In some species teeth also occur on the palatal ramus of the pterygoid. There may be 5-11 teeth on the premaxillae, although many genera are characterized by having fewer than nine premaxillary teeth. The number of teeth on the maxilla varies from 10-25. The skull may contain some pigment, although usually it does not. The surangular, articular, and pre- articular bones may be distinct or variously fused to one another. The splenial and angular are always distinct except in Gongylomorphus bojeri, where the angular is fused to the surangular, articular, and prearticular. Meckel’s groove is open in all but a few species. Description of external characters. The head scales in members of this subfamily are extremely variable. The external naris may be situated entirely in the rostral, between the rostral and one or more of the small head scales (diagnosis of Mittle- man’s subfamily Chalcidinae), between two or more smaller head scales exclusively, or, less frequently, entirely within a dis- crete nasal scale (diagnosis of Mittleman’s subfamily Scincinae ). An external ear opening may or may not 164 Figure 4. be present. The body scales are cycloid, imbricate, and generally smooth. The scales are disposed in 14-42 longitudinal rows at midbody, and there are two or more pre- anal scales. Limb reduction is a common trend in the subfamily, although only about 28 of the approximately 182 species totally lack any external trace of limbs. Bulletin Museum of Comparative Zoology, Vol. 139, No. 3 Distribution of the Scincinae (exclusive of the genus Eumeces), the most primitive subfamily of skinks. Mode of reproduction. Of the 44 species of scincines for which the mode of repro- duction is known, half lay eggs and half bear living young, and, as yet, only in the genus Eumeces is the mode of reproduction known to transcend taxonomic boundaries. Distribution. With the exception of Eumeces, the genera of scincines show a disjunct distribution in east and south central Asia. It is only in southwest Asia, north Africa and more especially in Africa south of the Sahara, Madagascar and the islands of the West Indian Ocean that scincines are widely distributed (Fig. 4) and constitute a significant part of the skink fauna. Eumeces is the largest and most wide- spread scincine genus (Fig. 5). The group is distributed along the northern periphery of the world distribution of skinks. The “cold” tolerance implied by this distribution has undoubtedly helped Eumeces cross the Bering Land Bridge into the New World and probably accounts for the group’s suc- cess in the high plateau country of Mexico. There are no scincines in the Indo- Australian Archipelago or the Australian Region. Genera. The following genera are in- cluded in the Scincinae. I have arranged them in geographic order proceeding west through North America into the Old World. Eumeces Wiegmann, 1834; approxi- mately 46 species; Bermuda, North and Central America; east and south- east Asia; southwest Asia; North Africa (see Fig. 5). Neoseps Stejneger, 1910; 1 species; south and central Florida. Brachymeles Duméril and Bibron, 1839; 13 species; Philippine Islands. Barkudia Annandale, 1917; 1 species; Calcutta and Chilka Lake Area. Sepsophis Beddome, 1870; 1 species; central and southern India. Nessia Gray, 1839; 8 species; Ceylon. Chalcidoseps Boulenger, 1887; 1 species; Ceylon. Ophiomorus Duméril and Bibron, 1839; 9 species; Greece through southwest Asia to northwest India. Chalcides Waurenti, 1768; southern Europe, southwest north Africa, Canary Islands. Scincus Gronovius, 1763; 12 species; north Africa to southwest Asia. 14 species; Asia, SUBFAMILIES OF SKINKS * Greer 165 Scincopus Peters, 1864; 1 species; north Africa from Khartoum, Sudan _ to Mauritania. Proscelotes de Witte and Laurent, 1943; 3 species; southeast Africa. Sepsina Bocage, 1866; 5 species; south- ern Africa. Scelotes Fitzinger, southern Africa. Scolecoseps Loveridge, 1920; 2 species, east central Africa. Fitzsimonsia de Witte and Laurent, 1943; 1 species; southern Africa. Typhlacontias Bocage, 1873; 5 species; central and southern Africa. Melanoseps Boulenger, 1887, 2 species; central east Africa and Cameroon. Pygomeles Grandidier, 1867; 3 species; 1826; 15. species; Madagascar. Pseudacontias Bocage, 1889; 1 species; Madagascar. Paracontias Mocquad, 1894; 2 species; Madagascar. Cryptoscincus Mocquad, 1906; 1 species; Madagascar. Grandidierina Mocquad, 1894; 4 species; Madagascar. Voeltzkowia Boettger, 1893; 1 species; Madagascar. Malacontias, new generic designation, see p. 162 above; 2 species; Madagas- car. Gongylomorphus Fitzinger, 1843; 1 spe- cies; Mauritius. Incertae sedis, 25 species of Malagasy “Scelotes” and 3 endemic “Scelotes” of the Seychelles. Discussion. The systematics of the Malagasy scincines is undoubtedly the big- gest problem remaining in the taxonomy of this subfamily. Many species of Mala- gasy scincines are known from only a few specimens—too few to allow skulls to be prepared. Unfortunately this dearth of specimens is not likely to be remedied in the near future, as many of the species are apparently very secretive in their habits, and Madagascar is not, at present, a popular place for collecting reptiles. 166 Distribution of the scincine genus Eumeces. Figure 5. The disjunct distribution of the scincine genera in east and central Asia implies, of course, that the scincine ancestors (pos- sibly, but not necessarily, Ewmeces) were more widespread at one time in the past. Just how widespread these scincines may have been is a very interesting question that future paleontological discoveries may answer, It would be interesting to know, Bulletin Museum of Comparative Zoology, Vol. 139, No. 3 for example, whether the scincines ever in- habited the Australian Region—a _ region where now only lygosomines are found. Specimens examined. I have seen the following scincine skulls: BARKUDIA: insularis (MCZ 54712). BRACHYMELES: bonitae (MCZ 20129), gracilis boulengeri (MCZ field tag 710, MCZ 20131, 26540, 26545, 54253, 26552, 26553, + 1 untagged specimen), gracilis taylori (AMNH 86661), vermis (MCZ 26587 ). CHALCIDES: bedriagai (MCZ 15692), mionecton (MCZ 7753, 25145), ocellatus (MCZ 9817, 9828, 9837, 9839, 9842, 9844, 9849, 9851, UMMZ 1930, CAS(SU) 18137), sepsoides (MCZ 27483, 18351, CAS(SU) 18143). EUMECES: algeriensis (MCZ 4281), anthracinus (MCZ 29312), brevilineatus (MCZ 79776), brevirostris (FMNH 111614), chinensis (MCZ 29005), copei (UINHM 33238), elegans (MCZ 28983, 28992, 29000), fasciatus (MCZ 54126, UINHM 133239, 33240), gilberti (USNM 5310), indubitus (FMNH 114201), inex- pectatus (MCZ 45498, 55506), kishinouyei (MCZ 55935), laticeps (MCZ 55505, + 1 untagged specimen), latiscutatus (FMNH 55511), longirostris (MCZ 20503, 20508), lynxe (MCZ 19086, 19087, 24533, 24534), marginatus (MCZ 57111, 57112 part, 7409), multivirgatus (UINHM 33244), obsoletus (MCZ 35547, 61366, 61367), ochoterena (FMNH 114493), schneideri (MCZ 6986, UMMZ 2119, 2148), schwartzei (USNM 113603), skiltonianus (MCZ 6617—2 speci- mens, 8887, + 1 untagged specimen, CAS 28138), stimsoni (CAS 21660), taeniolatus (FMNH 1868), tunganus (USNM 82751). FITZSIMONSIA: brevipes (MCZ 96702). GONGYLOMORPHUS: bojeri (MCZ 46677). GRANDIDIERINA: lineata (PM 3378). MALACONTIAS (palatal characters only ): hildebrandti (PM 99-376), holome- las (PM 95-215, 7792). MELANOSEPS: ater (MCZ 50955, 52487), occidentalis (BM 1907.5.22.6A). NEOSEPS: reynoldsi (MCZ untagged specimen ). NESSIA: layardi (MCZ 38174). OPHIOMORUS: _ brevipes (FMNH 141550), persicus (FMNH 141557), raith- mai (AMNH 85846), tridactylus (AMNH 75610). PROSCELOTES: aenea (MCZ 18709), SUBFAMILIES OF SKINKS © Greer 167 arnoldi (MCZ 55145), eggeli (MCZ 24217, 24218, 24220). PYGOMELES: braconnieri (PM 1715). SCELOTES: anguina (MCZ 96791), arenicolor (MCZ 14205), bidigittata (MCZ 96789), bipes (BM XVIIL2.F), brevipes (MCZ 21237), caffer (MCZ 96792), gronovi (BM 97.5.15.8), limpopoensis (MCZ 96906), mira (MCZ 96790), ulu- guruensis (MCZ 24206). SCINCUS: scincus (MCZ 27456—2 specimens, 27462, 27464). SCOLECOSEPS: boulengeri (MCZ 18357 ). SEPSINA: angolensis (AMNH 40734, FMNH 142793), bayoni (BM RR 1967.80), tetradactylus (MCZ 42885, 47770—3 speci- mens, 47775, 56963, 56965, 56967, 85536). TYPHLACONTIAS: | gracilis (USNM 159338), rohani (FMNH 142787). VOELTZKOWIA: mira (MCZ untagged specimen ). Incertae sedis: Malagasy “Scelotes”: astrolabi (MCZ 20953, 20955), melanura (MCZ 11733); splendidus (FMNH 72086); Seychelles “Scelotes”: braueri (BM 1910. 3.18.33), gardineri (BM 1910.3.18.91). Lygosominae Diagnosis. Frontal bone single (Fig. 1). Palatines usually in contact along ventral midline except in most Egernia and Corucia zebrata. Palatine making contact with ectopterygoid if at all only through an anteriorly projecting ectopterygoid process; palatal ramus of pterygoid but not palatine itself in broad contact with ectopterygoid along posterior edge of infraorbital vacuity (Fig. 2). Supratemporal arch complete, ie., postfrontal and squamosal always in contact directly or by way of postorbital bone; post-temporal fenestra obliterated in some species. Lateral descending processes from frontal not large when present; lateral descending processes from parietal only fingerlike projections to epipterygoid. Single discrete nasal scale (except in Sphenomorphus schultzei and Ateucho- 168 saurus) bearing the external naris; almost always some external indication of limbs (limbs totally lacking in only four of 600+ species in the subfamily); at least one pair of preanal scales; tail more than 30 per cent (usually 50 per cent or more) of total length. Description of skull. A number of de- scriptions of the skulls of single species in this subfamily have been published as follows: Ablepharus pannonicus (Haas, 1935); Didosaurus mauritianus (Hoffstet- ter, 1945 and 1949); Lygosoma sp. (Pear- son, 1921); Mabuya carinata (Rao and Ramaswami, 1952); Sphenomorphus quoyi (King, 1964). In addition, there are figures of whole skulls of Dasia smaragdina, Ctenotus leseuri, Mabuya_ multifasciata, Sphenomorphus australe, and S. quoyi in Siebenrock (1892), and Brihl (1886) figures the skull of Tiliqua rugosa. Waite (1929) figures ventral palatal views of Sphenomorphus quoyi, Egernia stokesi, and Tiliqua sp. and Duméril and Bocourt (1881) figure a ventral view of the skull of Mabuya mabouya. Mitchell (1950) also has line drawings of the palates of several Egernia and Tiliqua, and Greer (1967a and b) and Greer and Parker (1968) figure the palates of “Ablepharus” lineoocellatus, “Ablepharus” smithi, Carlia _ bicarinata, Emoia samoense, Eumecia anchietae, Geomyersia glabra, “Leiolopisma’ metal- lica, Leptosiaphos blochmanni, Lerista elegans, L. bougainvilli, Mabuya polytro- pis, Riopa punctata, and Sphenomorphus pardalis. The general shape of the skull is highly variable. In burrowing forms the post- orbital region may be elongate and the whole skull bullet-shaped, or conversely, in surface forms, the skull may be short and rather deep. In other instances the skull may be depressed. The palatines meet along the ventral midline to form a secondary palate, above which is the main air passage. Anteriorly the palatines arch over this passageway, but posteriorly most of the air passage is Bulletin Museum of Comparative Zoology, Vol. 139, No. 3 enclosed dorsally by a membranous tissue. This is in contrast to the condition in the Acontinae, where the entire dorsal arch of the air canal is formed by the palatine bones. In only a few species (most Egernia and Corucia zebrata) are the palatines sepa- rated ventrally along the midline. In these species the palatines are never separated by as great a distance as in most of the genera of the more primitive scincines (e.g., Scincus, Eumeces, and Chalcides). The palatine may or may not be in con- tact with the ectopterygoid. When the palatine is in contact with the ectoptery- goid, it is by means of an anteriorly pro- jecting process from the ectopterygoid. The palatal ramus of the pterygoid is the only bone in broad contact with the ectoptery- goid along the posterior edge of the infra- orbital vacuity (Fig. 2). The pterygoids may be in contact along the medial edge of their palatal rami or separated either by the interpterygoid vacuity or by two medioposterior processes of the palatines. The premaxillae and nasals are paired. The vomer may be single or divided. The frontal and parietal are single. There is a parietal foramen in the anterior part of the parietal except in Ateuchosaurus, where the parietal foramen is in the pos- terior part of the frontal or in a small, median azygous bone between the frontal and parietal. The frontal may form a surface suture with the maxilla and thereby separate the prefrontal from the nasal, or all four bones may meet at a common point, or the nasal and prefrontal may meet to separate the frontal and maxilla. There is a pair of descending lateral processes from the frontal in a few species, but these processes are never very deep. There is also a pair of usually well- developed, fingerlike lateral processes descending from the parietal to the ecto- pterygoid. In some species, however, these parietal processes are barely distinguish- able from the shallow crest from which they arise. The post-temporal fenestra is obliterated in burrowing species, but in other forms the fenestra is usually well developed. The primary elements of the supratemporal arch, ie., the postfrontal and squamosal, are always present and in contact with each other directly or by the intermediary of a postorbital when this bone is present. A well-developed jugal and epipterygoid are always present. The quadrate tends to become reduced to a short, rodlike element in forms lacking an external ear opening; this modification of the quadrate occurs in non-burrowing species as well as burrowing forms. Teeth are always present on the pre- maxillae, maxillae, and dentaries, and are found on the palatal rami of the pterygoids in a few species of several genera. The number of teeth on the premaxillae ranges from 6-15 (usually 9-11) and from 8-40 on the maxilla. The skull and mandible may or may not contain pigment. In the mandible the dentary, coronoid, splenial, and angular are always distinct, but the articular, prearticular, and sur- angular may be fused to various degrees. Meckel’s groove may be present or absent or in various states of closure. Description of external characters. The external naris is in a discrete nasal, al- though in Sphenomorphus schultzei and Ateuchosaurus the nasal is fused to the first supralabial. The dorsal head scales most commonly consist of the following scales from anterior to posterior: a single rostral; a single frontonasal; paired supranasals (present or absent); paired prefrontals (fused, meeting along the midline, sepa- rated or absent); a single frontal in contact with one to seven of the one to nine supra- oculars; paired (or fused) frontoparietals and a single (or fused to the frontopari- etals) interparietal, and paired (rarely fused) parietals. The parietal foramen is in the interparietal scale. SUBFAMILIES OF SKINKS « Greer 169 An external ear opening may or may not be present. Some species have auricular lobes along the edge of the external ear opening. The body scales are cycloid and imbricate (except in Tribolonotus, which has granular and tubercle-like scales on the dorsum and sides) and are either smooth or keeled. At midbody the scales are disposed in 18-112 rows. There are two or more preanals, except in a few species of Tropidophorus which have but a single, large preanal scale. Most of the species have both fore and hind limbs, although digit and limb re- duction is a recurrent trend in the group. The fore limbs and digits are usually re- duced and lost ahead of the hind limbs and digits. However, in only four of the 600+ species of the subfamily are limbs totally lacking. Mode of reproduction. The mode of re- production is known for 193 species of lygosomines; of these 193 species, 124 (64.3 per cent) lay eggs and the remaining 69 (35.7 per cent) are live-bearing. Distribution. The lygosomines, being the most numerous, in terms of numbers of species, and most diverse subfamily of skinks, are, not surprisingly, the most wide- spread subfamily (Fig. 6). Lygosomines are best represented both in numbers of species and in diversity of adaptations in the Australian Region. It is also this group, exclusively, that has seeded the far-flung islands of the Pacific, where they extend as far north as the Bonin Islands and Lay- san (Ablepharus boutoni) and as far south as Stewart Island (Leiolopisma lineoocel- lata and L. zelandica). The lygosomines seem to be absent from the more arid regions of central north Africa and the Arabian Peninsula. Genera. The genera in this subfamily are as yet too poorly known taxonomically to list. While I do not agree with many of Mittleman’s (1952) new generic groupings of lygosomines, his work does provide a convenient list of most of the species in the subfamily. In addition, he provides a In the oceans only the islands that 170 Bulletin Museum of Comparative Zoology, Vol. 139, No. 3 4 atl y) i ae Lol eae Figure 6. Distribution of the Lygosominae, the most advanced subfamily of skinks. bound the northern and southern limits of the range are marked with distribution lines. nearly complete primary generic synonymy. Mittleman (1952) did not regard the genera Tiliqua, Egernia, Corucia, and Mabuya as lygosomines. Mabuya and Tili- qua fit the diagnosis of the lygosomines as given here in every regard, but some Egernia and Corucia zebrata differ from other lygosomines in having the palatine bones slightly separated along the midline of the secondary palate. The two genera represented by these species are very closely related (Greer, unpublished work) and on all other characters, especially the fusion of the frontals, are lygosomines. At this point, the separation of the pala- tines seems most readily interpreted as either a secondary separation or, less prob- ably in my mind, as a primitive character retained from a scincine ancestry. Discussion. The greatest single taxonomic problem with the Lygosominae is the delimitation and relationships of genera. Recent work has shown that among the lygosomines, as well as in other major skink taxa, the correlation of skull and external morphology is a very effective method for recognizing taxa worthy of generic rank (Greer, 1967a and b, and Greer and Parker, 1968). Specimens examined. Listed below are the skulls of the lygosomine skinks I have examined. In this list I have tried to indi- cate my present opinions of species re- lationships by an informal nomenclature, e.g., alpha and beta Leiolopisma, and geography. Some of these groups have been discussed in other papers (Greer and Parker, 1967 and 1968). ABLEPHARUS: alpha Asian: brandti (MCZ 56533), deserti (MCZ 5364), gray- anus (MCZ 84084, CAS 99883), kitaibeli (CAS 47453), pannonicus (MCZ 3961); alpha Australian: boutoni (MCZ 31040), lineoocellatus (MCZ 33143, 33144, CAS 77404, BM XL4.A), spenceri (AMNH 83929), taeniopleurus (MCZ 35321); beta African: megalurus (MCZ 31065), smithi (MCZ 42880), wahlbergi (MCZ 55827). ANOTIS: mariae (MCZ 92393). ATEUCHOSAURUS: pellopleurus (MCZ 50925, 55927), sowerbyi (AMNH 34153). CARLIA: bicarinata (MCZ 64315), fusca (MCZ 49412, 49423, 73791, 73793, CAS 100777), novaeguineae (MCZ _ 83758), vivax (AMNH 82758). COPHOSCINCOPUS: durus (MCZ un- tagged specimen ). CORUCIA: zebrata (MCZ 68815, 72918, 77375, AMNH 69434). €TENOTUS: australis (MCZ 79537, CAS 76722), fasciolatus intermedius (MCZ 39442), labillardieri (MCZ 24730), leon- hardi (MCZ untagged specimen), leseuri (MCZ 74891); spaldingi (MCZ 35374), taeniolatus (MCZ 6302). SUBFAMILIES OF SKINKS * Greer ila DASIA: olivacea (MCZ ex 7726), semi- cincta (MCZ 26414), smaragdina (MCZ 4094—2 specimens), smaragdina moluc- carum (MCZ 7709), s. perviridis (MCZ 49315, 72275, 72508), s. philippinicum (MCZ 26429), vittata (MCZ 16352). EGERNIA: bungana (FMNH 35146), cunninghami (FMNH_ 31041), depressa (MCZ 33062), formosa (MCZ_ 33067, 33070, 33071, 33078, 33076), hosmeri (AMNH 87779), inornata (MCZ 35289, 30291, 35294, 35297), kingi (MCZ 33087), luctuosa (MCZ 33104), m. major (AMNH 69434), nitida (CAS 76619), stokesi (MCZ 33105; 33106, 33108) EMNH™ 51707); -s: striolata (MCZ 24552), whiti napoleonis (MCZ 24491). EMOIA: adspersa (AMNH 29227), atro- costata (MCZ 15074, 15080, 26476, 26479), boettgeri (MCZ 22074), callisticta werneri (MCZ 67203, 67308, + 3 untagged speci- mens), cyanogaster (MCZ 15121, 15135, 72278, 72287), cyanura (MCZ 14582, 14584, 14586, 75954, 75956), flavigularis (MCZ 65869), kordoana (MCZ 48603), kueken- thali (FMNH 134594), loveridgei (MCZ 49321), maculata (MCZ untagged speci- men), mivarti fuscolineata (MCZ_ 73807, 75984), nigra (MCZ 15153, 15157, 67770, 50 72514. (25a oli, 2523, hlool2ne p. pallidiceps (MCZ 79856), p. physicae (AMNH 95772), ruficauda (MCZ 26482— 2 specimens, 26492), samoensis (MCZ 16931), sanfordi (AMNH 40169), sorex (MCZ 7705), submetallica (AMNH 59015). EUGONGYLUS: albofasciolatus (MCZ 4097, 72703), rufescens (MCZ 49341). EUMECIA: anchietae (MCZ_ 41557, 41562). GEOMYERSIA: glabra (MCZ 87611). HEMIERGIS: decresiensis (MCZ 49173), initiale (MCZ 74976), peroni (MCZ 24648, 24652), quadrilineatum (MCZ_ 33210), tridactylum (MCZ 24595). LEIOLOPISMA: telfairi (MCZ 3077); alpha Asian: bilineata (MCZ 3923), hima- layana (MCZ untagged specimen), mo- desta (AMNH 23669), reevesi (MCZ 39234, 39237, 39236); alpha Southeast 172 Asian-New Guinea: cheesmanae (AMNH 62461), longiceps (MCZ 48585), miotis (MCZ untagged specimen), noctua (MCZ 76006, 76008), pulchella (MCZ 26440, + 1 untagged specimen), quadrivittata infra- lineolata’ (MCZ_ untagged specimen), q. quadrivittata (AMNH_ 86665), rabori (AMNH 93698), semperi (MCZ 20120); alpha Australian South Pacific: austrocale- donica (MCZ 15970), elegantoides (MCZ 80111), entrecasteauxi (MCZ 33223), mac- coyi (MCZ 33199), metallica (MCZ 67129), moco (MCZ untagged specimen), nigrofasciolata) (MCZ 27943), pretiosa (MCZ 10232), stanleyana (MCZ 47904, 47906), suteri (MCZ untagged specimen), zelandica (MCZ 92261); alpha North American: laterale (MCZ 2436, CAS 31123), cherrei cherrei (MCZ 29400), c. lampropholis (MCZ 15479); flavipes spe- cies group: flavipes (MCZ 22189, x-21440), prehensicauda (MCZ 85561 + 1 untagged specimen), virens (MCZ 76270, 76917, 76920); beta African: reichenovei (AMNH 11195); beta Australian: challengeri (MCZ 30455, AMNH 82792), guichenoti (MCZ 61379), mustelina (MCZ 61386), weeksae (MCZ 49190). LEPTOSIAPHOS: blochmanni (MCZ untagged specimen), graueri (MCZ 47662), kilimense (MCZ 24189, 41577), meleagris (MCZ 47676). LERISTA: bipes (AMNH 86089), bou- gainvillei (MCZ 61403), elegans (FMNH 11319), fragilis (MCZ 42988, CAS 77190), gerrardi (MCZ 33255), lineata (MCZ 33265), lineopunctulata (BM 1902.7.30.5), miopa (MCZ 33260), muelleri (MCZ 86699), planiventrale (BM 1954.1-2.21), praepedita (MCZ 33265), punctatovittata (MCZ 79494), tetradactyla (BM 1902.7.30 6), timida (MCZ 13246). LYGOSOMA: equale (MCZ 35344), quadrupes (MCZ 20518), verreauxi (MCZ 10263 ). MABUYA: aurata septentaeniata (MCZ 56550), bayoni (MCZ 39731), bensoni (MCZ).22583)),) *binotaias ( MCZ, 22421), Bulletin Museum of Comparative Zoology, Vol. 139, No. 3 blandingi (MCZ 55171), brachypoda- (MCZ 71410), brevicollis (MCZ 41306), capensis (MCZ 21433), comorensis (MCZ 24151—2 specimens, 24155), dorsovittata (MCZ untagged specimen), elegans (MCZ 67954), englei (MCZ untagged specimen), fasciata (MCZ 37835 + 2 untagged speci- mens), f. frenata (MCZ 49547), graven- horsti (MCZ 11609), hildebrandti (MCZ 70248, 70254), lacertiformis (MCZ untag- ged specimen), mabouya (CAS 71456, UMMZ s-1047), m. mabouya (MCZ 32040, 38935, 54201, 81182, 81184), m. sloani (MCZ 36617), macrorhyncha (MCZ 49551, 49552), macularia (MCZ untagged speci- men), maculilabris (MCZ 24820, 24821), megalura’ (MCZ 47611), multicarinata (CAS 60430), multifasciata (MCZ 25198, 25199, 37843, CAS 60692, 362 + 2 untagged specimens, UMMZ s-1831, s-1830), occi- dentalis (MCZ 43180), perroteti (MCZ 19711), planifrons (MCZ 54559, 85545), polytropis (MCZ 8103), quinquetaeniata margaritifer (MCZ 52424—2 specimens, 55179, 67838-67840 ), sulcata (MCZ 21645), striata (MCZ 74472-74474), varia (MCZ 18658—2 specimens, 18668, 50823, 50824, 85543). MENETIA: greyi (MCZ 79498). OPHIOSCINCUS: anguinoides 74098 ), roulei (MCZ 74099). PANAPSIS: breviceps (MCZ untagged specimen ). RIOPA: afer (MCZ 41517, 41519, 71881), albopunctata (MCZ 8360, UMMZ 122269), bowringi (MCZ untagged specimen, CAS 60737), fernandi (MCZ 49696), laeviceps (MCZ 71889), lineata (AMNH 46379), mabuiiformis (MCZ 40267), pembana (MCZ 46106), popae (MCZ 44706), punctata (MCZ 3238), sundevalli (MCZ 41537, 41543), tanae (MCZ untagged specimen), vinciguerrae (MCZ 17892). RISTELLA: beddomi (BM 82.5.22.152), guentheri (BM 82.5.22.137), rurki (BM 74.4.29.1329), travancorica (BM 74.4.29 ESTO): SPHENOMORPHUS: fasciatus species group: antimorus (MCZ 25374), australe (MCZ (MCZ 24568 ), cranei (MCZ 76048 ), crassi- caudus (MCZ untagged specimen, AMNH 82606), elegantulus (BM 83.4.14.20), emi- grans (MCZ 27043), fasciatus (MCZ 26357), maindroni (MCZ 64273, 72737), muelleri (MCZ untagged specimen), nigri- caudus (MCZ 35407), p. pardalis (MCZ 30413), pratti (MCZ 48596), pumilus (MCZ 48824), punctulatus (MCZ 5250), quoyi (MCZ 3301, 3307, 79549, 79552), rufus (MCZ 47064), solomonis (MCZ 72618, 72626, 72664, 72665, 77373, 77374), tanneri (MCZ 76507, 76509, 89126, 92227), tenuis (MCZ 35398), tryoni (MCZ 35387, 30388 ); variegatus species group: acutus (MCZ 20114), aignanus (BM 1946.8.15.48 ), anomalopus (MCZ_ 37849), boulengeri (AMNH 33180), concinnatus (MCZ 72732, 72733, 91426), cumingi (MCZ 20113), darlingtoni (MCZ 83965), dussumieri (BM 1946.8.15.42 ), florensis nitidus (MCZ 27018, 27019, 27022, 27024), formosensis (AMNH 34909), fragosus (MCZ 92267), granulatus (AMNH 95782), i. indicus (MCZ 44724), j. jobiensis (MCZ 44190, 99336, BM 1935.5 10.108), maculatus (MCZ 3336), melano- pogon (MCZ 68919, x-10113), nigrolabris (FMNH 14255, BM 96.4.29.19-21), sanctus (MCZ 7663), striolatus (MCZ 27034), taylori (MCZ 78090), tersus (MCZ 39284 ), variegatus (MCZ 25398 ); alpha SPHENO- MORPHUS: bignelli (MCZ 19602), min- utus (MCZ 54259), ornatus (MCZ 6154); incertae sedis: fallax (MCZ 19602), louisia- dense (BM 1946.8.19.25), monotropis (BM 1908.5.28.54-55), — striatopunctatus (BM 1948.1.7.60). TILIQUA: branchiale (MCZ 78652), nigrolutea (MCZ 1077—2_ specimens, FMNH 22498), occipitalis multifasciata (MCZ 35310), rugosus (MCZ 24456, UMMZ s-2346), scincoides (MCZ 65221), UMMZ _ s-1864, s-1863, FMNH_ 51702, 51710). TROPIDOPHORUS: beccari (MCZ 43524 ), laotus (MCZ 100512), misaminus (MCZ 44163—3 specimens), robinsoni (MCZ 39374). SUBFAMILIES OF SKINKS * Greer 173 TRIBOLONOTUS: | blanchardi (MCZ 72763), gracilis (AMNH 82364), novae- guineae (MCZ 21063), pseudoponceleti (MCZ, - 72766, 76424, 76425, 76456), schmidti (AMNH 66219). A KEY TO THE SUBFAMILIES OF SKINKS The following key is as much a review of the diagnostic characters of the sub- families of skinks as it is a device for their identification. In each section the char- acter states are listed in the order of their taxonomic importance. 1. Frontal bones separate (Fig. 1); pala- tine bones separated ventrally along mid- line of secondary palate (Fig. 2), except in some Scelotes, Proscelotes, and Gongy- lomorphus bojeri; supratemporal arch com- plete or incomplete; external naris often not in a discrete nasal scale; many species Without any, trace OF limbs 2. as 2, Frontal bones fused (Fig. 1); palatine bones meeting ventrally along midline of secondary palate (Fig. 2), except in some Egernia and Corucia zebrata; supra- temporal arch always complete; external naris almost always in a discrete nasal scale; rarely without any external trace of limbs LYGOSOMINAE 2. Palatine excluded from position on infraorbital vacuity by anteriorly projecting process from palatal ramus of pterygoid to maxilla (Fig. 2); nasal bones fused; jugal absent; supratemporal arch complete is is NOV oe RS ee Pate FEYLININAE Palatine borders edge of infraorbital vacuity (Fig. 2); nasal bones separate; jugal present except in Typhlacontias gracilis, T. rohani, and Fitzsimonsia brevi- pes; supratemporal arch complete or in- Completes = en se ee 3 3. Palatines in broad contact with ecto- pterygoid along posterior edge of infra- orbital vacuity to partial exclusion of palatal ramus of pterygoid (Fig. 1); supra- temporal arch incomplete except in Acon- tias plumbeus and Typhlosaurus lineatus; 4-6 teeth on premaxillae, 3-10 teeth on maxilla; limbless; a single transversely en- 174 larged preanal scale; tail 22 per cent or less of total length _.___- ACONTINAE Palatines separated from ectopterygoid by palatal ramus of pterygoid along pos- terior edge of infraorbital vacuity (Fig. 2) or palatines in contact with ectopterygoid by way of anteriorly projecting process from ectopterygoid that excludes palatal ramus of pterygoid from a position on infra- orbital vacuity, but palatine never excludes the palatal ramus of pterygoid from a major contact with ectopterygoid as in Acontinae; supratemporal arch complete; 5-11 teeth on premaxillae, 10-25 teeth on maxilla; only a few species lack any trace of limbs; at least two preanal scales; tail 30 per cent or more of total length SCINCINAE FOSSIL RECORD OF SKINKS In spite of the great diversity and abun- dance of skinks today, the family has a very poor fossil record (Hoffstetter, 1944 and 1961). Only two genera of pre-Pleisto- cene fossil skinks have been accurately reported. Sauriscus cooki (Estes, 1964a) is known from the late Cretaceous Lance Formation of eastern Wyoming. The diagnostic char- acters of the available fragments, iLe., weakly bifid teeth and striations on the lingual surface of the tooth crown, do not serve to align the species with any living skink. If the fossil is in fact a skink, rather than a representative of another scinco- morph family, its primary importance is that it indicates that skinks were extant by at least the late Cretaceous. All other skink fossils are referable to the scincine genus Eumeces and come from deposits within the present geographic range of the genus. The oldest of these fossils is Oligocene in age (Estes, 1964b ) and is as easily recognizable as a Eumeces as is any recent skull. The later, pre- Pleistocene fossils of Eumeces are dis- tributed as follows: early Miocene of Florida (Estes, 1963), Miocene of Morocco Bulletin Museum of Comparative Zoology, Vol. 139, No. 3 (Hoffstetter, 1961), Mio-Pliocene of Ne- braska (Estes and Tihen, 1964), and the late Pliocene of Kansas (Taylor, 1941, and Twente, 1952). PHYLOGENY OF THE FOUR SUBFAMILIES OF SKINKS Although the fossil record is thus of little value in elucidating the phylogeny of skinks, much can be inferred from the morphology and distribution of living spe- cies. The strength of this approach to the phylogeny of skinks, or any other group without a fossil record, for that matter, is, of course, only as sound as the reasons given for believing that one particular character state is historically antecedent to other alternative and contemporaneous character states. For the present, I will limit the discussion to the four subfamilies of skinks. The phylogenetic relationships of lower level taxa will be considered separately in papers on those taxa. The fusion and loss of bones in the skull, the reduction and loss of appendages, and the acquisition of live-bearing habits are general trends in vertebrate evolution and are accepted here as advanced character states for skinks. Thus the fusion of the nasals and frontals,! the loss of the jugal and postorbital bones, and an incomplete supratemporal arch are advanced char- acters, as are complete limblessness and a live-bearing mode of reproduction. Other clear evolutionary trends among ‘That paired frontals are antecedent to the single frontal in skinks is supported by evidence from the ontogenetic development of this bone in lygosomine skinks. In the embryos of the live- bearing lygosomines I have examined (Eumecia anchietae, Hemiergis tridactylus, Leiolopisma ele- gantoides lobula, Mabuya_ lacertiformis, Spheno- morphus australe, S. concinnatus, and S. quoyi; also Lygosoma sp. according to Pearson, 1921), there are two centers of ossification in the de- velopment of the single frontal of the adult. The frontals remain separated in these embryos until quite late in development (squamation and color pattern fully developed) but have fused by the time of hatching. squamates, such as the loss of pterygoid teeth and the loss of an external ear open- ing, also help in reconstructing the phylog- eny of skinks (Fig. 7). The formation of the bony secondary palate in the four subfamilies of skinks offers further clues to phylogeny. First, it should be pointed out that with the ex- ception of Anelytropsis and Dibamus, the bony secondary palate of skinks is unique among squamates and has surely been de- rived from the primary palate of other lizards. The secondary palates of the Feylininae and Acontinae are quite different from one another, and both palatal types are also quite complex, suggesting that they have been derived independently from some less complex palate. Of the two remaining subfamilies of skinks, the scincine secondary palate ap- pears to be more primitive than and an- cestral to the lygosomine palate. The basic differences between the palates of the two taxa are that, in general (the exceptions will be discussed below), the scincines have the palatine bones apposed but not meeting on the midline, and the palatal rami of the pterygoids are widely sepa- rated, whereas the lygosomines have at least the palatines meeting medially, and in some groups the palatal rami of the pterygoids meet as well. Three lines of evidence indicate that the sequence of palatines and pterygoids not meeting medi- ally (general scincine condition), palatines but not pterygoids meeting medially, and palatines and pterygoids both meeting medially (the two general lygosomine con- ditions) is in fact probably the actual de- velopmental sequence in the evolution of a complete secondary palate in skinks. (1) To derive the complete secondary palate of scincids from the non-scincid squamate primary palate, one would ex- pect a priori that a proto condition to the medial contact of the palatines and ptery- goids of the complete secondary palate would be the progressively closer medial SUBFAMILIES OF SKINKS « Greer 175 apposition of these bones, instead of the construction of a complete secondary palate in one macromutation. It would also be functionally more efficient first to appose the more anterior bones of the primary palate (the palatines) before the more posterior bones (the pterygoids) were in- corporated; that is, it is difficult to imagine the efficiency of a secondary palate con- sisting of the pterygoids in contact medially but the palatines widely separated. (2) The development of a secondary palate in different groups in the fossil record, e.g., turtles and crocodilians, in- volves the progressive incorporation of suc- cessively posterior bones of the primary palate. (3) The ontogenetic development of a complete secondary palate in lygosomines involves first the medial closure of the palatines, followed by the closure of the palatal rami of the pterygoids. The close correlation between the closure of the palatines on the midline of the secondary palate and the fusion of the frontals is the primary justification for recognizing the lygosomines as a distinct taxon of skinks. And in that the divided frontals and the separated palatine bones in the secondary palate of the scincines are primitive characters, the lygosomines must be considered an advanced group derivable from the scincines. Those few lygosomines which have the palatines not quite meeting along the midline of the palate and those few scincines which do have the palatines and sometimes the pterygoids meeting along the midline of the palate do no damage to the foregoing outline of the phylogeny of the scincines and lygosomines. Thus the lygosomines in which the pala- tines do not quite meet medially (most Egernia and Corucia zebrata) may be either very primitive lygosomines, in which the palatines have never met medially, or they may be advanced lygosomines, in which the palatines are secondarily sepa- rated. Although the two genera to which “IDW YsD|S YJ Jo sapis 4yBi pud 4ya] ayy UO sasayjuaiod Ul uaAlB a1 UOXDY yopa Joy saiseds pup psauab jo Jaquinu ey) *(4) yslajso uD Aq pajodipul 21D sayH4s JaJDIDYD paaliaq “SyUIYS JO SaljlWOYqns Inoy ayy yo AuabojAyd ooyaujodAy “7 aunbi4 Buljeaq art] pue Bulhe| bbe (% pg) salgads ysow ul juasaid squit| Aj je} paw Huljaaw you Ajjesauab splob{4ayd ‘Kyjelpaw Huyjyeew you Ajjesauab yng pasodde sauijejed ‘ayejed Asepuooes pasojd Jo uado aAcosb s,|ax9eW juasqe Jo juasasd UYjee} plobAsayd ( e4auab payed Ajasoja omy ul ydaoxe) juasadd jeonf yuasge Jo Juasasd jeyqsoysod ajajdwod yose jesodwayesdns yOUNISIP sjeseu JOUI}SIP S]e}U0J (081 / +82) JVNIONIOS Bulletin Museum of Comparative Zoology, Vol. 139, No. 3 , PUlJeaq AAl| . DULJeaq dA] » }Uasge squil| « UaSge Squil| « (}Xa} 99S ) Xajdwioo ayejed Asepuosas Huljeaq aAlj pue BulAe; ba « (}X8} 99S) Xajdwoo ayejed Asepuoras » pasojd aA00IH S,jaxIeW (% 66) salseds jo Ajsofew ul yuasaid sqwil| uado saoosh s,jaxda\y . Juasge jaa} plobAJayd » Ajjeipaw jaaw . Juasge Yee} plobAsayd juasaud jebn{ uayo splobAsayd ‘Ajjeipaw j9aw « Juasge jebn/ « Juasqe jeyqsoysod Ajjesauab saurljejed sazejyed Ajepuodas « Juasge eyquoysod , ajajdwoou! Ajjesauab yose jesodwayesdns pasojd 40 uado aAooib s,jayIaW ajajdwod yose jesodwajeidns yOUNSIP sjeseu juasqe 40 yUasaid Yyy90} plobAuayd « pasny S|eseu JOUISIP S]|e}UOAJ yuasaid jebnf JOUIISIP S]E}UO!Y juasqe JO JUasaid jeyquoj}sod (41 /€) AVNILNOOV ajyajdwoo yose jesodwazedns (p/2 40 |) SVNINITA4 JOUIYSIPD S|eseu « pasny sjeyuol 176 ( +009 /09 - 0h) JVNIWOSODAT these aberrant species belong are quite closely related, the diversity in morphology and behavior displayed by these species (within the group formed by the two genera) leads me to view them more as distantly related end products, in which the palatines have secondarily separated, than as a closely knit, basal lygosomine stock retaining the primitive scincine palate. Similarly, the few scincines with only the palatines meeting medially ( Proscelotes and Scelotes) appear to be a monophyletic group and could be viewed, on the basis of this character, as being either immedi- ately ancestral to the lygosomines or inde- pendently specialized scincines. Although it is difficult to make a decision between these two hypotheses, my feeling is that the latter hypothesis is correct. It seems fairly clear, however, that the three endemic scincine Seychelles “Scelotes’ and the Mauritian scincine Gongylomorphus bojeri have, as a group,! independently evolved a very advanced, complete secondary palate involving both the palatine and pterygoid bones. If this were not the case, and these four species were to be considered as ancestral to the lygosomines, then we would have to look upon those lygosomines with the palatines and pterygoids meeting medially as being primitive, and the lygosomines with only the palatines either meeting medially or separated as advanced. Such a hypothetical developmental sequence, however, has no evidence whatsoever to support it and is, in fact, refuted by the ontogenetic and fossil evidence discussed above. To be- lieve this hypothesis would require us to throw out the only evidence we have on the evolution of the secondary palate in 1On the basis of other characters, as well as the relationships of the bones of the palate, the endemic Seychelles “Scelotes” and the Mauritian Gongylomorphus bojeri appear to be each other's closest relatives and form a monophyletic group. This relationship and its interesting zoogeographic implications will be discussed elsewhere. WaT SUBFAMILIES OF SKINKS * Greer skinks and to accept the hypothetical alter- native purely on faith.” It would seem, then, that a complete secondary palate has evolved from an in- complete secondary palate at least twice and perhaps three times in skinks: once in the lygosomines (concomitant with the fusion of the frontals) and once or twice in the scincines (depending on the as yet unanswered question of whether the complete secondary palate in the Prosce- lotes-Scelotes group and the Seychelles “Scelotes’-Gongylomorphus bojeri group is due to relationship or convergence ). It thus seems fairly clear that scincines are ancestral to lygosomines, but we have yet to place the feylinines and acontines in the phylogeny of the subfamilies. These two groups are highly specialized bur- rowers and are unlikely to have been an- cestral to any major group of skinks living today. Their divided frontals and incom- plete secondary palates align them much more closely with the scincines than with the lygosomines. This notion is further supported by the fact that both acontines and feylinines are limbless, and it has been the scincines more than the lygosomines that have tended to lose the limbs entirely. The secondary palates of the acontines and feylinines are extremely complex and extremely unlike each other, which makes it seem very probable that the two taxa arose independently from a scincine an- cestry. It is difficult to distinguish the scincine relatives of the acontines, but the scincine Typhlacontias and Fitzsimonsia, 21T intend to refute this hypothesis only as a broad explanation of the evolution of the second- ary palate in skinks. Minor “reversals” in the general trend from an incomplete secondary palate to a complete and ever more extensive secondary palate might be expected and would not be strong enough evidence, in my mind, to offset the onto- genetic and fossil evidence in favor of this general trend. Indeed, as suggested above, it seems pos- sible that such a minor “reversal” in the general trend is what we see in the incomplete secondary palates of a few lygosomines (most Egernia and Corucia zebrata). 178 with their peculiar stapes-quadrate articu- lation (see page 163), and the absence of the jugal bone, are similar to the fey- linines. However, this similarity may well be the result of convergence (both taxa are burrowers) rather than relationship. The data discussed in this section are summarized in the phylogenetic tree of Figure 7. ZOOGEOGRAPHY OF THE MAJOR TAXA OF SKINKS The zoogeography of the major taxa of skinks can be readily understood on the basis of the morphological and distribu- tional data for the many living and very few fossil species summarized in the pre- ceding sections of the paper. Basic to the following discussion is the idea developed above that the scincines have given rise independently to the other three major groups of skinks, the Fey- lininae, Acontinae, and the most speciose and morphologically “advanced” group of skinks, the Lygosominae. The present distribution of the four subfamilies of skinks seems to support this broad phylo- genetic hypothesis. With the exception of Eumeces (the largest genus in the subfamily, 46 species ) and the monotypic Neoseps of Florida, the Scincinae are entirely Old World in dis- tribution and, again with the exception of the widespread Eumeces, show a relict distribution in south central and eastern Asia (Fig. 4). For example, the only scincine, with the exception of Eumeces, in eastern Asia is Brachymeles (13 species ) in the Philippines. As one moves west through Asia, no other scincines are en- countered until, on reaching India, the monotypic Barkudia is known from the regions around Chilka Lake and Calcutta. Further south in India there is a single species of Sepsophis in the central and southern part of the subcontinent and two endemic genera, Nessia (8 species) and Chalcidoseps (1 species), on Ceylon. It is not until one reaches southwest Asia Bulletin Museum of Comparative Zoology, Vol. 139, No. 3 and the Mediterranean area that one en- counters widely distributed genera with many species, e.g., Ophiomorus (9 species), Scincus (12 species), and Chalcides (14 species ). It is south of the Sahara Desert in Africa, Madagascar, and the islands of the western Indian Ocean that the scincines become an important part of the skink fauna (76 of the 136 species of non- Eumeces scincines occur in this area). Two of the other three subfamilies are also found in subsaharan Africa. The Acontinae, with approximately 15 species, and the Feylininae, with 4 species, are undoubtedly derived from scincines in Africa; this, along with the present large number of species and their extensive distribution in subsaharan Africa, Madagascar, and the islands of the western Indian Ocean, ap- pears to indicate that the scincines have been in subsaharan Africa for much, if not most, of their evolutionary history. Whether the scincines were ever in the Australian Region is an interesting ques- tion. The furthest east scincines range in the Old World today is the Philippines (Brachymeles, 13 species). It is, of course, possible that the scincines have been com- pletely replaced in the Australian Region by the lygosomines, although the total absence of any scincine relicts in Australia or the numerous island groups of the Region makes me believe that the scincines never reached this part of the world. The reasons for the relict distribution of the scincines in south and east Asia and their abundance in southwest Asia, Africa, and Madagascar are undoubtedly complex but may be due in part to the evolution and radiation of the Lygosominae in south- east Asia and the Australian Region. The lygosomines are clearly derived from scincines and are morphologically the most advanced skinks. This group is most numer- ous and diverse in southeast Asia and the Australian Region, and its expansion from this area of origin may account in part for the relict distribution of the scincines in south and east Asia. In southwest Asia, Africa, and Madagascar, the area of the Old World furthest from their area of origin, the lygosomines are fairly well represented by species, but they are not morphologically diverse (i.e., there are not many genera). Presumably the lygosomines are only recent arrivals in this area and have not yet swamped their ancestral scincine relatives. Perhaps if we could return in several million years, the scincines would show a relict distribution in Africa, Madagascar, and the islands of the west Indian Ocean as they do in southern and eastern Asia today. The overall geographic picture of skink evolution in the Old World is distinctly bipolar. The scincines appear to have had a long evolutionary history in Africa, giving rise to numerous genera and species as well as to two other subfamilies of skinks, whereas the spectacular radiation of the more advanced lygosomines appear to be predominately a phenomenon of the Aus- tralian Region—an area that was probably never reached by the scincines. The origin of the New World skink fauna is of special interest. In view of the great diversity and abundance of skinks in the Old World, the most remarkable aspect of the New World skink fauna is its paucity. This, plus the fact that three of the four genera in the New World are also wide- spread in the Old World, indicates that the Old and not the New World is surely the ancestral home of the family. Eumeces is represented by 31 species in the New World and 15 in the Old World. The genus has been in North America at least since the late Oligocene (see Fossil Record of Skinks, above) and in that time has successfully rafted to Bermuda (E. longirostris ), but, peculiarly, the group has not spread further south than Costa Rica. Eumeces undoubtedly arrived in the New World via a Bering Land Bridge. The group is very primitive even for scincines! ! Morphologically Eumeces is very possibly the most primitive living skink taxon and may, in fact, SUBFAMILIES OF SKINKS ° Greer 179 (Greer, in preparation), and its distribu- tion along the northern periphery of the range of skinks in the Old World (Fig. 5) implies a greater cold tolerance than in most other skinks. In both time and place, therefore, Eumeces would have been in a good position to take advantage of a Bering Land Bridge. The relationships of the New World Leiolopisma with each other and with their supposed Old World congeners is a major unsolved problem in skink systematics. For the moment I am treating the Leiolopisma of the New World as congeneric with a group of southeast and east Asian Leiolo- pisma. In east Asia, this group ranges as far north as about 41°N lat., which is only slightly further south than the northern limit of the range of Eumeces in Asia (about 45°N lat.). Thus, like Eumeces, these Leiolopisma would be “cold tolerant” enough to have taken advantage of the Bering Land Bridge during slightly warmer times. There is no fossil record for Leiolopisma in the New World, so the time of arrival of the group is unknown, but the few species in the New World and their ab- sence from islands like Bermuda may indicate that the group arrived in North America after Eumeces. Like Eumeces, however, Leiolopisma has not entered South America, although it is known as far south as Panama. The fact that both Eumeces and Leiolo- pisma come so close, but fail to enter South America, seems a bit peculiar to me and merits further discussion. The water gap that persisted through most of the Tertiary across Panama and_ northern be quite similar to the ancestor of all skinks. In view of this primitiveness, it might seem peculiar that the group should be so successful—if number of species is acceptable as a criterion for success (Eumeces, with about 46 species, is the most speciose genus of scincines). But the primitiveness discussed here is morphological, and on other characters, such as the maternal care of the eggs, Eumeces shows the greatest advancement of any lizard for which such data are available. 180 Colombia probably aided in excluding both genera from South America, but this can- not be the whole answer, as skinks have few peers among squamates in crossing water barriers and, regardless of the Tertiary water gap, both genera have had ample time since the closure of the isthmus to enter South America (as has apparently been the case with the genus Rana and the two genera of Bolitoglossine salamanders ). Competitive exclusion by resident South American lizards filling niches similar to those filled by Leiolopisma and Eumeces may be of as great importance in explaining the absence of these two genera from South America as is the Panamanian-Colombian water gap. For example, the micro-teiids, which probably arose in South America and have only recently invaded Central America, are very skink-like in their ex- ternal morphology and habits and may be South America’s candidate for the skink niche (along with the endemic Mabuya). The genus Mabuya is currently thought to comprise approximately nine species in the New World (Dunn, 1936), although this estimate may be low. These species are distributed throughout South America, the West Indies, Central America, and Mexico as far north as Veracruz and Colima. The lack of diversity of the New World species may indicate that the group has not been in the New World very long. There are many Mabuya both in Asia and Africa, so it is difficult to decide whether the group arrived from Asia via the Bering Land Bridge or from Africa by over-water rafting. Three bits of evidence make me favor the latter possibility. First, the genus is very good at crossing water gaps, as evidenced by the endemic Mabuya of the Cape Verde Islands, Madagascar, the Seychelles, and Fernando de Noronha and the Mabuya of the West Indies. Second, there are no Mabuya in the southeastern United States, unlike the case with many Asian immigrants (Magnolia, pattlefishes, Ophisaurus, Leiolopisma and Eumeces). And third, there are endemic South Ameri- Bulletin Museum of Comparative Zoology, Vol. 139, No. 3 can and West Indian species of other African lizard genera (Lygodactylus, Ta- rentola) whose ancestors almost surely arrived in the New World across the Atlantic. Also, Dunn (1936) thinks that “the nearest relationship of the mainland and Caribbean forms [of Mabuya] seems to be with the mainland African species of the raddoni-affinis group.” The fourth group of New World skinks is the monotypic scincine genus Neoseps of peninsular Florida. Neoseps is an attenuate burrower in sandy loamy soil and presents no great zoogeographic problem, as it has probably simply evolved from a Eumeces ancestor in situ (Telford, 1959). Not only is there a close morphological similarity between the two genera (Greer, personal observation), but they are also the only skinks yet known in which the female guards the eggs. ACKNOWLEDGMENTS A broad study covering the osteology and external morphology of the higher taxa of scincid lizards would be impossible without the cooperation and encourage- ment of many people and institutions. Foremost among these are the curators of the major herpetological collections of the world who have very generously sup- ported this study with the loan of important and, in many cases, extremely poorly- known species. In this regard I would like to extend special words of thanks to Miss A. G. C. Grandison of the British Museum (Natural History) (BM), Drs. Richard G. Zweifel and Charles M. Bogert of the American Museum of Natural History (AMNH), Dr. Robert F. Inger and Mr. Hymen Marx of the Field Museum of Natural History (FMNH), Dr. Jean Guibé of the Muséum National d'Histoire Natur- elle (PM), and Dr. James A. Peters of the United States National Museum (USNM). Other curators and_ institutions have been as generous as their collections have permitted and in many cases have supplied crucial specimens available nowhere else. For these considerations I would like to thank Drs. Alan E. Leviton and Steve C. Anderson of the California Academy of Sciences (CAS), Dr. W. D. Haacke of the Transvaal Museum, Dr. Donald G. Broad- ley of the Umtali Museum, Drs. Charles F. Walker and Arnold G. Kluge of the Uni- versity of Michigan Museum of Zoology (UMMZ), Dr. Edward H. Taylor of the University of Kansas, Dr. Hobart M. Smith of the University of Illinois Natural History (UINHM), Dr. William Duellman of the University of Kansas Natural History Museum, Dr. Jay M. Savage of the Uni- versity of Southern California, Professor George S. Myers and Dr. Warren C. Freihoffer of the Division of Systematic Biology, Stanford University (SU), and Dr. Konrad Klemmer of the Senckenberg Museum. I would like to express my special thanks to Dr. Ernest E. Williams of the Museum of Comparative Zoology (MCZ) for mak- ing available to me the extensive skink collections under his charge, and for ex- pending much time and energy in helping me to procure specimens. Dr. Williams’ interest in this work has been a constant source of encouragement and_ helpful criticism. Dr. Williams, along with Dr. Richard Estes of Boston University, Dr. Samuel B. McDowell of Rutgers, and Dr. Richard Zweifel of the American Museum of Natural History have read the paper, but the responsibility for the accuracy of the factual content and conclusions is solely mine. Mr. Arnold Clapman drew the skulls for Figures 1 and 2. LITERATURE CITED Aut, S. M. 1947. The dermal scutes of Mabuya dissimilis Hallowell. Curr. Sci. 16: 348. ANGEL, F. 1942. Les lézards de Madagascar. Mémoires de Académie Malgache 36: 1- 193. BOULENGER, G. A. in the British 1887. Catalogue of the lizards Museum (Natural History). Greer 181 SUBFAMILIES OF SKINKS ° Vol. II, 2nd ed., London, Taylor and Francis, xii + 575 pp. Brock, G. T. 1941. The skull of Acontias mele- agris, with a study of the affinities between lizards and snakes. J. Linn. Soc. (Zool.) 41: 71-88. Brtui, C. B. 1886. Zootomie aller Thierklassen fiir Lernende, nach Autopsien, skizzert. Lief. 38. Vienna, Alfred Holder. Corr, E. D. 1892. The osteology of the Lacer- tilia. Proc. Amer. Phil. Soc. 3: 185-221. DERANIYAGALA, P. E. P. 1953. A colored atlas of some vertebrates from Ceylon. Vol. II. Tetra- pod Reptilia. Ceylon National Museums Publication, Ceylon Govt. Press, 101 pp. . 1960. Some osteological characters of the apodal skink Anguicephalus hickanala Der- aniyagala. Spolia Zeylan. 29: 29-32. DumeriL, A., AND M. F. Bocourt. 1881. Mission scientifique au Mexique et dans Amérique Centrale. Part 3(7): 441-488. Dunn, E. R. 1936. Notes on American Mabuyas. Proc. Acad. Nat. Sci. Philadelphia. 87:533- 5b Ex-Tousi, M. R. 1938. The osteology of the lizard Scincus scincus (Linn.). Bull. Fac. Sci. Egyptian Univ. (Cairo) 14:1-38. Essex, R. 1928. Studies in reptilian degeneration. Proc. Zool. Soc. London, 1927, part 4:879- 945. Estes, R. 1963. Early Miocene salamanders and lizards from Florida. Quart. J. Florida Acad. Sci. 26:234-256. . 1964a. Fossil vertebrates from the Late Cretaceous Lance Formation, eastern Wyo- ming. Univ. California Publ. Geol. Sci. 49: 1-180. . 1964b. Notes on some Paleocene lizards. Copeia 1965(1):104—106. Esters, R., AND J. A. TrHeEN. 1964. Lower verte- brates from the Valentine Formation of Nebraska. American Midl. Nat. 72(2):453- 472. Erueriwce, R. 1967. Lizard caudal vertebrae. Copeia 1967(4):699-721. Fircu, H. S. 1954. Life history and ecology of the Five-lined Skink, Eumeces fasciatus. Univ. Kansas Mus. Nat. Hist. Publ. 8(1): 1-156. Ganapati, P. N., AND K. RAJYALAKSHMI. 1958. Bionomics and some anatomical peculiarities of the limbless lizard Barkudia insularis An- nandale. Rec. Indian Mus. 53:279-291. Gossr, P. H. 1848. On the habits of Mabouya agilis. Proc. Zool. Soc. London 16:59-62. Greer, A. E. 1967a. for some Australian scincid lizards. No. 267:1-19. A new generic arrangement Breviora 182 . 1967b. The relationships of the African scincid genus Eumecia. Breviora No. 276: 1-9. Greer, A. E., AND F. Parker. 1967. A new scincid lizard from the northern Solomon Islands. Breviora No. 275:1-20. 5 . 1968. Geomyersia glabra, a new genus and species of scincid lizard from Bougainville, Solomon Islands, with com- ments on the relationships of some lygosomine genera. Breviora No. 302:1-17. Haas, G. 1935. Zum Bau des Primordialcraniums und des Kopfskelettes von Ablepharus pan- nonicus. Acta Zool. 16:409-429. . 1936. Ueber das Kopfskelett von Chal- cides guentheri (Seps monodactylus). Acta Zool. 17:55-74. Hewitt, J. 1929. On some Scincidae from South Africa, Madagascar and Ceylon. Ann. Trans- vaal Mus. 13(1):1-8. Horrstetter, R. 1944. Sur les Scincidae fossiles. I. Formes européennes et nord-américaines. Bull. Mus. Natl. Hist. Nat. (2)16(6):547— 553: 1945. Sur les Scincidae fossiles. II. Formes subfossiles de L’Ile Maurice. Bull. Mus. Natl. Hist. Nat. (2)17(1):80-86. 1949. Les reptiles subfossiles de L’Ile Maurice. I. Les Scincidae. Annales de Paléontologie 35:45-72. . 1961. Squamates. In: Le gisement de vertébrés Miocénes de Beni Mellal (Maroc). Notes Mém. Serv. Géol. Maroc 155:95-101. 1962. Revue des récentes acquisitions concernant histoire et la systématique des squamates. In: Problems actuels de paléon- tologie (évolution des vertébrés). Coll. Intern. Centre Nat. Rech. Scient. 104:243- 279. Kamet, A. E. 1965. The cranial osteology of adult Chalcides ocellatus. Anat. Anz. 117: 338-370. Kinc, D. 1964. The osteology of the water skink Lygosoma (Sphenomorphus) quoyii. Aust. J. Zool. 12:201-216. KINGMAN, R. H. 1932. A comparative study of the skull in the genus Eumeces of the family Scincidae. Univ. Kansas Sci. Bull. 20(15): 273-295. Lovermce, A. 1957. A check list of the reptiles and amphibians of East Africa (Uganda; Kenya; Tanganyika; Zanzibar). Bull. Mus. Comp. Zool. 117(2):153-362. Merwe, N. J. vAN pe. 1944. Die Skedelmorfolo- gie van Acontias meleagris (Linn.). Tydskrif vir Wetenskap en Kuns, vol. number not available: 59-88. Minter, M. R. 1966a. The cochlear duct of Bulletin Museum of Comparative Zoology, Vol. 139, No. 3 lizards. Proc. Calif. Acad. Sci. 33(11):255- 359. . 1966b. The cochlear ducts of Lantha- notus and Anelytropsis with remarks on the familial relationship between Anelytropsis and | Dibamus. Occ. Pap. Calif. Acad. Sci. No. 60: 1-15. MircHELL, F. J. 1950. The scincid genera Egernia and Tiliqua (Lacertilia). Rec. South | Aust. Mus. 9(3):275-308. MiITTLEMAN, M. B. the lizards of the subfamily Lygosominae. Smithsonian Misc. Coll. 117(17):1-385. OuivEr, J. A. 1951. Ontogenetic changes in osteo- dermal ornamentation in skinks. Copeia 1951 (2):127-130. Ortro, H. 1908. Die Beschuppung der Brevilin- guier und Ascalaboten. Jena. Zeitschr. Natur- wiss. 54(37):1-61. Pearson, H. S. 1921. The skull and some related structures of a late embryo of Lygosoma. J. Anat. London 56:20-43. Peters, W. C. H. 1882. Reise nach Mossarnbique. Zoologie. III. Amphibien. lin, x + 191 pp., 26 pls. RasBanus, K. 1911. Uber das Skelett von Voeltz- kowia mira Bttgr. In: A. Voeltzkow, Reise in Ostafrika. Wiss. Ergebn. 4:279-330. Rao, M. K. M., anp L. S. RAaMaswamti. 1952. The fully formed chondrocranium of Mabuya with an account of the adult osteocranium. Acta Zool. 33:209-275. Ratuore, M.S. 1967. The bionomics and anat- omy of Ophiomorus tridactylus (Blyth) Boul- enger. Ph.D. Thesis, University of Rajasthan, Jaipur, India. Rice, E. L. 1920. The development of the skull in the skink Eumeces quinquelineatus L. J. Morph. 24:119-220. Romer, A. S. 1956. Osteology of the reptiles. Chicago, Univ. Chicago Press, xxi + 772 pp. SipTAIN, S. M. 1938 Studies on the caudal autot- omy and regeneration in Mabuya dissimilis Hallowell. Proc. Indian Acad. Sci. 13(1): 63-78. SIEBENROCK, F. 1892. Zur Kenntniss des Kopf- skelettes der Scincoiden, Anguiden und Ger- rhosauriden. Ann. naturhist. Mus. Wien 7 (3):163-196. SmitH, M. A. 1935. The Fauna of British India, including Ceylon and Burma. Reptilia and Amphibia. Vol. II. Sauria. London, Taylor and Francis, xii + 440 pp. 1937. A review of the genus Lygosoma (Scincidae: Reptilia) and its allies. Rec. Indian Mus. 39(3):213-234. Taytor, EF. H. 1941. Extinct lizards from Upper Pliocene deposits of Kansas. Univ. Kansas Publ., State Geol. Surv. Kansas, Bull. 38, Rep. of Studies, Pt. 5:165—-176. 1952. A generic synopsis of | SS G. Reimer, Ber- | TELForD, S. R. 1959. A study of the sand skink, Neoseps reynoldsi Stejneger. Copeia 1959 (2):110-119. Tirak, R., anv S. C. Rastocr. 1964. Dermal scutes in Sauria (Reptilia). Proc. Zool. Soc. Calcutta 17:183-191. TWENTE, J. W. 1952. Pliocene lizards from Kan- sas. Copeia 1952(2):70-73. Vintuiers, C. G. S. pE. 1939. Uber den Schidel des siidafrikanischen schlangenartigen Scin- ciden Acontias meleagris. Anat. Anz. 88: 320-347. Waite, E. R. 1929. The reptiles and amphibians SUBFAMILIES OF SKINKS « Greer 183 of South Australia. Adelaide, Harrison Weir, Govt. Printer, 270 pp. Witte, G. F. pe. 1953. Exploration du Parc National de PUpemba. Reptiles. Institut des Parcs Nationaux du Congo Belge, Fasc. 6: 1-322, Wirter, G. F. pE, AND R. Laurent. 1943. Con- tribution a la systématique des formes dé- gradées de la famille des Scincidae ap- parentées au genre Scelotes Fitzinger. Mém. Mus. Roy. Hist. Natur. Belgique, sér. 2, fasc. 26:1—44, (Received 17 May 1968.) New Species of Bottom-Living Calanoid Copepods Collected in Deepwater by the DSRV Alvin GEORGE D. GRICE AND KUNI HULSEMANN HARVARD UNIVERSITY VOLUME 139, NUMBER 4 CAMBRIDGE, MASSACHUSETTS, U.S.A. APRIL 9, 1970 PUBLICATIONS ISSUED OR DISTRIBUTED BY THE MUSEUM OF COMPARATIVE ZOOLOGY HARVARD UNIVERSITY BuLietin 1863-— Breviora 1952- Memorrs 1864-1938 JoHNnsoniA, Department of Mollusks, 1941- OccasIONAL Papers ON Mottusks, 1945- a i - Other Publications. Bigelow, H. B., and W. C. Schroeder, 1953. Fishes of the Gulf of Maine. Reprint, $6.50 cloth. Brues, C. T., A. L. Melander, and F. M. Carpenter, 1954. Classification of In- sects. $9.00 cloth. Creighton, W. S., 1950. The Ants of North America. Reprint, $10.00 cloth. 4 Lyman, C. P., and A. R. Dawe (eds.), 1960. Symposium on Natural Mam- malian Hibernation. $3.00 paper, $4.50 cloth. Peters’ Check-list of Birds of the World, vols. 2-7, 9, 10, 12, 15. (Price list on request. ) Turner, R. D., 1966. A Survey and Illustrated Catalogue of the Teredinidae oy (Mollusca: Bivalvia). $8.00 cloth. Fi Whittington, H. B., and W. D. I. Rolfe (eds.), 1963. Phylogeny and Evolution of Crustacea. $6.75 cloth. a Proceedings of the New England Zoological Club 1899-1948. (Complete sets 4 only. ) Publications of the Boston Society of Natural History. Publications Office Museum of Comparative Zoology Harvard University Cambridge, Massachusetts 02138, U. S. A. © The President and Fellows of Harvard College 1970. NEW SPECIES OF BOTTOM-LIVING CALANOID COPEPODS COLLECTED IN DEEPWATER BY THE DSRV ALVIN’ GEORGE D. GRICE? AND KUNI HULSEMANN? ABSTRACT The use of a deep sea submersible for collecting near bottom copepods is de- scribed. Thirteen new species of calanoid copepods were found in a plankton sample collected within 30 cm of the bottom at a depth of approximately 1800 meters on the continental slope south of Woods Hole, Massachusetts. The species are described and illustrated. INTRODUCTION Several families of calanoid copepods have species which live in proximity to the bottom. These species, termed planktoben- thos by Hutchinson (1967), are not usually collected in abundance by ordinary plank- ton collecting techniques, as it is not prudent to permit a towed plankton net to get very near to the bottom, especially in deepwater, where it might be fouled, torn or lost. Specialized collecting apparatus have been devised for collecting animals that live near the seabed. Matthews (1964) described three types of gear that he used to sample bottom-living calanoid copepods 1 Contribution No. 2144 of the Woods Hole Oceanographic Institution, Woods Hole, Massa- chusetts, 02543. This study was supported by National Science Foundation Grants GB-6052, 8612, and in part by the Office of Naval Research, Nonr-3484(00). * Woods Hole Oceanographic Institution, Woods Hole, Massachusetts, U.S.A. * Scripps Institution of Oceanography, La Jolla, California, U.S.A. Bull. Mus. Comp. Zool., 139(4): 185-230, April, 1970 which he referred to as epibenthos, at a depth of 240 m south of Bergen, Norway. In addition to Matthews’ methods, plankto- benthos have also been collected by at- taching a tow net above a bottom dredge or to a bottom trawl as Farran (1905), for example, did in his investigations of the copepods of the slope area off Ireland. Farran’s deepest bottom plankton collec- tion was obtained by a trawl in 382 fath- oms. Frolander and Pratt (1962) have described a “bottom skimmer” which they used for collecting planktobenthos at a depth of 40 feet in a lake. The recent acquisition of the Deep Sub- mergence Research Vehicle ALVIN by the Woods Hole Oceanographic Institution provided us with a means to sample plank- tobenthos with greater precision and in greater depths than we believe to have been hitherto practical. With plankton nets attached to ALVIN as shown in Figures 1 and 2, the pilot, by visual ob- servation, can keep the nets just above the bottom for prolonged sampling periods in depths down to approximately 1900 m, the maximum operational depth of ALVIN. SAMPLING PROCEDURE In the initial attempt to sample the planktobenthos from ALVIN, samples were collected by means of two nets attached to the submersible. The mouth opening of the nets (.233 mm aperture size) were “D” shaped with the straight side fastened 185 186 Bulletin Museum ‘ i = cS Figure 1. Position of plankton nets during descent of ALVIN. vertically to a hinge located on the forward end of the submersible’s collecting basket (Figs. 1, 2). During the descent of the submersible, the mouth was held sideways (Fig. 1) in order to reduce the possibility of obtaining plankton while on the way to the bottom. Upon reaching the bottom, the mechanical arm of ALVIN was used to swing the mouth of one net into the sam- pling position (Fig. 2) and to hold it there for the duration of the sampling period. of Comparative Zoology, Vol. 139, No. 4 The net was kept within 30 cm of the bottom while the submersible cruised at approximately 11/2 knots. At the end of one hour the mouth was allowed to swing back to its original position and the pursing line was tightly drawn around the net by means of the mechanical arm. The other net was then opened and a sample col- lected in the same manner as described for the first net. The collections were obtained by Dr. Howard Sanders and his associates. New SpeEcIes OF CALANOID Copepops « Grice and Hiilsemann Figure 2. One plankton net in sampling position. DESCRIPTION OF COLLECTING AREA The collections were made south of Woods Hole, Massachusetts, at 39°45.2’N 70°33.8’W on September 19, 1967 (ALVIN Dive 220). The water depth in the area varied between 1750 and 1822 m and the temperature was 3°C. The bottom was composed of fine sediment with a floc- culent zone at the surface which was readily stirred up. A slight current was noticeable near the bottom. 187 RESULTS Since the two samples came from ap- proximately the same depths and since the bottom topography appeared homogeneous in the area, the two are here treated as one. Sixty-five species of calanoid copepods were identified, including 13 new species that will be described below. These 13 species with the possible ex- ception of Aetideopsis magna, probably live in proximity to the seabed. From 188 dredge collections made near Norway, Sars (1902) described bottom-living species of Bryaxis (= Comantenna), Diaixis, Thary- bis, and Xanthocalanus, the first three being established as new genera. Except for Tharybis, we encountered undescribed species referable to these genera. Tharybis is now considered to be one of three genera in the Family Tharybidae. We found un- described species of the other two genera of Tharybidae, Undinella and Parundinella. Comantenna is closely associated with the seabed (Matthews, 1964), as are some species of Xanthocalanus (Sars, 1925). Species of Amallophora, of which we found two undescribed ones, have not been taken in abundance in mid-water plankton samples. Perhaps this genus, too, has bot- tom affinities. Representatives of typically planktonic species were also noted and include, for example, Calanus finmarchicus, C. hyper- boreus, Eucalanus elongatus, Clausocalanus furcatus, Microcalanus pygmaeus, Pseudo- calanus elongatus, Temoropia mayumbaen- sis, Metridia curticauda, Pleuromamma robusta, Centropages typicus, and Acartia danae. Some of these may have entered the net during the descent of the sub- mersible. As may be noted in Figure 2, the net is folded back (not pursed) during this phase of the collection. However, several of these very same species together with some others were reported from samples collected just over the sea floor by Matthews (1964) in his study on bot- tom-living copepods off western Norway. Type specimens have been deposited in the U. S. National Museum. FAMILY AETIDEIDAE Aetideopsis magna n. sp. Pl. I, figs. 3-19 Material examined: 6 males Diagnosis (male). Head and _ first thoracic segment fused, fourth and _ fifth thoracic segments partially fused. Abdomen consisting of five segments, second segment Bulletin Museum of Comparative Zoology, Vol. 139, No. 4 largest, anal segment very short. Rostrum 2-pointed, small. Posterolateral margin of fifth thoracic segment more or less rounded with posteriorly directed small spine. Antennule reaching third thoracic segment. In right antennule segments 8 and 9, 20 and 21 fused; in left antennule segments S and 9 only fused. Large aesthetasks on segments 2 to 9. Exopod of antenna shorter than endopod. Basal segment of mandible palpus devoid of setae. Mandible blade rudimentary. Maxillule and maxilla re- duced. Second basipodal segment of maxil- lipod slender, slightly longer than first. Endopod of first leg 1-segmented, of second leg 2-segmented, of fourth leg 3- segmented. Exopod of first leg 3-seg- mented, of other legs broken short. Pos- terior side of second and fourth legs with patches of short hair. Fifth legs asym- metrical, left side longer than right. Endo- pods slender, 1-segmented. Exopod of right fifth leg ending with slender point. Second exopodal segment of left fifth leg with setae distally, third segment small, with hair at tip. Total lengths 4.20-4.56 mm. Holotype No. 125135. The female is unknown. Remarks. Aetideopsis magna resembles A. multiserrata (Wolfenden), but is dis- tinguished by its larger size, the relatively shorter spine in the fifth thoracic segment, and the longer left fifth leg. No female belonging to this genus was found in the sample. The name magna refers to the relative size of this species. Comantenna recurvata n. sp. Pl. I, figs. 20-24; Pl. Il, figs. 25-35 Material examined: 4 females Diagnosis (female). Head and _ first thoracic segment fused. Fourth and fifth thoracic segments separate. Abdomen 4- segmented, anal segment small. Rostrum absent. Posterolateral corners of cephalo- thorax pointed and curved upward. EARLY ARCHOSAURIAN EVOLUTION Mi “yy ——— — Figure 1. Lower Triassic of South Africa); and Ela- phrosuchus (1 species, from the Lystrosau- rus Zone, South Africa). The Erythrosuchidae includes the fol- lowing genera: Garjainia (Fig. 2) (1 species, from the Russian Zone V, lower- most Triassic); Erythrosuchus (1 species, from the Cynognathus Zone, late early Triassic, South Africa); Vjushkovia (Fig. 4) (1 species, from the Russian Zone VI, late early Triassic); and Shansisuchus (1 Dorsal view of the skull of Chasmatosaurus vanhoepeni Haughton. - Reig 235 (From Broili and Schroder.) or 2 species, from the Chinese Ermaying Series, late early Triassic). Cuyosuchus (1 species, Cacheuta beds, Lower Triassic, Argentina) must be con- sidered as Proterosuchia incertae sedis, as the material is not sufficient for family allocation. Ankistrodon, Arizonasaurus, Dongusia, Seemania, and Ocoyuntaia are generic names applied to material that may prove to be referable to the Protero- suchia, but which must be considered 236 Bulletin Museum of Comparative Zoology, Vol. 139, No. 5 \ [_ 1OCM Figure 2. nomina dubia for the present because the specimens are extremely fragmentary. As these last remarks imply, not all the above-mentioned genera are really well known, and some are based on material too incomplete for adequate knowledge of all relevant characters. All evidence considered, however, we have a_ fairly good knowledge of at least the genera Chasmatosaurus, Erythrosuchus, Vjush- kovia, Shansisuchus, and Cuyosuchus, from all of which a good part of the postcranial skeleton is known. The other genera that permit family allocation are known from less complete material. They are very use- ful, however, either to infer phylogenetic conclusions, as in the case of Elaphrosuchus and Garjainia, or to improve knowledge of the temporal and geographical distribution of the groups concerned. Nevertheless, we must admit that we know only a very small part of the actual proterosuchian array, and this must be carefully kept in mind when discussing early archosaur evolution. It must be taken for granted that many _ proterosuchians existed that are at present unknown, and Lateral view of the skull of Garjainia prima Ochev. (From Ochev.) that among them might lie the direct ancestors of later archosaurs, which are not easily to be detected among the forms we know at present. This kind of assump- tion is the very basis of paleontological inference. THE INTENSION OF THE PROTEROSUCHIA-CONCEPT The Proterosuchia are such a puzzling group that von Huene was inclined, in one of his first works (1911), to place one of the included genera, Erythrosuchus, in an order of its own, sharing pseudosuchian and pelycosaurian features. As stressed by Hughes (1963), they combine some truly — archosaurian peculiarities in the skull and — other parts, with primitive, non-archosau- | rian characteristics in the limbs and girdles. | As we shall see below, some non-archo- saurian features are also present in the | skull structures. Hughes made a careful analysis of the peculiarities of the Proterosuchia, but he | emphasized primarily postcranial morphol- | Romer (1956, 1967), on the other | ogy. hand, pointed out the significance of very Figure 3. Cervical vertebrae and ribs of Chasmatosaurus vanhoepeni Haughton. (From Broili and Schroder.) peculiar proterosuchian skull characters, neglected by Hughes and other authors. Charig and Reig (in press) list the state of many characters in this taxon, but they do not discuss thoroughly their evolution- ary significance. A further analysis, there- fore, seems necessary. Statement and analysis of the proterosuchian character-states Following Sokal and Sneath (1963), we shall use the character-state terminology in our present analysis. For these authors, a character is a variable that can occur in different states from one kind of organism to another. These character-states are the relevant features that taxonomists deal with in comparing different taxa. For instance, “dermal ossifications” is a char- acter, and “dermal ossifications absent” is a character-state. Since they belong to a taxon of higher rank, the subclass Archosauria, the Protero- suchia have a set of character-states shared by all archosaurs. We shall refer to this set of character-states as the “All-Archo- saurian set of character-states” (AA). This AA set represents the intension of the taxon-concept Archosauria, and should not afford a relevant basis for elucidating the concept of Proterosuchia, though its assess- ment is very important to support the EARLY ARCHOSAURIAN E\VOLUTION * Reig 237 Figure 4. Lateral view of the pelvis of Vjushkovia_ tripli- costata von Huene. (From von Huene.) inclusion of the Proterosuchia in the Archo- sauria and for an enquiry regarding the origin of the whole subclass. The following list includes the character-states that we consider as belonging to this set: i) Two-arched skull (diapsid condition ) ii) Antorbital fenestra present iii) Mandibular fenestra present iv) Laterosphenoid ossified v) Skull metakinetic vi) Quadrate-squamosal articulation move- able vii) Supratemporal and tabular bones absent ) Posttemporal fenestrae small ) Vertebrae not notochordal ) Ribs with capitulum and tuberculum ) Rib facets of dorsal vertebrae on transverse processes, becoming closer to a complete fusion posterad Capitular facets for cervical ribs situ- ated well anteriorly and ventrally on the centrum; tubercular facets for the same ribs at the tip of transverse process Posterior limbs longer than anterior (limb disparity ) xiii ) 238 / ENT TAAT L{ A VAVAGTA / Bulletin Museum of Comparative Zoology, Vol. 139, No. 5 fl (XK (\ ' WS i j yj (RS CSM \ aN N Oh As i mat \ isd. EOE ese }_-___________ 10 cm Figure 5. Lateral view of the skull of Chasmatosaurus vanhoepeni Haughton. (From Broili and Schroder.) Some allegedly characteristic archosau- rian character-states, such as upright stance and bipedalism, are not included in this list. As has been suggested by Charig (1965), they are neither characteristic nor widespread archosaurian features. The core of our discussion should be connected with those character-states that would help to define the Proterosuchia as distinct from other taxa included in the Archosauria. These character-states may be grouped in four different classes: (a) the All-proterosuchian-No-other-archo- saurian set of character states (AN), which includes peculiarities shared only by the proterosuchians, absent in any other archosaurian taxon; (b) the Some-protero- suchian-No-other-archosaurian set (SN), comprising characters that are present in the described state only in some of the proterosuchians, while present in a differ- ent state in other proterosuchians and in all the other archosaurs; (c) the All-protero- suchian-and-Some-other-archosaurian set (AS), including character-states shared by all the members of the extension of the Proterosuchia, but also present in some other non-proterosuchian archosaurs; (d) the Some-proterosuchian-and-Some-other- archosaurian set (SS), referring to those character-states shared by some, but not all the members of the Proterosuchia, and also by some, but not all, archosaurian groups not belonging to the Proterosuchia. The following list attempts to synthesize the relevant character-states of the Protero- suchia. The letters preceding each state- ment refer to the above-defined sets. 1. (AS) A single median postparietal bone present 2. (AS) Small postfrontal bones pres- ent 3. (SN) A small pineal foramen present 4. (AN) A typical otic notch not present 5. (AN) The posterior border of the infratemporal fenestra nearly straight (without the V- shaped contour characteristic of most archosaurs ) 6. (AS) The jaw articulation well be- hind the level of the occiput 7. (AS) Antorbital fenestra of moder- ate size, not opening as a part of a more extended, basin-like depression Nares of moderate size, sub- terminal, fairly well separated from the antorbital fenestra 9. (AS) Pterygoids not meeting in the midline, bordering a long and narrow interpterygoid vacuity extending forward be- tween the vomers Saa(AS) 10. Tale 12. 13. 14. Ig) 20. 22. 23. 24. 25. 26. 27. 28. ee) . (AN) Posterior (SS) Palate with teeth in the ptery- goid flanges Occipital plane rather con- cave, slanting forward to- wards the skull table Prefrontal bones large, pro- jecting laterally to form a ridge that makes an abrupt limit between the roof of the skull and the lateral antorbital region Marginal teeth isodont and acrodont or subthecodont in implantation (SS) Intercentra usually present behind the axis, more com- monly between the cervical vertebrae (AS) Gait quadrupedal (AN) (AS) (AN) . (AN) Propodials horizontal in posi- tion (sprawled stance ) limbs moderately longer than the front ones (primitive limb disparity ) . (AN) Femur bearing a large in- ternal trochanter (AN) Intertrochanteric fossa of the femur present (SS) Humerus with wide and twisted ends . (AN) Pes with mesotarsal ankle joint (proximal tarsals with- out specializations ) (AS) Iliac blade with anterior spine absent or only moderately developed (AS) Posterior expansion of the iliac blade narrow and long (AS) Acetabula completely closed, only moderately excavated, and relatively far apart one from the other (AS) Pubis and ischium compar- atively short (AS) Coracoids large (SN) Scapulae broad and short (AS) Dermal elements of the pec- toral girdle well developed EARLY ARCHOSAURIAN EVOLUTION * Reig 239 29. (AS) Dermal armor of any sort absent From the above list of character-states, interesting conclusions can be drawn, but it is first necessary to make a brief analysis of them. (1) The possession of postparietal bones (Fig. 1) (interparietal, dermosupraoccip- ital) is a primitive condition for reptiles, and is widespread in such primitive groups as the cotylosaurs, the pelycosaurs, the eosuchians, and the millerettids. This character-state is shared by all the genera assigned to the proterosuchia, in the form of an unpaired postparietal. However, this is not an exclusive proterosuchian condition among the archosaurs, as a postparietal is also present in the pseudosuchian theco- dont Euparkeria. (2) Postfrontal bones (Fig. 1) are also present in most primitive reptile groups and in all the proterosuchians so far known. As in the former case, other non-protero- suchian archosaurs retain this primitive state, as postfrontals are present not only in Euparkeria but also in the phytosaurs, the stagonolepidid pseudosuchians, and the rhamphorhynchoid pterosaurs. (3) A pineal foramen is, as far as is known, present only in all the known specimens of the erythrosuchid genus Erythrosuchus, in the primitive erythrosu- chid Garjainia (see Tatarinoy, 1961: 121), and in one of three known skulls of Chas- matosaurus. Other proterosuchian genera either have been reported as not possessing this character, or cannot be checked due to the nature of the material. Among other non-proterosuchian archosaurs, this char- acter is absent, save in one doubtful genus, Mesorhinosuchus (=Mesorhinus auct.), currently considered the only Lower Trias- sic phytosaur. We are also dealing here with a very primitive state of a character, present as such in the earliest reptilian groups. 240 (4) Romer pointed out (1956, 1967) the absence of a typical otic notch in the Proterosuchia. He based his statement on the genera Chasmatosaurus (Fig. 5) and Erythrosuchus. Garjainia (Fig. 2), Shan- sisuchus, and Vjushkovia give support to the same view. The latter genus has indeed been reconstructed by von Huene (1960) as having a well-developed otic notch, but this reconstruction is purely hypothetical and is not supported by the morphology of the surrounding parts. Tatarinov (1961) has indicated that the posterior border of the infratemporal opening was straight in Vjiushkovia, as in Erythrosuchus, a feature correlated, in other proterosuchian genera, with the absence of a defined otic notch. In all proterosuchian skulls, therefore, the construction of the otic region is very primitive. This recalls the pelycosaurian and captorhinomorph condition and differs from all remaining archosaurs and from lepidosaurs (including millerettids and eosuchians, in which a distinct lepidosau- rian otic notch is clearly present). In all non-proterosuchian archosaurs the _ otic notch is clearly defined by a curved pos- terior border of the quadrate and by a projection of the squamosal, which extends posteriorly above the head of the quadrate to form the dorsum of the notch. The character-state “absence of the otic notch” hence belongs obviously to the AN set. (5) Linked with the otic notch is the shape of the posterior border of the infra- temporal fenestra. The V-shaped contour of this border, with the apex of the V facing forward, is common to all the non-protero- suchian archosaurian genera (save those with secondary modifications from a primitive V-shaped condition). In con- nection with the posterior position of the mandibular articulation, the quadrate of the proterosuchians slants sharply backwards. The ascending ramus of the quadratojugal and the descending ramus of the squamosal follow the quadrate in this position. In more advanced archo- Bulletin Museum of Comparative Zoology, Vol. 139, No. 5 saurs, the jaw articulation moved forward, apparently in connection with the develop- ment of a more efficient biting mechanism (Ewer, 1965), and the quadrate acquired a more vertical position. In this position of the quadrate, the V-shape of the quad- ratojugal and squamosal arms is obligatory, and, consequently, room is developed for an otic notch, further enlarged by the backward projection of the squamosal. The proterosuchian condition of this character is again a primitive one, as this is the state shown by the pelycosaurs, especially by the varanopsid pelycosaurs. The assump- tion that this condition is shared by all the proterosuchians is safe, and the same is valid for character-state 4, as it is present both in primitive (Chasmatosaurus) and advanced genera in which the skull is known (Erythrosuchus, Shansisuchus). Therefore, this is to be considered an AN character-state. (6) As far as the position of the jaw articulation is concerned, this character obviously belongs to the same cluster as the two previously described. All the proterosuchian skulls so far known show a backward position of the suspensorium (Figs. 1, 2, 5), the articular condyles for the mandible lying in a line well posterior to the line of the occipital condyle. This condition is distinctly different in the non- proterosuchian archosaurs, save the primi- tive crocodile Proterochampsa and, in a lesser degree, some phytosaurs. Character- state 6 belongs therefore to the AS class. Romer (1967) pointed out that this long- jawed condition is characteristic of very primitive reptiles and is reminiscent of the captorhinomorph skull architecture. In primitive pelycosaurs of the ophiacodont- varanopsid group this character-state is even more pronounced, but both the millerettids and the eosuchians are more progressive in this respect. (7) The presence of an antorbital fe- nestra is a characteristic archosaur char- acter-state. It is safe to consider the condition of the character in the protero- suchians as primitive, as in them the fe- nestra does not reach a large size and, especially, as it does not lie in a depression with sharp borders, as is the case in most other thecodonts and other archo- saurs. Though the function of this fenestra is not completely clear (Ewer, 1965; Walker, 1961), it is obvious that whatever its function may have been, its increase in size, and the development of a basin-like structure to contain it are to be considered as an intensification of the function; the structure was not fully developed in the proterosuchian level of archosaurian evo- lution. The described proterosuchian state of this character seems to be shared by all the known skulls (Figs. 2, 5) referred to this taxon, with Shansisuchus as an atypical example, since this genus has the peculiarity (also present in some saurischian dinosaurs) of having an ad- ditional opening, though not a basin-like depression. Vjushkovia has been restored by von Huene with a great antorbital open- ing, but again this seems clearly to be a quite tentative reconstruction, as most of the borders of the fenestra are not pre- served in the known specimens. The fact is that other, non-proterosuchian, archo- saurs share this state of the character, as is shown in the primitive crocodile Protero- champsa, in the peculiar pseudosuchian Rhadinosuchus (=Cerritosaurus), in Cla- renceia, and in the phytosaurs. This char- acter-state is therefore to be considered as belonging to the AS class. It is indeed very suggestive that an antorbital fenestra, elsewhere only an archosaurian character- state, is present in the varanopsid pely- cosaurs (Olson, 1965, and see also below). (8) The described state of the external nares is shared by all the proterosuchian genera (Figs. 1, 2, 5). More advanced thecodonts usually have the external nares larger and nearer to the antorbital vacuity, or else posterior in position (phytosaurs ). Subterminal, small nares well separated EARLY ARCHOSAURIAN EVVOLUTION * Reig 241 from the antorbital opening are also present in Rhadinosuchus and Clarenceia, and the situation in Euparkeria is best considered reminiscent of the proterosuchian state. This character-state must therefore be grouped in the AS category. (9) This character-state is inferred from the condition in Chasmatosaurus, the only proterosuchian in which the palate is well known. Inasmuch as the same condition is shared in such a probable erythrosuchid- derivative as Euparkeria, it is safe to con- clude that this state was widespread among the proterosuchians. Among other archo- saurs, it is shared not only by Euparkeria, but also by Proterochampsa, so that the character-state must tentatively be con- sidered as belonging to the AS class. (10) The presence of palatal teeth in the pterygoid flanges has been verified in Chasmatosaurus and Proterosuchus among the proterosuchids, but no erythrosuchid has given any evidence of them. Palatal teeth are known among archosaurs, other than proterosuchians only in Euparkeria and in Proterochampsa (Sill, 1967). This state of the character is obviously a primi- tive one, as palatal teeth are present in millerettids, younginids, procolophonids. pelycosaurs, and captorhinomorphs among the primitive groups. It must hence be placed, so far as present knowledge allows, in the SS class. (11) This is a peculiar, primitive, and pelycosaur-like state of the occipital region. All the proterosuchian genera in which the character can be checked show this state clearly; it is especially evident in Chasma- tosaurus. No other archosaur shows a similar condition, so that this feature is to be allocated to the AN class. (12) This state of the prefrontal is not a proterosuchian peculiarity, as it is also characteristic of many thecodonts that are not proterosuchians and of some saurischi- ans. The condition is also shared by some non-archosaurian reptiles, such as the 242 ophiacodont and varanopsid pelycosaurs. This fact suggests that we are confronting a primitive character-state that evolved slowly within the archosaurs. As it is shared by all the proterosuchians so far known, it must be placed in the AS class. (13) In all proterosuchians so far known, the marginal teeth are isodont and either acrodont (proterosuchids ) or subthecodont (erythrosuchids); true heterodonty and thecodonty are not clearly developed in either group. All non-proterosuchian archo- saurs are definitely thecodont in tooth implantation, and their teeth are primi- tively heterodont or subheterodont. The proterosuchian condition is also a primi- tive one, widespread among the earliest reptiles and their first derivatives. This character-state must hence be placed in the AN class. (14) Another primitive condition rem- iniscent of the seymouriamorph, cap- torhinomorph, pelycosaurian, and_ early lepidosaurian condition, is the presence of intercentra. This has been clearly demon- strated in the neck vertebrae of Chasmato- saurus vanhoepi (Fig. 3), and Young (1963) has described the same situation in the trunk vertebrae of Chasmatosaurus yuani. Neck intercentra have been re- ported in Erythrosuchus, but seem not to be present in Shansisuchus, Garjainia, Vjiushkovia, and Cuyosuchus. In later archosaurs, intercentra have not been re- ported in any genus save Euparkeria, where they seem to be present all along the presacral region of the column. An- other (abnormal) exception is the raui- suchid Ticinosuchus, which is alleged to have had an intercentrum associated with one of the caudal vertebrae (Krebs, 1965). We are dealing therefore with a feature of the SS class. (15) The quadrupedal gait is, of course, a character-state shared by all the known proterosuchians, but obviously common, too, in many non-proterosuchian archo- saurs, such as the euparkeriids, the raui- Bulletin Museum of Comparative Zoology, Vol. 139, No. 5 suchids and the stagonolepidids among the thecodonts, the crocodiles and phytosaurs, and many groups of saurischians and orni- thischians. This is obviously a primitive reptilian feature, and must hence be placed in the AS class. . (16) The position of the propodials has been inferred by Hughes (1963) to be horizontal in the known _ proterosuchians. Nevertheless, Young’s (1964) reconstruc- tion of the skeleton of Shansisuchus shows the propodials in a vertical position, which is probably also reasonable. Completely sprawled legs would not have allowed large terrestrial animals such as the erythro- suchids to be successful predators, and the evidence seems to indicate that they had a time of success during the Lower Triassic. It is probable that all the proterosuchians had a sprawled stance most of the time, as indicated by the anatomical data, but that at least the advanced erythrosuchids could proceed in a largely upright stance for short distances. In any case, it is obvious that the proterosuchians sprawled more than any later archosaur, and that this state was shared by all the genera that afford relevant evidence in the girdle and limb skeletons. As stated by Ewer (1965), Euparkeria also seems to have had a sprawled stance, but this genus seems to have been far more advanced than the proterosuchians as far as locomotion is con- cerned. This feature can therefore safely be considered to be in the class of the AN character-state. (17) This character-state is a_ typical archosaur one, though it has been exagger- atedly associated with bipedalism, which is not only not a widespread condition in archosaurs, but is not even a_ primitive archosaurian characteristic (Charig, 1965). Charig has named this condition limb- disparity, and though characteristically archosaurian, it must be noticed that this is also present in the ophiacodontid and varanopsid pelycosaurs. Limb disparity may be considered a preadaptation for bipedalism, but is less marked in the Proterosuchia than in more advanced archosaurs. In the known cases, for in- stance, the humerus/femur ratio is never lower than 77.7 in the _ proterosuchians, and is always lower than 67 in the non- proterosuchian thecodonts. This might be therefore considered an AN character-state. (18), (19) The possession of an internal trochanter and of an intertrochanteric fossa is alleged by Hughes (1963) to be a full indication of the sprawled position of the legs. As far as is known, all proterosuchian femora share in the possession of these characters. The pelycosaurs and _ capto- rhinomorphs share the same character-state, but none of the known non-proterosuchian archosaurs have either an internal tro- chanter or an_ intertrochanteric fossa. Hughes assumed that the Argentinian rauisuchid Saurosuchus shared the protero- suchian state of these characters, but this is a misinterpretation of the illustrations given by Reig (1961), as Charig and Reig (in press) have already made clear. These character-states hence belong to the AN class. (20) The structure of the humerus is well known in Chasmatosaurus (Young, 1963), Erythrosuchus, Shansisuchus, Vjushkovia, and Cuyosuchus (Rusconi, 1961, wrongly described this bone in Cuyosuchus as the femur of the labyrinthodont Chigu- tisaurus). In all these genera the ends are twisted, but in the last they are not typi- cally wide, as is the case in the other four genera. Humeri with wide and _ twisted ends are also present in the rauisuchid Stagonosuchus (von Huene, 1938; Boonstra, 1953) and in the problematic Argentinian Middle Triassic genus Argentinosuchus (Casamiquela, 1961). This may be con- sidered a primitive character-state, as it is also present in the pelycosaurs and cap- torhinomorphs. In any case, the exception of Cuyosuchus and the presence of the same state in other non-proterosuchian EARLY ARCHOSAURIAN E;VVOLUTION * Reig 243 thecodonts, indicate that it is convenient to place this feature in the SS class. (21) The structure of the feet in the proterosuchians has been elucidated by Hughes (1963) with the help of new ma- terial. Work by Ewer (1965) and Krebs (1963, 1965) on Euparkeria and Ticino- suchus respectively, offers additional sup- port to Hughes's conclusions. In the proterosuchians the foot anatomy is only known to an appropriate degree in Chas- matosaurus and = Erythrosuchus, but it seems safe to infer that the condition in these genera was widespread among. all the proterosuchians. The state is that of a tarsus without “crocodiloid” or “dinosau- rian” specializations in the proximal tarsals (astragalus and calcaneum), and with a primitive, mesotarsal ankle joint. All other archosaurs show some type of tarsal modi- fications from this primitive condition, which is, by the way, like that in primitive lepidosaurians, such as Youngina, and in captorhinomorphs and pelycosaurs. All evi- dence indicates the convenience of placing this character-state in the AN class. (22) The shape of the anterior spine of the iliac blade (Fig. 4) varies among the different proterosuchian genera from al- most obsolete in Chasmatosaurus to moder- ately developed in genera like Cuyosuchus, but it is never highly developed, as it is in some pseudosuchians and “dinosaurs.” The proterosuchian type of anterior spine of the ilium is very similar to that of the varanopsid pelycosaurs. At the same time, this same feature is also present in some non-proterosuchians, as is the case in Euparkeria and the rauisuchids, and for this reason it must be considered an AS character-state. (23) The posterior spine of the iliac blade is long and narrow in all the known proterosuchian genera that afford evidence in this regard. Among the non-protero- suchian thecodonts, Euparkeria and_ the rauisuchids share the same condition, so 244 that this is also a character-state of the AS class. (24) The fully closed condition of the acetabula is a proterosuchian character, associated with the amount of space be- tween them; both conditions are related to the generally sprawled position of the posterior propodials. All the thecodonts show a closed acetabulum, and in most of them these are relatively far apart. Open and more closely approximated acetabula were developed in the sauris- chian and ornithischian dinosaurs in con- nection with the advanced bipedal stance. This is also an AS character-state. (25) The relative length of the ventral pelvic bones varies within narrow limits in the proterosuchians, never reaching the development shown in more advanced archosaurs with triradiate pelves (Fig. 4). In the primitive forms the triradiate trend is only incipient, although it is more obvious in terminal forms like Erythro- suchus. In forms like Chasmatosaurus and Cuyosuchus, features of the very primitive puboischiadic plate can also be observed. Euparkeria shows in this respect a con- dition more proterosuchian than typically pseudosuchian, and Ticinosuchus seems to be transitional in this regard. This char- acter-state must thus be considered to be in the AS class. (26) Coracoids are known in Chasmato- saurus, Cuyosuchus, Erythrosuchus, Shan- sisuchus and Vjushkovia. In the first two they are obviously larger and more primi- tive than in the latter, but in any case, the proterosuchian coracoids are to be con- sidered as large in comparison with those of most later archosaurs. Among the Pseu- dosuchia, large coracoids are present in Euparkeria, the rauisuchids Ticinosuchus and Proterosuchus, and the stagonolepidids. We must hence place this character-state in the AS class. (27) The scapular blade is short and broad, and primitive in general shape, in Bulletin Museum of Comparative Zoology, Vol. 139, No. 5 both Chasmatosaurus and Cuyosuchus (Fig. 1). In the genera Erythrosuchus, Shansisuchus, and Vjushkovia it is higher and narrower, with both ends more ex- panded than the median “shaft.” Short and broad scapulae are to be considered as primitive, and the shape of this bone in the erythrosuchids is obviously an improve- ment, which becomes more fully developed in pseudosuchians and later archosaurs. This character-state is to be placed in the SN set. (28) The presence of dermal elements of the pectoral girdle is now known in Chasmatosaurus, Shansisuchus, Erythro- suchus, Vjushkovia, and Cuyosuchus. The first had been assumed to have a clavicle and interclavicle because of the presence of these bones in more advanced thecodonts (Hughes, 1963), but Young (1963) actu- ally found a clavicle associated with other bones of Chasmatosaurus yuani. It is safe to conclude that dermal bones of the shoulder girdle were present in all the members of the Proterosuchia. same time, this primitive feature is also shared by many pseudosuchians, such as the rauisuchids, the stagonolepidids, Eu- parkeria, and even Ornithosuchus (see Walker, 1964: 110). We are dealing there- fore, with a character-state of the AS class. (29) As far as dermal armor is con- cerned, the Proterosuchia, in lacking any indication of it, are clearly different from all other thecodonts (Charig and Reig, in press). The only doubtful case in this respect is Cuyosuchus, as among the At the. original material some atypical scutes were © found. Since these could belong to the labyrinthodont found associated with the | Argentinian proterosuchian, it is better not to consider this case as an actual exception. Crocodiles, phytosaurs, and ornithischians have osteoderms, but they are missing in saurischian dinosaurs (see below) and pterosaurs, so that the present condition | must also be considered as an AS character- state. t Evolutionary and taxonomic significance of the proterosuchian character-states The foregoing analysis indicates that the Proterosuchia-concept is not a fully poly- thetic one, as only five among twenty-nine peculiarities are not shared by all the members of its extension. But, by the same token, it is not a monothetic concept. More significant is the fact that eighteen of the twenty-nine character-states are shared by non-proterosuchian archosaurs. A com- pletely phenetic classification, based on overall similarity, would indeed include some other taxa in the extension of the Proterosuchia-concept, a procedure that we believe would be misleading from the evolutionary point of view. This analysis supports the inference that characters evolved at different rates in the early evolution of archosaurs. Some char- acters changed in state within the group Proterosuchia itself, as reflected by all characters in the SN set. In both cases of SN character-states, we are dealing with very primitive reptilian heritages, hardly to be considered of positive selective value at the archosaurian level of evolution, and their persistence should have been dis- advantageous for the changes that the proterosuchians developed in skull archi- tecture and locomotor improvements. Other characters changed only little beyond the proterosuchian threshold; they are our AS set. As in the former case, these are also primitive characters, most of which are maintained in some families of primitive pseudosuchians, in the first crocodiles, or in the phytosaurs, and only exceptionally in more advanced archosaurs. They seem to indicate that the achievement of a pro- gressive archosaurian stage was, for more than half of the characters involved, a process of gradual evolutionary change. There are also those characters of our SS set that changed both within the protero- suchians and beyond them. They have the combined meaning of both the previous cases, and indicate that some _protero- suchians evolved beyond the level reached EARLY ARCHOSAURIAN EVOLUTION * Reig bo 4! Ol by some of their first derivatives. These characters are useful, indeed, to infer phylogenies: no proterosuchian descendant can be supposed to have evolved from a proterosuchian ancestor that had evolved a different state in a character belonging to the SS class, if it maintains the same character in the state described in that class. There remains, finally, a set of characters that show little or no change within the Proterosuchia, but that behave differently beyond the proterosuchian threshold (the AN class). Nine of the twenty-nine analyzed belong to this group. In most of the cases, the change in these characters in proterosuchian descendants may be interpreted as improvements linked with the emergence of new evolutionary possibilities, as we will attempt to demon- strate below. The general pattern of character-state changes within and beyond the protero- suchians is obviously indicative of the process known as mosaic evolution (de Beer, 1954), heterobathmy of characters (Takhtajian, 1959), or stepwise evolution (Bock, 1965 presents an_ illuminating analysis of the process). As a matter of fact, characters involved in mosaic evolution do not afford any basis for a clear-cut distinction of a taxon from its close descendent relatives. In our case, this is especially obvious for the characters belonging to the SN, AS, and SS sets of character-states. On the other hand, char- acter-states of the AN class actually do afford a clear-cut distinction of the Protero- suchia from the Pseudosuchia, the Croc- odilia, the Parasuchia, and the other more advanced archosaurian groups. An Aristo- telian-minded taxonomist would very easily find the clue for what in the context of his philosophy should be a mere pseudo- problem: he would choose only the AN character-states as the sufficient and neces- sary features that determine the “es- sence” of the Proterosuchia. This procedure will not satisfy the purposes of evolutionary taxonomy, as in this universe of discourse 246 we are not trying to grasp the essence of any static entity, but to discover how to evaluate evolving characters in order to define evolving entities. As far as the characters belonging to the SN, AS, and SS classes are concerned, the question could be raised whether they are not better excluded from the definition of the intension of the Proterosuchia-concept, as they are either shared by other non- proterosuchian archosaurs or not shared by all the proterosuchians. It could also be questioned whether the very existence of this kind of character-state is not an in- dication that the proterosuchian-concept is an artificial construct without any real referent in the objective world. We think that the answer to both questions must be negative, but in any case, it is true that we are facing a common and one of the most difficult of taxonomic problems: namely that of tracing borderlines (needed because of the requirements of taxonomy, but also, alas, because the human brain does not seem to be capable of functioning without categorizing ) in ancestor-descend- ant series that evolve gradually from one state to the other. From the point of view of the logic of the system, an analysis of the “core” and the “fringe” of the taxo- nomic set represented by the protero- suchian-concept (as these terms have been defined and used by J. H. Woodger, 1952) would indeed help very much in a full elucidation of this problem. Such a sophisticated formal treatment is, however, beyond the aim of the present essay. We must keep in mind only that a fringe of vagueness seems to be unavoidable in any concept having evolving entities as re- ferents; the peculiarities involved in such a vagueness are not to be excluded from the definition of this concept, if they are relevant for an adequate understanding of the evolutionary meaning of the entity we are dealing with. The polythetic nature of the proterosuchian-concept, with its fringe of vagueness, must be considered, on the contrary, an inherent quality of the con- Bulletin Museum of Comparative Zoology, Vol. 139, No. 5 cept, one which affords plenty of infor- mation for a better understanding of the features of early archosaurian evolution, a point which we will attempt to stress in the following part of this article. But we must first refer to the following point: we have already said that Simpson and Gisin stressed the importance of alleged discontinuities arising during the process of detachment of a new taxon (as it shifts into a new adaptive zone) for the task of establishing non-arbitrary limits be- tween major taxa. In Gisin’s terms: “Um auch hier ‘nattirliche’ Einheiten zu erhalten, missen deren Grenzen den in der Natur objektiv gegebenen Diskontinuitéten, und diese einer bestimmten Qualitaét entspre- chen” (Gisin, 1964: 9). These discontinui- ties given objectively in nature are believed to be the result of the threshold transition arising from a faster evolution between two major adaptive zones, a situation in which selective pressures act upon one character or a set of characters very strongly, making them evolve at a faster speed (the quantum effect). Should the explanation be correct, we would have a clue with which to trace borderlines between a series of ancestor- descendant major taxa, provided that we are able to discover which are the relevant characters involved in such a_ threshold effect, ice., the “key innovations” respon- sible for the emergence of a new taxon. Whatever the relativity of the discontinuity, it should be possible to discover these characters if we have a complete enough fossil record. The situation is perhaps less simple, however. Bock (1965) has contended that to postulate that in the origin of a major taxon (and hence in its delimitation) the operating process is a single-phase change, involving a switch from one major adaptive zone to another, implies an oversimplifi- cation not supported by any positive evi- dence. For him, the process is better thought of as a stepwise one, through which minor radiations occurred in the transitional adaptive zone. Key innovations and preadaptations are involved in this process, but there is no special reason to assume that evolution is greatly speeded up in the intermediate area. The stepwise character of the transition between major taxa is exemplified for Bock by the mosaic pattern of character changes occurring in the known cases of the emergence of major taxonomic groups. This view seems to discourage any attempt to look for natural boundaries between major taxa and, hence, to get an accurate assessment of the inten- sion of their concepts. It should be very interesting, therefore, to investigate just how the evidence from early archosaur evolution does match each of these views. But such an_ investi- gation will require, first of all, a new evaluation of the evidence, for the assess- ment we have made of the proterosuchian character-states will have new conse- quences for the explanation of the origin and early evolution of archosaurs. How- ever, before discussing our main topic, we must refer to the origin of the protero- suchians, and to the proterosuchian de- scendants. THE ORIGIN OF THE PROTEROSUCHIA Obviously, if the Proterosuchia are the first and the most primitive archosaurs, the problem of the origin of the Proterosuchia is to be identified with the problem of the origin of the Archosauria. The latter has been considered a difficult matter and has been generally approached in a very broad context, usually in connection with the dis- cussion of the alleged early split of the reptiles into two main branches, the Sauropsida and the Theropsida. A special account of this general question is beyond our present aim and we must restrict our- selves to the points more closely connected with archosaur ancestry [for a general survey of the whole matter, see Vaughn (1955), Watson (1954, 1957), Parrington (1958), Tatarinov (1959), Olson (1962) ]. The fact that archosaurs and lepidosaurs have two-arched skulls led to their being EARLY ARCHOSAURIAN E\VOLUTION + Reig 247 grouped in one single taxon, the Diapsida, in early classifications. This taxon-concept has been generally abandoned since Romer (1956) advanced the current classification. But the general idea of a close relationship between archosaurs and lepidosaurs_ sur- vives, and the concept of Diapsida is fre- quently used in phylogenetic discourse, although devoid of any explicit taxonomic intention. How close this relationship is is a matter of the disagreement, but little doubt has been cast upon the assumption that the two groups had a common origin, or that archosaurs are derived from early lepidosaurians. The critical groups for the enquiry into archosaurian ancestry usually have been considered to be: the younginid eosuchians, the millerettiforms, and the captorhino- morph cotylosaurs. As far as the different possible hypotheses of archosaurian an- cestry are connected with these three groups, we can speak of the younginid hypothesis, the millerettiform hypothesis, and the captorhinomorph hypothesis. In a recent paper (Reig, 1967), I have briefly discussed these different hypoth- eses, pointing out that the proterosuchian character-states make it necessary to rule out both the younginid and the milleretti- form hypotheses. Each of these groups is more advanced than the first archosaurs (the proterosuchians) in relevant char- acter-states. The younginid hypothesis was first ad- vanced by Brocm (1914, 1922, 1924a, 1946 ) and has been subsequently adopted by such authors as Camp (1945), Piveteau (1955) and von Huene (1956). This hypoth- esis maintains that the archosaurs, the rhynchocephalians, and the squamates took their origin from the younginids, repre- sented by the small South African Ciste- cephalus Zone reptiles Youngina, Young- oides, and Youngopsis, known mostly from skull material. The family Younginidae forms the central group of the suborder Younginiformes of the Lepidosauria in Romer’s (1956) classification, the other 248 families of the same _ suborder being Paliguanidae, Prolacertidae, and Tanga- sauridae. The younginids have both the diapsidan temporal opening fully de- veloped (character-state i of our AA class) and the typical lepidosaurian otic notch, formed by a curved posterior border of the quadrate and defined above by a small spur of the squamosal (in disagreement with our proterosuchian character-state 4). At the same time, the suspensorium is nearly at the same level with the occipital region (contradicting our character-state 6), and the quadrate is attached by suture with the squamosal in a monimostylic way (in contrast with character-state vi of our AA class). It is now generally accepted that the younginids can be considered as the stem group of the Rhynchocephalia and that the origin of the Squamata is better sought in the Prolacertidae (Camp, 1945; Par- rington 1935; Kuhn-Schnyder, 1954, 1962). As far as the archosaurs are concerned, the younginid ancestry has been seriously questioned by Romer (1946, 1956). And apart from the arguments of this author, it is clear that the younginids cannot be considered ancestors of the proterosuchians because of the structure of the quadrate, as even the first proterosuchians (i.e., Chasmatosaurus, Brink, 1955) show a movable quadrate, articulated with the squamosal through a head, a condition which has been established in the mil- lerettids (Watson, 1957). But in addition, the lack of any sort of otic notch and the very backward position of the mandibular articulation of the quadrate (shown al- ready in the most primitive protero- suchians ) definitely preclude the idea of any kind of younginiform ancestry for them. The proterosuchian character-states 4 and 6 constitute a serious objection to the younginid hypothesis, and this is better abandoned. The core of the Millerettiformes (also a suborder of the Eosuchia of the Lepi- dosauria in Romer’s classification of 1956) Bulletin Museum of Comparative Zoology, Vol. 139, No. 5 is formed by several genera described by Broom (1938, 1940, 1948) from the same Cistecephalus beds of South Africa and placed in the family Millerettidae. Earlier genera of the same group are usually re- ferred to different families. The whole taxon has been carefully surveyed by Watson (1957) who maintained that these are sauropsid reptiles possessing very primitive qualities, though not having al- ready developed the two-arched condition. He suggested (1957: 388) that the the- codonts could have come direct from the Millerettiformes (called by him Millero- sauria), and, in the chart of figure 23 of the same work, he derives the Pseudosuchia plus later archosaurs and the “Erythro- suchia” (= Proterosuchia), as a separate branch, from the “millerosaurs.” The im- plication is that the proterosuchians do not belong in the ancestry of later archo- saurs (a contention not expressed in his text), but that both pseudosuchians and proterosuchians evolved independently from “millerosaurs.”. As we _ shall make more evident below, no relevant evidence exists ruling out the proterosuchians from the ancestry of the pseudosuchians and, on the contrary, the presence of such intermediate forms as Euparkeria suggests that proterosuchians actually were the an- cestors of the pseudosuchians. As far as proterosuchian origin from the millerettids is concerned, it is highly im- probable that at least any of the small genera of the Cistecephalus Zone could be in the line of proterosuchians. All of them have an otic notch already developed, and the quadrate in an upright position, with the mandibular articulation close to the occipital plane. These are character-states that are not expected to be found in any proterosuchian ancestor. It is true that the millerettids are more plausible archosaur ancestors than are the younginids, because the former have a movable quadrate-squa- mosal articulation, but, at the same time, the millerettids had not reached the diapsid condition already developed in the young- inids. Furthermore, the millerettids could hardly be considered as adequate fore- runners of the contemporaneous Archo- saurus from the Russian Upper Permian Zone IV. This genus indicates that, at the time the millerettids thrived, the protero- suchids were fairly large animals which had already developed their typical char- acter-states. However, discarding the millerettids as direct proterosuchian ancestors is not the same as discarding the millerettiform hypothesis, since the group is not restricted to millerettids of the South African Ciste- cephalus Zone. The older Tapinocephalus Zone of the Karroo succession has yielded Broomia, a genus tentatively placed in a family of its own, and the still older strata of the Mesen River in Russia (Upper Ka- zanian, Zone II of the Russian Permian ) afforded Mesenosaurus, a genus considered of pelycosaur affinity by Efremov (1938) and by Romer and Price (1940), but more correctly placed in the Millerettiformes as the type of a family of its own (Watson, 1957; Romer, 1956; Tatarinov, 1964). Romer (1967) has stressed the phylo- genetic importance of the Millerettiformes. They are likely to have been a widespread group, both in time and in space. Can it be supposed, therefore, that the Protero- suchia evolved from some early milleretti- form population? This is hardly probable, as such an early member of this taxon as Mesenosaurus had already acquired, ac- cording to published descriptions, a perfect otic notch. The Millerettiformes are better considered as forerunners of the Lepido- sauria, not as a group having direct relationships with the archosaurs. Romer (1956: 519) suggested that the archosaurs might have arisen independently from cotylosaur ancestors. It is obvious that the captorhinomorphs are here im- plied, as he did not consider other cotylosaur groups as being close to the archosaurs. The two-arched temporal re- gion of archosaurs and lepidosaurs would in this view be another case of parallelism, EARLY ARCHOSAURIAN E;VOLUTION + Reig 249 which, by the way, might also be the case if one advocated a millerettiform ancestry. The first adequately known captorhino- morph, and also the earliest adequately known reptile, comes from the Lower Penn- sylvanian (Westphalian A) of the Port Hood formation in Nova Scotia. This is the genus Romeriscus, a limnoscelid re- cently reported by Baird and Carroll (1967). Remains of two romeriid capto- rhinomorphs and one pelycosaur have also been described from the Joggins of Nova Scotia, a slightly higher level in the Lower Pennsylvanian (Westphalian B) (Carroll, 1964). Romeriids are represented also by dubious remains from the Middle Penn- sylvanian, and they are better known through their last representatives in the Lower Permian (Romeria, Protorothyris ). The other captorhinomorph family, namely the captorhinids, has its first members in the Lower Permian Leonardian stage (see Table I), with Captorhinus as a well-known representative. Members of this family are, moreover, the latest captorhinomorphs, reaching the early Guadalupean and early Kazanian (Rothia, Kahneria, etc.). The limnoscelids departed very early from the main line of reptilian evolution (Baird and Carroll, 1967), so that only romeriids and captorhinids could be relevant in the dis- cussion of archosaur ancestry. It is clear that both romeriids and capto- rhinids would make better archosaur an- cestors than younginids, prolacertids, or millerettids, in the sense that they do not contradict the requirement of the absence of an otic notch as demanded by the pro- terosuchians. They are, however, very archaic, fully anapsid, and with the sus- pensorium not primarily posterior in position. The form and the relationships of the quadrate, moreover, are more archo- saur-like in the millerettids than in the captorhinomorphs. However, Parrington (1958) has demonstrated that the mil- lerettid condition of the quadrate is easily derived from that of Captorhinus. But, as the same arguments used by Parrington bo Ol A, lateral view of the Figure 6. Varanodon agilis Olson. skull; B, dorsal view of the skull; C, series of cervical verte- brae. (From Olson.) could be applied to derive the archosaurian condition of the quadrate from that of the captorhinids, this does not run counter to the possibility of captorhinomorph deri- vation of the archosaurian skull. In fact, no theoretical objection can be raised against the contention that the protero- suchian skull, diapsid, without otic notch, and with a very posterior suspensorium could be derived from a romeriid or captorhinid skull. Furthermore, the post- cranial skeleton is so primitive in these cotylosaurs that practically every protero- suchian character-state of that part of the body could easily be thought of as having evolved from a captorhinomorph state. But it is clear that too large a morpho- logical gap exists between even the more primitive proterosuchians and the more advanced captorhinomorphs, and_ neither romeriids nor captorhinids show any defi- nite trend towards some of the peculiar 0 Bulletin Museum of Comparative Zoology, Vol. 139, No. 5 archosaurian character-states. Even if in- termediate forms should be discovered between captorhinomorphs and early archo- saurs, the amount of difference between the ancestor and the descendent groups would necessarily be so great that the linking group might better be considered as a major taxon of its own. In this case, the captorhinomorph hypothesis should be transformed into one arguing for ancestry from this intermediate taxon. Another objection to the captorhino- morph hypothesis is the lack of explanatory value, as it can be agreed that many reptilian groups could eventually have stemmed from captorhinids or romeriids. Moreover, it becomes clear that this hypoth- esis should be abandoned if another reptilian group more closely related to the first archosaurs exists. As I have already proposed (Reig, 1967), I believe that a strong case exists for assigning this role to a definite group of pelycosaurs; this makes it necessary to put forward a new hypoth- esis, namely the pelycosaurian hypothesis. This idea is not completely new. The notion of pelycosaur and archosaur re- lationships was first expressed by von Huene (1911), when he discussed the position of Erythrosuchus. He found that this genus shared with pelycosaurs so many features in skull and_ postcranial morphology, that he created for it an order of its own, Pelycosimia, a name coined with the evident purpose of expressing the idea of pelycosaur relationships. He later abandoned the idea of the Pelycosimia as a separate order, and the name has been used in its original spelling, or as Pely- cosimioidea, as an equivalent of Protero- suchia, or Proterosuchoidea, and, hence, as a taxon subordinated in the Thecodontia. More recently, Rozhdestvenskii (1964: 204) suggested plainly the pelycosaur origin of the archosaurs, when he said: “The mammal-like reptiles, and particu- larly the pelycosaurs, are also to be con- sidered as archosaur ancestors. The earliest archosaurs, the Triassic thecodonts, are RUSSIAN STANDARD CONTI- RUSSIAN NENTAL ZONES NORTH- CENTRAL TEXAS MAJOR STANDARD DIVISIONS ecicee SCALE STAGES OCHOAN CISTECERHAEUS N WHITE ~% HORSE GROUP CAPITAN Zz ANI, YOUNGINIFORMES a LJ jal oO 7 |) Se = a Wy jo qa |W = oO =) CAPTORHINOMORPHA SS DAE” ID p VARANO N | a PENNSYLVANIAN L COMMON MISSISSIPPIAN ANTHRACOSAUR ANCESTORS Figure 9. Phylogenetic diagram of the suggested ancestry of the Archosauria and the probable relationships among cap- torhinomorphs, synapsids, lepidosaurs and archosaurs. (Modified from Reig, 1967.) the mandibular articulation, are not de- veloped in the more advanced pelycosaurs, we can agree with Olson’s suggestion that the Varanopsidae have departed from the main lines of pelycosaur evolution (Olson, 1965). Romer and Price (1940), however, maintained that the Varanopsidae are an- | cestral sphenacodontians, a contention that | does not seem to be supported by the specialized, archosaur-like features shown by the known members of this family. The | occurrence of true sphenacodonts as early | as the Lower Pennsylvanian (Carroll, 1964; Baird and Carroll, 1967) clearly indi- cates, moreover, that the hypothesis of derivation of sphenacodontids from varan- opsids should be at least submitted to a critical reappraisal. In our present state of knowledge, I think it is more reasonable to place the Varanopsidae in the Ophiaco- dontia, as a family in which at least the known members separated from the main direction of synapsid evolution to follow their own evolutionary course, a course that eventually led to their transformation into the proterosuchians. The possibility should not be discarded, however, that very early, unknown varanopsids could be the common ancestors of both sphenaco- dontians and proterosuchians. Mention must also be made here of the problematic late Pennsylvanian — reptile Petrolacosaurus (Peabody, 1952). On the basis of strong similarities in the palatal structure with the eosuchian Youngoides and rather less relevant postcranial fea- tures, Peabody interpreted this genus as being a primitive eosuchian and proposed a diapsid reconstruction of its skull. This reconstruction is obviously quite hypotheti- cal, but the material seems to suggest, at least, that it possessed a lower temporal opening. Analyzing the quadrate region of the skull and other cranial features, Watson (1954) contended that Petrolaco- saurus is to be considered a theropsid rep- tile, a contention that Vaughn (1955) is inclined to accept. In agreement with these views, Romer (1966b) places Petrolaco- saurus as a probable member of the prim- itive edaphosaurian family Nitosauridae. It seems to me highly probable that this genus belongs to the Pelycosauria, the data af- forded by Peabody giving strong support to this interpretation. If this is the case, it must be noted that the structure of the palate and the elongated cervical centra shown by Petrolacosaurus are character- states suggestive of archosaurian ancestry. But in other respects, this genus is so primitive that it cannot successfully con- EARLY ARCHOSAURIAN E;VOLUTION * Reig 259 tend with the known varanopsids as a proterosuchian ancestor, the geological oc- currence of the varanopsids being also more consistent with the idea that they make better forebears of the archosaurs. I believe that the body of evidence supporting the pelycosaurian hypothesis (Fig. 9) is stronger by far than that sup- porting any alternative view, and I have not been able to find any serious evidence against it. Apart from its empirical foun- dations, it can also be said that the hypothesis is also supported by such at- tributes as explanatory value and sim- plicity. It is able both to explain the until now obscure question of archosaurian origin in a simple way, and also to explain the reasons for seemingly aberrant features of the late Varanopsidae and the peculiar characteristics of the proterosuchians. It is also rich in suggestions that explain the ecological factors underlying early archo- saurian evolution, and is in agreement with other cases of emergence of major groups, namely a pattern of steady development of features of the evolving group. ECOLOGICAL AND EVOLUTIONARY FEATURES WITHIN THE PROTEROSUCHIA We have already suggested in the intro- duction that the proterosuchians represent the first step in an exploratory radiation performed by the thecodonts before the complete dominance of the archosaurs at the end of the Triassic. Now, it will be of prime interest to investigate what conclu- sions can be drawn about the pattern fol- lowed by early archosaurian evolution during this first phase. For this, knowledge of the ways of life and the ecological roles of the proterosuchians can afford important data. Not much doubt can be cast upon the conclusion that the proterosuchids were mostly aquatic, predaceous reptiles living in ponds, lakes, and rivers, using swimming as their main form of locomotion, and preying upon other vertebrates. This con- 260 clusion is based on the similarity that they display in body form and proportions to modern crocodiles and in the character- istics of the skull and the dentition. Tatari- nov (1961: 130) suggested that big forms like Chasmatosaurus fed upon fishes, and that the small forms like Chasmatosuchus might have been invertebrate eaters (how far invertebrates contributed to the diet of the proterosuchids is not clear). More- over, the fact that proterosuchids have been found associated with unquestionable water dwellers, gives additional support to this conclusion. Hughes (1963: 221) affirms that in South Africa “bones of Lystrosaurus and Chasmatosaurus may be found side by side,” and although Robinson (fide Hughes, 1963, same reference) cast doubts about the association of these two genera in the Panchet beds of India, this association, with the presence of labyrin- thodonts as an additional element, has recently been reported by Satsangi (1964) in the Raniganj coal field. Moreover, Young (1936) reported the same fact in China. It must be recalled that Lystro- saurus is a dicynodont very specialized for an aquatic way of living, as indicated by the dorsally placed nostrils, the orbits projecting above the level of the roof of the skull, and the features of the carpus and tarsus. Lystrosaurus seems to have been an herbivorous animal not unlike the mod- ern hippopotamus in habits, and its fre- quent association with the carnivorous Chasmatosaurus can be interpreted as an indication of food chain relationships be- tween the two genera, the former playing the food role of a primary consumer fed upon by the latter, which played the role of a secondary consumer in the freshwater communities in which they lived. The pat- tern would, of course, be more complicated, since fishes and labyrinthodonts probably provided an additional food supply for the maintenance of the Chasmatosaurus popu- lations, and since Lystrosaurus could have provided food for other pond _ predators, such as the big rhinesuchids that have been Bulletin Museum of Comparative Zoology, Vol. 139, No. 5 recorded in the Lystrosaurus Zone (see Watson, 1962). But the widespread oc- currence of the Lystrosaurus-Chasmatosau- rus association and the relative abundance of the former in the deposits are to be con- sidered as good indications that the re- lationships of both these genera represented the dominant channel of energy flow in the food web of the communities to which they belonged. Garjainia has been found in the deposits of the Russian Zone V, which is considered equivalent to the Lystrosaurus Zone. It is, in our belief, the first known erythro- suchid, and its position in the fossil record agrees with its possession of several inter- mediate features between proterosuchids and erythrosuchids (Charig and Reig, in press). The dentition is more carnivorous, and the skull shows modifications for a more efficient biting mechanism. The post- cranial skeleton is unfortunately very little known. The skull characteristics of this genus are better developed in later erythro- suchids. The way of life of more advanced erythrosuchids may be inferred from the skeletal morphology of the upper Lower Triassic genera (Erythrosuchus, Shansi- suchus, Vjushkovia). Von Huene (1911: 20) pointed out that Erythrosuchus should be considered a mainly aquatic predator (“ein sich viel im Wasser aufhaltendes Raubtier”), maintaining that its enormous head can hardly be supposed to belong to an entirely terrestrial animal and that the same conclusion is supported by the struc- ture of the remainder of the body (“Der plump Korper, der kraftige, aber relativ nicht lange Schwanz und namentlich der des grossen Schadels wegen aussergewohn- lich kurze Hals unterstiitzen die Annahme, das Erythrosuchus sich meist im Wasser aufhielt [Fliisse oder Tumpel].”). Tatarinov (1961: 131), on his part, although accept- ing that “the general proportions of its body, with a relatively huge head and short legs” indicate that erythrosuchids were tied to the water, seems inclined to believe that they were relatively more terrestrial than the proterosuchids, and stressed the car- nivorous specializations of these animals, saying: “The main difference of the erythrosuchids with respect to the protero- suchids is related to the passage to an active carnivorous way of life” (Tatarinov, 1961: 130). We doubt that bulky and clumsy animals like Erythrosuchus or Shansisuchus should be considered very active animals, a point that has been em- phasized by Young (1964: 146). It is more likely that they were inhabitants of swamp marshes, able to prey upon big, slow herbivorous vertebrates, inhabiting the same environments, which could be caught by a relatively slow and heavily built predator. In this connection, we may explore the question of what animals were the prey of the erythrosuchids. Although evidence of certain association is not abundant, it is meaningful that the erythrosuchids can be considered animals that belonged to the same communities inhabited by the big, upper Lower Triassic dicynodonts of the families Kannemeyerii- dae and Shansiodontidae (for a modern survey of these dicynodonts, see Cox, 1965). The most reliable association data are probably those coming from the de- posits of the Ermaying Formation in China (Young, 1964; Sun, 1963). In several localities of this formation, bones of Shansisuchus and of Erythrosuchus were found, although not in actual association. Pearson (1924: 851) maintains that Kanne- meyeria was a terrestrial animal that prob- ably used its well-developed paws for digging or scraping in order to obtain its food, and she reported that Watson sup- posed that Dicynodon and Kannemeyeria lived on dry land. The origin of the giant dicynodonts of the Kannemeyeriidae is not well known but, as Cox (1965) has stated, the dicynodonts are hardly derivable from the aquatic and specialized lystro- saurids of the earlier level of the Lower Triassic. More probably they originated from some member of the vast array of E|ARLY ARCHOSAURIAN EVOLUTION * Reig 261 Upper Permian dicynodontids, which are commonly considered herbivorous reptiles well adapted to living in terrestrial environ- ments (see Watson, 1960: 201). The Middle Triassic representatives of the same group (kannemeyeriids and_ stahleckeriids) pro- vide good evidence of association with terrestrial reptiles. It can be argued that if the giant kan- nemeyeriids are derivable from the ter- restrial herbivorous dicynodonts of the Upper Permian, the Lower Triassic Kanne- meyeriids and shansiodontids should be also considered as upland dwellers. We believe, however, that this conclusion is not necessarily valid, and that the heavily- built and big-headed kannemeyeriids may be better thought of as inhabitants of shallow waters. Moreover, there is no reason why, if the Upper Permian terrestrial dicynodontids should have been able to evolve into the fully aquatic lystrosaurids, they could not also have been the ancestors of semi- aquatic marsh dwellers. Therefore, Pear- son’s interpretation of the habits of Kanne- meyeria cannot be taken as conclusive. If this reasoning is correct, proterosuchian evolution during Lower Triassic times can be interpreted as a shift from the aquatic and swimming predaceous way of life as represented by the proterosuchids, towards a shallow-water predaceous way of life, the shallow-water predators being adapted for slow walking in swamps. In the first case the main prey was the aquatic lystrosaurids, in the second case, the giant marsh-dwelling herbivorous kanne- meyeriids. In support of this conclusion, it is mean- ingful that the high point of the protero- suchids occurs in the Lystrosaurus Zone and equivalent levels of the lowermost Triassic, and that the erythrosuchids began to be abundant once Lystrosaurus itself became extinct. This seems to indicate that the shift in proterosuchian evolution from an aquatic towards a lowland marsh environ- ment was necessitated by the extinction i) bo 6 of the main source of food of the protero- suchid populations: the aquatic lystro- saurids. Once these became extinct, the originally aquatic proterosuchians were forced to look for their prey in the large herbivorous dicynodonts inhabiting the lowland marsh regions. This triggered the development of improvements for a walk- ing locomotion and for large animal pre- dation, both of which are characteristics of erythrosuchids. The sprawled condition of the legs is less efficient than the upright stance in a walking animal, but the latter is not completely necessary for slow animals hunting in shallow water environ- ments for sluggish herbivores. This may explain how the erythrosuchids were suc- cessful animals in spite of the fact that they were sprawled and not very active pred- ators and, at the same time, why they developed improvements for a_ walking locomotion as compared with the protero- suchids. In this sense, the changes in appendicular skeleton shown by _ the erythrosuchids, which do not reach a full de- gree of fitness for a terrestrial active loco- motion, can be satisfactorily explained as an adaptive level suitable for a marsh dweller, and as a prospective adaptation (or a “pre- adaptation”) for future terrestrial loco- motion. The fossil record also indicates that the proterosuchids did not become completely extinct after the Lystrosaurus zone and the extinction of the lystrosaurids, as one species of Chasmatosaurus has been re- ported in beds equivalent in age to the Cynognathus Zone (Young, 1964). Seem- ingly, the proterosuchids remained in their old environment as such, but were reduced in number and variety and played a second- ary role in the aquatic communities. These aquatic proterosuchids from the upper part of the Lower Triassic, surviving after the detachment of the erythrosuchids, may well be the source of the other aquatic groups of archosaurs present in the record at later levels in the Triassic period. The erythrosuchids seem to have become Bulletin Museum of Comparative Zoology, Vol. 139, No. 5 extinct by the end of the Lower Triassic. From the very beginning of the Middle Triassic other large predaceous archosaurs have been found in different parts of the world, representing a more terrestrial type; most of these belong to the family Rau- isuchidae of the pseudosuchian thecodonts. At the same time, the evidence seems to indicate that at least some kannemeyeriids shifted towards a more terrestrial life in middle Triassic times, as their remains have been found associated with typical upland reptiles. The extinction of the erythrosuchids, however, and their replace- ment by more terrestrial thecodonts better adapted for upland and active locomotion could also be explained by a change in habitat of the animals representing the main source of food for carnivorous archo- saurs. But in this case, the replacing group is not derivable from the replaced one, as the rauisuchids seem to have evolved from another group of Lower Triassic theco- donts, the pseudosuchians of the family Euparkeriidae. It will be of interest now, to review our knowledge of the protero- suchian descendants. PROTEROSUCHIAN DESCENDANTS It is here maintained that the Protero- suchia may be considered the stem archo- saurian group, in which most of the subsequent evolution of archosaurs is rooted. The ways in which descent took place remain, however, rather obscure. The taxa which seem most likely to have been derived directly from the protero- suchians are the Pseudosuchia and_ the Crocodilia. Saurischians and phytosaurs are also likely to be direct derivatives of the proterosuchians, but the evidence is far from being conclusive. The Ornithischia and the Pterodactyla are better thought of as descendants of the Pseudosuchia, but we are lacking the relevant data to advance any more secure opinions about them. This theory does not agree with the classical view, which considers the pseu- dosuchian thecodonts as the ancestral group of later archosaurs, claiming that a tiny and bipedal pseudosuchian was the prototypical archosaur forebear from which the various dinosaurs, the pterodactyls, the crocodiles, and even the birds could have arisen. According to this view, bipedalism and small size, combined with fully ter- restrial habits, are to be considered as primitive archosaur characteristics. We be- lieve this widely-accepted hypothesis to be outdated and in direct contradiction to the evidence gathered in recent years. We shall develop our points of view in a brief analysis of some of the critical details. Classification and evolutionary significance of the Euparkeriidae The origin of the Pseudosuchia from the Proterosuchia is strongly supported by the existence of such an intermediate thecodont genus as Euparkeria, from the Cynognathus beds of the South African Karroo succes- sion. Euparkeria has been recently revised by Ewer (1965) in an elegant work that added a great deal of information to our previous knowledge of it. Its evolutionary significance has been also discussed by this author and by Hughes (1963). It is profit- able to make an additional analysis of the bearing of Euparkeria upon the classifi- cation and phylogeny of the thecodonts. Ewer emphasized the intermediate nature of Euparkeria. This genus is remarkable for the fact that it shares proterosuchian and pseudosuchian character-states, which, of course, is the reason for the different familial allocations given to it by various authors. Both Ewer and Hughes are in- clined to place Euparkeria within the Proterosuchia as a member of the family Erythrosuchidae. Previous authors gener- ally placed Euparkeria within the Pseudo- suchia (1) as a member of the family Ornithosuchidae (Tatarinov, 1964), (2) in a family of its own, Euparkeriidae (von Huene, 1920; Romer, 1956; von Huene, 1956), or, (3) rather oddly, in the family Sphenosuchidae (von Huene, 1962). Broom EarLy ARCHOSAURIAN EVOLUTION * Reig 263 (1913), Heilman (1926), and Watson (1957) emphasized its central position among the Pseudosuchia, and thought of Euparkeria as a genus typifying the group from which the main lineages of the later archosaurs could have arisen. Euparkeria shares with the Proterosuchia the following character-states of our list: 1, 2, 8, 9, 10, 12, 14, 15, 22, 23, 24 26; and 28. This means that it has in common with the proterosuchians thirteen of the twenty-nine items of our analysis, and that it differs in the remaining sixteen. If we should apply a taxonomic criterion based on overall re- semblance, Euparkeria would have to be placed in a taxon distinct from the Protero- suchia. Our approach is not, however, a phenetic one, and we are more attracted toward an evaluation of the character-states of this genus from an evolutionary point of view. Eleven of the thirteen character-states shared by Euparkeria with the protero- suchians belong to our AS class. They are primitive archosaurian (and _pre-archo- saurian) features that evolved slowly dur- ing the first states of the archosaurian evolution. On the other hand, as these character-states are present in all the proterosuchians, they do not afford clues by which to investigate the affinities of Euparkeria within the Proterosuchia. More significant is the agreement of this genus with the proterosuchians in two of the three SS character-states: the presence of palatal teeth and the presence of inter- centra. Palatal teeth are known to be possessed by the proterosuchids, but not by the erythrosuchids. Intercentra are present in Euparkeria through all the length of the presacral vertebrae, just as in Chasmato- saurus. Erythrosuchus is the only erythro- suchid having intercentra, and they are present only in the cervical region of the column. These facts could be interpreted as an indication that the erythrosuchids were not the ancestors of the euparkeriids, and that the latter arose somewhere within 264 the proterosuchids as a separate lineage. However, the erythrosuchids show features in the dentition, the skull, and the ap- pendicular skeleton, that relate them more closely to the euparkeriids than to any of the proterosuchids. If one were to infer relationships by overall resemblance, it would be safe to conclude that the eupar- keriids are more closely related to the erythrosuchids than to the proterosuchids. Palatal teeth and intercentra are, in spite of that, a true challenge to erythrosuchid derivation. An additional hint in the same direction is afforded by the presumed way of life of Euparkeria. As Ewer pointed out, this genus was a predator upon tiny verte- brates and invertebrates living in upland regions, and, as such, was capable of rapid locomotion in a _ terrestrial environment. This kind of animal is hardly derivable from such bulky and sluggish marsh dwel- lers as the contemporary erythrosuchids seem to have been. These contradictions can be overcome if we visualize the origin of the euparkeriids as an event that took place during the transitional phase of the proterosuchid-erythrosuchid descent. At this stage, the transitional forms should have retained some of the primitive protero- suchid character-states, and they should also have acquired some of the morpho- logical and ecological traits of the erythro- suchids. These proterosuchians would have lived in a transitional ecological zone where selective pressures would have rewarded any acquisition for a better adaptation as predators of great size dwelling in lowland marshes, and also any change improving upland fast locomotion, air-wave hearing, biting efficiency, and water economy, all of which are necessary acquisitions for active terrestrial predators. Directional selection would have created, in the first case, the typical erythrosuchids; in the second case, the euparkeriids. It is meaningful in this connection that the euparkeriids differ from both erythro- suchids and _ proterosuchids precisely in those characters that can be correlated Bulletin Museum of Comparative Zoology, Vol. 139, No. 5 with functions linked with upland rapid locomotion, air-wave hearing, masticatory efficiency, and, presumably, water econ- omy. Euparkeria shows changes to a dif- ferent state in, among others, items 16, 17, 18, 19, 20, 21, and 25 of our list of protero- suchian character-states. In all those cases, the changed state of the character in Euparkeria was evidently linked with im- provements for a more efficient terrestrial locomotion: upright stance; hind limbs longer than the fore limbs to a greater degree than in the proterosuchians; femur without intertrochanteric fossa or internal trochanter; humerus with less expanded ends; tarsus with incipient specializations in the ankle joint, thus anticipating de- velopments in the later pseudosuchians; a longer pubis and ischium representing a more advanced type of triradiate pelvis. At the same time, the development of a fully evolved otic notch shown by Eupar- keria, distinct from proterosuchian char- acter-state 4 and correlated with changes in the state of character-states 5 and 6, is to be interpreted as an improvement for better air-wave hearing, the otic notch being obviously an improved device in this direction, as it gives room for, and enhances the function of, the tympanic membrane. Concerning the changes in the biting mechanism, Watson (1957) and Ewer (1965) demonstrated how far the shifting forward of the suspensorium, moving the quadrate towards a more vertical position, is a necessary development toward increas- ing the height of the temporal region and correlatively toward lengthening the fibers of the temporal musculature for a more efficient biting action. This development is fully attained in Euparkeria, and in this genus it is correlated with an enlargement of the upper temporal opening, which pro- vides additional area for the insertion of the pseudotemporalis muscle, and with the development of a dentition more spe- cialized for a predaceous way of life. Ewer has convincingly argued against the interpretation of the antorbital fenestra as an area of insertion of the pterygoideus D. muscle maintained by Dollo, Gregory and Adams, and Walker. She stresses the possibility that this fenestra might have housed a large salt gland, as suggested by Broom (1913). It is now well known that not only several marine vertebrates (Schmidt-Nielsen, 1958) but also desert lizards such as Ctenosaura and Sauromalus (see Templeton, 1964, 1966) have nasal salt glands that play an important role in removing chloride salts from the body, with a small loss of water, thus acting as an extrarenal mechanism for salt excretion and water economy. The known cases of the presence of nasal salt glands of this sort in living vertebrates do not show this gland housed in an antorbital fenestra, but we do not believe that this fact need be a serious challenge to the interpretation of Broom and Ewer. Though admittedly highly speculative, the following reason- ing is presented as a possible explanation of the known facts concerning this prob- lem. As the mammals are urea-secreting animals derived from the pelycosaurs through the therapsids, it can be assumed that the pelycosaurian ancestors of the archosaurs were also ureotelic animals, and that uricotelism developed only later in their archosaurian descendants (the birds are typically uric acid-secreting animals). Uricotelism being related with water econ- omy in animals living in dry conditions, the lack of this metabolic device in the increasingly upland dwelling archosaurs may have been balanced by the develop- ment of an extrarenal salt-secreting device. If the antorbital fenestra is actually the site for a salt gland, this may explain the characteristic development of such an opening in all the archosaurs. In this con- nection, Euparkeria clearly shows an im- provement beyond the proterosuchian level, as it has a larger antorbital fenestra lodged in a basin-like depression, which indicates a bigger size, and hence, an intensification EARLY ARCHOSAURIAN EVOLUTION * Reig 26 Ul of the function of the salt gland. This intensification of function of an extrarenal salt-secreting organ can be thought of as an improvement of the adaptation to up- land, dry environments, in ureotelic animals coming from a freshwater environment in which economy of water was not necessary. The presence of a small antorbital fenestra in Proterochampsa and later crocodiles agrees with this argument; the presence of a large antorbital fenestra in phytosaurs, however, is not consistent with it. For all these reasons, it seems evident that Euparkeria has departed from the proterosuchian level of evolution in sig- nificant respects. As most of its innovations are also well developed in the pseudo- suchian thecodonts, it is reasonable to think of it as a member of the group representing the early shift of the thecodonts towards the upland life to fulfill the roles of ter- restrial carnivorous reptiles, a shift that triggered the radiation of the Middle and Upper Triassic pseudosuchians. In_ this sense, the new character-states shown by Euparkeria in locomotion, biting mecha- nism, hearing, and water economy are to be interpreted as key innovations opening up new evolutionary possibilities and en- hancing the emergence of a new major taxon, which in this case is the suborder Pseudosuchia of the Thecodontia. In spite of the fact that Euparkeria (with Browniella as a junior synonym) is the only Lower Triassic slightly-built pseudo- suchian known from skeletal remains, the available evidence shows that thecodonts that had already attained the same level of evolution were widespread in upper Lower Triassic and lower Middle Triassic times. This evidence comes mainly from ichnological data, which indicates that quadrupedal, lightly built, and small-sized pseudosuchians flourished by that time in North America (Peabody, 1948). As con- tended by this and other authors, it is quite probable that the large manus foot- prints of the chirotheriids of small size were actually made by euparkeriid the- 266 codonts. At the same time, it is also pos- sible that some dubious skeletal remains of the same general age could in the future be demonstrated as belonging to the same family. Wangisuchus, a genus based on fragmentary remains of various individuals, has been referred by Young (1964) to this family. The basis for this assignment is not clear, however. The known skeletal structure of Eupar- keria makes it clear that this genus had not attained certain of the specializations that are full-fledged in the Middle and Upper Triassic pseudosuchians that are probably euparkeriid derivatives. This fact supports the splitting off of Ewparkeria into a family of its own, distinct from the remaining families of the Pseudosuchia. As far as the relationships of the euparkeriids with the other pseudosuchians are concerned, one could say that with respect to the remain- ing pseudosuchians, the euparkeriids hold the same relationship that the Protero- suchians hold with respect to the whole of the non-proterosuchian archosaurs. Relationships with the Pseudosuchia The remaining Middle and Upper Trias- sic thecodonts are far from affording a clear-cut picture of their evolutionary relationships and classification. It has been said that the Pseudosuchia are a sort of waste-basket, a statement that seems to cast serious doubts about the naturalness of the group. The Pseudosuchia seem to be, however, a natural group, but it is evident that the whole taxon is in need of a thorough revision. Some recent papers by Krebs (1963, 1965), Reig (1961), Sill (1967), Walker (1961, 1964, 1966), and others have already contributed to a great extent to clearing up the status of parts of this taxon. It is now agreed that the Elachistosuchi- dae must be ruled out of the Pseudosuchia, as Elachistosuchus has been demonstrated by Walker (1966) to belong to the rhyn- chocephalians. At the same time, Sill (1967, see also below) suggested that the Bulletin Museum of Comparative Zoology, Vol. 139, No. 5 crocodiloid thecodonts usually placed in the superfamily Sphenosuchoidea of the Pseudosuchia, are better considered as belonging to the protosuchian crocodiles. After these deletions, the main subordinate taxa of the Pseudosuchia are the Lower (and Middle?) Triassic Euparkeriidae, the Middle Triassic Rauisuchidae, the Middle and Upper Triassic Stagonolepididae (see below) and the probably related Upper Triassic Stegomosuchidae,! the Upper Triassic Ornithosuchidae, and the Upper Triassic Scleromochlidae. It will now be useful here to assess the main conclusions that can be drawn from present knowledge of the pseudosuchians (Fig. 10). All pseudosuchian families share the fol- lowing characters: possession of an otic notch; suspensorium shifted forward; V- shaped contour of the posterior border of the lower temporal opening; large ant- orbital fenestra lying in an extended basin- like depression (with the exception of Rhadinosuchus and Clarenceia, see later ); fairly large nares close to the antorbital fenestra’ (same exceptions); pterygoids joined at the midline; palatal teeth absent (with the exception of Euparkeria); mar- ginal teeth subheterodont and thecodont; intercentra absent (with the exception of Euparkeria); advanced quadrupedal or bi- pedal gait; posterior limbs somewhat longer than the front ones; propodials vertical in position; pes “crocodiloid,” with astragalo- crural—calcaneum-tarsal ankle joint (in- cipiently so in Euparkeria); calcaneum with a tuberosity; long pubis and ischium; well-developed dermal armor (except in Scleromochlus, surely a secondary loss). It seems clear that the above intension of the concept of Pseudosuchia makes this taxon a well-defined one with respect to the Proterosuchia. The — pseudosuchian — character-states evolved seemingly as an adaptation to 1 Walker (1968), however, has recently main- tained that the Stegomosuchidae are crocodiles; see Addendum. EARLY ARCHOSAURIAN EVOLUTION * Reig p , Y z g Y : is So MIDDLE TRIASSIC 267 © Ww wn = oa bE oO LJ 5 ° PROTERO™ ac © CB = = or LJ a or J a oO =) Figure 10. other thecodonts. terrestrial life, and for the most part they were already established in the eupar- keriids. The rauisuchids probably evolved as a branch divergent from the euparkeriid stock in the early Middle Triassic or upper- most Lower Triassic. Their first well- documented representative is Ticinosuchus from the Anisian of Europe (Krebs, 1965). Young (1964) referred to the same family the upper Lower Triassic Chinese genus COMMON PERMIAN PEEYCOSAURIAN ANCESTORS Phylogenetic diagram of the suggested relationships among the various families of the Pseudosuchia and the Fenhosuchus because of some. similarities in vertebral morphology, shape of the scutes, and other dubious characters. This genus is known from fragmentary bones of various individuals, and its status is far from clear. Nevertheless, the presence of raui- suchids in the Lower Triassic is suggested again by the ichnological evidence, as large-sized quadrupedal chirotheriids of probable rauisuchid relationships have been 268 found in beds of Scythian age in Germany, North America, and South America (see Peabody, 1948, 1955; Krebs, 1965). Apart from those mentioned above, rauisuchids are known in Middle Triassic (lower Ladi- nian?) beds of Africa (Stagonosuchus of the Manda beds of Tanganyika) and Brazil (Rauisuchus, Prestosuchus from the Santa Maria beds of Rio Grande do Sul) and in the upper Middle Triassic (upper Ladi- nian?) of Argentina (Saurosuchus from the Ischigualasto beds of San Juan Province). The rauisuchids seem to have been reptiles well adapted for terrestrial life, and they reached a great size. They were surely huge predators more active and efficient than the erythrosuchids, but they remained quadrupedal like the latter, perhaps be- cause of the attainment of a bulky body and a great weight before the full acqui- sition of the necessary limb modifications for bipedal stance and locomotion. Advance beyond the euparkeriid level is shown, however, in the full development of a cruro- tarsal crocodiloid ankle joint, the great elongation of the ventral pelvic bones, the loss of palatal teeth, and the pterygoid union at the midline (as shown in Sauro- suchus, unpublished personal data), the loss of postparietal and postfrontal bones, and large size. The rauisuchids became extinct at the end of the Middle Triassic, apparently without giving rise to any other group, and perhaps because of the compe- tition of the carnosaurian saurischians. It is also probably meaningful that their spread and diversification from the be- ginning of the Middle Triassic can be correlated with the extinction of the erythrosuchids at the end of the Lower Triassic. Another well-defined family of pseudo- suchians is the Stagonolepididae.! Reig 1T agree with Walker in including in one family all the genera of thecodonts currently referred to the families “Stagonolepidae,” Aétosauridae, and Desmatosuchidae. The correct familial name for this assemblage is Stagonolepididae Lydekker, July Bulletin Museum of Comparative Zoology, Vol. 139, No. 5 (1961), Walker (1961), and Krebs (1965) have demonstrated that the stagonolepidids are not as closely related to the rauisuchids as is maintained by some authors. Never- theless, Reig’s contention that the two families must be placed in different sub- orders now appears too exaggerated a view, as it is quite possible that the two families originated in the euparkeriids. The stago- nolepidids are, of course, a very clear-cut group, as their specializations in bony armor and in skull and dentition are unique among the thecodonts. That the family was fully established in upper Middle Triassic times is demonstrated by Aéto- sauroides from the Ischigualasto beds of Argentina (Casamiquela, 1961). They may have separated from the euparkeriid stock in early Middle Triassic times, evolving as an independent lineage that played its own distinct ecological role. Aétosaurus from the German Keuper, Stagonolepis from the Elgin Sandstones of Scotland, and Typo- thorax, Desmatosuchus, Acompsosaurus, and Stegomus from the Upper Triassic of North America demonstrate that the family was rather widespread in Keuper times. Though the way of life of the stagono- lepidids is still a matter of controversy, it is evident at least that the members of this family were completely terrestrial pseudo- suchians and that they are to be regarded as the first archosaurs that were not pred- ators. Walker has supposed that they were mostly herbivorous, while Sawin (1947) maintained that they were scaveng- ers. It is interesting to realize that the stagonolepidids share some general resem- blance with the dasypodids, both in the possession of dermal armor and in the general shape of the skull and dentition, a point that would bolster the scavenger hypothesis, but which does not necessarily exclude the assumption of a rather com- 1887, a name that antedates Aétosauridae Baur, September 1887. Von Huene’s “Stagonolepidae” (1908), so frequently encountered in the litera- ture, is etymologically incorrect. posite and variable diet, with vegetables and arthropods as usual components. Stegomosuchus and Dyoplax, from the Upper Triassic of North America and Europe, respectively, are rather poorly known genera showing several resem- blances to the stagonolepidids in armor de- velopment and other features. They may be closely related to the aétosaurids in origin, but if they are really related to each other, they should be placed in a separate family Stegomosuchidae. The taxonomic status and the relation- ships of the remaining pseudosuchians are less clear. Most of the non-rauisuchid and non-stagonolepidid genera are commonly grouped in the family Ornithosuchidae, which is supposed to include small or medium-sized, bipedal predators, of which Ornithosuchus would be a typical example. However, this genus has been recently demonstrated by Walker (1964) to include fairly large animals, and the large Dasy- gnathus from the same Elgin Sandstones that yielded the original remains of Ornitho- suchus is placed by him in its synonymy. Walker also arrives at the odd conclusion that Ornithosuchus is neither a pseudo- suchian nor any other kind of thecodont, but that it is better placed within the order Saurischia. This latter view is rather dif- ficult to agree with, and the present author has not found in Walkers new data and appraisals sufficient supporting reasons for such an astounding upheaval of the current arrangement. It is true that Ornithosuchus looks like the carnosaurian dinosaurs in several re- spects, but the instances of resemblance are better ascribed either to the sharing of general archosaurian features or to the fact that Ornithosuchus and the carnosaurs attained, in parallel, specializations for bi- pedal locomotion and a predaceous way of life. On the other hand, Walker did not attempt to demonstrate that this genus is not a pseudosuchian, his argument being directed to support of the view that it is a carnosaur. We think that important rea- EARLY ARCHOSAURIAN EVOLUTION * Reig 269 sons are at hand for keeping Ornithosuchus in the Pseudosuchia. One of them is the possession of the double line of paramedia] scutes, a character-state shared by the euparkeriids, the rauisuchids, and some genera referred to the ornithosuchids, and which is to be considered as an original pseudosuchian feature from which evolved the armor of such heavily armored forms as the stagonolepidids. No certain evidence of dermal armor is known for the Carno- sauria; the alleged carnosaurian scutes from the Upper Cretaceous of India are better referred to ornithischian dinosaurs (see Walker, 1964: 117-119). Another im- portant point is that Ornithosuchus has, almost surely, a typical pseudosuchian ankle joint. The carnosaurs, like all the saurischians, have a completely different type of ankle joint, which is hardly deriv- able from such a specialized structure as the pseudosuchian-crocodiloid tarsus (see below). In other respects, Ornithosuchus agrees perfectly with the pseudosuchian character-states. It seems rather bizarre to claim that it is a carnosaur when it is not really separable from the thecodonts. Walker admits that “it might ultimately prove necessary to retain Ornithosuchus in the Pseudosuchia” (1964: 110), a statement that does not seem to fit very well with his previous affirmation that only the coeluro- saurs and the carnosaurs “need be seriously considered in a discussion of the affinities of Ornithosuchus” (1964: 105). Walker also maintains that Ornitho- suchus lies morphologically close to the boundary between the pseudosuchians and the carnosaurs, and that phylogenetic re- lationships are more clearly expressed by placing it with the carnivorous dinosaurs. In fact, this seems not to be the case, as typical carnosaurian and other saurischian dinosaurs have been found in beds defin- itely earlier than the Elgin Triassic (see Reig, 1963a; Charig, Attridge and Cromp- ton, 1965; Ellenberger and Ginsburg, 1966). These finds clearly prove that by the Middle Triassic several lineages of 270 saurischians were already differentiated, and this suggests that the origin of the group is to be sought as early as the Lower Triassic. The Upper Triassic Ornitho- suchus cannot be considered as intermedi- ate for temporal reasons, and there are no cogent grounds for placing it anywhere but in the Pseudosuchia. It is more reason- able to believe that within that suborder of thecodonts, one family attained bipedal- ism and other carnivorous specializations, paralleling some lineages of contemporary dinosaurs with which it entered in compe- tition. If we retain the family Ornitho- suchidae and include in it not only the large-sized Ornithosuchus, but also the tiny genera Saltoposuchus and Hespero- suchus, we may agree that the ornitho- suchids paralleled both the coelurosaurs and the carnosaurs in general appearance and ecological roles. The curious Scleromochlus may be con- sidered as an arboreal derivative of the Ornithosuchidae, distinct enough to war- rant familial separation. There remain, however, other pseudosuchian genera that are less clear as to family allocation. Erpetosuchus, from the Upper Triassic of the Elgin Sandstones, has been commonly classified with the ornithosuchids, but other opinions have resulted in the erection of a family of its own for this genus. Walker (1961) places Erpetosuchus, Dyoplax, and probably Stegomosuchus in the family Erpetosuchidae, an arrangement that seems unnatural to the present author. The place of this genus is better considered as un- settled until a modern revision is under- taken. As far as Cerritosaurus (Price, 1946) from the Santa Maria Middle Triassic of Brazil is concerned, it is almost surely, as suggested by Hoffstetter (1955), a junior synonym of Rhadinosuchus von Huene. This genus is very peculiar in the small size of the antorbital fenestra, the size and the position of the external nares, the obliteration of the postemporal fenestra, and the straight posterior border of the Bulletin Museum of Comparative Zoology, Vol. 139, No. 5 lower temporal opening. These features make this genus hardly derivable from the euparkeriids, and some of them are actually proterosuchian, non-pseudosuchian — char- acter-states. Nevertheless, it has acquired pseudosuchian status in such characters as the absence of postfrontal and postparietal bones, the presence of an otic notch, and the thecodont and subheterodont dentition. If Rhadinosuchus is actually a pseudo- suchian, it could represent a family of its own, Rhadinosuchidae, as proposed by Hoffstetter (1955) and accepted by Kuhn (1961). This family might have originated independently within the proterosuchians, reaching the pseudosuchian level in its own way. Another poorly known genus from the Upper Triassic of South Africa, Clarenceia (see Brink, 1959), agrees with Rhadinosuchus in the structure of the ant- orbital fenestra and the form of the maxilla, and might belong to the same _ family (Romer, 1966b, makes this genus a dubious member of the Ornithosuchidae, a position that seems to lack relevant foundations). If our interpretation of Rhadinosuchus is right, the implication is that either we ac- cept the Pseudosuchia as a polyphyletic assemblage, or we must allow for the in- convenience of erecting a new suborder to accommodate Rhadinosuchus and allies. Our knowledge of these forms is, however, too imperfect to support any formal pro- posal of changes in the system of the The- codontia. The origin of the crocodilia The crocodiles have been classically considered as descendants of the Pseudo- suchia. Within the latter, the Sphenosuchi- dae from the Upper Triassic of South Africa were considered to be the ancestral group. Primitive crocodilian archosaurs such as Notochampsa and Pedeticosaurus (from the Cave Sandstone beds of the Stormberg Series of South Africa), Erythro- champsa (from the underlying Red Beds, which also yielded Sphenosuchus), and Protosuchus (from the later Triassic or earliest Jurassic of Arizona), commonly grouped in the crocodilian suborder Proto- suchia, have been regarded as transitional between the ancestral sphenosuchids and the later typical crocodiles (Mesosuchia, Sebecosuchia, Eusuchia). According to this conception, the assumption is made that the crocodiles evolved from primi- tively bipedal pseudosuchians, and_ that they returned to a quadrupedal gait as an adaptation to the amphibious way of life (for broader information on these ideas on crocodilian origins, see Haughton, 1924; von Huene, 1925; Colbert and Mook, 1951; Kelli 11955) Recently, Sill (1967) has made a thorough reappraisal of the question, on the basis of the bearing of Proterochampsa upon crocodilian origins. Proterochampsa (Reig, 1959) (Fig. 11) is an obvious crocodile from the late Middle Triassic Ischigualasto beds of Argentina, showing a remarkable combination of primitive, transitional, and advanced character-states. It is the earliest crocodile so far known, and it is definitely earlier than the spheno- suchids reported to be the pseudosuchian ancestors of the crocodiles. The crocodilian nature of Protero- champsa is evident from the morphology of the dorsal surface of the skull, the presence of a rudimentary secondary palate built up by the premaxilla and the maxilla, the sculptured bones of the roof of the skull, and the structure of the vertebral apophyses. Besides this, it is noteworthy that the anterior foot shows the typical carpal specializations of modern crocodiles: elongated radiale and ulnare carpal bones. This is demonstrated by a nearly complete anterior leg found in association with the remains of a coelurosaurian dinosaur in the Ischigualasto beds (Reig, 1963a).! The femur and the humerus, known to the author through undescribed specimens as- sociated with skull remains, are also typi- cally crocodiloid. Unfortunately, bones of 1 See, however, the Addendum. EARLY ARCHOSAURIAN EVVOLUTION * Reig Ventral and dorsal views of the skull of Pro- Figure 11. terochampsa barrionuevoi Reig. (After Sill.) the girdles have not been found so far. As pointed out by Sill (1967), it is meaningful that Proterochampsa is in several respects more crocodilian than the later genus Protosuchus.” The implication of the discovery of Proterochampsa is that the sphenosuchids can no longer be considered as the theco- dont ancestors of the crocodilians, nor can Protosuchus and its allies be thought of as a transitional group between the pseudosuchians and the later full-fledged crocodiles. Sill has made a_ suborder Archaeosuchia to group together both the Middle Triassic monotypic family Protero- champsidae and the Upper Triassic Noto- champsidae (including Notochampsa and 2For another view on the place of Protero- champsa and other early crocodiles, see Walker (1968) and the Addendum. OT: Bulletin Museum of Comparative Zoology, Vol. 139, No. 5 UD a << ISS VeMer DISSWINL YaddNn DISSVINL J1GGCIW AS S ss IN SS Ne DISSVINL YSMO1 NVIWYad Yaddn Erythochampsa). He _ believes that this suborder is the ancestral group of the Mesozoic and modern crocodiles of the suborders Mesosuchia, Sebecosuchia, and Eusuchia (Fig. 12). Protosuchus, on the other hand, would represent a suborder, the Protosuchia of Mook (1934) and later authors, that has departed from the main direction of crocodilian evolution by adap- tating to a more terrestrial way of life. As Sill has proposed and Romer (1966b) has accepted, the Sphenosuchidae and such dubious genera as Pedeticosaurus and Platyognathus are better grouped within the Protosuchia, since they agree with Protosuchus in the. sharing of an early crocodilian heritage with adaptations for a more terrestrial life. Referring to these animals, Sill uses an expression coined by Kermack: they are “crocodiles trying to be dinosaurs.” This meaningful expression de- scribes perfectly the evolutionary trend in these atypical crocodiles for a dinosaur-like (i.e. terrestrial and predaceous) way of life. Sill advances two alternative hypotheses for crocodilian origins: either they origi- nated from a non-pseudosuchian group of aquatic thecodonts, or they descended from a primitive group of terrestrial the- codonts, possibly early pseudosuchians. As we have already seen, the euparkeriids make perfect early pseudosuchians in their organization. Proterochampsa is, however, hardly derivable from euparkeriid ancestors for the following reasons: (1) it has not developed the typical pseudosuchian otic- notch; (2) it has a primitive and small antorbital fenestra; (3) it has not acquired the pseudosuchian V-shaped contour of the posterior border of the lower tem- poral opening; and (4) it has the suspen- sorium placed backwards. These are actually proterosuchian character-states, and Proterochampsa is also proterosuchian in the possession of palatal teeth and in the shape and proportional size of the temporal openings. This gives support to the first of Sill’s EARLY ARCHOSAURIAN EVOLUTION * Reig 273 two alternative hypotheses, suggesting that the Archaeosuchia (and through them, all the crocodiles) might have been derived from the aquatic proterosuchians of the Lower Triassic. It should be remembered that after the separation of the erythro- suchids, proterosuchids were represented in beds equivalent to the Cynognathus Zone. These late aquatic proterosuchians could have been the ancestors of other lines of aquatic archosaurs. Nevertheless, one important point re- mains unexplained if we accept Sill’s first alternative. Crocodiles and pseudosuchians (and probably phytosaurs) share the possession of a peculiar type of ankle joint, the so-called “crocodiloid” tarsus, in which the functional joint lies between the astragalus and calcaneum, these being articulated by means of a ball-and-socket type of joint. As we have already seen, this kind of tarsus is not a primitive archo- saur characteristic, as both proterosuchids and erythrosuchids show quite another, more primitive, type of ankle. Walker's belief (1964: 110) that the crocodilian ankle-joint “may after all represent a basic archosaurian pattern,” is therefore lacking a serious basis. Krebs (1963) has pointed out that the resemblance between pseudo- suchians and crocodiles in tarsal structure is so great that it is difficult to think that such a tarsus arose independently in both groups by convergent evolution. It must be realized that the hypothetical common ancestral group for both crocodiles and pseudosuchians, required by tarsal struc- ture, could not be identical with the euparkeriids, as Euparkeria has not reached full development of such a type of ankle joint. This means either that the supposed common ancestor should be sought at a post-euparkeriid level of thecodont evolu- tion or that it must be accepted that the character-state under discussion developed independently in pseudosuchians and crocodilians. The first possibility seems to be ruled out, as the characteristics of the archaeosuchians do not permit thinking of 274 a common ancestry even at the level of the euparkeriids. It would be very useful to have information about the structure of the ankle in Proterochampsa, which, unfortu- nately, is not available thus far. In our present state of knowledge it seems best to adhere to the hypothesis of the proterosuchian origin of the croco- dilians, and to accept the idea of the convergent evolution of the type of ankle found in both crocodiles and pseudo- suchians. It must be admitted, however, that the evidence is still too incomplete to permit a fully satisfactory explanation of crocodilian origins, and that a better knowl- edge of Lower and Middle Triassic theco- donts may make it necessary in the future to introduce changes in the present ex- planation. At this point, it is interesting to recall the Rhadinosuchidae, a Middle Triassic group of scarcely known theco- donts that seem to have reached the pseu- dosuchian level from an ancestry distinct from the euparkeriids. It will not be sur- prising if a better understanding of these forms throws light on questions of the kind raised here. Saurischian ancestry The ancestry of the saurischian dinosaurs is also commonly explained by hypotheses that advocate that the pseudosuchian the- codonts were the ancestral group. Until recently, the first unquestionable sauris- chians were known only from beds _ of Upper Triassic age; indeed the presence of dinosaurs has been considered conclusive evidence for dating Triassic strata of dubious age as Upper Triassic. Coeluro- saurs, carnosaurs, and prosauropods were known from the Upper Triassic, and all three groups were supposed to derive from a single source in the Upper Triassic, namely allegedly tiny, bipedal, carnivorous pseudosuchians similar to the ornitho- suchids. According to this conception, the quadrupedalism of the sauropods was secondary and derived from a_ primitive bipedal condition. Bulletin Museum of Comparative Zoology, Vol. 139, No. 5 Our intent here is not to essay an ex- haustive look at the rather confusing situ- ation of the Triassic saurischians. This task has been partially carried out by Charig, Attridge, and Crompton (1965), Colbert (1964), and Walker (1964), and work by these and other authors will surely contribute to a better understanding of the group. We need, however, to present a very general survey of the present status of knowledge about Triassic saurischians in order to frame the question of sauris- chian origins as coherently as possible in terms of its factual foundations, and thus to check to what extent the existing stereo- typed opinions on saurischian origins are supported by the available evidence. The Upper Triassic faunas of the world differ sharply from the Middle and Lower Triassic ones in the abundance and variety of their dinosaurs. Romer (1966a) re- cently made it clear that in spite of semantic discussions on the rather conventional question of the boundary between Middle and Upper Triassic, the faunas currently referred to the Middle Triassic are distinct from those usually referred to the Upper Triassic by the fact that their dominant groups are different. Gomphodonts and rhynchosaurs are dominant in the B assemblages (Middle Triassic); dinosaurs are the dominant group in the C faunas (Upper Triassic). The same synecological criterion has been used in Reig’s (1963a) discussion of the age of the Ischigualasto beds, a criterion that seems not to have been sufficiently grasped by Bonaparte (1966) in his recent discussion of the Argentinian vertebrate-bearing — Triassic. These Upper Triassic faunas are known in the European Keuper, the Red Beds and Cave Sandstones of South Africa, the Forest Sandstones of Southern Rhodesia, the Dockum and Chinle of North America, and the Lufeng Series of China. The Los Colo- rados beds and the E] Tranquilo Formation of Argentina, the faunas of which are now being studied by Bonaparte and Casami- quela, probably belong to the same group. Faunas of the B type are known in South America (Santa Maria, Ischigualasto, Cha- nares ), Africa (Manda beds, Molteno beds, Ntaware Formation), and India (Maleri beds). Some faunas, such as those from the Elgin Sandstones (Scotland) and Maphutseng (Basutoland), seem to be transitional between the B and C assem- blages. The saurischians of the late Triassic faunas belong to three different infra- orders, which are clearly recognizable at the time of their first appearance in the Lower Triassic, namely the Coelurosauria, the Sauropoda, and the Palaeopoda (I use here Colbert’s [1964] new name _ in- stead of Prosauropoda, as this last con- cept is confusing both in intension and in extension). The coelurosaurians are represented in the Upper Triassic by the family Podokesauridae, Hallopidae, and Segisauridae (the second not surely distinct from the first). They were slightly-built upland predators, distinguished from other contemporaneous dinosaurs by the “doli- choiliac” pelvis (Colbert, 1964), advanced bipedal gait, birdlike feet, calcaneum usu- ally with a tuber, long neck, relatively elongated skull. It is now clear that the true Carnosauria of the Jurassic and Cre- taceous are an offshoot of the Coeluro- sauria, with which they share the same type of pelvis, the birdlike feet, and many other features. Both infraorders are there- fore grouped in the suborder Theropoda of Marsh, giving to this taxon-concept a nar- rower extension than that in the current conservative classification. The Sauropoda are represented from the very beginning of the Upper Triassic by the Melanorosauridae. This family is usu- ally placed within the “Prosauropoda” (= Palaeopoda). Recent work by Ellenberger and Ginsburg (1966) demonstrates that they are quadrupedal and very close to the true sauropods. These authors and Attridge (1963) suggested that the mel- anorosaurids should be considered true sauropods, a suggestion that seems very Ne) Ol Ti EARLY ARCHOSAURIAN EVOLUTION + Reig reasonable to me. Though disregarding the melanorosaurids as direct ancestors of the sauropods, Charig et al. have convincingly demonstrated that “the line of evolution which led from the pseudosuchians to- wards the sauropods was entirely quadru- pedal; thus the sauropods were not, as commonly supposed, quadrupedal rever- sions from bipedal forebears. “The various families of prosauropods were offshoots from this main, quadru- pedal sauropodomorph line, representing adaptations to different habitats which dif- fered especially in their degree of bi- pedality; none survived the Trias” (1965: 205). From the new evidence provided by Ellenberger and Ginsburg (1966), one arrives at the conviction that the melanoro- saurids should belong to this “main, quadru- pedal sauropodomorph line” which, from its very beginning, was part of the evo- lution of the true sauropods. Melanoro- saurids are known from the Middle-Upper Triassic boundary, as represented by the remains referred to Euskelosaurus by Ellen- berger and Ginsburg (1966), which come from the “Passage beds” of Basutoland (the “Maphutseng dinosaur” of Charig et al., 1965); a hind leg from the same_ beds described by Crompton and Wapenaar (in press) (reported by Charig et al. as the “Blikana dinosaur”); and the “Soebeng trackways,” footprints of a big quadru- pedal dinosaur, mentioned by the above authors and by Ellenberger and Ginsburg (1966). Besides these early finds, melano- rosaurids are relatively abundant in the Red Beds of South Africa. The Melanoro- sauridae are likely to have been herbivores and swamp-dwellers; the possibility that the family would include carnivorous forms has been suggested by Charig et al. (1965), but there are good reasons to doubt this. The evidence supporting such a view is far from conclusive and it is not very likely that these enormous quadrupedal marsh- dwellers could have been sustained by any food other than large amounts of green matter. 276 The Palaeopoda are represented by the Thecodontosauridae, the Plateosauridae, and the “Triassic carnosaurs.” This last group has been demonstrated (Colbert, 1964; Charig et al., 1965; Walker, 1964) not to have any relationships with the true, post-Triassic carnosaurs, and to be closely connected with (or even inseparable from, as maintained by Charig et al., 1965) the first two families. The thecodontosaurids are medium-sized bipedal or semi-bipedal upland herbivores, known from different levels of the Upper Triassic of South Africa, China, Europe, and North America. The plateosaurids are large European and Asi- atic (probably also South American) bi- pedal plant-feeders dwelling in lowlands. The carnivorous palaeopods are here con- sidered as belonging to one distinct family, for which the name Gryponychidae must be used.!. Though the facts of association of skull and postcranial bones are scarce and dubious, there is enough evidence to show that carnivorous palaeopods were living in the Upper Triassic. The conve- nience involved in placing these forms in families containing herbivorous dinosaurs is not very great, as one of the current criteria for family separation is distinction in ecological type. It is therefore preferable to separate the gryponychids as a carnivo- rous offshoot of the palaeopods, though recognizing that they are close to the other two families with which they share the same type of pelvis, tarsus, and limb structure. All the palaeopods are closely related, and they are also very similar to the melanorosaurids and later sauropods, so that it makes sense to group both palaeo- pods and sauropods in a suborder Sauro- podomorpha as proposed by Charig et al. (1965) and accepted by Romer (1966b). 1 Both Walker (1964) and Charig et al. (1965) have indicated that the name Palaeosauridae can- not be used, as Palaeosaurus Riley and Stutchbury is preoccupied by Palaeosaurus Geoffroy; Kuhn (1959) created the name Palaeosauriscus to re- place the first name. Bulletin Museum of Comparative Zoology, Vol. 139, No. 5 Charig et al. make a convincing case in claiming that this term, coined by von Huene (1932), is preferable to Pachypodo- sauria of the same author, a name applied to the unnatural assemblage of sauropods, “prosauropods,” and carnosaurs. Within the Sauropodomorpha, the distinction of pa- laeopods and sauropods as infraorders is meaningful, as it adequately expresses the evolutionary situation. The sauropods seem to have played a secondary role during Triassic times, only evolving to full-fledged diversity and abundance after the close of that period. The palaeopods, most prob- ably derived from a quadrupedal pro- melanorosaurid or melanorosaurid stock, represent the main radiation of Triassic Sauropodomorpha, and they evolved into both upland and lowland plant-eaters, and upland bipedal carnivores. What do we know about the probable origin of the three groups of dinosaurs al- ready well established at the very be- ginning of the Upper Triassic? Not too much, but at least enough to reveal that the history of the sauropodomorphs and coelurosaurs must be traced well back into the Triassic. Saurischian remains are known from the Middle Triassic of Argentina (Reig, 1963a) and Brazil (von Huene, 1942). The Argentinian fossils are rather abundant, and they come from the Ischi- gualasto beds, a formation that, following Romer (1966a) and Reig (1963a), contains a fossil assemblage that clearly belongs to the B type of faunas representing, perhaps, an upper Ladinian stage (i.e., the latest Middle Triassic). The Brazilian remains occur from the Santa Maria beds, which are generally agreed to be older than the Ischigualasto and roughly equivalent to the Manda beds of Tanganyika. According to our present knowledge, the Argentinian material represents at least four genera of saurischians, only three of which have been described (Reig, 1963a ). One genus is a very small, undescribed coelurosaur. Another coelurosaur is repre- Sa = | 5 than 55° (Bufonidae, Brachy- cephalidae); (2) muscle originating from squamosal and dorsal fascia, squamosal angle 45°-50° (Ranidae, Microhylidae, Rhacophoridae, Leptodactylidae, Hylidae); (3) muscle originating only from dorsal fascia, squamosal angle < 45° (Discoglossi- dae, Pelobatidae ). He noted that all groups passed through condition (1) in their de- velopment and that care should be taken in using this character because of the pos- sibility of parallel paedomorphy. In specimens I measured, the squamosal angle was 45° or less only in Megophrys; but in Eopelobates guthriei nov. (see be- low), E. hinschei, and Scaphiopus skinneri nov. (see below), the angle fell between 45° and 50°. All other pelobatines were between 56° and 73°, the highest in S. couchi. This change in the squamosal angle suggests that the development of a higher skull and larger orbit in pelobatines (dis- cussed above) may involve a paedomor- phic trend. Ossified Sternum Kluge (1966, p. 17) noted that Griffiths (1963, p. 271) was incorrect in stating that all pelobatids have an ossified sternal ap- paratus. Zweifel (1956, p. 24) states that the sternum is cartilaginous in Scaphio- pus. This seems to be true in general, but a specimen of S. couchi chosen at random 300 Bulletin Museum of Comparative Zoology, Vol. 139, No. 6 Figure 3. ventral view of ethmoid, ethmoid cartilage stippled; both * 3. T= turbinal fold in cartilage; VY = vomer; W = lateral wing. lateral process; P = palatine articulation surface; (MCZ 64374, cleared and stained) has an irregular sternal ossification (Fig. 9d) in the stylar region, and an ossified, paired omosternum as well. Although this con- dition has not yet been described in a fossil Scaphiopus and I have not checked it in S. holbrooki, it is possible that some ossifi- cation is the primitive condition in Scaphio- pus. Ethmoid The ethmoid shows considerable inter- generic variation in general shape, and since it is often found in fossils it can be useful in identification. I lack sufficient material for a meaningful study on intra- generic variation, but the material available seems to be relatively consistent and to demonstrate that some species may be identifiable on this basis as well. In Megophrys the ethmoid is pinched-in ventrally, but develops lateral wings dor- sally, giving a rhombic shape to the dorsal surface of the bone. In Leptobrachium no lateral wings are present and the ethmoid is hour-glass shaped. The lateral processes (Fig. 3) are prominent, but are not strongly separated from the anterior process by emargination in the choanal region. The (A) Megophrys carinensis, AM 23965, ventral view of ethmoid and vomer; (B) Megophrys robusta, MCZ 25735, —.—.—.—.—=dorsal border of ethmoid roof; L = palatines underlie the lateral processes and the vomers lie along the lateral sides of the anterior process. Internally there is only a faint development of a turbinal fold between lateral and anterior processes, if it is present at all (Fig. 4); however, a turbinal fold is present in cartilage. The internal surface is flattened dorsoventrally | and the capsule area is completely roofed by the ethmoid; only at the anterior end is it covered by the nasal. In Pelobates cultripes and P. syriacus, the anterior proc- ess is moderately developed, but the end of the process is relatively blunt with only a slight median projection. The turbinal fold is moderately developed. In Pelobates fuscus and especially in | Scaphiopus, there is marked separation of | the anterior and lateral processes turbinal fold projects strongly in ventral view as the capsular process (Fig. 5), and | the anterior process itself has two separate projections. The capsular process is much better developed in Scaphiopus (again, by | emargination. In the emarginated area be- tween those processes, P. fuscus has a. moderately developed turbinal fold, and | Scaphiopus a very well developed one. In | both species (except S. holbrooki), the Figure 4. Ethmoids in anterior view; a, Megophrys monti- cola, AM 23964; b, Eopelobates grandis, PU 16441; c, Mac- ropelobates osborni, AM 6252; d, Pelobates cultripes, UMMZ S-2630; e, Pelobates fuscus, MCZ 1012; f, Scaphiopus couchi, AM 56284; a-d, & 3; e-f, & 6; diagonal hatch- ing = broken surface, dashed line = restoration, stippled area = cartilage attachment surface; A = anterior process; T = turbinal C = capsular L = lateral fold. process; process; except in S. holbrooki) and is somewhat different than in Pelobates fuscus. In Eopelobates intermediate conditions prevail, so far as this can be determined in the fossil material. There is definite sepa- ration of lateral and anterior processes by emargination in E. bayeri, although the general configuration is more Megophrys- like than pelobatine. The anterior process as shown in E. guthriei n. sp. and E. bayeri ossifies very little (see p. 312 and Fig. 6), and remains broad as in megophryines. Fossi. PELOBATID FRocs « Estes 301 ANTERIOR Figure 5. Pelobatine ethmoids in ventral view; a, Pelobates fuscus, MCZ 1012; b, Scaphiopus couchi, AMNH 56284; c, S. holbrooki, MCZ 25577; d, P. cultripes, UMMZ S-2730; e, P. varaldii, MCZ 31970, with ethmoid cartilage in stipple; all xX 2. Irregular line = depression; —.—.—.—.—= dorsal border of bony ethmoid; — —.— — . — — = dorsal border of ethmoid cartilage. A = anterior process; C = capsular process; L = lateral process; P = palatine articula- tion surface; PM = premaxillary articulating surface. This situation is approached in P. varaldii (separated from P. cultripes by Pasteur and Bons, 1959; Fig. 5e, this paper). A separate anterior process is not present on E. grandis (Fig. 7) and is not visible in the other species. In the ventral view of E. bayeri, a depression develops between lateral and anterior processes, reflecting a weak turbinal fold development like that of Megophrys and Pelobates, but not as distinct as in Scaphiopus. The ethmoid of Macropelobates is as in P. cultripes, as far as can be determined (cf. Figs. 7b; 5d). In all pelobatines, the dorsal ethmoid roof of the nasal capsule is absent and the entire capsule is then roofed by the nasal (Fig. 5), but in Megophrys the ethmoid floor and roof are of about equal extent and the nasal provides cover for the cap- sules only anteriorly (Fig. 7). The extent of roofing by ethmoid in Eopelobates can be seen only in E. grandis, and is approxi- mately as in the megophryines. In the subgenus Spea of Scaphiopus, the anterior process may become extremely large and 302 Bulletin Museum of Comparative Zoology, Vol. 139, No. 6 Figure 6. Ventral view of ethmoid of (A) Eopelobates guthriei, MCZ 3493, X 3; (B) E. bayeri, CUPI 6.874, * 5.5. Dashed line = restoration, dotted line = broken bone outline; P = palatine articulation; VB = boss for vomerine teeth. flared anteriorly (e.g. S. intermontanus), producing the most extreme pelobatine condition. The bony ethmoid is, of course, merely an ossified portion of the ethmoid cartilage and not coextensive with it. The cartilage itself is also quite different in the two modern subfamilies (cf. Figs. 3b, 5e) and within that cartilage, the above-noted variations in ossification occur. The re- treat of the bony roof of the pelobatine ethmoid is accompanied by regression of the cartilage to a partial ring surrounding the naris and a thin, membranous cover over the main unossified part of the cap- sule. Figure 7. (A) Eopelobates grandis, PU 16441, ventral view of ethmoid and vomer; (B) Macropelobates osborni, AM 6252, ventral view of ethmoid; both & 3. Dashed line = restoration; dotted line = broken bone surface; —.—.—.— = dorsal border of ethmoid; V = vomer; P = palatine articulation surface. Fossiz PELOBATID FRocs + Estes 303 Figure 8. Eopelobates anthracinus, type, BM R-4841; left, restoration of dorsal and lateral views of skull; right, camera lucida drawing of vertebral column, posterior skull roof outline shown anteriorly; X 6. Without the knowledge that the large rodlike anterior process is present in carti- lage in Pelobates cultripes, the similarities of Pelobates fuscus and Scaphiopus in ethmoid construction might seem to indi- cate that the spadefoot genera are closely related through P. fuscus, but the latter is not likely to be ancestral to the North American spadefoots, as is discussed further below. Scaphiopus holbrooki, the most primitive member of the genus, is inter- mediate between P. cultripes (or P. syria- cus) and other Scaphiopus in this regard; S. couchi, S. (Spea), and P. fuscus have independently ossified the anterior process of the ethmoid as far anteriorly as the pre- maxillae. It would be of considerable interest to study olfaction within the pelobatines; their nasal capsules indicate some strong adap- tive trends not seen in the aquatic Megophrys and Eopelobates. 304 Bulletin Museum of Comparative Zoology, Vol. 139, No. 6 Figure 9. Scapulae and sternal styles of pelobatids. a, Pelobates cultripes, UMMZ S-2629; b, P. syriacus balcanicus, MCZ — 50690, style only; c, Megophrys monticola, AM 23964; d, Scaphiopus couchi, MCZ 64374; e, Eopelobates hinschei, MME — 6692, scapula only; f, E. grandis, PU 16441; g, E. anthracinus, BM R-4841, scapula only; h, E. bayeri, CUPI 6.874; a-g, X 3; h, X 4.5. Chronological Review of Described Eopelobates CLASS AMPHIBIA SUPERORDER LISSAMPHIBIA ORDER SALIENTIA Family Pelobatidae Eopelobates anthracinus Parker 1929 Parker’s account is good, but better knowledge of other species allows some additional discussion. In the skull, the pat- tern is approximately as Parker described it, but contrary to the implication of his figure, there is no process of the squamosal leading towards the frontoparietal; this is partly the result of the bone being under- lain by the pterygoid and partly the result of crushing in the area. Also, the squamosal is more hatchet-shaped posteriorly than in his figure. The frontoparietal shows prom- inent, well-defined pits on the lateral edges, and sculpture is more apparent laterally than medially. Because of crush- ing, the exact shape of the frontoparietal — is difficult to determine, but it is about © as indicated in Figure 8. There is a groove between the two halves of the frontopari- etal that probably indicates a suture, but since all adult E. bayeri specimens appear fused, this cannot be certain. There is a complete maxillary arcade; the quadrato- jugal can be seen clearly on the photograph (Fig. 1), and there is a strong quadrato- jugal process of the maxilla. The teeth are pedicellate. The bone in the left orbit that —— ¢ mas 4 Figure 10. Table 1, 8a. Eopelobates hinschei, MME 6692; X 3; see Parker thought was the dentary is actually the prearticular. The anterior tip of the parasphenoid appears to be visible near the anterior end of the left frontoparietal, but the impression is vague. In the post- cranial skeleton, imprints of transverse processes on all vertebrae occur on the matrix, contrary to Parker’s statement: these are long on the anterior vertebrae but short and anteriorly directed on the posterior ones (Fig. 8) in accord with other species of Eopelobates, Pelobates, and some Megophrys. Again contrary to Parker, the cleithrum is visible on the morphological left side. Parker remarks (1929, p. 280) that the skull “appears to have been almost identical with that of the recent Pelobates.” In fact, the skull differs from that of Pelobates and Scaphiopus and resembles that of other Fosst. PELoBATID FRocs « Estes 305 TABLE | SYNONYMY OF EOPELOBATES HINSCHEI Eopelobates hinschei (Kuhn) 1. Halleobatrachus hinschei, type, MME 1312, Kuhn, 1941, p. 353, pl. I, fig. 1. Parabufella longipes, type, (unique specimen, no number? ), ibid., p. 358, pl. 4, fig. 5. Palaeopelobates geiseltalensis, type, MME 6695, ibid., p. 360, pl. 1, fig. 5. Archaeopelobates efremovi, type, (no num- ber), ibid., p. 361, pl. 3, fig. 6. 5. A. eusculptus, type, MME 6728, ibid., p. 362, pl. 4, fig. 1. 6. Amphignathodontoides eocenicus, type, MME 6744, ibid., p. 364, pl. 6, fig. 1. 7. Germanobatrachus beurleni, 6719, ibid., p. 368, pl. 2, fig. 4. 8. The following specimens referred by Kuhn to the above genera are also referable to E. hinschei: a. Palaeopelobates geiseltalensis, MME 6692, jolly dhs rake, 2) b. P. geiseltalensis, pl. 2, fig. 5. c. P. geiseltalensis, MME 6696, pl. 3, fig. 2. d e f= co type, MME . P. geiseltalensis, pl. 3, fig. 7. . cf. Archaeopelobates eusculptus, pl. 2, sateq, Jl f. cf. A. eusculptus, MME 6762, pl. 4, fig. 3. g. PA. efremovi, MME 1572 h. Opisthocoelellus weigelti, pl. 4, fig. 2 (not the holotype). i. O. weigelti, MME 4995, pl. 5, fig. 2 (not the holotype). Eopelobates in having a flattened or concave skull table and in having approxi- mately subequal orbit and temporal open- ings. The dermal sculpture is coarse and open, more or less as in the other European Eopelobates. There is an anterior lamina on the scapula (Fig. 9). The urostyle is separate and there were two, perhaps three, post- sacral vertebrae, although crushing makes the exact number uncertain (Fig. 8). The skull restoration of Eopelobates anthracinus (Fig. 8) was made from camera lucida tracings of the individual bones; the tracings were then fitted to- gether. Since the bones were all flattened after burial, their somewhat different shape in the restoration results from curvature incorporated into the three dimensional 306 Bulletin Museum of Comparative Zoology, Vol. 139, No. 6 Lew’ Ley A | are Figure 11. Right posterior half of pelobatid skulls, dorsal view. guthriei, MCZ 3493; c, Pelobates fuscus, MCZ 1012; d, Megophrys lateralis, AM 23549; all X 3. f = frontoparietal; p = prootic;= s'=—"squamosals 7-2 = margin of prootic covered by squamosal; cartilage stippled. model. The skull height (especially an- teriorly ) is the major feature in doubt, but as given it is approximately intermediate between the flattened skulls of Megophrys and the domed skulls of Pelobates and Scaphiopus. The bone outlines do not allow much deviation either way from the outline suggested here. There is a well- defined groove between the frontoparietals, but a distinct suture cannot be seen. The exact shape and placement of the nasals is conjectural, but the arrangement given is consistent with what remains of the bones. The photograph of the specimen (Fig. 1) does not allow confirmation of all bone outlines; this was only made possible by comparing many photographs taken with light coming from different angles and from drawings made at the time of study of the original specimen. Eopelobates hinschei (Kuhn, 1941) This species was originally described as Halleobatrachus hinschei by Kuhn (1941, p. 353) from the middle Eocene Geiseltal deposits near Halle, Germany. As Spinar (1967, p. 218) correctly pointed out, this species belongs to the Pelobatidae rather than to the Palaeobatrachidae. Much of the other material described by Kuhn also belongs to the genus Eopelobates. All the a, Scaphiopus h. holbrooki, MCZ 58003; b, Eopelobates | characters of the genus are clearly visible — in this series of specimens. The photograph — given here (Fig. 10) shows one of the best skulls available. Kuhn gave six generic and seven specific names to this sample, but on the basis of proportions alone, the fossils can easily be related and demon- strated as a growth series (Fig. 25). Hecht (1963, p. 23) has already commented ac- curately on the reliability of Kuhn’s study, but contrary to Hecht, however, Spinar | (1967) has shown the presence of palaeo- batrachids at Geiseltal. I think it unlikely that Eopelobates bayeri (Spinar, 1952) is conspecific with E. hinschei. As Figures 19 and 20 show, the squamosals are different, and there are proportional differences of the nasals. | However, the two species are related and | both have rather elongated frontoparietals, | though that of E. bayeri is fused (Fig. 12). Their scapulae are also similar (Fig. 9e, h), as is their ratio of tibiofibula-femur to head-vertebral column length (Fig. 29). Prof. Spinar is presently studying the speci- mens of E. hinschei and E. bayeri, and his report will deal with this matter more fully. Table I lists the synonymy of Eopelo- | bates hinschei as I interpret the Geiseltal remains. Figure 12. Estes Fossi. PELoBATID FROGS ° 307 Skull roof of (A) Eopelobates hinschei, MME 6692 (8a, Table 1), X 4.5; (B) E. bayeri, CUPI 6.874; xX 4.8; dashed line = restoration; dotted line = broken bone outline. Eopelobates bayeri Spinar 1952 As the figure shows, the late Oligocene —middle Miocene Czechoslovakian species E. bayeri has all of the characters of the genus noted above (Figs. 2, 12b). Vari- ation may exist with respect to fusion of urostyle and sacrum; in the type specimen of E. bayeri, they appear to be separate (perhaps because of poor preservation), but in the new complete specimen are ap- parently fused. They are separate in E. bayeri tadpoles as in tadpoles generally. E. bayeri has a somewhat similar squa- mosal to E. anthracinus, but other features, such as frontoparietal shape and ratio of limb to body (Fig. 29), are different. Both species have more sculpture laterally than medially on the frontoparietal, but E. bayeri lacks the large pits seen in E. anthracinus. The two species seem quite clearly different. The Czechoslovakian material confirms the absence of a spade, and the orientation and shape of the trans- verse processes is in accord with those of the other specimens of Eopelobates, some Megophrys, and Macropelobates. Of special interest is the shape of the ethmoid, which is well shown on the new specimen of Eopelobates bayeri (cf. Figs. 2, 6). It is similar to that of E. guthriei n. sp. (see p. 312) but differs from that of E. grandis. The exact contour of the nasals is con- jectural. They have been thrust backward over the frontoparietals, and their relations to the latter in the restoration have been determined by triangulation with other skull parts and by comparison with other Eopelobates specimens (including the type of E. bayeri). On the left side of the restoration (morphological right; the speci- men is an imprint), the two parts of the 308 nasal thrust apart by crushing have been rejoined. Compensation for flattening of the nasals in preservation has been made laterally in the restoration in order to make all restorations comparable. Eopelobates neudorfensis (Wettstein- Westersheimb, 1955) was based on dis- articulated elements derived from a Middle Miocene (Helvetian) fissure filling in southern Czechoslovakia. Most of the diag- nostic elements are preserved. The fronto- parietal is fused except at the anterior margin and is indistinguishable from that of the new specimen of Eopelobates bayeri. The squamosal has a hatchet-shaped tym- panic process as in E. bayeri and E. an- thracinus (Fig. 19c). The maxilla has a strong posterior process for the quadrato- jugal. Urostyle and sacrum are separate. The close association of this species with E. bayeri in morphology, time, and ge- ography indicates that it is a synonym of the latter. Eopelobates grandis Zweifel 1956 A few additions and corrections can be made to Zweifel’s excellent account of this early Oligocene North American species (Zweifel, 1956). Although the maxilla and squamosal are in firm contact, there is no contact of squamosal and frontoparietal as Zweitel indicated (1956, p. 5). The right squamosal, on which he apparently based this interpretation, has been rotated and displaced up against the frontoparietal. Normal relationships to the frontoparietal are retained by the left squamosal, as con- firmed by Eopelobates anthracinus, E. bayeri, and E. guthriei n. sp. (see p. 311). The squamosal shape is more rounded than Zweifel’s figure indicates, and is essentially a deeper version of the E. guthriei squa- mosal (cf. Figs. 19d and 20d). The fronto- parietal differs from that of E. guthriei and E. anthracinus, but, except for being rel- atively short, it is in accord with that of other Eopelobates (Fig. 13a). The quadratojugal (identified as stapes by Bulletin Museum of Comparative Zoology, Vol. 139, No. 6 Zweifel) is present and is excavated for a posterior projection of the maxilla as in Megophrys. The vomer is now exposed (Fig. 7a) and is like that of Pelobates in having a rather expanded anterior wing, an almost transversely-oriented tooth row (rather than a patch), and a dorsal flange clasping the side of the ethmoid as in P. cultripes. The ethmoid is more megoph- ryine than in any other Eopelobates. It is flattened and dilated anteriorly, and has prominent lateral processes that are deeply notched on their ventral surfaces for the palatines (Fig. 7a). The dorsal surface of the ethmoid is little emarginated. The order of difference from ethmoids of other Eopelobates is about the same as between those of the modern species Megophrys carinensis and M. robusta (Fig. 3). The scapula has a_ well-developed anterior lamina (Fig. 9f), which has a straight anterior border as in E. anthracinus. The wide posterior extent of the nasal resembles that of E. guthriei n. sp. (see Fig. 13) and the pelobatines. This resemblance tends to link the two American species, but I believe it unnecessary to distinguish them generically. Zweifel’s reference of this species to Eopelobates is undoubtedly correct; it is probably a distinct species because of ethmoid shape, wide fronto- parietal, and rounded tympanic process of the squamosal. Hecht (1963, p. 23) has suggested that this animal is a distinct genus, but it differs no more from other Eopelobates than the Recent Megophrys carinensis differs from M. lateralis, for in- stance. Eopelobates sp. Hecht (1959, p. 131) described a mego- phryine sacrum from the middle Eocene Tabernacle Butte local fauna of Wyoming and correctly noted a close resemblance to Eopelobates grandis Zweifel. It is reason- able to refer the Tabernacle Butte specimen (AMNH 3832) to Eopelobates without spe- cific designation. Fossi. PELOBATID FRocs « Estes 309 A. Figure 13. Skull roof of (A) Eopelobates grandis, PU 16441, X 1.8; (B) E. guthriei, MCZ 3493, XX 3. Dashed line = restoration; dotted line = broken bone outline. Mlynarski referred to Eopelobates sp. material from the Pliocene of Poland. The specimens consist only of sacra having separate urostyles. Other fused sacra and urostyles and characteristic skull elements he referred to Pelobates cf. fuscus. Since, however, Eopelobates is otherwise un- known later than middle Miocene, and since Pelobates cultripes often has partially or completely separated urostyles, it seems unlikely that Eopelobates is represented in the Polish material, at least in the absence of characteristic skull elements. These elements may be referable to Miopelobates (see below). Since the salamander Andrias is now known to occur in the European Pliocene (Westphal, 1967) there is no ap- parent reason why Eopelobates might not also have persisted, but at present there is insufficient reason to confirm its extension beyond the middle Miocene. DescriPpTION OF NEw MATERIAL OF EOPELOBATES Eopelobates guthriei, n. sp. Type: MCZ 3493, nearly complete skull and associated fragmentary scapula. Diagnosis: Differs from other species of Eopelobates in having a narrow tympanic process of the squamosal combined with a triple emargination of the frontoparietal margins and a relatively short skull. Etymology: Patronym for Dr. Daniel Guthrie, who collected the unique speci- men in 1962. Locality: NE 1/4, SE 1/4, Sect. 16, T 39 N, R 90 W, Fremont County, Wyoming. Horizon: Upper part of the Lysite mem- ber, Wind River Formation. Age: Early Eocene (Lysitean, Sparnacian equivalent). Preservation: Only the skull, portions of the prearticular region of the jaws, and an associated fragment of left scapula are present (Fig. 14). The slightly crushed skull is well preserved on the right side, but on the left the temporal region is missing. The premaxillae, the anterior por- tions of the nasals, and the anterior part of both maxillae are missing. Although the skull is slightly flattened, distortion is limited for the most part to the peripheral tooth-bearing and temporal bones. The ventral borders of the maxillae late Bulletin Museum of Comparative Zoology, Vol. 139, No. 6 310 *xiuyow — ad -diys asip0d 'Z X ‘BulwokAA ‘aua09 AjJDa ‘sMaIA [D’JUaA ‘p PUD ‘|DSIOP ‘OD ‘o4idio00’q ‘jo1ayo] yyBu ‘oO ‘espe ZOW ‘[|Ny¥s edAjojoy ‘ds -u ‘ialsyjnB sayoqojadoz “py, ainbi4 are warped laterally, a condition especially true of the left maxilla, whose lateral aspect now faces almost dorsally. Ventrally, the left palatine has been pushed dorsally away from the ethmoid, but it seems to have retained its natural relationship to the latter. Description: The skull indicates a rather broad-headed animal with subequal dorsal temporal excavations and orbits, separated by postorbital processes. The dorsal skull region is flattened and concave medially and bounded by weak crests. The skull as a whole is covered with a well-developed dermal sculpture. Posteriorly the nasals meet on the mid- line, diverge at their posterior borders to expose the ethmoid, barely meet the fronto- parietals, and extend laterally to meet the maxillae. The nasals are weakly crested in the area continuous with the lateral borders or the frontoparietals, and slope toward the midline between these crests. The nasals are sculptured on their entire preserved surface. Located between the frontoparietals and the nasals is a smooth, somewhat diamond- shaped portion of the ethmoid, which is the center of a depression bounded an- teriorly by the nasal crests mentioned above and posteriorly by the lateral borders of the frontoparietals. The depression ex- tends to the posterior border of the skull. The paired frontoparietals are subrec- tangular in shape and prominently sculp- tured. The postorbital processes are situated anteriorly about two-thirds the fronto- parietal length from the apex of the fora- men magnum. The anterior tip of the left frontoparietal is missing, increasing the apparent depth of the ethmoid depression. The anterior tip of the right frontoparietal touches the nasal at its lateral border. The undistorted occiput, the lateral crests of the frontoparietals, and the symmetry of the cranial roof indicate that the midline de- pression of the frontoparietal, ethmoid, and nasals is natural. The postorbital processes are the widest points on the frontoparietals Fossi. PELOBATID FRocs « Estes alt except for the posterior tips, which extend onto the paired projections of the paroccip- ital processes on the occiput dorsal to the condyles. Posteriorly the frontoparietal reaches the apex of the foramen magnum, from which point lambdoidal crests form concave curves, extending towards the paired projections noted above. In occipital view the median skull roof is depressed; the highest points are on its lateral borders. The occipital surface of the skull is well preserved and relatively little distorted; there is little breakage ex- cept for the missing left temporal region. The most prominent bones are the otoccip- itals, which meet above and below the triangular foramen magnum. The large circular foramina for the ninth and tenth cranial nerves are recessed at the base of the prominent hemispherical occipital condyles. Lateral to these foramina, the otoccipital forms the posterior border of the fenestra ovalis, forming a prominent rounded process underlain by the para- sphenoid. Laterally the otoccipital forms a prominent knobbed paroccipital process, which is capped by the frontoparietal. The stapes is forked proximally and is closely appressed to the ventral surface of the lateral extension of the otoccipital. The fenestra ovalis is open ventral to the proximal end of the stapes and dorsal to the rounded process of the otoccipital mentioned above; a large opercular space is present as in recent spadefoots, and since the very delicate stapes is preserved in place, a calcified operculum was probably absent. The right squamosal has been displaced dorsally at its posterior articulation with the otoccipital; in fact it has pivoted some- what (along with pterygoid and maxilla) on the lateral tip of the otoccipital, so that the greatest dorsal displacement is at the medial end of the squamosal, and the de- scending (quadrate) process of the squa- mosal has been rotated mediad, carrying with it the remains of the lower jaw. The quadrate is represented by a small sliver 312 clasped between squamosal and pterygoid. The posterior end of the lower jaw is miss- ing, as are the tip of the quadrate and the posterior border of the maxilla; apparently the quadratojugal and posterior process of maxilla (if present) were broken off in the dislocation of the temporal region. In the ventral view, the posterior portion of the parasphenoid is well preserved, but the cultriform process is faulted by the right scapula and then terminates by break- age at the ethmoid border. The parasphe- noid extends anteriorly from the border of the foramen magnum to the _ posterior border of the ethmoid. The lateral arms of the parasphenoid form the floor of the fenestra ovalis region. Prominent nuchal, pterygoid, and retractor bulbi muscle scars, set off a trapezoidal, flattened area midway between the lateral arms of the para- sphenoid. The otoccipitals extend posteriorly some- what beyond the posterior borders of the parasphenoid, completing the fenestra ovalis region ventrally. There is a large opening in the posterior braincase region, bounded anteriorly by ethmoid, ventrally by parasphenoid, pos- teriorly by otoccipital, and dorsally by frontoparietal. The major cranial nerves emerged through this opening, but only the prootic foramen has any individuality. It is a narrow suboval notch, open an- teriorly. The ethmoid is broadly exposed between the parasphenoid and the vomers, and ventral processes of the frontoparietals clasp it laterodorsally. It sends _ broad, crested processes laterally toward the maxillary arcades, and posterodorsal to each of these open the foramina for the anterior (orbital) extensions of branches of the occipital arteries. Anterior to each lateral ethmoid process is a depression, from which bone is missing as a result of erosion and breakage. Anterior to these depressions, the curved choanal borders of the vomers are still preserved in natural position. A raised area over the left an- Bulletin Museum of Comparative Zoology, Vol. 139, No. 6 TABLE 2 MEASUREMENTS OF EOPELOBATES GUTHRIEI The following measurements (in mm) are rela- tively unaffected by crushing or distortion: 1. posterior median height of the skull from the most dorsal point on the frontoparie- tals to the most ventral point on the mid- line of the parasphenoid 49 2. height of foramen magnum ___---- 2.0 3. width of foramen magnum -__-_---------- 4.0 4, maximum width across the paroccipital processes: 2 ta ee Jalal 5. maximum width across occipital condyles 6.2 6. maximum length of stapes as preserved 4.7 7. maximum length of frontoparietal from apex of foramen magnum to anterolateral tip! She 12.0 8. length from apex of foramen magnum to postorbital process 2 8.3 9. maximum anteroposterior length of right squamosal) Wu 2 eee WIL 10. length of posterior projection of squa- mosal behind anterior margin of tym- panic: cavity. ee 5.0 11. maximum width across postorbital proc- €ssesior frontoparietals) == ae 9.0 12. maximum height of posterior process of squamosall i220 ee 2.0 terior part of the ethmoid probably repre- sents the left vomerine tooth plate. The other parts of the vomers are not preserved. Laterally, an irregular, broken bar of bone seen on the left side probably represents the palatine. The pterygoid is present as a complete bone only on the right side, and is strongly curved, bending broadly toward the quad- rate region on the one hand, and toward the otoccipital and maxilla on the other. In lateral view the relationships of the maxilla, squamosal, quadrate, pterygoid, and prearticular are undistorted on the right side. On the left side, only the middle part of the maxilla is present; the temporal region and premaxilla are missing. The maxillae bear pedicellate teeth and are heavily sculptured in a pattern similar to that of the frontoparietals. On the right side, the posterodorsal corner of the maxilla meets the squamosal in a broad horizontal suture. f MS Figure 15. oO ‘ lo); G: | d. Estes i Fossi. PELOBATID FROGS = 313 Orbitotemporal opening relationships in pelobatids; all are of right side, anterior towards the top. a, Eopelo- bates anthracinus, BM R-4841; b, E. grandis, PU 16441; c, Pelobates fuscus, MCZ 1012; d, Scaphiopus skinneri, FAM 42920; e, S. holbrooki, MCZ 58003; f, Megophrys carinensis, AM 23965; g, E. hinschei, MME 6692; h, E. guthriei, MCZ 3493; i, S. couchi, AM 14478. Not to same scale; O = orbit; P-S = prootic and squamosal roof of ear region; dashed line = pos- terior border of orbit in all, and restored portion of frontoparietal in d. The T-shaped squamosal is well pre- served on the right, and, like the maxilla, is sculptured on the crossbar of the T. Anteriorly the bone is much broader than it is posteriorly. The posterior process of the squamosal curves posteriorly over the tympanic cavity, expands slightly at its posterior border, and forms an acute angle with the descending process of the squa- mosal. The latter process is flattened an- teroposteriorly and has a sharp crest sepa- rating the tympanic cavity from the lower temporal excavation. The descending proc- ess is closely applied to the posterolateral border of the pterygoid, and is separated from it ventrally by the sliver of quadrate noted in the description of the occipital view. The ventral portion of the quadrate is lost, as is the articular. Pieces of the prearticulars indicate the position of the lower jaws, and lie in their natural positions ventromedial to the maxillae. The crushed and fragmentary left scapula has been rotated 180° and now lies on the right side. Its posterior border is broken and little, if any anterior lamina appears to have been present. Discussion: Because of the possession of a concave skull roof, approximately sub- equal orbital and temporal openings, and the distinctive shape of squamosal and ethmoid (Figs. 14, 13b, 19, and 20), refer- ence of this specimen to Eopelobates seems clear. In the proportions of nasals and fronto- parietals, E. guthriei shows the relatively short skull characteristic of pelobatines and E. grandis, whereas the European 314 species, except for E. anthracinus, are more elongated and megophryine in these char- acters. E. anthracinus also shows the triple frontoparietal emargination of E. guthriei, but in squamosal shape there is close agree- ment between E. guthriei and the middle Eocene E. hinschei from the Geiseltal. In both of the latter, the anterior maxillary process of the squamosal is more expanded than the tympanic process, which is narrow and forms a wide, laterally visible roof to this part of the tympanic cavity. This roof lacks dermal sculpture (Fig. 20b, d, R). In dorsal view, E. guthriei resembles Scaphio- pus and E. grandis in the excavation of the posterior border of the otoccipital and squamosal (Fig. 15). The ethmoid of E. guthriei is incomplete and poorly preserved anteriorly but seems to resemble that of E. bayeri and (so far as can be seen in the crushed material) E. hinschei; it is relatively shorter as a result of the less elongate skull of the American form. The vomer has a broad, flat process on the posterior border of the choana as in Leptobrachium hasselti, the most primitive megophryine (Inger, 1966, p. 21) rather than a short, pointed process as in Megophrys. E. grandis has a similar proc- ess to E. guthriei, but it is relatively smaller and closer to the Megophrys con- dition. The occiput of E. guthriei is quite pelo- batine in its well-ossified paroccipital proc- esses and tubera, its general proportions and relatively simple stapes. Unfortunately, the occiput is not known in any other specimen of Eopelobates. Comparison of Figures 12, 13, and 17-23 shows that, in combination, squamosal and frontoparietal shape distinguish the modern pelobatid species. Since the specific status of the latter is based on many other criteria not available in fossils, these characters can be confidently applied to fossil samples. Hither character separately may be useful, but wherever possible the two should be used together. By this criterion the separate species Bulletin Museum of Comparative Zoology, Vol. 139, No. 6 b bela = Figure 16. Anterior parts of left prootic bones showing prootic foramina. a, Megophrys carinensis, AM 23965; b, M. monticola, AM 23964; c, Eopelobates guthriei, MCZ 3493; d, Pelobates fuscus, MCZ 1012; e, the same, MCZ 1353; f, the same, right side (reversed); g, Scaphiopus h. holbrooki, MCZ 58003; h, S. skinneri, FAM 42920; i, S. h. hurteri, AM 44244. j, S. couchi, AM 57642; k, S. couchi, AM 14478; |, S. intermontanus, AM 16916. a-b, X 2; c-l, & 4. status of E. guthriei and the Gieseltal E. hinschei is shown by their different fronto- parietal proportions. Their squamosals are very similar and show Eocene transatlantic similarities, a phenomenon already ob- served in many fossil mammals and lizards. Yet there are minor proportional differ- ences between the squamosals of the two Eocene species that are of the order of magnitude seen in modern species such as Scaphiopus holbrooki and S. couchi. The frontoparietals of Eopelobates guth- riei, however, are relatively shorter than in either E. hinschei or E. bayeri, and are very similar to those of E. grandis and E. anthracinus (cf. Figs. 8, 12, 13). The general proportions of the posterior end of the skull are more as in megophryines than as in pelobatines (Fig. 11); the pos- terior border of the prootic part of the otoccipital, however, is expanded _posteri- orly as in Scaphiopus (and to a lesser degree in Megophrys) but not as in Pelo- bates, in which the tip of the prootic is narrow as in Macropelobates (Fig. 11). | Unfortunately, this condition is not known | in other Eopelobates. The prootic foramen | of E. guthriei (Fig. 16) resembles that of Megophrys carinensis and most Scaphiopus (Scaphiopus) in its rather elongate, simple, and unrestricted opening; there is no ap- proach to the restricted or closed opening seen in Pelobates and S. (Spea). Figure 17. Left squamosals of pelobatids. Fossiz PELOBATID FRoGs + Estes 315 a, Scaphiopus couchi, AM 56284; b, the same, AM 57641; c, the same, AM 14478; d, S. intermontanus, AM 16916; e, S. holbrooki hurteri, AM 44244; f, S. h. holbrooki, MCZ 58003; all & 6. ?Eopelobates sp. In 1964, I described disarticulated and questionably pelobatid remains from the late Cretaceous Lance Formation of Wyo- ming. These elements included humeri, ilia, a urostyle, and a maxilla. The ilia (Estes, 1964, fig. 3lc) closely resemble those of most pelobatids and the superior acetabular expansion is relatively small as in Pelobates cultripes, Macropelobates, some Eopelobates, and the discoglossids. The urostyle is megophryine in possessing a single articular cotyle and transverse processes; discoglossid and ascaphid uro- styles also have the latter but have a double condyle as well. The squamosal cited as hylid-like (Estes, 1964, fig. 3la-b) closely resembles that of Eopelobates guthriei and E. hinschei and is probably pelobatid rather than hylid. In 1964, I recognized resemblances of this squamosal to those of pelobatids (p. 60), but lacking knowledge of Eocene Eopelo- bates, I was reluctant to refer a squamosal of such unusual shape to the Pelobatidae. The maxilla (Estes, 1964, fig. 3ld-e) lacks sculpture and may not be referable to the pelobatids. A fragment of a maxilla that has sculpture like that of the squamosal is now known (AMNH 8133, V5620, Lance Formation, Wyoming ). The nasal question- ably referred to the Hylidae (Estes, 1964, p. 60) may also be pelobatid on the basis of sculpture similarity to the other speci- mens. It is possible that the Lance Formation specimens may be an early record of either Eopelobates or of a related pelobatid per- 316 Bulletin Museum of Comparative Zoology, Vol. 139, No. 6 Left squamosals of pelobatids. 1012; d, P. cultripes, UMMZ S-2630; e, Macropelobates osborni, AM 6252; f, Scaphiopus skinneri, FAM 42920. line = restoration; a-c, X 6; d-f, &K 3. Figure 18. haps nearer to the discoglossids. Unfortu- nately, without articulated or at least more extensive material the reference must re- main tentative. New material from the Lance Formation and from other late Cretaceous localities has made the asso- ciation of the remains somewhat more assured now than in 1964. Several disco- glossids are present in these localities (Estes, 1969) and are represented by well- preserved and distinctive skeletal elements different from those considered here. Tue RELATIONSHIPS OF EOPELOBATES In his original discussion, Parker (1929, p. 280) suggested that Eopelobates anthra- cinus was a late representative of Noble’s (1924, p. 9) “first stage” of pelobatid evo- lution, one in which ribs, an acromion, reduction of pubis, and expansion of sacral diapophyses were found. Parker also noted a, Pelobates fuscus, MCZ 1353; b, the same, MCZ 1013; c, the same, MCZ Dashed a close similarity in the proportions of E. anthracinus to those of Macropelobates. The latter genus exemplified Noble’s “sec- ond stage” of pelobatid development by development of prehallux, dermal skull casque, and further expansion of the sacral diapophyses. In 1952, Spinar made more explicit the relationship of Eopelobates to Megophrys in his discussion of a second species of Eopelobates. Zweifel (1956), in describing a third species, E. grandis, suggested that Eopelobates might be in- cluded as a subgenus of Megophrys, but that such a course would involve “investi- gation of other units within Megophrys probably worthy of subgeneric rank.” The description here of a new Eocene species of Eopelobates, the recognition of the excellent series of Eopelobates specimens (here referred to as E. hinschei) from the Geiseltal middle Eocene, and the new Figure 19. “‘neudorfensis'' (= bayeri); d, E. grandis, PU 16441. specimen of E. bayeri make it possible now to take a closer look at the relationships of Eopelobates. Redefinition of Macropelo- bates has also been necessary, and this will be discussed below. Eopelobates was a relatively widespread and common early and middle Cenozoic frog first known with certainty from early Eocene of North America and middle Eocene of Europe. These two forms, E. guthriei and E. hinschei, respectively, show relationship to each other in their squamosal shape, although E. hinschei has already developed the long skull table seen later in E. bayeri. The relationship between the two Eocene forms is probably real, however, and demonstrates another simi- larity in early and middle Eocene con- tinental transatlantic vertebrate faunas (Simpson, 1947). This similarity first ap- 317 Estes Fossir PELoBatip FRocs - Left squamosals of pelobatids. a, Megophrys lateralis, AM 23549; b, ?Eopelobates sp., UCMP 44707; c, E. Dashed line = restoration; a-b, X 6; c, X 10; d, X 3. pears in the late Paleocene mammalian and lower vertebrate faunas (Russell, 1964; Estes, Hecht, and Hoffstetter, 1966). Yet the time difference and the differenti- ation into long and short-headed forms indicate that the intercontinental similarity is not so specific as to imply direct con- nection. It is possible, as noted above, that Eopelobates (or an ancestor) was already present in the late Cretaceous of North America. Relevant material is very frag- mentary, however, and the record must be used with care. Eopelobates does not recur in Europe until the middle Oligocene of Germany, when E. anthracinus indicates the presence of the short-headed lineage. The long- headed line begun by E. hinschei in the Eocene leads directly to the late Oligocene 318 Bulletin Museum of Comparative Zoology, Vol. 139, No. 6 Figure 20. Left squamosals of pelobatids. a, Eopelobates bayeri, CUP! 6.874; b, E. hinschei, MME 6753; c. E. anthracinus, BM R-4841; d, E. guthriei, MCZ 3493. Dotted line = restoration; a, d, X 6; b, X 5; ¢, X 9.5. Figure 21. Skull roof of (A) Eopelobates anthracinus, BM R-4841, about X 10; (B) Pelobates cultripes, UMMZ S-2631, X 3; dashed line restored. Figure 22. restoration; dotted line == broken bone outline. (or early Miocene) and middle Miocene E. bayeri from Czechoslovakia. In North America, E. grandis continues the short- headed line into the early Oligocene but then apparently becomes extinct. Eopelobates is characterized by a num- ber of features listed at the beginning of this paper, the most distinctive being gen- Figure 23. Fossiz PELOBATID F’RoGs + Estes 319 Skull roof of (A) Scaphiopus h. holbrooki, MCZ 58003; (B) S. skinneri, FAM 42920; both & 3. Dashed line = erally long-limbed proportions, absence of dermal head casque fused to the skull. The body proportions differ from those of most megophryines in having a greater relative elongation of the vertebral column and urostyle as well as a lengthening of limb segments, especially the tibiofibula, which is significantly longer than the femur. In Vag B. bad Skull roof of (A) Megophrys lateralis, AM 23549; (B) Pelobates fuscus, MCZ 1012; both X 3. 40 30 Scaphiopus couchi 20 Bulletin Museum of Comparative Zoology, Vol. 139, No. 6 ——=Scaphiopus h. holbrooki ---- =$. bombifrons 10 yh 10 a 27 ~ 15 31 H VU U HU RU F T TF H VU U HU RU [P T TF H = head length RU = length of radioulna A VU = length of vertebral F = length of femur H \ 60 po\ column aind'uinostyle + “\tength of tibiofibula iiae U = length of urostyle TE SMengtheot tibiae H ‘ menses = Macropelobates osborni HU = length of humerus and fibulare 50 j \ —— =Pelobates cultripes ! \ 0 s------ =P. fuscus / \ 3 oS | \ i Y \ ee 407 ae \ Fi ‘ Pelobates syriacus (*specimens from H \ / \ f \ ; 4, p. 241 d \ 30 Basoglu and Zaloglu, 1964, p ) A | \ i \ Ne \ XN / 3 / =: / \ VAN > Ne / \ Je \ Sey \ 204 207) EN Si ; Wie \ — . f/ \ a Qe ye \ We SS 4 7° Nie fe PFA 2 rier ae So ii BAS 6 10 19% 10 Ne S73 Sy 7 13° : bal 10 H vu U HU RU IP T TF H VU U HU RU F T TF Figure 24. Body proportions of pelobatids. Measurements in millimeters. pelobatines, the tibiofibula is always mainland species M. aceras shows an a spade, and possession of a well-developed shorter than the femur. With increasing body size, all pelobatids show allometry in the vertebral column and hind limbs relative to other parts of the skeleton (Figs. 24-25), and the allo- metric pattern is distinctive for the in- dividual groups. Within the megophryines, the primitive Leptobrachium (see Inger, 1966) has head and vertebral proportions as in Pelobates rather than as in Mego- phrys; some similarity to Eopelobates (especially E. anthracinus) occurs as well. So far as my few specimens indicate, the Eopelobates-like elongation of the tibio- fibula whereas the East Indian M. monti- cola and Leptobrachium hasselti have a subequal femur and tibiofibula. The Bur- mese specimen of M. carinensis has a tibiofibula slightly shorter than the femur, a proportion reminiscent of pelobatines. Two groups within Megophrys seem dis- | tinguishable on the basis of the few species and specimens available to me: the one | group having relatively short, anteriorly- directed posterior transverse processes, | fused urostyles, and body proportions like | those of Eopelobates hinschei; the other Foss. PELOBATID FrRocs + Estes 321 A vvoeos- = Barbourula busuangensis pf \) eeeeec = Discoglossus pictus 50 Eopelobates hinschei ( Table | ) 50 ] Lh AN Mareen eee = Alytes obstetricans (« = estimated measurement) | / See = Bombina orientalis ADs! f = Zaphrissa eurypelis 40 N 30 \\; : \\ 7054 —— =Megophrys spp. (Appendix 1!) A =Eopelobates anthracinus = gg] [x4 \ ----- = Leptobrachium hasselti B=E. bayeri G =E. grandis 504 (e = estimated measurement) 407 304 G 20 B 10 A T = a H VU U HU RU F T TF Figure 25. Body proportions of pelobatids and discoglossids. Measurements in millimeters. group having loose urostyles, elongated, lesser degree) as in pelobatines. The first ‘more perpendicular posterior transverse group of Megophrys noted above is more processes, and limb proportions closer to like Leptobrachium in this regard; the those of the pelobatines. The latter group second and more terrestrial group is distant is less Eopelobates-like in the last two from the latter and approaches the terres- features. trial pelobatines in limb proportions. _ Leptobrachium is primitive in having Zweifel (1956, p. 13) emphasized the short, anteriorly-directed posterior trans- relationship of Eopelobates and Mego- verse processes as in Eopelobates and (to phrys first noted by Spinar (1952, p. 487). The characters used by Zweifel require some qualification and are discussed seriatim: (1) “...only the complete postorbital arch will distinguish [Eopelobates] from [Megophrys].” As noted above, a squa- moso-frontoparietal arch does not exist in Eopelobates. In pelobatines this arch is present only in Pelobates cultripes (lacking in small individuals), and in most P. syria- cus (Basoglu and Zaloglu, 1964, p. 239). This condition is discussed more fully in the section on anatomical features at the beginning of this paper. (2) Long transverse processes of the second, third, and fourth vertebrae are present in Eopelobates and in most Mego- phrys. In Eopelobates their breadth is equivalent to the length of from five to seven vertebrae; the greater the number, the larger the specimen. In Megophrys the range is from four to seven vertebrae, again increasing with size. In Pelobates this breadth covers only from four to five verte- brae; even the large P. cultripes and Macro- pelobates do not exceed this figure. In Scaphiopus the range is from four to six vertebrae, and the entire range is encom- passed by the S. couchi specimens in my sample. This character is therefore not entirely clearcut, but Eopelobates and Megophrys show the greatest general similarity. (3) The greatly expanded sacral di- apophyses common to Eopelobates and some Megophrys can be duplicated in Pelobates cultripes. The length of the diapophyses in the latter is equivalent to the length of about four or five presacra] vertebrae, in Eopelobates the range is about four to seven vertebrae, and in no Megophrys available to me does it exceed 3.0 vertebrae. (4) The shape of the bony sternal style is similar and the bone is elongated in both Eopelobates and Megophrys. However, in Pelobates cultripes the shape is close to that of E. bayeri and E. grandis and is relatively wider throughout its length in Bulletin Museum of Comparative Zoology, Vol. 139, No. 6 these three species than in Megophrys of equal size (Fig. 9). (5) The free urostyle with transverse processes is similar in some Megophrys and some Eopelobates, and fusion is variable in both genera. The urostyle of Pelobates cultripes is also suturally separate, although partial fusion may have taken place in- ternally. As has been pointed out by many authors (most recently Kluge, 1966), this character is of little value as presently understood. However, some of the intra- specific variation noted in Megophrys by various authors was the result of incorrect identification; this character may deserve more careful study. (6) The great posterior extent of the ischium is similar in both Megophrys and Eopelobates. Some approach to this con- dition may be found in Macropelobates — but the latter more closely resembles Pelo- bates cultripes in this regard (Fig. 26). In | this respect the megophryine resemblance is more clear cut. Thus only 1 and 6 are clear cut resem- blances (but to different subfamilies), 2 is perhaps megophryine, 3 and 4 resemble both subfamilies, and 5 is inconclusive. The following characters further emphasize the mosaic of megophryine and_ pelobatine characters of Eopelobates. The Eopelo- bates ethmoid resembles that of the mego- phryines; in the prootic foramen (known only in E. guthriei) and orbitotemporal opening there are resemblances to both sub- families; in body proportions the variation in pelobatines and Eopelobates is encom- passed by that found in Megophrys. | Eopelobates (except E. guthriei) has a | broad, thin, anterior lamina on the scapula | that is well developed even in the small E. anthracinus. Among pelobatines, only | Pelobates has such a structure, although it is less well developed. Intrageneric Classification From the above it is clear that Eopelo- | bates is not a subgenus of Megophrys as | Zweifel (1956, p. 13) suggested. Although Fossi. PELOBATID FrRocs « Estes 323 Figure 26. Pelves in left lateral view. a, Leptobrachium hasselti, MCZ 22626, * 3; b, M. monticola nasuta, MCZ 22640, x 1.8; c, Pelobates cultripes, UMMZ S-2631, X 3; d, Macropelobates osborni, AM 6252, X 1.8. Dashed line indicates restoration. it is related to the latter genus, it also re- sembles pelobatines in many features. Examination of the various species of Eopelobates might indicate to some work- ers that several genera rather than one are included. E. hinschei and E. anthracinus, for example, might be referred to two genera if the other species were unknown. Hecht (1963, p. 23) has already suggested that E. grandis is “probably another genus distinct from the European [E. anthra- cinus],” and that at least two types of pelobatids are present in the Geiseltal frog fauna. As far as the latter case is con- cerned, after examining the Geiseltal col- lection in 1965 and 1967, I found no reason to recognize species additional to E. hinschei, although it is possible that I over- looked another form. E. grandis is similar in body proportions to E. anthracinus, as is E. guthriei in frontoparietal shape; these three species seem to form a short-skulled lineage. E. bayeri and E. hinschei, on the other hand, are relatively long-skulled forms, at least as far as proportions of nasal and frontoparietal are concerned. These two lineages appear to be linked by the distinctive squamosal shape of E. hinschei and E. guthriei on the one hand, and of E. anthracinus and E. bayeri on the other. In addition, E. grandis, E. bayeri, and E. hinschei show similarities of the fronto- | : parietal border. The rather granular dermal sculpture pattern of E. grandis is super- ficially different from the open, ridged pat- tern of E. hinschei, but these intergrade through the other species. The list of similarities given at the be- ginning of this paper indicates that for the present it is best to include all of these species in one genus; I believe that no useful purpose would be served by dis- tinguishing the two lines within Eopelo- bates generically. The situation is some- what similar to that in the Scaphiopus-Spea complex, and the morphological differences are nearly of the same order. Since most recent workers who have dealt with both recent and fossil forms have preferred only subgeneric distinction of Scaphiopus and Spea (Zweifel, 1956; Kluge, 1966), re- tention of the fossil species in one genus, Eopelobates, makes the internal classifi- cation of pelobatids more consistent. I prefer not to apply subgeneric distinctions to the two inferred fossil lineages without better knowledge of the record, however. Adaptation and Intrafamilial Classification In the final analysis of Eopelobates, it is clear that its position cannot be defined in terms of the archetypal and_hierarchial series of stages proposed by Noble (1924, p. 9) and utilized by Parker (1929, p. 280). 324 Kluge’s statements on generic definition (1966, p. 18) are pertinent to this problem. Rather than giving unnecessary emphasis to either a “classical morphotype” or an “adaptive” approach, he shows that both approaches produce similar results when treated in an evolutionary context incor- porating the pattern of variation displayed by the organisms. Eopelobates or any other fossil must, of course, be defined on ob- servable, hence morphological criteria. Yet when it is compared with living representa- tives whose adaptive characteristics may be more fully known, its own adaptive features may be assessed more meaning- fully. In this context, it is a frog having a tendency towards elongated limb and body segments, especially those of the distal hind limb. This produces an adaptation, similar to that of many species of Rana (e.g. R. pipiens), as a semiaquatic, salta- torial animal. It is even more similar in proportions to the living Discoglossus (also semiaquatic and saltatorial) than it is to the other discoglossids, Bombina and Alytes, which are more terrestrial and have more compact, pelobatine proportions (Figs. 24, 25). The fused dermal skull casque is remi- niscent of such fossil discoglossids as Latonia and Zaphrissa and may have been derived from some common ancestor, al- though as noted at the beginning of this paper it may be a separately derived con- dition. The thin anterior lamina on the scapula also occurs in discoglossids, al- though the scapula is much shorter. Eopelobates can thus be viewed as a primitive pelobatid, and in the light of the characters discussed above, one not easily relegated to either of the living subfamilies. In an evolutionary approach, subfamilial or other taxonomic boundaries are by definition arbitrary. Eopelobates is inter- mediate between megophryines and _ pelo- batines, and Macropelobates connects it with the latter. The Megophryinae are defined by characters not found in fossils Bulletin Museum of Comparative Zoology, Vol. 139, No. 6 (Beddard, 1907), but should these become known for Eopelobates, it is possible that the fossil genus would show an_inter- mediate condition here as well. For the sake of convenience, distinction between the two subfamilies can be maintained by the presence or absence of a spade; in this context Eopelobates becomes the most pelobatine member of the Megophryinae; Macropelobates the most megophryine of the Pelobatinae. An alternative position would be to place Eopelobates in a monotypic subfamily an- cestral to the two Recent subfamilies. I have emphasized the position of Eopelo- bates as intermediate between the two currently recognized groups, yet I have also attempted to show that it is more closely related to the Megophryinae. In part the decision is determined by one’s philosophy of classification. I prefer to emphasize the megophryine relationships here, but it is quite possible that more detailed study of the Czechoslovakian specimens will show that there is sufficient justification for separate subfamily status of Eopelobates (Spinar, in litt., 1969). THE PELOBATINAE The most primitive known spadefoot toad is Macropelobates osborni Noble (1924), from the Hsanda Gol Formation of Mongolia. Originally believed to be of late Oligocene age, the associated fauna is now thought to be at about the boundary between early and middle Oligocene (Mel- lett, 1968). Recent preparation of the unique speci- men of Macropelobates has shown features that further confirm its primitive pelobatine position, and which must be discussed be- fore considering spadefoot evolution as a whole. Macropelobates osborni Noble 1924 The skull of Macropelobates is somewhat | dislocated, but it is possible to restore its general proportions with fair certainty. The Fosst. PELoBATID FROGS ° 325 Estes Figure 27. Left, skull roof of cleared and stained Pelobates syriacus balcanicus, MCZ 50690, X 3; right, restoration of skull roof of Macropelobates osborni, AM 6252, X 1.8. Dashed line = restoration; dotted line = broken bone edge; dotted and dashed line = bone border covered by other bone in life. breadth across the back of the skull can be determined since the dorsal part of the squamosal is present and the otoccipital is complete laterally. Most of the ethmoid is present, and, by comparison with all other pelobatids, it seems clear that the skull did not exceed 30-32 mm in length. The sug- gested proportions are compared with (e.g.) Pelobates syriacus in Figure 27. The dorsal surface of the skull is flat- tened or slightly concave, as in most megophryines, including Eopelobates. The rounded tympanic process of the squamosal is pelobatine rather than Eopelobates-like. There is a posterior process on the maxilla (the latter bone is forced into the left orbit and was called the ethmoid by Noble) indicating the probable presence of a quadratojugal and hence of a complete maxillary arcade. The smooth and essen- tially complete borders of the fronto- parietal and the posterior part of the squa- mosal indicate that no postorbital bone bridge was present between these two bones. As in Pelobates cultripes, P. fuscus, and small P. syriacus, there is an opening on the midline between frontoparietals and nasals, and, as is common in P. cul- tripes, a small separate nubbin of dermal ossification is present. The nasals are miss- ing but the facet for the posterior border is present on the left side of the ethmoid, and a faint impression occurs medially on the ethmoid. The medial part of the pos- terior border of the otoccipital is expanded posteriorly as in Pelobates, and the tip of the prootic part of the otoccipital is narrow, also as in Pelobates. In general shape and lack of a thickened and projecting anterior process, the ethmoid is like P. cultripes and P. syriacus rather than P. fuscus or Scaphio- pus (Figs. 5, 7). A moderately developed turbinal fold is present as in Pelobates, and in anterior view the ethmoid is similar to that of P. cultripes (Fig. 3). The tarsus is completely pelobatine (Fig. 28). The tibiale and fibulare are about the same length as the radius, as in P. cultripes, P. fuscus, Leptobrachium, and some primi- tive Megophrys, rather than being signifi- cantly longer as in Eopelobates or shorter as in some Scaphiopus and P. syriacus. The tibiale is strongly expanded distally as in all pelobatines. A sickle-shaped, enlarged prehallux (spade) is present and closely bound to a large proximal element or pretarsal. Lateral to this is a large centrale 1, followed by distal tarsal 1. The well- ossified tarsus of megophryines includes a large fused distal tarsal 2 + 3, even in small 326 \ \ \ \ \ | ‘s wt i eae ! ! ll Ill IV @E 1) v t f H if / / A. Hi \ y I i hi Figure 28. Macropelobates osborni, AM 6252; a, plantar view of right ankle; b, lateral view of spade; & 3. I-V = metatarsals; c = centrale 1; d = distal tarsal 1; f = fibu- lare; p = prehallux; pt = pretarsal; + = tibiale. individuals. In pelobatines the latter bone does not ossify, but the other bones occur in all species. In Scaphiopus (Spea) and S. (Scaphiopus) couchi the pretarsal and prehallux fuse. The tibiofibula is shorter than the femur as in all pelobatines. The length from the dorsal border of the acetabulum to the anterior tip of the ilium approximately equals that of the femur. This is greater ilial elongation than is common in pelobatines but such pro- portions do occur in large Pelobates cul- tripes. The ischial projection posteriorly is more as in Pelobates than in Eopelobates or Megophrys (Fig. 26). The sacral diapophyses are expanded to about the length of 4.5 presacral vertebrae as in Pelobates. The forward inclination of the transverse processes of the posterior vertebrae is not quite so extreme as in Pelobates and is more like that of most Megophrys and Eopelobates. The urostyle is elongate, exceeding the length of the sacral diapophyses and about equalling or exceeding the length of the skull. In this feature it is in general agree- ment with that of megophryines and, to a lesser degree, Scaphiopus; it is unlike that of Pelobates, contrary to the statement of Zweifel (1956, p. 12). Bulletin Museum of Comparative Zoology, Vol. 139, No. 6 The flatness of the skull surface, the lesser inclination of posterior transverse processes, and the elongated urostyle are the only features that distinguish Macro- pelobates from Pelobates. These features are similar to those of Eopelobates and some Megophrys, and are probably primi- tive for the Pelobatidae. The other features of the skeleton relate Macropelobates closely to Pelobates (especially P. cul- tripes) and to the new Scaphiopus de- scribed below; this serves to clarify and expand Noble’s concept of this genus as differing only slightly from the modern forms. Zweifel (1956) and Parker (1929) have cited a similarity of proportions of Eopelobates, Megophrys, and Macropelo- bates. As Figures 24, 25, and 29 show, the latter is clearly on the pelobatine growth curve. Only the elongated urostyle can be cited as a megophryine proportional feature. Macropelobates seems to represent an early member of the pelobatines, in diag- nostic ways characteristic of that group. but possessing a few features relating the spadefoots more closely to the mego- | phryines. It is closest, however, to Pelo-— bates, especially P. cultripes, and can only be separated from it by the megophryine primitive characters noted above and by the absence of the enlarged dermal cover- ing of the squamosal and the squamosal- frontoparietal bridge. Pelobates The fact that Macropelobates seems to have its closest relationships to P. cul-— tripes' probably indicates the primitive © position of the latter species. Gislén (1936) | has already considered P. cultripes primi- tive on the basis of size, parasphenoid | teeth, and frontoparietal-squamosal con- | nection. The first two characteristics, how- ever, are of little value. The frontoparietal- 1 Here, as elsewhere in this discussion unless otherwise stated, the conditions closely related Pelobates varaldii (Pasteur and Bons, 1959) are as in P. cultripes. of the very | 120 v v v v 110 a ® 100 Vv Vv RATIO VALUE © Oo 80 Fossi. PELOBATID FRocs + Estes 327 -Eopelobates grandis bayeri . anthracinus hinschei Ceptobrachium hasselti Megophrys aceras M. monticola M. P Be P u tl u ImIlim|m M. carinensis elobates fuscus P. cullitiripels Po SYVEIACUS -Scaphiopus couchi SS holbrookt Macropelobates osborni @ @OH0C@OO*¥XG O4BRE 20 30 40 HEAD LENGTH IN MM Figure 29. Ratio of tibiofibula-femur length to head-body (skull-urostyle) length plotted against head length for various pelobatids. squamosal connection was shown to be a secondary condition in the discussion of anatomical features at the beginning of this paper. Pelobates syriacus is most closely related to P. cultripes. Both P. cultripes and P. syriacus, as well as the primitive Scaphio- pus holbrooki and S. skinneri described below, have an ethmoid with little ossifi- cation of the anterior process, but P. fuscus 328 has developed a complex anterior process similar to that of the specialized Scaphio- pus couchi. While P. fuscus has an unusual prootic foramen (Kluge, 1966, p. 13; Fig. 16, this paper), P. cultripes and P. syriacus have one of more open, megophryine type as in Figure 16b. P. syriacus has a tibiale and fibulare shorter than the radius, a condition advanced over that of S. holbrooki and more like that of the specialized S. couchi. Thus both P. syriacus and P. fuscus appear to be advanced over P. cultripes, although in different ways and to different degree; P. fuscus is the more specialized of the two former species. Zweifel (1956) has suggested that P. fuscus is most like S. holbrooki, but as the description (see be- low) of the new Oligocene Scaphiopus material shows, Macropelobates is prob- ably phyletically closer to the ancestral spadefoot than is the relatively specialized P. fuscus. Miopelobates robustus (Bolkay, 1913) Pelobates robustus Bolkay (1913), from the Lower Pliocene of Hungary, was de- scribed on the basis of maxillae, pre- maxillae, angular, thyroid process of hy- oid, and ilium, all fragmentary. Bolkay noted that the maxillae were not completely covered with osteoderms, the anterior part being relatively smooth and separated from the sculptured posterior area by a “bifur- catediiurrowe (1913) on 219 soll ties) Wettstein-Westersheimb (1955) described Miopelobates zapfei on the basis of fronto- parietals, nasals, maxillae, sacra, urostyles, and vertebrae from the Middle Miocene fissures near Neudorf, Czechoslovakia. The material is dissociated although some of it, designated “Typen” by Wettstein (1955, p. 812), may be from the same individual. The paired frontoparictals are in contrast with those of Pelobates, Macropelobates, and Eopelobates bayeri. The nasals are compact and Scaphiopus-like in appear- ance, although there was apparently a dorsal exposure of the ethmoid. The maxillae differ from those of Eopelobates Bulletin Museum of Comparative Zoology, Vol. 139, No. 6 bayeri in lacking a lobed squamosal proc- ess and sinuous posterior border. The expanded squamosals are most like those of Pelobates cultripes. The relatively straight borders of the sacral diapophyses are more as in Pelobates than in Eopelo- bates. The peculiar smooth anterior portion of the maxillae, the suborbital sculptured area, and the bifurcated furrow (for facial blood vessels and nerves) separating these two areas are clearly visible on Wettstein’s specimens (1955, pl. 2, fig. 3a) and there seems little doubt that Wettstein’s species zapfei is a synonym of robustus. The very Pelobates-like ilium that Bolkay associated with P. robustus suggests that Miopelo- bates is a pelobatine. This is supported by the configuration of the squamosals, the sacra, and the nasals as noted above. Be- cause of the paired frontoparietals and the peculiar ossification pattern, this species is retained in Wettstein’s genus Miopelobates. Kluge (1966, p. 16) allied Miopelobates with the Megophryinae, but for the above reasons I believe it to have been a spade- foot. It may have been a somewhat aberrant offshoot from the ancestral Pelo- bates type, and may be near P. cultripes as indicated by the expanded squamosals. Mtynarski (1961) has cited a possible occurrence of Miopelobates from _ the Lower Pliocene of Poland; this is very likely in view of its now recognized occur- rence in the Lower Pliocene of Hungary. Scaphiopus Since both Zweifel (1956) and Kluge (1966) recently discussed the evolution of the North American spadefoots, discussion here will be limited to the pertinence of the new Oligocene Scaphiopus described below to their scheme of spadefoot diversi- fication. Scaphiopus skinneri, n. sp. Type: FAM 42920, complete skull and vertebral column, left scapula, right cora- coid, left? thyroid ossification. Referred specimens: FAM 42921, one left and one right frontoparietal, both fragmentary, and a partial vertebral column with adherent tibiofibular fragment. Etymology: Patronym for Mr. Morris Skinner, Frick Laboratory, American Mu- seum of Natural History, who collected the type specimen in 1950. Locality: Leo Fitterer Ranch, Sect. 7, T 137 N, R 97 W, 13 miles South, 8 miles west of Dickenson, Stark County, North Dakota. Horizon: First banded zone, 15 feet above base of channel deposits, Unit no. 6A (Skinner, 1951, p. 53). Age: Middle Oligocene, Orellan (Europ- can equivalent = Helvetian). Preservation: The skull, vertebral column and girdle elements are associated and in almost natural position. The skull has been separated from the vertebral column for study. The skull is well preserved on the right side, but on the left, part of the pos- terior region of the squamosal and the left frontoparietal are lost. The left otoccipital had been dislocated at the time of burial (probably when the squamosal and fronto- parietal were lost) but has been prepared free and replaced in its natural position. Otherwise the skull is undistorted and un- crushed. The atlas and the neural arch of the fourth vertebra are lost, as are the tips of the transverse processes of all vertebrae. Description: In posterior view the skull roof appears essentially flat but is slightly depressed medially. The occipital canal opens just medial to the prominent par- occipital process. The foramen magnum is a flattened oval; its apex is directed dor- sally. The occipital region is well pre- served, although the left frontoparietal, left stapes, and lateral edges of the otoccip- ital are missing. The otoccipitals extend laterad to form the border of the fenestra ovalis. Dorsally they articulate with the frontoparictal and ventrally with the para- sphenoid, which is excluded from _ the fenestra ovalis. The foramen for the ninth and tenth cranial nerves opens prominently Fossi. PELoBAtip FRocs + Estes 329 just lateral to the large, rounded occipital condyles. The paroccipital process has a prominent boss on its lateral tip, just lateral to the frontoparietal and the occipital canal. The prootic is notched laterally, and forms the medial border of the foramen for the maxillomandibular branch of the trigeminal nerve. The stapes is just pos- terior and dorsal to this foramen, and has a forked head fitting into the anterodorsal part of the fenestra ovalis. A large oper- cular space remains, but if a calcified operculum was present, it has been lost. Since such delicate structures as tooth crowns, septomaxillae, and stapes remain, it is likely not to have been present. The prominent descending suspensorium _ is formed by the pterygoid medially, and the squamosal laterally, which clasp between them the well-developed quadrate. Dorsally the premaxillae are unsculp- tured; the right bone is well preserved but the nasal process of the left is missing. The nasals are prominently sculptured and complete except for their pointed anterior processes above the nasal openings. They articulate on the midline where they form a slight depression, and also laterally with the maxillae. There is no open groove or unsculptured area in the nasomaxillary suture. The frontoparietals are also sculp- tured and have a prominent postorbital projection (broken except on FAM no. 42921a, Fig. 30). Anteriorly they articulate with the nasals but leave a small trape- zoidal area of the ethmoid uncovered on the midline. Posteriorly their borders are rounded, curving into the postorbital pro- jection. A tiny, pointed, and unsculptured process of the frontoparietal extends onto the paroccipital process. Maxillae and squamosals are also completely covered by dermal sculpture; the latter articulate firmly with the former but there is no con- nection or process of squamosal to or to- ward the frontoparietals. The tympanic process of the squamosal is prominent and rounded, and a broad _ prootic process covers the tip of the otoccipital. The latter 330 Bulletin Museum of Comparative Zoology, Vol. 139, No. 6 Figure 30. Scaphiopus skinneri, n. sp. A-E, dorsal, right lateral, ventral, anterior, and occipital views of skull; F-G, vertebral column, dorsal and ventral views; H, scapula, coracoid, and thyroid process ossification; FAM 42920, * 2. I, | right frontoparietal, anterior end broken, FAM 42921a; X 2. | bone is exposed to its tip on its posterior end. Laterally the maxillae are deep and sculptured over all their surface except for a narrow band immediately dorsal to the teeth. The latter are pedicellate, and most of the narrow, spatulate crowns are pre- served. The rounded tympanic process of the squamosal extends almost to the occip- ital condyles posteriorly and is notched ventrally for the tympanic membrane. No quadratojugal is present, although there is, on the lateral surface of the quadrate, a tiny projection that may represent its fused remnant. In palatal view the vomers have strong processes anterior to the choanae, there are small tooth patches medially, the bones do not meet on the midline, and slim lateral processes to the palatines al- most reach the pterygoids. The palatines are completely fused to the maxillae. The ethmoid has strong lateral processes, and well-developed concavities behind the vomers indicate a prominent “turbinal” fold. The anterior tip of the ethmoid is broken away. The pterygoids have a long suture with the maxillae and end in small unossified spaces separating them from the vomers. The parasphenoid wings clasp the ptery- goids laterally; anteriorly the cultriform process lies smoothly on the ethmoid with- out developing a channel, and posteriorly there are well-defined crests for nuchal and retractor bulbi muscles, and for the eustachian tubes. The prootic foramen is elongated and Open anteriorly; its dorsal and ventral borders are approximately parallel. The oculomotor and optic foramina are not out- lined in bone. The mandibles are broken away pos- teriorly. Anteriorly the symphysial (men- tomeckelian) bones are present, separated from the prearticulars by unossified spaces and clasped by the dentaries. In anterior view the premaxillae are well preserved but loosely attached. On the Foss. PELOBATID FRoGs «+ Estes 331 right, the ascending process contacts the small septomaxilla. The anterior process of the ethmoid is broken away but was apparently not thickened; a well-defined capsular process with a prominent turbinal fold is visible. The atlas is missing, as is the neural arch of the fourth vertebra. The vertebrae are procoelous, and, posteriorly, the ninth (sac- ral) vertebra has well-defined, hatchet- shaped diapophyses. The main postsacral foramina are relatively small, and there appears to have been a smaller second pair as well as considerable webbing (about as in Zweifel, 1956, fig. 19g). The urostyle is broken off but the narrowness of the remaining portion and the presence of two pairs of postsacral foramina indicate with- out much question that it was fused with the sacrum. The scapula, coracoid, and ossified thy- roid cartilage are all robust but display no unusual characteristics. The disarticulated vertebral column (FAM 42921c) is similar to that of the type. Discussion: Scaphiopus skinneri, in pos- sessing the following characters, is clearly referable to the subgenus Scaphiopus: (1) presence of squamoso-maxillary contact, (2) widely emarginate prootic foramen, (3) absence of frontoparietal fontanelle, (4) extensive dermal skull, (5) probable absence of calcified operculum, (6) pres- ence of pterygoid process of maxilla, (7) presence of palatine, (8) large size. These characters are as given by Kluge (1966, p. 19) except that the condition of the oper- culum (his character no. 6) is reversed in his table for the two subgenera, although given correctly in the text (1966, p. 10). In general skull proportions, Scaphiopus skinneri is similar to the most primitive living species, S. holbrooki. It has a broader skull when compared with length of pre- sacral column: 1/2 skull breadth = 5.5 presacrals as opposed to 4.5 presacrals in a random sample of S. (Scaphiopus) at hand, although this relationship may be the result of large size of the fossil. It resembles S. 332 holbrooki in orbitotemporal opening, al- though its orbit is not relatively as large (see Fig. 15d, e). As shown by the re- ferred frontoparietals, the postorbital pro- jection is rounded and _ relatively far forward as in S. holbrooki. However, the tympanic process of the squamosal is longer, the posterior extent of dermal bone on squamosal and frontoparietal is greater than in any modern pelobatine, and the skull as a whole is slightly more flattened than in S. hammondi. In these characters it resembles Eopelobates, Macropelobates, and Pelobates cultripes. The tendency in other species of Scaphiopus and in Pelo- bates is to develop a more domed skull, although that of P. cultripes is flatter than it is in any other living pelobatine. The persistence dorsally of a small area of ethmoid not covered by dermal bone is also a character reminiscent of Eopelobates, Pelobates fuscus, and P. cultripes. Usually in all Scaphiopus (Scaphiopus) and in most P. syriacus, the dermal covering of the frontoparietals fills this space. The vertebral column is not unusual except that the second vertebra has the condyle of the atlas fused to it and is hence bicondylar. This fusion is irregular, however, and does not appear to be the usual condition, although it was certainly functional in this individual. A variety of articulations have been noted in pelobatids; Boulenger (1908) found both opisthocoely and procoely in Megophrys, and Griffiths (1963) found free intervertebral discs in an adult Megophrys major as did Noble (1926). My observations are in accord with theirs, and in addition, I have found free intervertebral discs in a large, cleared and stained adult Pelobates syriacus (MCZ 50690). Thus, no significance should be attached to the bicondylar fossil vertebra; all the other vertebrae are procoelous. The length (expansion) of sacral diapophyses in this specimen is equal to the length of nearly three presacral vertebrae, and I have found this to be the case in all in- dividuals in my sample of Scaphiopus Bulletin Museum of Comparative Zoology, Vol. 139, No. 6 (Scaphiopus), contrary to the statements | of Kluge (1966, p. 17) and Zweifel (1956, | Table 1). The girdle elements and thyroid cartilage | ossification resemble those of Recent S. | (Scaphiopus) and are in about the same size proportion to the skull. Scaphiopus skinneri is a_ primitive Scaphiopus as shown by the generally more depressed skull, relatively small orbit, flat skull roof, large rounded tympanic process of the squamosal, low squamosal angle (50°; Griffiths, 1963, fig. 2, p. 248 and see section on this character-state at beginning of this paper), posterior extent of dermal bones on frontoparietal, and dorsally ex-_ posed ethmoid. Yet, as noted above, it possesses all of the characteristics of the subgenus Scaphiopus. In orbitotemporal proportions, degree of expansion of the anterior process of the ethmoid, and short quadrate process of the squamosal, it re- sembles the most primitive living Scaphio- pus, S. holbrooki. It is also similar to Macropelobates in the large, rounded tympanic process of the squamosal and the shape of the posterior part of the frontoparietal. These are prob- ably primitive pelobatine characters. Eopelobates guthriei resembles Scaphio- pus in having a relatively short skull, strongly concave posterior border of the prootic part of the otoccipital, long narrow prootic foramen, and relatively great pos- terior extent of nasals. It is perhaps the closest to the spadefoot line of any known megophryine. Possibly the two groups had their common ancestor in the Paleocene or perhaps even in the Cretaceous. The fact that the well-defined Scaphiopus skinneri occurs in the early Oligocene indicates that the spadefoot line is at least as old as Eocene, and perhaps older; S. skinneri also occurs in the early Oligocene of Saskatchewan; this material is being described by Dr. J. Alan Holman.’ As 1 Holman, 1969. Quart. Jour. Florida Acad. Sci. | 31:273-289; received after this paper went to | press. with the living S. holbrooki, S. skinneri and probably all primitive Scaphiopus were associated with deciduous forests and an essentially humid warm-temperate or sub- tropical climate (in the sense of Dorf, 1959). The development of the Spea complex was probably correlated with the semiarid open woodland scrub and grass- lands that were beginning to develop in midcontinental North America by the middle and late Oligocene (Dorf, 1959, p. 189). This is essentially the picture already set forth by Zweifel (1956, p. 41) and supported by Kluge (1966, p. 21). SPECIES REMOVED FROM THE PELOBATIDAE Zaphrissa eurypelis Cope 1866, described from the Middle Oligocene lignite beds of Rott, near Bonn, Germany, is usually con- sidered a discoglossid (Friant, 1960). Kuhn (1938, p. 20) synonymized it with Pelo- bates on the basis that Wolterstorff (1929, p. 931) believed it to be “identisch mit Pelobates decheni Tr.,” but later (Kuhn, 1962) replaced it in the Discoglossidae. Friant (1960) suggested that it might be a juvenile of Latonia, a giant discoglossid from the Miocene deposits at Oeningen. The type specimen of Zaphrissa was re- cently rediscovered (Baird, 1970). It has ribs, opisthocoelous vertebrae, a relatively large atlas, a very short scapula, and a double condyle on the urostyle. These characters in combination indicate that the specimen is discoglossid. It has a well- developed dermal skull casque rather like that of Pelobates cultripes. The fronto- parietal fenestra cited by Cope, and used as an indication of juvenility by Friant (1960), is actually an area where the dermal bone has been broken away before burial, although such a fenestra does occur occasionally even in such a heavily en- crusted skull as that of P. cultripes (UMMZ S-2630). I have not seen the material of Pelobates decheni noted above, but if Wolterstorff Fossiz PELOBATID FRoGs + Estes 333 was correct, then the material is incor- rectly referred to Pelobates and the proper name for this animal would be Zaphrissa decheni. Nevo (1956) gave a preliminary notice of fossil frogs from the early Cretaceous of Israel and stated that the specimens dis- played some pelobatid features. Griffiths (1963, pp. 276, 282, 283) later referred to these specimens as pelobatids. A more detailed paper by Nevo (1968) shows these specimens to be members of the Pipidae. EVOLUTION AND ZOOGEOGRAPHY OF THE PELOBATIDAE If the late Cretaceous Lance Formation specimens from Wyoming are properly re- ferred to Eopelobates (p. 315), then this earliest pelobatid was associated with a humid, subtropical, coastal plain environ- ment in North America (Estes, 1964). The paucity of the Cretaceous record in Europe precludes knowledge of a possibly wider distribution of the group. In any case, the extensive epicontinental seas characteristic of the Northern Hemisphere Cretaceous would probably have hindered or pre- vented such movement. Holarctic conti- nental connections seem not to have been re-established until the late Paleocene (Russell, 1964), and strong interconti- nental faunal similarities persisted until the end of early Eocene time. By this time, Eopelobates guthriei was already estab- lished in North America and this form may be near the point of divergence of the spadefoot line. E. guthriei was associated with a climate essentially like that of the late Cretaceous of Wyoming. Although there is floristic evidence for a period of cooling at the beginning of the Cenozoic (Dorf, 1959), much of the lower vertebrate fauna already established by late Creta- ceous time persisted through the Paleocene in Wyoming (Estes, 1962). Not later than late Paleocene or early Eocene time, Eopelobates must have achieved a Holarctic distribution. By mid- 334 Eocene time, it was well established in Europe in the swamps of the Geiseltal in what was an essentially tropical environ- ment (Krumbiegel, 1959, p. 116). The Geiseltal species, E. hinschei, was the most specialized member of the group in that it had developed relatively long posterior limb segments like those of mainland popu- lations of the Recent Megophrys aceras, but since the latter is montane the ecology of the two forms must have been quite different. These proportions, in E. hinschei, were probably adaptations for an amphib- ious existence much like that of some species of Rana, e.g. R. pipiens, which re- mains on moist banks and uses its long limbs for jumps either for food or to regain the safety of the water. Although in squamosal shape Eopelo- bates hinschei shows resemblance to E. guthriei in North America, it seems to have been the ancestor of a relatively long- headed European line that persisted until at least the middle Miocene. Eopelobates was also present in North America during the middle Eocene, al- though the remains are fragmentary. A subtropical climate still persisted in the midcontinental area at this time, but a slight cooling effect has been noted (Dorf, 1959). North American and Eurasian Eopelobates must have been pursuing separate evolutionary paths at this time, for faunal interchange was now relatively restricted. The next record of Eopelobates is in the early Oligocene of North America. This animal, E. grandis (Zweifel, 1956), is the largest known member of the genus. It resembles E. guthriei in having a_ short, wide frontoparietal, and was almost cer- tainly an autochthonous element. Early Oligocene also saw the appear- ance of the first spadefoot toads: Scaphio- pus skinneri is more primitive than, but is closely related to, the most primitive living species, S. holbrooki. Climatic changes were beginning to take place at this time (early-middle Oligocene ); Bulletin Museum of Comparative Zoology, Vol. 139, No. 6 Eopelobates grandis and Scaphiopus skin- | neri probably lived in a warm-temperate — rather than subtropical climate (Dorf, | 1959; Clark et al., 1967). The warm tem- | perate flora extended into Alaska (Dorf, — 1959) and there was a period of strong faunal interchange (Simpson, 1947). A form close to Pelobates was already estab- lished in the early Oligocene of Belgium (Hecht and Hoffstetter, 1962). It is pos- sible that spadefoots were derived from Eopelobates in the Eocene in North | America, or even in the Paleocene. Skull proportions of American Eopelobates sug- gest a closer approach to spadefoot pro- portions than do those of the long-headed © European forms. E. anthracinus indicates that the short- headed lineage was also present in Europe, however, where it appeared in the middle Oligocene of Germany. It is a relatively short-headed form bearing frontoparietal similarities to the Eocene North American E. guthriei. Its squamosals resemble those of the somewhat later E. bayeri of Czecho- slovakia (cf. Figs. 8, 12, 13, 19, 20), while the body proportions appear similar to those of E. grandis (Fig. 24). This may | indicate that it was derived from short- | headed North American populations that | migrated to Europe not later than the early | Oligocene, and probably earlier. It might be assumed that its body proportions are | the result of its small size, but even small members of E. hinschei have body pro- | portions related to those of the large specimens (Fig. 29). On the basis of the short, emarginated frontoparietal, I prefer | the first alternative. | At the end of the Oligocene, Eopelobates | bayeri appears in Central Europe. It per- | sists into the middle Miocene, and _ is closely related to the Eocene E. hinschei, and is also a long-headed form. It prob- ably lived under subtropical conditions in | the late Oligocene, which became more | warm-temperate in the Miocene (Dorf, | 1959). These changing conditions seem | to have been related to the disappearance | teens marareees: of Eopelobates in Europe by middle Mio- cene time. Pelobates-like fossils are present in France in the late Miocene (Hecht and Hoffstetter, 1962). The same deteriorating climatic con- ditions that caused the eventual extinction of Eopelobates were favorable to the con- tinued development of the essentially warm-temperate spadefoot line. The first known spadefoot, Scaphiopus skinneri, oc- curs at a latitude transitional at that time between subtropical and warm-temperate conditions (Dorf, 1959). It is probable that this transitional climate was the site of original evolution of the spadefoot type, and that they spread northward from the transition into Temperate regions. The Eocene of North America was a time of the gradual rise of the midconti- nental region. Mountain building activity associated with this rise exposed granitic rocks, whose erosion produced the sandy soils preferred by spadefoots, as well as by other burrowing animals. These soils were (and are) used by spadefoots as a retreat from aridity and because of ease of burrowing. Not only did the mountain building itself cause the developing aridity, but it also produced the soils favoring the fossorial adaptation. Because the early and middle Oligocene Scaphiopus skinneri was already a primi- tive but well differentiated member of the North American spadefoot line, the Hol- arctic spread of the spadefoot group must have been no later than the late Eocene, when faunal similarities (principally mam- malian) indicate that migration was taking place again between Old and New Worlds. The Holarctic radiation was also possible during the early Eocene, and because of the spadefoot resemblances of Eopelobates guthriei I favor this alternative (Fig. 31). Since we have no Eocene record of the spadefoots, another possible alternative is that Scaphiopus originated in Asia after the early Eocene spread of the ancestral type. In view of the present inadequate Foss. PELOBATID FRoGcs ° Estes 335 fossil record, the simplest explanation is an autochthonous origin of Scaphiopus. Macropelobates, the primitive Pelobates- like spadefoot, appears in Asia by the mid- dle Oligocene. Although it is closer to Pelobates, in certain features Macropelo- bates shows some similarities to Scaphiopus skinneri, demonstrating some intermediacy between Old and New World forms. The ancestral Pelobates populations probably spread westward into Europe no later than early Oligocene, if the material noted by Hecht and Hoffstetter (1962) is indeed Pelobates or its ancester. Populations of the genus extended through Northern Europe into the Iberian Peninsula, and evolved into a group ancestral to P. cul- tripes and P. syriacus. At the eastern edge of its range, this ancestral group probably formed northern and southern sections on each side of the late Cenozoic Aralocaspian sea-lake (Gislén, 1936; Gignoux, 1955); the modern species had probably evolved by Miocene time. During the Pleistocene, the advancing ice sheets restricted P. cul- tripes and P. syriacus to the Iberian Penin- sula and Asia Minor, respectively. P. fuscus, derived probably from northern popu- lations of P. syriacus, remained in Europe wherever the advancing ice sheets per- mitted, and as Gislén (1936) has already noted, again spread widely over northern Europe during the thermal maximum. Fossils of Pelobates have been found in various localities in Europe (see Mtynarski, 1961) from at least as far back as the early Pliocene, and other possible occurrences go back to early Oligocene (Hecht and Hoffstetter, 1962). These remains have not been studied carefully by anyone who had an adequate sample of all three Recent species; such a study would be very help- ful towards understanding the diversifi- cation of the European spadefoots. It seems clear, however, that P. fuscus is the most recent and specialized of the three species and that it is not directly related to Scaphiopus holbrooki, its ecological 336 Bulletin Museum of Comparative Zoology, Vol. 139, No. 6 EURASIA NORTH AMERICA EUROPE S.couchi S. intermontanus PLEIST- | Pcultripes P fuscus Psyriacus S.holbrooki S.bombifrons S. hammondi S. diversus S.studeri S. pliobatrachus | . S. alexanderi S.wardorum —— S.wardorum ? MIOCENE PLIOCENE Macropelobates lei luli See skinneri E. anthracinus S.skinneri WwW CS uJ oO (e) (©) =!) oO €. grandis E O:P se Or BrARnEERS / Eopelobates sp. E. hinschei E. guthriei $ ancestral spade foot TO MODERN MEGOPHRYINES oes. aE: Figure 31. Temporal, geographical, and inferred phyletic relationships of pelobatids. On the vertical lines separating the continental areas, the times of major faunal exchange (based primarily on mammals; Simpson, 1947) are indicated by broken lines. ae PALEOCENE EOCENE equivalent in North America. The findings of this study support Zweifel’s contention that Scaphiopus and Pelobates had a com- mon fossorial ancestor, and that Macro- pelobates is close to that ancestor although too late in time. The patterns of diversifi- cation within Scaphiopus suggested by Kluge (1966) and Zweifel (1956) are con- sistent with the known fossils. The position of the modem mego- phryines is not directly clarified by this study. It is a diverse group and seems to include animals spanning the range of body proportions found in other pelobatids. Because of the peculiar nature of their dermal ossification, their primitive eth- moid, and the similarity of nasal-fronto- parietal relationships to the early Cenozoic long-headed European lineage of Eopelo- bates, I consider them to be of very ancient origin from a common stock with Eopelo- bates. As noted in several places above, the primitive megophryine genus Leptobra- chium is the closest to Eopelobates of any of the modern forms, yet the resemblance is not especially strong. Eopelobates may have been in existence in the late Creta- ceous, and since all Cenozoic members show pelobatine features not found in modern megophryines, I believe that their common ancestor cannot have been later than the late Cretaceous. Leptobrachium and its relatives were probably tropical differentiates of the ancestral pelobatids. Whether or not the resemblance between the long-skulled European Eopelobates and the Recent southeast Asian forms implies an origin of pelobatids in the Old World Tropics is conjectural. Zweifel (1956, p. 15) has properly emphasized the caution necessary in making such inferences. Dar- lington (1957) favors the origin of many groups in the Old World tropics and such an origin has been often assumed by authors dealing more specifically with amphibians (e.g. Noble, 1924). Yet it is perfectly plausible to imagine a common ancestor of megophryines and pelobatines living in relatively high-latitude Holarctic Foss, PELoBATID FRocs + Estes 337 tropics of the late Mesozoic, and differenti- ating into tropical Leptobrachium-like forms (their descendants remaining still in present day tropics), tropical and sub- tropical Eopelobates (now extinct) and the temperate geographical replacements of the latter, the pelobatines. In this latter scheme, Leptobrachium and its relatives became restricted to the Old World tropics during the early Ceno- zoic, and subsequently differentiated into a number of island and montane (temper- ate) forms. Eopelobates diversified into mainly subtropical environments, but also extended into tropical areas (E. hinschei). With the progressive restriction of high latitude tropical climates during late Ceno- zoic time, some warm-temperate forms developed into pelobatines, adapting pro- gressively to increasing aridity in both Old and New Worlds by developing a burrow- ing habitus. They now have a complemen- tary, Holarctic distribution. Eopelobates itself was perhaps unable to compete with more successful ecological analogues that were becoming widespread by the Miocene, such as some species of Rana, and therefore became extinct. APPENDIX I: LIST OF RECENT COMPARATIVE MATERIAL Numbers refer to measured specimens, Figures 24, 25. Numbers in parentheses indi- cate that more than one specimen is listed under a given museum number. Pelobatidae Megophryinae 1. Leptobrachium hasselti, MCZ 22626, Bor- neo. 2. Megophrys monticola nasuta, MCZ 22640, Borneo. 3. M. m. nasuta, MCZ 19756, Sumatra. 4, M. monticola, AM 24786, Java. 5. M. lateralis, AM 23549, Kuang China. . aceras, AM 23964, Burma. . Paceras, MCZ 23436, Burma. . Paceras, MCZ 23437, Burma. . carinensis, AM 23965, Burma. . robusta, MCZ 25735, Thailand. SOMND SERRE 338 1l. Scutiger mammatus, MCZ 17422, Szechuan, China. Pelobatinae 1. Pelobates cultripes, UMMZ S-2629, no data. P. cultripes, UMMZ S-2630, no data. P. cultripes, UMMZ S-2631, no data. P. cultripes, BM 682, Spain. P. cultripes, BM 233, Spain. P. cultripes, S-002 (Coll. Spinar), France. P. cultripes, S-001 (Coll. Spinar), France. P. varaldii, MCZ 31970, Morocco. P. syriacus balcanicus, MCZ 50690, Ro- mania. 10. P. fuscus, MCZ 1012, Italy. 1l. P. fuscus, MCZ 1013, Italy. 12. P. fuscus, MCZ 1353, Italy. 13. P. fuscus, MCZ 1012-b, Italy. 14. P. fuscus, MCZ 1013-c, Italy. 15. Scaphiopus h. holbrooki, MCZ 25577, Massachusetts (2). 16. S. h. holbrooki, MCZ 17420-1, Massa- chusetts (2). 17. S. h. holbrooki, MCZ 17418-9, Massa- chusetts (2). 18. 5S. h. holbrooki, MCZ 28786, Florida. 19. S. h. holbrooki, AM 58003, Florida. 20. S. holbrooki hurteri, AM 44244, Texas. 21. S. couchi, AM 14478, Baja California. 22. S. couchi, MCZ 64374, Arizona (cleared and stained). 23. S. couchi, AM 56284, Arizona. 24. §. couchi, AM 57641, Arizona. 25. §S. couchi, MCZ 3079, Texas. 26. S. couchi, MCZ 6710, Texas. 27. S. couchi, MCZ 44335, Mexico. 28. S. couchi, MCZ 44336, Mexico. 29. S. intermontanus, AM 16918, Utah. 30. S. intermontanus, AM 16916, Utah. 31. S. bombifrons, MCZ 32912, Texas. 32. S. bombifrons, MCZ 32913, Texas. 33. S. bombifrons, MCZ 32911, Texas. 34. S. bombifrons, MCZ 32914, Texas. Discoglossidae (only specimens used in Fig. 25 listed ) 35. Discoglossus pictus, MCZ 3196, Corsica. 36. Alytes obstetricans, MCZ 904, France. 37. Bombina orientalis, MCZ 19722, Korea. 38. Barbourula busuangensis, MCZ 25656, Philippines. Extensive comparison has been made with many specimens of other families of frogs too numerous to mention here. All specimens examined (other than those noted in this appendix from other institu- tions) are in the collection of the Museum Bulletin Museum of Comparative Zoology, Vol. 139, No. 6 of Comparative Zoology, Harvard Univer- sity. A list of specimens available in the skeletal collection is available on request from the Curator of Reptiles and Am- phibians. REFERENCES CITED Barrp, D. cene frog Zaphrissa eurypelis Cope, 1886. Copeia, 1970, no. 2, pp. 384-385. BasoGiu, M., AND S. ZaLoGiu. 1964. Morpho- logical and osteological studies in Pelobates syriacus from Izmir Region, Western Anatolia. Senckenb. Biol. 45: 233-242 (R. Mertens Festschrift ). BEDDARD, F. E. 1907. Contributions to the knowl- edge of the anatomy of the batrachian family 1970. Type specimen of the Oligo- | 7 Pelobatidae. Proc. Zool. Soc. London, 1907: © 871-911. Boxkay, S. J. 1913. Additions to the fossil herpe- tology of Hungary from the Pannonian and Praeglacial Periode. Jahrb. kg]. ungar. Geol. Reichsanst. 20: 217-230. Bou.enceEr, G. A. 1908. A revision of the Oriental — pelobatid batrachians (genus Megalophrys). | Proc. Zool. Soc. London, 1908: 407—430. Ciark, J., J. R. BEERBOWER, AND K. K. KIETZKE. 1967. Oligocene sedimentation, stratigraphy, paleoecology, and paleoclimatology in the Big Badlands of South Dakota. Fieldiana: Geol. Mem. 5: viii + 158 pp. Corr, E. D. 1866. On the structures and dis- tribution of the genera of the arciferous Anura. Journ. Philadelphia Acad. Nat. Sci. (N.S.) 6: 67-112. DarLINGTON, P. J., Jr. 1957. Zoogeography: the geographical distribution of animals. New York: John Wiley and Sons, xi + 675 pp. Dorr, E. 1959. Climatic changes of the past and present. Contr. Mus. Paleon. Univ. Michigan, | 8: 181-210. Estes, R. 1962. Late Cretaceous and early Ceno- | zoic lower vertebrate faunas in North America. | Program Geol. Soc. Amer. Natl. Meetings, — 1962: 46a. . 1964. Fossil vertebrates from the late Cretaceous Lance Formation, eastern Wyo- ming. Univ. Calif. Publ. Geol. Sci. 49: 1-180. Harvard Univ., Breviora No. 328: 1-7. Estes, R., M. Hecut, AND R. HoFrrsTEeTTeR. 1966. Paleocene amphibians from Cernay, France. Amer. Mus. Novit., 2295: 1-25. Friant, M. 1960. Les batraciens anoures. Caractéres ostéologiques des Discoglossidae | dEurope. Acta Zool. 41: 113-139. . 1969. A new fossil discoglossid frog from | Montana and Wyoming. Mus. Comp. Zool., | Gicnoux, M. 1955. Stratigraphic Geology. San Francisco: W. H. Freeman Co., (trans. G. Woodford), xvi + 682 pp. GisLEN, T. 1936. On the history of evolution and distribution of the European pelobatids. Zoogeographica, 3: 119-131. GriFFitus, I. 1963. The phylogeny of the Salientia. Biol. Rev. 38: 241-292. Hecut, M. K. 1959. Amphibians and Reptiles. In: McGrew, P. O. et al., The Geology and Paleontology of the Elk Mountain and Taber- nacle Butte Area, Wyoming. Bull. Amer. Mus. Nat. Hist., 117: 117-176. . 1963. A reevaluation of the history of the frogs, II, System. Zool. 12:20-35. Hecut, M., AND R. Horrstetrer. 1962. Note préliminaire sur les amphibiens et squamates du Landenien supérieur et du Tongrien de Belgique. Bull. Inst. Roy. Sci. Nat. Belgique 38: 1-30. IncErR, R. F. 1966. The systematics and zooge- ography of the Amphibia of Borneo. Fieldi- ana: Zoology, 52: 1-402. _Kuuce, A. G. 1966. A new pelobatine frog from the Lower Miocene of South Dakota with a discussion of the evolution of the Scaphiopus- Spea complex. Los Angeles County Mus. Contr. Sci. 113: 1-26. KruMBIEGEL, G. 1959. Die tertiare Pflanzen- und Tierwelt der Braunkohle des Geiseltales. Die Neue Brehm-Biicherei, Heft 237, Wittenburg Lutherstadt: A. Ziemsen Verlag, 1-156. Kuun, O. 1938. In: Fossilium Catalogus, I: Animalia, Pars 84: Stegocephalia, Urodela, Anura. 98 pp. . 1941. Die Eozainen Anura aus dem Geisel- tale. Nova Acta Leop. (n.f.) 10: 345-376. 1962. Die vorzeitlichen Frésche und Salamander, ihre Gattungen und Familien. Jahrb. Ver. Vaterl. Naturkde. Wiirttemberg. 1962: 327-372. MELLETT, J. S. 1968. The Oligocene Hsanda Gol Formation, Mongolia: A revised faunal list. Amer. Mus. Novit., 2318: 1-16. Mertens, R. 1923. Beitrage zur Kenntnis der Gattung Pelobates Wagler. Senckenbergiana, 5: 118-128. Mrynarski, M. 1961. Plazy (Amphibia) z Pliocenu Polski. Acta Palaeont. Polonica, 6: 261-282. | Nevo, E. 1956. Fossil frogs from a Lower Cretaceous bed in southern Israel (central Negev). Nature (London), 178: 1191-1192. . 1968. Pipid frogs from the early Creta- ceous of Israel and pipid evolution. Bull. Fossi. PELoBAtTip FRocs «+ Estes 339 Mus. Comp. Zool., Harvard Univ. 136: 255-318. Nose, G. K. 1924. A new spadefoot toad from the Oligocene of Mongolia with a summary of the evolution of the Pelobatidae. Amer. Mus. Novit., 132: 1-15. . 1926. An analysis of the remarkable cases of distribution among the Amphibia, with descriptions of new genera. Amer. Mus. Novit., 212: 1-23. Parker, H. W. 1929. Two fossil frogs from the Lower Miocene of Europe. Ann. Mag. Nat. Hist., (10)4: 270-281. PAsTEuR, G., AND J. Bons. 1959. Les batraciens du Maroc. Trav. Scient. Chérifien, Ser. Zool., 17: xvi + 240 pp. RussELL, D. E. 1964. Les Mammiféres paléo- cenes d’Europe. Mem. Mus. Natl. d’Hist. Nat., Paris (n.s.) 13: 1-324. SKINNER, M. F. 1951. The Oligocene of Western North Dakota. In: Bump, J. D., ed., Guide Book, Fifth Field Conference, Soc. Vert. Paleon., Mus. Geol. So. Dakota School Mines Tech., Rapid City, So. Dakota. Simpson, G. G. 1947. Holarctic mammalian faunas and continental relationships during the Cenozoic. Bull. Geol. Soc. America, 58: 613-688. Spinar, Z. V. 1952. Eopelobates bayeri, nova zaba z ceskych tretihor. Sbornik Ustredniho Ustavu Geologického, 19: 457-488, English summary. 1967. Neue Kenntnisse uber den strati- graphischen Bereich der familie Palaeo- batrachidae Cope, 1865, ibid., 42: 217-218. WestpPHaL, F. 1958. Die Tertidren und Rezen- ten Eurasiatischen Riesensalamander (Genus Andrias, Urodela, Amphibia). Palaeonto- graphica., 110: 20-92. . 1967. Erster Nachweis des Riesensala- manders (Andrias, Urodela, Amphibia) im europaischen Jungpliozin. Neues Jahrb. Geol. Paliont., Monatsh. 1967: 67-73. WETTSTEIN-WESTERSHEIMB, O. 1955. Die Fauna der miozinen Spaltenfiillung von Neudorf a. d. March (CSR.). Amphibia (Anura) et Reptilia. Sitz.-Ber. Osterr. Akad. Wiss., Math.-Natl. Kl. 164: 801-815. WotterstorrF, W. 1929. Uber fossile Frésche aus der Papierkohle von Burgbrohl (Laacher See), Jahrb. Preuss. Geol. Landesanstalt. 49: 918-932. ZWEIFEL, R. G. 1956. Two pelobatid frogs from the Tertiary of North America and their re- lationship to fossil and Recent forms. Amer. Mus. Novit., 1762: 1-45. Rulletin OF THE Museum of Comparative Loology The Galaxiid Fishes of New Zealand R. M. McDOWALL I I HARVARD UNIVERSITY VOLUME 139, NUMBER 7 CAMBRIDGE, MASSACHUSETTS, U.S.A. JUNE 12, 1970 PUBLICATIONS ISSUED OR DISTRIBUTED BY THE MUSEUM OF COMPARATIVE ZOOLOGY HARVARD UNIVERSITY BULLETIN 1863- BrEvIoRA 1952— Memorrs 1864-1938 JounsontA, Department of Mollusks, 1941- OccasIoNAL Papers oN Mo.wusks, 1945— Other Publications. Bigelow, H. B., and W. C. Schroeder, 1953. Fishes of the Gulf of Maine. 3 Reprint, $6. 50 cloth. Brues, C. T., A. L. Molander, and F. M. Carpenter, 1954. Classification of In- sects. $9.00 cloth. ; Creighton, W. S., 1950. The Ants of North America. Reprint, $10.00 cloth. Lyman, C. P., and A. R. Dawe (eds.), 1960. Symposium on Natural Mam- malian Hibernation. $3.00 paper, $4.50 cloth. Peters’ Check-list of Birds of the World, vols. 2-7, 9, 10, 12, 14, 15. (Price list on request. ) Turner, R. D., 1966. A Survey and Illustrated Catalogue of the Teredinidae (Mollusca: Bivalvia). $8.00 cloth. Whittington, H. B., and W. D. I. Rolfe (eds.), 1963. Phylogeny and ide of Crustacea. $6.75 cloth. Proceedings of the New England Zoological Club 1899-1948. (Complete sets only.) 4 Publications of the Boston Society of Natural History. Publications Office Museum of Comparative Zoology Harvard University Cambridge, Massachusetts 02138, U. S. A. © The President and Fellows of Harvard College 1970. THE GALAXIID FISHES OF R. M. McDOWALL” CONTENTS PAlos trac eee ee ee eee eee ee 341 Tismf SRG XG UGH Cop a: ee en ee 342 Materials and Methods) == 2 ee 343 Material “examimed. 229 ee 343 Samplingstechmiques =e ee 345 Measurements and counts __.___--------- 345 Clearinesandystaiming se ee ee 347 Sy/SteIMatiCS maser eens cnet coe ee eke Sees 347 tennlby (Gallenaie bye 347 General and diagnostic characters —_- 347 Generic classification — 350 Keyatongencra psc sn we tL 352 Calaxiask@uvicre ee eee Le 352 Key to species of Galaxias _____.__--_-----.- 302 Galaxias argenteus (Gmelin) —___---__- BD2 @alaxiasiasciatuse Gray 355 Galaxias postvectis Clarke __._____________ Sal Galaxias brevipinnis Gunther __________.__- 363 Galoxias culgaris Stokell= 2) Sie Galaxias maculatus (Jenyns) ___.---_._- 378 Galaxias usitatus McDowall _______--------- 382 Galaxias gracilis McDowall 385 Galaxias divergens Stokell _____-__._.. 385 Galaxias paucispondylus Stokell 390 Galaxias prognathus Stokell ____.________-. 393 INiecochannasGunther sees eee 394 Key to species of Neochanna 395 Neochanna burrowsius (Phillipps) — 395 Neochanna apoda Giinther __........------------ 398 Neochanna diversus Stokell —...------_- 402 SPeClesmincentacysedisn a eeeamn uum eee 404 Galaxias kaikorai Whitley = 404 Galaxiasyabbreviatus Clarke 2 406 IDiSCUSSIO Tee ee ee eee eee eee 406 i This paper is based on a thesis presented to the Department of Biology, Harvard University, in partial fulfillment of the requirements for the Ph.D. degree. * Fisheries Research Division, New Zealand Marine Department, Box 19062, Aro St. Welling- ton, New Zealand. NEW ZEALAND’ Identification of diadromous whitebait (MENT CS pr tik) ees pe RS ee eRe, 406 Life history and distribution patterns — 409 Origin and age of the New Zealand galaxiid. faunay le sega seen ea ee ee 412 Ndaptivies ra ciation: esses m ee Beene 414 Species groups and phylogeny —-----__------ 418 Acknowledgements, 22 ss es eee en 427 Iiteratuxesi@ited) 218 5. ee ea ee 427 INGE TIC aa 5 se SA ie ee i eee 431 ABSTRACT Fourteen species in the family Galaxiidae are recognized from New Zealand, three fewer than in previous works. These are placed in two genera —Galaxias and Neochanna—as follows: Galaxias argenteus (Gmelin), G. fasciatus Gray, G. post- vectis Clarke, G. brevipinnis Giinther, G. vulgaris Stokell, G. maculatus (Jenyns), G. usitatus Mc- Dowall, G. gracilis McDowall, G. divergens Stokell, G. paucispondylus Stokell, G. prognathus Stokell, Neochanna burrowsius (Phillipps), N. apoda Gimther, and N. diversus Stokell. This arrangement of taxa differs from previous arrange- ments in that lacustrine populations formerly known as G. lynx Hutton and G. koaro Phillipps are treated as synonyms of G. brevipinnis, G. anomalus Stokell is found to be a synonym of G. vulgaris and, although formerly placed in Galaxias, Neochanna burrowsius is regarded as showing much greater similarity to and affinity with the other neochannoid species and is accordingly placed in Neochanna. Study of samples of the migratory juveniles of the diadromous species (G. argenteus, G. fasci- atus, G. postvectis, G. brevipinnis, and G. macu- latus) showed that although clear diagnostic characters for the juveniles of these species do not emerge, it is possible to distinguish species in mixed samples by means of modal differences in length at migration, head length, and body depth. The diadromous species were found to have numerous small to moderate-sized eggs, to spawn mostly in the autumn and early winter, to spend larval and early juvenile life in the sea, and to Bull. Mus. Comp. Zool., 139(7): 341-432, June, 1970 342 migrate into fresh water during the subsequent spring. G. usitatus and G. gracilis have forsaken the marine migratory habits (because of landlock- ing), but have numerous small eggs. The re- maining seven species have few, larger eggs, spawn mostly in the winter and spring, and com- plete their entire life histories in fresh water. The correlation between egg size, egg number, and life history pattern suggests selective ad- vantage in having many small eggs, in species living initially in productive, marine plankton, and fewer, larger eggs in species living in flowing fresh water and not subject to the same type of dispersal away from the natal habitat. There is a very obvious relationship between range and life history pattern—those species with marine life history phases are widespread in the New Zealand region and may occur on offshore islands and also in other, more distant land areas (Australia, South America). These species tend to have easily determined phylogenetic relation- ships with species outside the New Zealand region. Species restricted to fresh water have a much more restricted range and have largely cohesive distribution patterns, which can be mostly ex- plained simply by known changes in New Zea- land’s geomorphology. The age of the New Zealand galaxiid fauna is unknown. The family seems to have evolved in the Australasian region, since about 90 per cent of the species occur there. Phylogenetic relation- ships with the Retropinnidae and Aplochitonidae and a common origin for the three families in some early Northern Hemisphere salmoniform stock are suspected. Phylogenetic relationships between Australian and New Zealand species can in many cases be established, and this, together with known marine life history phases, indicates that the New Zealand fauna is derived by transoceanic dispersal. The East Australian ocean current seems to provide a suitable mechanism for dispersal from Australia to New Zealand. Although the New Zealand freshwater fish fauna is very small, there is no evidence that the present fauna represents only a fragment of a formerly larger fauna, reduced by marine trans- gressions that occurred during the early and mid- Tertiary, or by the glaciations of the Pleistocene. Though the fauna is small, and though the Galaxiidae represent a large proportion of the fauna, the family shows little evidence of radiation to fill the New Zealand freshwater habitats. Galaxiids are mostly solitary, stream dwelling, benthic, invertebrate feeding predators. They seem to show considerable sensitivity to alterations in the nature of the stream catchment and _ its vegetation cover. The galaxiid fauna is easily and naturally di- visible into a series of small species groups. G. Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 argenteus, G. fasciatus, and G. postvectis are clearly closely related to each other, and to G. truttaceus in Australia. G. brevipinnis is very similar to, perhaps conspecific with G. weedoni in Tasmania, and is also probably ancestral to G. vulgaris. G. maculatus is common to Australia, New Zealand, and South America, and gave rise in New Zealand to G. usitatus and G. gracilis. G. divergens, G. paucispondylus, and G. prognathus form a very compact species group of an origin at present undetermined. N. burrowsius, N. apoda, and N. diversus are similarly very closely related and are perhaps derived from the Tasmanian neochannoid species, G. cleaveri and G. anguilli- formis. INTRODUCTION The fishes of the family Galaxiidae are mostly small and scaleless, more or less benthic in habit, with rounded trunks and somewhat depressed, broad heads. Nearly all the species are secretive and solitary and have thick, fleshy fins. Some species are nocturnal, with free-ranging, pool- dwelling habits, and may exhibit some deepening of the trunk. A few species have mid-water shoaling habits, and these tend to have membranous fins and a more slender form. The family Galaxiidae is very widespread in the Southern Temperate Zone, species occurring in Australia, New Zealand, South America, and South Africa, as well as on many islands in the vicinity of these land areas. Species abundance is greatest in Australia and decreases eastwards to New Zealand, South America, and South Africa in the pattern described by Fell (1962: 761). One species, G. maculatus (Jenyns), is found in Australia, Tasmania, Lord Howe Island, New Zealand, Chatham Islands, Chile, Patagonia, and the Falkland Islands, and is one of the most widely dis- persed species of freshwater fish. The family Galaxiidae is currently con- sidered to belong to the order Salmoni- formes (Greenwood et al., 1965: 394), comprising, with the families Aplochitoni- dae, Retropinnidae, and Salangidae, the suborder Galaxioidei. These four families are considered to constitute a distinctive radiation within the Salmoniformes. Var- ious of the three southern families— Galaxiidae, Retropinnidae, Aplochitonidae —have at some time been related to the salmonoid or the haplomous fishes (Regan, 1909; Berg, 1940; Chapman, 1944; Gosline, 1960); the present consensus agrees that they have very definite salmonoid affinities (Weitzman, 1967; McDowall, 1969). From the beginnings of galaxiid taxon- omy late in the 18th century, the family has been a difficult and confused one. The morphology of the New Zealand species is plastic, and in many localities and some species groups, active speciation is occurring. The failure of earlier work- ers to take into account the rather dis- tinctive juveniles, and the allometric growth that may succeed the juvenile stages, has led to repeated descriptions of some species. Lack of knowledge of the life history patterns and their relation to dispersal has resulted in description of fishes from apparently isolated localities as new. Repetitive description of well- defined species due to ignorance of earlier descriptions or mistaken identity has added to the problems, and confusion in the application of existing names has been considerable; e.g., Powell (1869), discussing the young stages of some Galaxias species, called them “smelt’— properly Retropinna in New Zealand—and published a figure that is clearly G. macu- latus (Jenyns), labeling it G. fasciatus Gray. Apart from a small paper by Hutton (1896) and Regan’s (1905) revision of the whole family, the works of Stokell (1945, 1949) were the first serious attempt to define the New Zealand galaxiid species, and for the first time it became possible to identify adults of most of the species occurring in New Zealand. As a result of these and later papers by Stokell (1954, 1959b, 1960) and one by the writer (Mc- Dowall, 1967a), there are currently 17 galaxiid species recognized from New Zealand. Studies of a New Zealand fishery based New ZEALAND GALAXIIDAE °* McDowall 343 on species of Galaxias (McDowall, 1964b, 1965a, 1968b) showed that more meristic data and clearer diagnostic characters should be sought for adequate identifi- cation of some of the species, especially in their juvenile stages. Subsequent collec- tions of many large samples of all the New Zealand species from a wide range of localities also suggested that there were some irregularities in their taxonomy. As a result of the present review, the number of species recognized is reduced to 14. An attempt to determine species groups, phylogenetic patterns, and the evolution of the New Zealand galaxiid fauna is long overdue. It is also time that an attempt be made to relate the New Zealand fauna to the galaxiid faunas in Australia and South America. It is the objective of this study to attempt a synthetic analysis of the New Zealand Galaxiidae, to examine the manner in which galaxiid fishes appear to have invaded New Zealand’s fresh waters and speciated there, and to determine the phylogenetic relationships of the species. Unfortunately, the systematics of the Aus- tralian and South American galaxiid faunas are not well known; studies of the species in these two areas will be necessary before the desired synthesis of the whole family can be accomplished. MATERIALS AND METHODS Material examined. A large collection of New Zealand Galaxiidae was studied, much of which was collected during a study of the biology of G. maculatus (McDowall, 1968b) or on specific field trips to collect certain species. Further material was col- lected by technicians at the Fisheries Re- search Division of the New Zealand Marine Department, and this was supplemented by samples in the collection of the New Zealand Dominion Museum. Neochanna burrowsius is a rare species that is difficult to collect, and my samples of this species were small; examples in the fish collections of the University of British Columbia and the National Museum of Canada were also 344 Figure 1. Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 North Auckland 7% Auckland Bay of Plenty NORTH ISLAND Taranaki LIVYLS Se LEWIS PASS BANKS PENINSULA SOUTH ISLAND Otago fo} u thy, Ny STEWART ISLAND New Zealand place names—regions and physiographic features mentioned in text. EAST CAPE studied. For most species, large series were examined from a broad range of geographi- cal localities. I have not listed in detail the material examined in the study, but in the distributional data for each species an asterisk is inserted by localities from which specimens were examined for meristic or morphometric data. The identifications of species from each locality are my own, except for a number reported by Fisheries Research Division biologists and_ tech- nicians. Museum abbreviations. In the listing of type specimens, the institutions at which the types are held are indicated by the following abbreviations: AMS Australian Museum, Sydney, Aus- tralia. BMNH_ British Museum (Natural His- tory), London, England. CMCNZ Canterbury Museum, — Christ- church, New Zealand. DMNZ_ Dominion Museum, Wellington, New Zealand. GMUO_ Geology Museum, University of Otago, Dunedin, New Zealand. MCZ Museum of Comparative Zool- ogy, Cambridge, Mass., U.S. A. MNHNP Museum National d'Histoire Nat- urelle, Paris, France. NZMD New Zealand Marine Depart- ment, Fisheries Research Divi- sion, Wellington, New Zealand. USNM _ United States National Museum, Washington, D.C., U.S. A. Sampling techniques. Galaxiid fishes are usually secretive, occupy deep cover, and are fast swimming; many species occur in very rapid, turbulent water. Thus they are usually difficult to capture. The principal tool used for collection was a small, back- portable electric fishing machine, which was used in all waters except estuaries where high salinities sometimes rendered it inoperable because of high water con- ductivity. Also, in some very pure moun- tain streams conductivity was very low and the effectiveness of the machine greatly reduced. The normal running time for one New ZEALAND GALAXUDAE * McDowall 345 set of batteries—a pair of six-volt motor- cycle batteries—was one and a half to two hours, although this depends on water con- ductivity. With two sets of batteries it was possible to spend a full day in the field without recharging. The effectiveness of the machine was greatest in shallow water, up to about 24 inches, and for the capture of solitary, cover-dwelling species. How- ever, using the machine in conjunction with small seine nets, shoaling species were easily captured in large numbers. Paralyzed fish were usually retrieved with small metal gauze dip nets, but in torrential streams it was necessary to place a barrier, like a large dip net, a bag net, or a small seine across the stream flow, and chase the fish downstream towards the barrier with the electrode. For capturing shoaling fishes, a small, five-foot, one-man seine was con- structed from fine-mesh mosquito netting strung between two bamboo poles; a length of light chain was used to weigh down the lower edge of the net. Captured fish were immediately placed in a pail of water containing a narcotic— usually chlor-butol, occasionally “MS 222.” Narcotizing the fish as they were caught prevented distortion due to asphyxiation and allowed long collection runs in the field without delays for fixing specimens. Whenever possible, the fish were fixed in the field, in shallow plastic photographic trays. The fish were spread out in the trays with minimal overlap and sufficient 10 per cent formalin poured on to cover but not float them. They were bottled when they had begun to harden. By this simple expediency, the difficulty of working with bent, twisted, and otherwise distorted specimens was almost completely avoided, and in general, the specimens were in ex- cellent condition. After fixation for four or five days in formalin, the fish were washed for a similar period in several changes of tap water and transferred to 40 per cent isopropyl alcohol for storage. Measurements and counts. Methods of measurement used were largely those de- 346 scribed by Hubbs and Lagler (1947: 13-15, figs. 3-5), in a few cases adapted to the particular morphological characteristics of the fishes studied. In most cases, measure- ments were taken with needle point di- viders, although in large species vernier calipers were found to be more effective. In general, dimensions were determined to the nearest half millimeter. In small fish, and in measuring small dimensions, usually those less than 15 mm, and whose reference points are well defined, measurements were estimated to the nearest quarter millimeter. Frequently, accuracy of this degree is not warranted since the reference points are not clearly defined, and variations in body flexure at fixation and types of preserva- tive used modify the body dimensions to an extent that makes accuracy of a quarter of a millimeter, and sometimes half a millimeter, quite meaningless. Measurements were taken as follows: total length—either length to caudal fork (L.C.F.), or if the caudal is rounded, to posterior extremity of fin (T.L.); standard length (S.L.); body depth at vent (B.D.V.) —used instead of greatest body depth be- cause the latter is greatly affected by sexual maturity and distension of the stomach after feeding; depth of caudal peduncle (D.C.P.); length of caudal peduncle (L.C.P.); predorsal length (Pre-D.); pre- anal length (Pre-A.); length of bases of dorsal and anal fins (D.F.B. and A.F.B.); maximum length of dorsal and anal fins (D.F.M. and A.F.M.); pectoral fin length (Pec.); pelvic fin length (Pel.); pre-pelvic length (Pre-Pel.); pectoral-pelvic length (Pec.-Pel.); head length (H.L. )—measured to edge of opercular membrane; head depth (H.D.)—an uncertain measurement, but taken vertically at the ridge across the nape which represents the posterior margin of the cranium, the position of which can be determined by running the finger for- wards across the top of the head; head width (H.W.); snout length (Sn.L.); post- orbital head length (P.O.H.L. ); interorbital width (lo.W.)—fleshy interorbital; diam- Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 eter of eye (D.E.)—horizontal fleshy eye diameter, not bony orbit; length of upper jaw (L.U.J.); length of mandible (L.M.); width of gape (W.G.). The following structures were counted: fin rays in the dorsal, caudal, anal, pelvic, and pectoral fins; vertebrae; gill rakers on the first arch; branchiostegals; pyloric caeca. Counting fin rays in galaxiid fishes presents a minor problem, since a variety of types of soft rays occurs. As in all the salmonoid fishes, procurrent rays are pres- ent in the dorsal, anal, and caudal fins. Hubbs and Lagler (1947: 9) recommended the inclusion of these rays in the counts of the dorsal and anal fins of salmonoids. In the Galaxiidae, such a procedure is a problem, since the anteriormost rays are usually deeply embedded in the opaque, fleshy fin bases and accurate counts are impossible without staining. Use of alizarin stain techniques showed that in the dorsal and anal fins there are from one to five of these rays, varying in size from a tiny, little-ossified splint, to a strongly-ossified but unbranched and unsegmented ray. Accordingly, the counts given in the sub- sequent descriptions are, in all cases, of segmented rays, whether branched or not. This procedure, which enables accurate and standardized counts, is more or less equivalent to the principal ray count, although sometimes a segmented, un- branched ray is counted, which does not quite reach to the distal margin of the fin, as it should to be counted as a principal ray. In the paired fins, the situation is a | little simpler and more stable. Occasionally | one unbranched but segmented ray is pres- — ent in the medial border of the pectoral | and pelvic fins, and this was counted, | together with the larger branched rays. The small, unsegmented splintlike ray, more rarely present in these fins, was not counted. Vertebral counts were taken as excluding | the urostylar vertebra and hypural plate. | Their inclusion would increase the count by one or two, depending on whether the urostylar elements were fused or not; the condition was found to be variable. All branchiostegals, including those which do not have a definite attachment to the hyoid arch, were counted. Gill rakers were counted in the conventional manner, the raker at the angle between the epibranchial and ceratobranchial, which does not asso- ciate with either bone, being counted with the lower limb. All counts were made using a dissecting stereomicroscope. In many cases, samples were sufficiently large to allow preparation of cleared, stained skeletal preparations. X-rays were also used extensively for verte- bral counts. Clearing and staining. During the early part of the study, the potassium hydroxide clearing technique of Hollister (1934) was employed. Later, the clearing method de- veloped by Taylor (1967), in which trypsin is utilized for the digestion of body tissues, was used. The principal advantage of this method is that the problem of explosion and distortion of fish and the fragmen- tation of old specimens is largely avoided. In addition, clearing is accomplished more rapidly than in other techniques, and dis- torted, asphyxiated specimens are often partially relaxed. SYSTEMATICS FAmMILy GALAXUDAE The family Galaxiidae was formed by Miller (1844) to contain the genus Gal- axias Cuvier, 1817. Osteological study is becoming imperative for understanding the relationships of the species within the family, as well as among the Galaxiidae and the Retropinnidae and Aplochitonidae, and the relationships of the three families with the broader sphere of the isospondy- lous fishes. The present synopsis hopefully forms an initial basis for determining the limits of the family. Diagnosis. Medium-sized to small fishes (3-60 cm) with 0-3 rudimentary to well- developed pyloric caeca. Both gonads de- veloped, although the left may be larger New ZEALAND GALAXHDAE * McDowall 347 than the right, ovaries gymnoarian. Urino- genital aperture on a papilla set in a post- anal depression. Sexes similar, male nuptial tubercles not present, but in many species sensory tubercles present on the head and pectoral fins in both sexes. All the species except one are believed to breed in fresh water, the exception in river estuaries. Some species are confined to fresh water, either lacustrine or fluviatile, others are amphidromous with marine juveniles. Seales lacking, lateral line well de- veloped, an accessory lateral line present dorsolaterally in some species. Pelvic fins abdominal, 4-8 rays, usually 7, or fin absent. Caudal fin emarginate to rounded, rarely forked, usually 16 principal rays (14 branched), procurrent rays well developed along caudal peduncle and an- terior to dorsal and anal fins, dorsal and anal fins originate well back on trunk. Vertebrae 37-64, branchiostegals 5-9. Maxilla partly included in gape, toothless; teeth on premaxilla and dentary uniserial, mesopterygoidal, basihyal, and pharyngeal teeth developed (reduced or absent in Neochanna). No supramaxilla; no vomer- ine teeth. Parietals large, uniting broadly in a median suture, supraoccipital not in con- tact with frontals and excluded from fora- men magnum. Posterior myodome open. Orbitosphenoid, basisphenoid, and proeth- moids absent; supraethmoid and _ ventral ethmoid present. Posttemporal simple; no mesocoracoid; postcleithrum present or absent. Epipleural and = epineural ribs present (except in Neochanna and Neso- galaxias); neural and haemal arches autog- enous, anterior uroneural not fused with terminal vertebra, none of terminal verte- brae upturned. Caudal neural and haemal spines much compressed. General and diagnostic characters Fishes of the family Galaxiidae present a varied but distinctive facies. The first observers (Forster, 1778; Bloch and Sch- neider, 1801) saw a resemblance to the 348 Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 Figure 2. Northern Hemisphere pikes (family Esoci- dae), probably because of the extremely posterior position of the dorsal and anal fins and the long jawed appearance of the single species with which they were ac- quainted. However, these similarities are superficial. The fishes in the family are scaleless, with thick, highly mucigerous, leathery skins. The head is usually moder- ately large, with prominent jaws, the upper and lower varying in proportional develop- ment so that the lower may protrude, recede, or be equal in length to the upper. Lateral line pores on the head are well developed. In New Zealand representa- tives, the disposition of these pores is fairly constant, with only occasional individual variations in pore number (Fig. 2). Their disposition can be related to the supra- orbital, infra-orbital, and hyomandibular branches of the lateral line system of the head (see Lagler et al., 1962: 391). The lateral line is well developed on the trunk from the upper edge of the opercular aperture to the middle of the tail base. It consists of a series of superficial papillae, set in a midlateral furrow, which may be well defined, especially caudally. In some species groups there is a dorsal accessory lateral line along the dorsolateral trunk, evident as a more or less distinct linear series of small, widely-separated papillae from the occiput to about the dorsal fin. The nostrils are well developed, the anterior one set in a small depression and Distribution of laterosensory pores in a generalized galaxiid. A, Lateral head; B, Dorsal head. tubular. In Neochanna it is especially well developed, sometimes extending forward beyond the upper lip. The posterior nostril is a simple aperture. The form of the mouth varies, the profile of the jaws from the ventral aspect varying from deep and narrow, U-shaped, to broad and shallow, with depth much less than breadth (Fig. 3). In the adults of most species, the head, anterior trunk, and pec- toral fins and fin bases are covered with small papillae. These are unlike the papillae of the Percidae, which are some- what horny in nature, or those in the Retropinnidae, which are much better de- veloped and more widespread. They are unusual in that they are present equally in both sexes. Although a directly reproduc- tion-related function cannot be ruled out, the bisexual occurrence of these papillae, apparently in all seasons of the year, sug- gests a sensory function, which may or may not be related to reproduction. These papillae do not appear to be connected with the lateral line system, as is the case in the head papillae in many fishes. Papillae of this type do not appear to have been discussed in the literature, and their function is at present obscure. The dorsal and anal fins are positioned posteriorly, and when depressed against the trunk, may overlie the base of the caudal fin; the anal fin is more or less be- low the dorsal. These fins are variable in their size and shape, usually short-based, New ZEALAND GALAxUDAE * McDowall Figure 3. Ventral profile of jaws. fasciatus. sometimes high and rounded, but in other cases much lower and not extending back much beyond the fin base. The caudal fin varies from well-forked to much-rounded. The pelvic fins are usually rounded in shape and expansive. The pectoral fins are variable in length and position and may be quite high laterally, with the blade of the fin vertical, or low lateroventrally, with the fin lamina more or less horizontal. In most species all the fins are thick and fleshy, especially at the bases. Teeth are present on the premaxilla, mandible, basihyal, mesopterygoid, the pharyngobranchial of the third, the epi- branchial of the fourth, and the cerato- branchial of the fifth branchial arches. The basibranchial plate is toothless. Teeth on all but the basihyal and the pharyngeal bones are uniserial; regularly in some spe- cies and in occasional individuals in others, however, there is a tendency for teeth to be displaced laterally from the primary row, appearing biserial. Mesopterygoidal teeth are reduced or absent in the neochannoid species. The teeth are usually conical, but in Neochanna apoda the mandibular and maxillary teeth are peculiarly flattened and incisorlike. This condition does not occur in any other galaxiids. In many species the jaw teeth are enlarged laterally as opposing groups of canines. Associated with the toothed bones are unattached, or A, Broad and shallow—as 349 in Galaxias divergens; B, Narrow and U-shaped—as in G. decumbent teeth, which usually lie freely in the tissues covering the bones. The structure of the ovaries in the Galaxiidae was described as gymnoarian by Hoar (1957: 289). Kendall (1922: 202) examined the “oviducts” of some salmo- noids and concluded that they are shallow, open troughs and not entirely lacking, and that they are not radically different from those of other isospondylous fishes. How- ever, the reduced condition of the oviducts, as in the Salmonidae, persists throughout the salmoniform fishes, and the condition is sufficiently distinct for Hoar to dis- tinguish them from other ovarian types. Breder and Rosen (1966: 614) followed Hoar, stating that in the Galaxiidae and other salmonoids, the “ova pass into the peritoneal cavity and thence through the pores to the exterior.” Henderson (1967: 447) concluded that the eggs of Salmoni- dae are discharged into the abdominal cavity, and that proper oviducts are lack- ing. The New Zealand Galaxiidae exhibit considerable morphological plasticity. Most characters were found to vary from species to species, and even usually — stable characters, like pelvic and caudal fin ray number, were found to differ in several phylogenetic lines. Stokell (1945:475) con- sidered vertebral number to be the most important taxonomic character. This has 350 proved to be a useful character, but it is very important to bear in mind the temper- ature differences that occur along the 900 mile north-south axis of New Zealand— 34 1/2 to 47 degrees south latitude—and the effect of temperature on vertebral number. Apart from vertebral number, important meristic characters included number of caudal, anal, pelvic, and to a lesser extent, pectoral and dorsal fin rays. The number of gill rakers and branchiostegals exhibits interspecific variation. The most important morphometric char- acters were the following: length and depth of caudal peduncle, relative positions of the dorsal and anal fins and their basal and maximal lengths, lengths of pectoral and pelvic fins, head length, eye diameter, lengths of upper and lower jaws, width of gape. The development of canine and mesopterygoidal teeth, pyloric caeca, and gill rakers exhibits interspecific variation. In some species groups, color pattern is diagnostically important: e.g., G. fasciatus Gray, G. argenteus (Gmelin) and G. post- vectis Clarke are similar in form but can be separated immediately and reliably by color pattern alone. In other species groups, specific differences are clearly indicated by fundamental differences in the life history pattern: e.g., G. brevipinnis Ginther has marine or lacustrine whitebait juveniles, whereas G. vulgaris Stokell, which is morphologically quite similar, has no whitebait stage. Most of the taxonomic characters used are completely conventional in ichthyology, but the morphological plasticity of the Galaxiidae results in a considerable di- versity of such characters. Some of these are stable and unimportant throughout much of the family although they show significant variation in certain species or species groups (e.g., pelvic fin ray number, snout length), but other characters vary widely throughout the New Zealand mem- bers of the familv. Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 Generic classification Seven generic names have been applied to New Zealand galaxiids. Two of these involve now obvious errors—the use of Esox by early workers and the failure of Jenyns (1842: 118) to recognize the pre- viously published genus Galaxias when he described galaxiid species in a new genus Mesites, a name further invalidated by preoccupation for a genus of beetles (Scho- ennherr, 1838). These two names are clearly not applicable to galaxiid fishes and present no nomenclatural or taxonomic difficulties. The type genus for the family is Galaxias Cuvier, 1817, for which the type species is G. fasciatus Gray, 1842 (see McDowall, 1967b). Giinther (1867: 306) described a galaxiid mud-fish in a new genus Neo- channa, which was distinguished chiefly by the absence of pelvic fins. In 1899, Ogilby placed G. attenuatus (Jenyns) in a new genus Austrocobitis, distinguished from Galaxias by the form of the trunk, the small fins, and the forked caudal. Whitley (1935, 1956a, b) has consistently used Austro- cobitis, but Stokell (1945: 124) claimed that these characters are widespread amongst divergent groups of galaxiids and that these species do not form a natural grouping. G. attenuatus [=G. maculatus (Jenyns)] and its New Zealand and Aus- tralian derivatives do have characters that set them apart from the rest of the family. However, at present I think that a broad generic revision of the family is necessary, and, thus, that it is inappropriate to make generic changes of this type here. Scott (1936) proposed a reorganization of the family at subfamilial, generic, and subgeneric levels. In this paper he placed G. burrowsius Phillipps in a new genus Saxilaga, distinguished by the lack of mesopterygoidal teeth and the presence of pelvic fins (cf. Neochanna, which usually lacks both, and Galaxias, which has both). Stokell (1945: 129) correctly showed that G. burrowsius sometimes has mesoptery- goidal teeth, though they are reduced in size and number. In this paper (p. 134) he listed Saxilaga as not recognized and later (1949: 481) described Phillipps’s spe- cies in the genus Galaxias. Phillipps (1940: 39) included this species in the genus Para- galaxias Scott, but this is clearly an error since Scott (1936: 87) defined Paragalaxias as having the dorsal fin well forward, over the pelvic fins. The allied problem of the validity of the genus Paragalaxias need not be considered here. Scott (1966) reasserted the validity of his generic arrangement of the family. For Saxilaga, he noted (p. 250) that “further investigations have shown that certain diag- nostic features originally described as absolute probably are not so.” He main- tained that Saxilaga is a good genus for G. burrowsius Phillipps, G. globiceps Eigen- mann, and G. anguilliformis Scott, but noted that “if Saxilaga is to be maintained, modal and not absolute criteria are to be accepted for these features.” He assembled a series of characters that he considered to unite the three species—“elongate body, small eye, small head, reduced number of pelvic fin rays (modally five or six), paired fins short, vertical fins low, squarish, their rays compressed with or without branch- ing .... anal continuous or sub-continuous with caudal ridge which is well developed, high, caudal rounded or sub-truncate, fish heavily pigmented . . . . taken collectively, they appear to constitute a significant con- stellation.” Scott’s practice of basing generic di- visions on apparently plastic, adaptive, and often widespread characters leads to prob- lems that suggest that such generic di- visions are better abandoned. Examination of the distribution of galaxiid genera as he uses them produces the following patterns: Saxilaga—Tasmania, New Zealand, South America; Brachygalaxias Eigenmann, ac- cording to Scott's arrangement—South America, Australia. If we are to use the genus as a collective grouping for species comprising several similar independent radiations from the central stock of the New ZEALAND GALAXxHDAE * McDowall 351 family, then Scott's genera are proper. However, I think that our understanding of the family is better served if we use the taxon to express phyletic relationships. If the species in these two genera are phy- letically related, these patterns raise con- siderable zoogeographic problems, since all the species included belong to groups which, now at least, are found only in fresh water and none of which belong to the much more easily dispersed diadromous species groups, This association of morpho- logical peculiarity with restriction to fresh water has important implications. First, since these species are restricted to fresh water, their ability to disperse is probably lower than that of diadromous species. Second, the fact that they are restricted to fresh water suggests that their common morphological peculiarities may be related to independent development of adaptations to specialized freshwater habitats, as is possibly the case in the mud fishes in Tasmania and New Zealand. I think that some of the similarities that Scott has used to draw species into generic groups are convergent adaptations to similar modes of life (however, see p. 425, where dis- persal and phylogeny are discussed). Scott (1966: 253), discussing the sub- familial classification of the family, sug- gested that “a more natural division of the family would appear to involve the associ- ation on one hand of forms with more than 50 vertebrae and on the other hand of forms with fewer than 50 vertebrae.” G. gracilis McDowall from New Zealand has 47-50 vertebrae, the lowest number recorded for the family in New Zealand. This species is almost certainly derived from G. macu- latus (Jenyns), which has 59-64 vertebrae, the maximum for the family in New Zealand. Thus, in this simple case of land- locked speciation, G. gracilis has tra- versed the full range of vertebral number for Galaxias in New Zealand. According to Scott’s proposal, it has thus moved from one subfamily to the other. Bearing in mind the effect of temperature on verte- bral number, it is clear that this is not a useful character at the subfamilial, or even generic level. Scott (1966) also made use of pelvic fin ray number, combining species that exhibit reduction in the number of rays from the usual seven. I do not think that this is a useful character either. Within the New Zealand Galaxiidae alone, re- duction in pelvic fin ray number has almost certainly taken place in three widely di- vergent lines—those leading to N. burrow- sius (Phillipps), G. divergens Stokell, and G. usitatus McDowall. I think there is a need for the generic classification of the family to be based on more fundamental char- acters than vertebral and pelvic fin ray number, and for the classification to better express natural groupings and phylogeny. For these reasons, only two genera are recognized for the New Zealand Galaxiidae in the present study—Galaxias Cuvier and Neochanna Giinther, following — Stokell (1945, 1949). KEY TO GENERA Mesopterygoidal teeth, epipleural ribs, supraeth- moid and ventral ethmoid present, pelvic fins Sie Ol@NINOLCH LAYS] @ cee en a noe Galaxias Mesopterygoidal teeth reduced or absent, epi- pleural ribs, supraethmoid and ventral ethmoid absent, pelvic fins five or fewer rays, or fins and Sirdlemabsentycre yeu ee aan ne ee Neochanna GALAXIAS CUVIER Galaxias Cuvier, 1817: 183 (type species Galaxias fasciatus Gray by subsequent monotypy ). Mesites Jenyns, 1842: 118 (type species Mesites attenuatus Jenyns 1842 by subsequent desig- nation, Jordan, 1919: 212, preoccupied by Mesites Schoennherr, 1838, Coleoptera). Austrocobitis Ogilby, 1899: 158 (type species Mesites attenuatus Jenyns, 1842 by subsequent designation, Whitley, 1956a: 34). Diagnosis. Trunk cylindrical to a little compressed, naked; dorsal fin origin very posterior, about above vent. Pelvic fins present, six to eight rays, commonly seven; pectoral fin positioned laterally to low lateroventrally. Caudal fin usually with 16 principal rays, sometimes reduced to 15 or fewer. Jaw teeth conical, uniserial, with or without canines; mesopterygoidal teeth Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 well developed to rudimentary, uniserial; lingual teeth biserial. Median supraeth- moid and ventral ethmoid present; post- cleithrum present or absent; epipleural ribs present. Key TO SPECIES OF GALAXIAS This key is adapted from McDowall (1966b), incorporating taxonomic changes made since that time and those proposed in the following pages. 1. Lower jaw much longer than upper —— Eire ae Penne Hg G. prognathus, p. 393. Lower jaw not much longer than upper — 2. 2. Lower jaw much shorter than upper, tucks behind upper when mouth closed _ 3. Jaws sub-equal, lower if shorter not tucking behind upper == =a 4, 3. Canine teeth strongly developed —_- SO ie MADD FU e's eed G. brevipinnis p. 363. Canine teeth lacking _.. G. postvectis p. 361. 4. Pyloric caeca long, length much greater than! breadth)... ee Be Pyloric caeca short to absent, usually short stubs 5. Vertebrae 49-57, gill rakers 9-13, depth of caudal peduncle usually much less thanslength) == soe G. vulgaris p. 372. Vertebrae 58-61, gill rakers 14-17, depth of caudal peduncle sub-equal to length esi Dal ee abe a eae. oye G. argenteus p. 352. 6. Canines well developed — G. fasciatus p. 355. Canines lacking or weak 22 fe 7. Gill rakers very short, 11 or fewer __...... 8. Gill rakers long, 11 or more __....- 9. 8. Usually six pelvic rays, 15 caudal rays Hee ie Re a G. divergens p. 384. Usually seven pelvic rays, 16 caudal | RAY See et ehe oe ae dace G. paucispondylus p. 390. 9. Gill rakers up to 17, vertebrae 54 or more Gill rakers 18-23, vertebrae 47-50 _.... pha Bias Sn Ue Dea ee G. gracilis p. 384. 10. S.L./H.L. 22.0-24.6%, 54-59 vertebrae WEED) Bik ORR ie Le G. usitatus, p. 382. S.L./H.L. 18.5-21.6%, 59-64 vertebrae Be Lal aes, AR SAA A ESET G. maculatus p. 378. Galaxias argenteus (Gmelin, 1789) Figure 4 Esox argenteus Gmelin, 1789: 1393 (holotype: unknown; locality: a small lake in Dusky Bay (Dusky Sound?), New Zealand. ) Esox alepidotus Bloch and Schneider, 1801: 395 (replacement name for E. argenteus Gmelin, 1789): Cuvier, 1817: 184; Forster, 1844: 142. Galaxias alepidotus: Richardson, 1843: 25; Dief- fenbach, 1843: 219; Richardson, 1848: 77; Gunther, 1866: 208; Hutton, 1872: 58; 1889: New ZEALAND GALAXIIDAE * McDowall 353 Figure 4. 284: 1896: 317; 1904: 51; Regan, 1905: 375; Waite, 1907: 12; Phillipps, 1927a: 13; Stokell, 1949: 493; 1954: 419. Galaxias forsteri Valenciennes, In Cuvier and Valenciennes, 1846: 531 (replacement name for Esox alepidotus Bloch and Schneider, 1801). Galaxias grandis Haast, 1872: 278 (holotype: apparently lost, see Stokell, 1949: 493; locality: creeks near Lake Ellesmere); Hutton, 1874: 107; 1904: 51. Galaxias kokopu Clarke, 1899: 88 unknown; locality: | western Island); Hutton, 1904: 51. Galaxias argenteus: Whitley and Phillipps, 1940: 230 (partim); Stokell, 1960: 235. Diagnosis. Differs from G._ fasciatus Gray (Fig. 6) in coloration and in having very strongly developed pyloric caeca, longer head, more posterior pelvic insertion, higher depth of caudal peduncle/length of caudal peduncle ratio and jaw in head ratio (i.e., longer jaws), eye further for- ward in head and somewhat higher pec- toral fin ray counts. Overlap in most of these characters is considerable and color- ation is the most useful character. G. ar- genteus has numerous, small, irregular, gold spots on the dark trunk, while G. fasciatus has more regular vertical pale bands. Differs from G. postvectis Clarke (Fig. 9) in coloration, in having stronger de- velopment of canine teeth in the jaws, much longer head and jaws, the jaws sub- equal, the eye further forward in the head, longer anal fin base, more posterior pelvic fin insertion, more anal fin rays, and some- (holotype: slopes, South Galaxias argenteus (Gmelin), 280 mm L.C.F., Little Waitangi Stream, Pavatahanui Inlet. what higher numbers of branchiostegals and gill rakers. Coloration is again the most useful means of differentiating these species, jaw length, especially the short- ened lower jaw in G. postvectis also ena- bling easy separation. Description. Stout bodied, trunk some- what rectangular in section and flattened dorsally, mid-dorsal groove moderately to well developed. Trunk deep, deeper than broad, greatest depth at or in front of pelvic fins. Depressed dorsally on head, considerably compressed posteriorly on caudal peduncle which is very short and deep, usually deeper than long. Lateral line an indistinct lateral furrow; accessory lateral line present but difficult to dis- tinguish. Head very long, a little broader than deep. Eye moderately large and set moderately deep on lateral head, eye diam- eter/head length ratio not high because of great length of head; interorbital convex, very broad; jaws about equal, prominent. Lips thick and fleshy; cleft of mouth mod- erately oblique, reaching to between mid- dle and posterior margin of eye. Profile of lower jaw from ventral aspect deep and rather narrow, U-shaped. Canine teeth well developed laterally in both jaws; mesopterygoidal teeth strongly developed; gill rakers long; pyloric caeca strongly de- veloped. Unpaired fins well developed, with thick fleshy bases; dorsal base of moderate 354 length, anal base long; both fins have greatest fin length much exceeding basal length, with rounded distal margins. Dorsal fin set well back, anal origin below or a little behind dorsal origin. Pectoral fins moderately long but not expansive, fleshy, inserted moderately low lateroventrally. Pelvic fins very long, expansive, and fleshy. Caudal fin long, thick and fleshy, depth noticeably less than body depth; truncated or a little emarginate in small specimens; caudal peduncle flanges well developed, extending forward almost to anal fin in- sertion. Variation. Meristic: dorsal 10 (16), 11 (21), 12 (2); caudal 15 (1), 16 (38); anal 12S) Ss (20) 41S) SG) pelviers, (38), 8 (1): pectoral) 13(7))14 (27). 1s (5); branchiostegals 7 (2), 8 (33), 9 (4); vertebrae 58 (4), 59 (16), 60 (15), 61 (2); gill rakers 4-10 (3), 4-11 (7), 4-12 (1), 5-10 (4), 5-11 (21), 5-12 (2). Morpho- metric: see Table 1, p. 358. Coloration. Often dark, a deep gray- brown, sometimes paler, approaching a buff color. The head, dorsal and _ lateral trunk, and fin bases are profusely covered with delicate, gold spots, lines, crescents and rings. These tend to be finer dorsally and on the head, coarser and bolder on the trunk. The belly is usually paler, bluish gray in dark examples, correspondingly lighter in paler specimens. A bluish blotch is present above and behind the pectoral fin base, but in heavily pigmented fishes it is of similar color to the trunk and is indistinct. Size. Clarke (1899: 83) reported that G. kokopu (= G. argenteus) grows to 23 inches (584 mm) and a weight of six pounds. Haast (1872: 278) recorded a specimen 19.3 inches (490 mm) and Stokell (1949: 494) one of 17 inches (432 mm). G. argenteus is thus reliably reported to grow to much greater size than any other galaxiid, although large examples are now very The largest specimen I have seen 330 mm and others were com- monly up to about 280 mm long. rare, WaS Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 50 PS a ss 3) m fo) 0 = m — 50 2 b 58 59 60 61 VERTEBRAE Figure 5. Variation in vertebral number in Galaxias ar- genteus. A, Localities in the southwest of the North Island —Wellington District (20 examples); B, Localities in the west of the South Island (17 examples). Population differences. There was little scope amongst the few specimens examined for revealing regional variation in G. argenteus. From grouping samples from the Wellington Province, and those from the west coast of the South Island, there appears to be slight displacement in dorsal fin ray number, the northern examples having modally fewer rays than those from the south. A similar displacement in verte- bral number is evident (Fig. 5). These differences are slight, but further study of more adequate samples may confirm the southward increase in meristics hinted at here. Habitat. G. argenteus is exclusively low- land in range, inhabiting pools in lowland swamps and streams. It is an uncommonly seen species, usually lurking beneath cover, and is probably nocturnal, feeding in more open water during the night. Haast (1872: 278) reported catching G. argenteus of large size from very small streams, and such waters appear to be a characteristic habitat, especially when overgrown with flax (Phormium tenax) and raupo (Typha angustifolia). Collections from the west coast of the South Island have shown that G. argenteus is sometimes common in flax swamps and also occurs in the bush-stained, tea-colored streams there. It has also been taken from Lake Brunner (G. A. Eldon, pers. comm.) and may be more common in shallow weedy lakes than present records indicate. Life history. Specimens of G. argenteus collected in March included a female ap- proaching maturity and a ripe male. Others collected during September to December had very immature gonads. These data suggest autumn or early winter spawning. G. argenteus has a whitebait juvenile com- parable with juveniles of G. fasciatus and G. postvectis, which migrates into fresh water during the spring. These species with whitebait juveniles migrate upstream to- gether, and this suggests that they may spawn at about the same time, during the autumn or early winter. Nothing is known of the spawning lo- cality, although spawning migrations are not suspected. The eggs of the female approaching maturity were too small for useful size determination, but this fish, 251 mm long, contained about 11,000 eggs. After hatching, the larvae are probably carried downstream to the sea, where larval and juvenile development occurs. At their subsequent upstream migration, the white- bait of G. argenteus are transparent, with very little pigmentation (Fig. 43) and are comparatively large (50-55 mm). Soon after migration, trunk pigmentation in- creases and intensifies to a dark gray- brown, eight to ten pale blotches or bands develop across the lateral trunk, and the adult pattern finally becomes superimposed on the juvenile banding. The stout, deep- bodied form of the adult is rapidly at- tained. Distribution. G. argenteus is widely dis- tributed in lowland localities that are accessible from the sea. It is known from the following: Mokau River (Fig. 7: 25); Pokaka Stream (27); Waikawa Stream (32*); Waikanae River (36*); Whareroa Stream (39*); Horokiri Stream (40); Little Waitangi Stream (41*); Trotter's Gully Stream (44) and Hawkin’s Gully Stream Stream (45*), Makara System; Belmont Stream (48) and Moonshine Stream (42), Hutt System; York Bay Stream (51*); * From P. 345. New ZEALAND GALAXIIDAE *° McDowall 355 Days Bay Stream (53); Wainuiomata River (50); Wairarapa (55, Stokell, 1949: 494); tributaries of Lake Ellesmere (89, Haast, 1872: 278); Mokihinui River (69); Buller River (71); Grey River (72); Lake Haupiri (73); Lake Brunner (74*); Lake Kaniere (77); Lake Ianthe (79*); What- aroa River (80* ); Lake Paringa (81, Haast, 1872: 278); Moeraki River (82); Dusky Bay (86, Dusky Sound?, type locality, Forster, 1778: 159); Stillwater River (87); Milford Sound (85, Hutton, 1896: 317); Southland (91, Stokell, 1949: 494); Horse- shoe Bay Creek (93); Chatham Islands (94, Skrzynski, 1967: 95). Galaxias fasciatus Gray, 1842 Figure 6 Galaxias fasciatus Gray, 1842: 73 (syntypes (3): BMNH_ 1967.8.14.9-11, not seen; locality: River Thames, New Zealand); Dieffenbach, 1843: 221; Valenciennes, In Cuvier and Valen- ciennes, 1846: 350; Richardson, 1843: 25, 1848: 77; Gunther, 1866: 209; Kner, 1865: 319; Hut- ton, 1872: 59; Clarke, 1899: 90; Hutton, 1904: 51; Regan, 1905: 374 (partim); Phillipps, 1926b: 293, 1927a: 13, 1940: 15; Stokell, 1949: 492. Galaxias reticulatus Richardson, 1848: 76 (syn- types (3): BMNH 1967.8.14.12-14, not seen; locality: Auckland Islands? ). Galaxias brocchus Richardson, 1848: 76 (holo- type: BMNH 1855.9.19.800, not seen; locality: Auckland Islands? ). Galaxias argenteus: Whitley and Phillipps, 1940: 230 (partim ). Diagnosis. Differs from G. argenteus (Gmelin) (Fig. 4) in characters discussed in the diagnosis of that species (p. 353); differs from G. postvectis Clarke (Fig. 9) in coloration, in the absence of pyloric caeca and the presence of better-developed canine teeth in the jaws, in its longer and more slender head, and in its longer jaws, particularly the lower jaw. G. post- vectis has somewhat fewer anal fin rays and branchiostegals, and more vertebrae and gill rakers. Taxonomy. Two names that Stokell (1949) failed to apply to any New Zealand galaxiids are G. brocchus Richardson and G. reticulatus Richardson, described from 396 Bulletin Museum of Comparative Zoology, Vol. 139, No: 7 Galaxias fasciatus Gray, 155 mm L.C.F., Makahika Figure 6. the Auckland Islands. They were treated as synonyms of G. fasciatus by both Giin- ther (1866: 209) and Regan (1905: 374). Fishes of this type have not otherwise been recorded from the sub-Antarctic islands of New Zealand, the only species there being G. brevipinnis Ginther. Both Giinther and Regan based their identifications on re- examination of Richardson’s material, and the excellent likenesses of G. fasciatus in Richardson's figures (his plates 42 and 43) definitely support the view of Giinther and Regan. Despite several collections from these islands (e.g., the Cape Expedition, 1941-45, see Stokell, 1950), G. fasciatus has not been re-collected there, and there is no evidence to suggest it is present. It is possible that it has become extinct in the islands since Richardson’s fishes were col- lected, or that the material he studied was incorrectly labeled. Since G. fasciatus has marine larvae and juveniles, its dispersal to the sub-Antarctic islands is comprehen- sible, although its temperature preferences appear to be higher than those of G. brevi- pinnis (McDowall, 1965a: 299), indicating that these islands are probably less suited to G. fasciatus than to G. brevipinnis, which is present there. The possibility that the specimens were incorrectly labeled re- mains nothing more than a_ possibility. Naxeart aAlaaa >] ey ay . Nevertheless, G. brocchus and G. reticu- latus are probably best regarded as syno- nyms of G. fasciatus, and as not occurring Stream, Ohau River System. on the sub-Antarctic islands of New Zea- land, until further collections indicate otherwise. Description. Stout bodied, trunk squar- ish to rounded in section, somewhat flat- tened dorsally with middorsal furrow present; trunk deep, greatest body depth at or a little in front of pelvic fins, de- pressed anteriorly on head and much com- pressed on caudal peduncle, which is short and about as deep as long. Lateral line a distinct lateral groove; accessory lateral line present. Head prominent, broader than deep and somewhat depressed; eye large, towards upper head profile, inter- orbital convex, very broad. Jaws about equal, lips prominent; cleft of mouth reach- ing beyond middle of eye, oblique. Profile of lower jaw from ventral aspect deep and narrow, U-shaped. Canine teeth strongly developed laterally in both jaws, meso- pterygoidal teeth well developed; gill rakers well developed; pyloric caeca lack- ing. Median fins well developed, with thick fleshy bases; prominent, greatest fin length much greater than base length, distal mar- gin of fin much rounded; anal origin about below dorsal origin. Pectoral fin prominent and fleshy, rounded in outline; insertion moderately low. Pelvic fin expansive and long, inserted behind mid-point of standard length. Caudal fin fleshy, long, emarginate or truncated, emargination usually becom- NEw ZEALAND GALAxHDAE * McDowall Byoy Tf @ Galaxias fasciatus = cco G._argenteus 4 ¥ * G. postvectis Figure 7. Distribution of Galaxias argenteus, G. fasciatus, and G. postvectis (numbers in figure as in text pp. 355, 360, and 363). 358 Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 TABLE 1. MORPHOMETRIC VARIATION IN LARGE, STOUT-BODIED SPECIES ( FIGURES GIVEN AS PERCENTAGES OF DENOMINATOR OF RATIO). G. argenteus Min. Mean Max. Giles sen Gale 83.3 85.5 87.0 BaDAVe/ Sales 18.7 21.0 23.4 IL GHP Salles 98 114 13.5 ID (Cpe Le Cele 100.0 113.6 125.0 Pre D./S.L. 74.6 76.3 les) Pre D./Pre A. 94.3 98.0 102.0 D.F.B./S.L. 94 119 13.4 D.F.B./D.F.M. 45.7 55.6 68.5 A.F.B./S.L. 13.5 15.4 17.8 A.F.B./A.F.M. 55.3 61.7 76.3 Pre Pel./S.L. 53.5 56.4 58.1 Pec.Pel./S.L. 26.8 29.1 31.3 Pec./Pec.Pel. 48.2 60.7 68.5 Pel.An./S.L. 20.8 BOT 25.0 Pel./Pel.An. 56.8 68.5 77.8 ieee Salen 27.0 29.1 30.5 H.D./H.L. 50.0 54.6 61.0 H.W./H.L. 57.5 64.9 74.1 Sn.L./H.L. 26.6 29.6 32.4 P.O.H.L./H.L. 50.8 55.6 61.4 Jo.W./H.L. 40.0 42.5 46.1 D.E./H.L. 14.8 ile DPD? L.U.J./H.L. 41.2 43.7 45.7 L.M./H.L. 37.7 41.3 43.9 W.G./H.L. BSH CEO MANA: Fish examined 36 G. fasciatus G. postvectis Min Mean Max. Min. Mean Max. SIGS aID = onan 84.0 87.0 88.5 SON Lie OR IeA 16.4 19.6 22.4 10.8 12.0 14.9 12.1 13.2 14.4 61.0 76.9 114.9 95.2 102.0 113.6 73.0 76.5 79.4 10.9)" Aa eiG29 96.2 100.0 102.0 92.6 97.1 100.0 95 108 #£11.9 102) Se ae 47.6 546 64.1 50.0 58.1 65.4 11.9 14.0 16.1 11.1 13.4 144 54.4 614 #&70.4 53.8 61.7 66.2 48.8 53.2 56.8 50.3 52.9 56.2 96.8 29.4 32.7 28.9 31.8 35.3 50.6 60.5 72.4 41.3 53.1 67.3 914 2241 27.4 93.1 248 26.5 SoD) OO sold me Olly LD 22.8 25.9 28.6 OAT BO)» S40) 46.1 53.5 60:2 56.2 64.1 72.5 55.9 71.4 78.1 59.5 70.4 78.1 Se Sills). BO) Sls) eso S/H 45.7 50.5 55.0 49.0 50.8 54.1 40.3 43.9 47.4 40.5 448 47.4 NG 20.20 a e2Ar> 17.8 19.9 23.5 42.4 485 51.8 37.5 40.7 43.3 40.8 45.5 50.0 27.5 33.0 35.7 33.3 40.8 46.5 34.4 38.8 42.6 60 25 ing reduced with growth; fin depth usually somewhat less than greatest body depth; caudal peduncle showing considerable de- velopment of flanges. Variation. Meristic: dorsal 9 (15), 10 CATO a ALS ye (C2) Sea 2))encaucdaly i's (Gl) GH (C60)) e712) canal lula 3)) 5 IC 20) 13 (42), 14 (10), 15 (1); pelvie=7 (63); pectoral 12 (20), 13 ce 14 (7), 15 (4); branchiostegals 6 (3), eee 8 Sau 9 (3); vertebrae 56 (1),-57 (ll) (538.(22))- 59 oe 60 (2), 61 (1); gill rakers 4-8 (1), 4-9 (5), 4-10 (3), 4-11 (3), 5-9 (5), 5-10 (19), 5-11 (1), 5-12 (1). Morpho- metric: see Table 1 Coloration. Trunk color a dark purplish gray, banded dorsally and laterally with a series of narrow, pale, vertical bands. The bands are numerous in young fish, becom- ‘ower and more restricted to the the trunk as the fish grow. ing nari posterior of Lateroventrally, the trunk coloration alters quite abruptly to a dull purplish brown. There is a prominent, dark, blue-black blotch above and behind the pectoral fin base. Frequently living in small, bush- covered creeks and streams, G. fasciatus appears well adapted to broken lighting conditions. Size. G. fasciatus is one of the largest species of Galaxias and is known to grow to 260 mm. It commonly reaches 200 mm. Population differences. Regional char- acter differences were not found in diadro- mous populations of G. fasciatus from widely separated localities, although more | intensive studies may show that they do occur. Taxonomically interesting differences be- tween diadromous and lacustrine popu- lations were found. The fishes from Lake Okataina and the Kaihoka Lakes were New ZEALAND GALAXIIDAE * McDowall 359 50 a SO z 50 ‘S b Mm 8 0 50 Cc (6) : 56 57 58 59 60 61 VERTEBRAE 50 B a 0 x 71 A mM © 50 rm 2 b i?) < 0 50 C 0 I] 12 13 14 ANAL FIN RAYS Figure 8. Variation in meristics in Galaxias fasciatus. A, Vertebrae; B, Anal fin rays; a, Lake Okataina (8 examples); b, Kaihoka Lakes (8 examples); c, Diadromous populations (47 examples). found to be considerably more slender amples had fewer vertebrae and fewer than sea-going fishes, and those in the rays in the anal fin (Fig. 8), although Kaihoka Lakes had a shorter head (mean overlap with diadromous fishes was sub- H.L./S.L. 25.9 in diadromous and 24.3 in — stantial. Fewer branchiostegals were pres- Kaihoka populations). Lake Okataina ex- ent in the fishes from the Kaihoka Lakes, 360 again with overlap (mean number, Kaihoka 6.75, diadromous 7.83). As such landlocked populations become better understood, it may become necessary to recognize them as separate taxa at the species or subspecies level, as has been done with G. maculatus, but existing data do not justify this. Habitat. G. fasciatus is essentially a low- land species, although it shows moderate penetration of river systems, even where substantial falls are present. Adults have been collected from small pools on the faces of high waterfalls and the juveniles are known to be able to climb wet, smooth surfaces with ease and rapidity. The lo- cality furthest from the sea where G. fasciatus is known is the Kahuterawa Stream, a tributary of the Manawatu River about 40 miles upstream from the sea (C. L. Hopkins, pers. comm.). G. fasciatus is found mostly in small, quiet, winding creeks in coastal and lowland bush, usually hiding beneath cover such as logs, over- hanging banks, tree roots etc., or amongst rock aggregations at the bases of small pools and cascades in the streams. This species is also quite common in the tannin- stained waters of flax swamps on the west coast of the South Island. Sea-going popu- lations occur in Lakes ITanthe and Mapour- ika, in addition to the previously mentioned landlocked populations. Life history. The breeding site of G. fasciatus is undescribed, but since ripe adults were collected in typical adult habi- tat, it seems unlikely that there is an adult breeding migration. Ripe males were col- lected with milt running as early as the end of February and study of gonad maturity suggests that breeding takes place mostly during the autumn and early winter (Feb- ruary to May or June). The eggs are of moderate size, 1.3-1.6 mm in diameter, and numerous; a female 160 mm long contained 9,100 eggs. The larvae are apparently car- ried to sea after hatching, and a sub- sequent upstream migration of the juveniles occurs the following spring, together with Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 the whitebait of other diadromous species, although relatively late in the overall migration period (McDowall, 1965a). At migration the young G. fasciatus are trans- parent, little pigmented (Fig. 41), and measure 38-48 mm. Trunk pigmentation develops quickly after the fish enter fresh water, beginning as a general covering of melanophores; later a series of narrow, alternating light and dark bands develops. The slender juveniles become much stouter and the banding bolder, as the number of bands along the trunk increases. Eventu- ally they extend over the dorsum of the trunk, where they form a reticulum of lighter markings on the more intense trunk coloration. With increasing size, the band- ing decreases in boldness and finally be- comes obliterated along the anterior two- thirds of the trunk, especially in very large adults. Distribution. The range of G. fasciatus is very extensive, especially on the western coasts of New Zealand. It is known from the following localities: Awanui River (Fig. 7: 1); Cavalli Islands (2); Kerikeri River (3); Mangamuka Stream (4*); Wainui River, tributary of the Orouaiti River (5); a stream at Waiomio (6); Mero- whanara Stream, Waipoua System (7*); tributary of the Wairoa River at Tangaihi (8); tributaries of the Hakaru River at Mangawai (9); Chicken Islands (10, Stokell, 1949: 493); Little Barrier Island (11); Makarau River (12); a stream at Atkinson’s Park, Titirangi (13); Whanga- marino Stream (14); Mauku Stream (15); Waihou River (16, ? = Thames River, type locality); tributary of Kauaeranga River (17); Tairua River (18*); Waimai Stream | (19); Pikowai Stream (20); Whakatane | River (21); Lake Okataina (23*); Whana- | rua Stream (24*); Mokau River (25); Rangitikei River (28); Manawatu River | (29); Kahuterawa Stream (30); Makahika Stream, Ohau System (31*); tributary of Otaki River at Otaki Forks (33*); Man- | gaone Stream, Te Horo (34); Waikanae River (36); Whareroa Stream (38*); Horo- New ZEALAND GALAXIIDAE * McDowall 361 Figure 9. kiri Stream (40); Little Waitangi Stream (41); Tawa Stream (43); Makara Stream (45*), Hawkin’s Gully Stream, a Makara tributary (44*); Kaiwharawhara Stream (46); Hutt River (49); Catchpool Stream, Wainuiomata System (50); Days Bay Stream (53*); tributary of Lake Onoke (54); Whangamoana Stream (56*); Kapiti Island (57); Ngaruroro River (58); Pon- goroa River (59); Arapawa Island (60); D’Urville Island (61); Momorangi Bay Stream (63*); Wairau River (64); “Nel- son” (65, Stokell, 1949: 493); Kaihoka Lakes (66*); Karamea River (67); Little Wanganui River (68); Ngakawau River (70); Buller River (71); Grey River (72); Hokitika River (75); Taramakau River (76); Lake Ianthe (77*); Wanganui River (78); Lake Mapourika and Whataroa River (80*); Lake Paringa (81); Moeraki River (82); Jackson Bay Stream (83*); Awarua River (84); Waitati River (90*); Banks Peninsula (88); Stewart Island (92, Stokell, 1949: 493); Chatham Islands (94, Skrzyn- ski, 1967: 95). These localities show that G. fasciatus occurs commonly in western areas, and in the east in the North Auckland—Bay of Plenty districts. The general absence of the species from the east coast is probably the result of a combination of little suitable habitat and the fewer collections made, especially along the east coast of the North Galaxias postvectis Clarke, 178 mm L.C.F., stream at Otaki Forks, Otaki River System. Island. No localities are known to me from Southland and this is probably also due to the lack of collection. Galaxias postvectis Clarke, 1899 Figure 9 Galaxias postvectis Clarke, 1899: 88 (holotype: unknown; locality: “western slopes,’ South Island ); Stokell, 1960: 237. Galaxias fasciatus: Regan, 1905: 374 (partim). Galaxias charlottae Whitley and Phillipps, 1940: 230 (holotype: DMNZ 981, seen; locality: Queen Charlotte Sound). Galaxias argenteus: Whitley and Phillipps, 1940: 231 (partim). Diagnosis: G. postvectis differs from G. argenteus (Gmelin) (Fig. 4) and G. fasci- atus Gray (Fig. 6) in characters noted in the diagnoses of these species (pp. 353 and 355 respectively ). Description. Stout bodied, trunk rounded in section, not flattened dorsally, with no middorsal groove, rather turgid-looking. Trunk deeper than broad, depressed an- teriorly on head, which is not much flat- tened dorsally, compressed on caudal peduncle, which is short and deep, depth about equal to length. Lateral line an in- distinct midlateral furrow; accessory lateral line present. Head prominent, a_ little broader than deep. Eye large, moderately deep set, interorbital convex, very broad. Jaws well developed, lower much shorter than upper, cleft reaching to about anterior 362 third of eye, oblique. Profile of lower jaw from ventral aspect rather deep and nar- row, U-shaped. Canine teeth poorly devel- oped in jaws, or lacking; mesopterygoidal teeth moderately developed; gill rakers and pyloric caeca moderately long. Unpaired fins well developed with thick, fleshy bases, greatest fin length much greater than basal length; anal origin a little behind dorsal origin. Pectoral fin well developed, inserted moderately high, somewhat triangular in shape, with longest rays near upper margin. Pelvic fins long and expansive, inserted behind midpoint of standard length. Caudal fin rather fleshy, moderately long, emarginate, tend- ing towards truncation in very large in- dividuals, fin depth a little less than greatest body depth; caudal peduncle flanges well developed. Variation. Meristic: dorsal 9 (6), 10 (15), 11 (4); caudal 16 (25); anal 10 (1), 11 (10), 12 (14); pelvic 7 (25); pectoral 13 (2), 14 (14), 15 (9); branchiostegals 6 (2), 7 (21), 8 (2); vertebrae 59 (5), 60 (3) Gil(3) 62 (1); cil rakers 4=12" (1), 5-11 (3), 5-12 (1), 5-13 (2), 6-11 (10), 6-12 (5), 6-13 (2), 7-12 (1). Morpho- metric: see Table 1, p. 358. Coloration. Usual body color a deep brownish blue, with paler, indistinct, ir- regular marbling of slightly darker shade on the dorsal and dorsolateral trunk, and extending on to the lateral and lateroventral trunk or resolving into faint, slightly oblique bands. A purplish blotch is present above the pectoral fin base. Ventrally, the trunk is paler, more brownish in color, but nevertheless intensely pigmented. In trans- mitted light the fin bases appear a rufous color. Between the fin rays on the distal two-thirds to half of the median fins, there is bold and distinctive brown-black band- ing, which fades as the fleshiness of the fin bases develops. Size. The largest individual examined measured 250 mm, but Stokell (1960: 238) listed one at 261 mm. Individuals 180-200 mm long were relatively abundant. Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 Population differences. Insufficient large samples were available to enable meaning- ful comparison of samples from different localities. Habitat. G. postvectis is almost always collected from small, heavily bush-covered streams that are unmodified by agricultural development. These streams are usually stable, with small pools, often overhung with tree roots, or containing fallen trees. Logs in the streams often form small pools and cascades, and G. postvectis occurs in these. It is probably very sensitive to re- moval of bush cover and stream modifi- cation. Life history. The life history pattern of G. postvectis is similar to that of the other stout-bodied species. The eggs appear to be relatively small, although the only ripe individuals examined were somewhat de- hydrated, preventing satisfactory measure- ment of the eggs. In these fishes, the eggs were 1.0-1.5 mm diameter. A female 205 mm long contained about 13,000 eggs. The breeding site is undiscovered, but is likely to be close to the normal adult habitat. The larvae are apparently carried downstream to the sea, and develop and grow in the sea during the winter. The upstream mi- gration of the whitebait juveniles occurs concurrently with the other species, al- though probably towards the end of the migration period, along with G. fasciatus and G. argenteus. Adults collected from the Ohau River system in February showed gonads to be at an early stage of matu- | ration; one fish from the Mangaone Stream, | taken in late May, and another from the Waikanae River in early June had ova about ripe. Further examples collected from the Waikanae River in early September were spent or in early stages of gonad rejuvenation. Thus spawning probably oc- curs in autumn or early winter. The transparent whitebait of G. post- vectis (Fig. 44) develop a covering of | melanophores soon after entering fresh water, but do not develop pale bands like the young of G. fasciatus and G. argenteus. New ZEALAND GALAXIDAE * McDowall 363 Figure 10. Dusky brown bands develop along the sides of the trunk as the fishes grow and attain the stout-bodied form of the adult. Distribution. Like the other species with marine juveniles, G. postvectis is rather widely distributed in New Zealand. Col- lection localities are few, even though they extend along the west coast of New Zea- land from the Waipoua Kauri forests in North Auckland to the Awarua River, Big Bay, in the south. Localities in the Bay of Plenty and the Marlborough Sounds are also known. As yet, G. postvectis has not been recorded from the east coast between East Cape and Southland. This probably represents a real distributional gap for much of the Canterbury area, but G. post- vectis seems likely to be found in some Southland streams and on Banks Peninsula, where overgrown bush streams with stable, rocky beds are present. Very little col- lection has been done on the east coast of the North Island, and G. postvectis may also be found to occur there. It is presently known from the following localities: Mero- whanara Stream (Fig. 7: 7*); Waimana River (22); Ratapiko (26); Kahuterawa Stream, Manawatu System (30); Makahika Stream, Ohau System (31*); a stream at Otaki Forks, Otaki River (33*); Waikanae River and its tributaries, the Mangakotuku- tuku (35*) and Ngatiawa Streams (37*); Whareroa Stream (38); Little Waitangi Stream (42); Days Bay Stream (53); Owhiro Bay Stream (47*); Queen Char- lotte Sound (62, Whitley and Phillipps, Galaxias brevipinnis Gunther, 185 mm L.C.F., stream at Erua, Upper Wanganui River System. 1940: 230); Mokihinui River (69); Buller River (71); Waimea River (76*); Awarua River (84). Galaxias brevipinnis Gunther, 1866 Figure 10 Galaxias brevipinnis Giinther, 1866: 213 (syntypes (3): BMNH 1853.2.14.5-7, not seen; locality: New Zealand); Hutton, 1872: 59, 1896: 317, 1904: 51; Regan, 1905: 374; Waite, 1909: 586; Rendahl, 1926: 2; Phillipps 1927a: 13, 1940: 21: Stokell, 1954: 415, 1960: 236. Galaxias olidus: Hutton, 1872: 270 (not G. olidus Ginther, 1866: 209). Galaxias campbelli Sauvage, 1880: 229 (syntypes (4): MNHNP A-2381, not seen; locality: Campbell Island); Stokell, 1949: 487, 1950: 8. Galaxias lynx Hutton, 1896: 317 (holotype: CMCNZ 70 seen, paratypes: CMCNZ 71 (3) seen, AMS IB-435-6 (2) not seen; locality: Lakes Coleridge and Wakatipu); Stokell, 1949: A86. Galaxias robinsonii Clarke, 1899: 89 (holotype: unknown; locality: western slopes, South Is- land); Phillipps, 1926a: 98. Galaxias bollansi Hutton, 1901: 198 (holotype: BMNH 1905.11.30.23, not seen; locality: Auck- land Islands). Galaxias huttoni Regan, 1905: 373 (syntypes (7): 1905.11.30.27-33, seen; locality: “Lake Raini- era,” an unknown New Zealand place name); Phillipps, 1924b: 190. Galaxias castlae Whitley and Phillipps, 1940: 229 (holotype: DMNZ 2070, seen; locality: Lake Waikaremoana ). Galaxias koaro Phillipps, 1940: 35 (holotype: unknown; locality: Lakes Rotoaira and Roto- pounamu); Stokell, 1949: 487. Diagnosis. Differs from G. vulgaris Stokell (Fig. 16) in having more vertebrae (especially in the south, where the two 364 species are sympatric) and somewhat higher fin ray counts in the dorsal, anal, and pectoral fins. The gill rakers are much better developed in G. brevipinnis than in G. vulgaris, and the lower jaw recedes further in the former species. G. brevi- pinnis breeds in the autumn and has migra- tory marine or lacustrine whitebait juve- niles; in contrast, G. vulgaris breeds mostly in the spring and has no migratory juvenile. Taxonomy. Nine nominal species of the G. brevipinnis type have been described. Stokell (1949: 486-490, 1954: 413, 1960: 236) reduced the number recognized to three, viz. G. brevipinnis Ginther, G. lynx Hutton, and G. koaro Phillipps. Exami- nation of many large samples from a great variety of localities has shown that these populations represent a single, variable species. From the description of Regan (1905: 377), there appears to be little, if any, dif- ference between G. brevipinnis and G. weedoni Johnston. In G. weedoni, Regan recorded canine teeth in the jaws, the cleft of the jaw extending below the eye, long, low-placed pectoral fins, a long, slender caudal peduncle, and a blue-black blotch above the pectoral fin base. Meristic data— vertebrae 57-60, dorsal fin rays 10-11, anal rays 10-12, pectoral rays 14-15, branchios- clusion of G. weedoni in G. brevipinnis. I have seen only juveniles of G. weedoni, but their coloration is identical with that of juvenile G. brevipinnis and different from that of any other galaxiid I have seen. And they are long and slender, have a much shortened lower jaw, and have the anal fin set back below the middle of the dorsal, just as in G. brevipinnis. Accord- ingly, I think that the two species are con- specific, although formal synonymy of G. weedoni in G. brevipinnis must await ex- amination of adult specimens. Description. Elongate and slender-bodied, trunk rounded in section, somewhat flat- tened dorsally, with slight development of middorsal furrow; trunk much depressed Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 anteriorly on head, compressed behind vent; dorsal and ventral trunk profiles about parallel. Caudal peduncle moder- ately long and slender, substantially longer than deep. Lateral line a somewhat in- distinct lateral crease; accessory lateral line present. Head moderately long, much broader than deep, cheeks broadening be- low eye; jaws long, lower markedly shorter than upper and tucked behind it when mouth is closed, lips prominent. Snout short. Cleft of mouth slightly oblique, ex- tending to about middle of eye, profile of lower jaw from ventral aspect deep and rather narrow, U-shaped, but gape broad in head length. Eye rather small, deep on lateral head, interorbital convex. Jaws with prominent canines laterally; mesoptery- goidal teeth moderately well developed; gill rakers and pyloric caeca well de- veloped. Fins well developed, thick and fleshy; dorsal and anal short based but extending back well beyond bases, distal margins much rounded; anal origin usually well behind dorsal origin. Pectoral fin expansive, inserted low latero-ventrally, with lamina of fin directed ventrally; pelvic-anal inter- val rather short, fin expansive and long, inserted at about midpoint of standard length. Caudal fin truncated to slightly emarginate, fin tips somewhat rounded, depth about equal to body depth; peduncle flanges weakly to moderately developed. Variation. Meristic: dorsal 9 (30), 10 (126), 11 (54), 12 (7); caudal 15. (5), 16 (206), 17 (3), 18 (1); anal: Ox@) lO nes 11 (113), 12 (61), 13 (18); pelvic 6 (4), 7 (201), 8 (12); pectoral 13 (4), 14 (50); 15 ee 16 (48), 17 (2); branchiostegals 6 (1), 7 (54), 8 (125), 9 (36); vertebrae Smells ae te ), 54 (9), 55 (16), 56 (30), 57 (45), ee 59 ee. 60 (77), 61 o 62 cone 3 (5), 64 (2); gill rakers 9 (4), 3-10 (5 ) 4-8 (2), 4-9 (67), 4-10 ie 4-11 (9), 4-12 (1), 5-8 (1), 5-9 (4), 5-10 (13), 511 (6), 5-12 (1) cumieny Morphometric: see Table 2, p. 365. Coloration. Usually dark colored, the New ZEALAND GALAXIIDAE * McDowall TABLE 2. DENOMINATOR OF RATIO). 365 MORPHOMETRIC VARIATION IN LARGE, SLENDER SPECIES ( FIGURES GIVEN AS PERCENTAGES OF G. brevipinnis G. vulgaris Min. Mean Max. Min. Mean Max. Salen le CSE 84.8 87.0 89.3 84.0 87.0 89.3 B.D.V./S.L. 11.0 13.2 15.3 11.0 12.9 15.4 [Cae s/Ssle: 11.8 13.1 15.8 11.9 14.1 16.5 D.C.P./L.C.P. 59.5 71.9 88.5 56.2 69.4 94.3 Pre D./S.L. 68.5 73.5 81.3 67.6 71.9 75.8 Pre D./Pre A. 90.9 95.2 99.0 90.1 95.2 100.0 D.F.B./S.L. 7.8 9.4 10.8 CH 8.9 10.7 D.F.B./D.F.M. 47.6 56.8 70.9 47.6 56.5 64.5 A.F.B./S.L. 8.8 10.5 DE, 8.9 10.5 13.4 A.F.B./A.F.M. 50.0 61.7 73.0 51.0 61.4 74.1 Pre Pel./S.L. 46.3 52.2 57.8 49.3 53.0 56.5 Pec.Pel./S.L. 27.2 30.7 36.8 28.2 32.0 36.4 Pec./Pec.Pel. 43.1 55.6 68.3 36.8 49.6 61.5 Pel.An./S.L. 20.9 25.7 30.4 19.6 230 26.5 Pe]./Pel.An. 44.4 58.9 77.4 43.2 57.7 TOU. H.L./S.L. 20.7 23.6 28.7 20.5 93.4 27.0 H.D./H.L. 41.7 49.3 55.9 45.9 52.6 59.9 H.W./H.L. 56.2 67.6 78.7 57.5 67.1 78.1 Sn.L./H.L. 26.3 30.7 34.4 26.7 31.6 36.2 P.O.H.L./H.L. 48.5 50.0 70.4 43.7 51.8 58.1 Io.W./H.L. 32.5 37.7 44,1] 34.5 38.6 43.7 D.E./H.L. 13.9 17.8 25.0 14.8 17.6 22.0 L.U.J./H.L. 37.3 42.9 48.1 33.3 43.1 51.0 L.M./H.L. 32.4 38.5 43.7 Bx0}83 38.5 A472, W.G./H.L. 40.0 42.7 56.2 35.1 43.5 55.6 Fish examined 160 215 basic body color a dark gray-brown, the dorsal and lateral trunk covered with ir- regular greenish brown to gold vermicu- lations, sometimes as a_ coarse, bold reticulum or varying to dense, fine speck- ling. Belly paler, a smokey gray. A promi- nent blue-black blotch is present above and behind the pectoral fin base. Size. G. brevipinnis is one of the larger Galaxiidae, the largest examined by the writer being 220 mm long. An example described by Phillipps (1926a: 99) as G. robinsonii Clarke was 9.6 inches (240 mm) and one described by Clarke (1899: 99) 8.2 inches (213 mm) long. G. brevipinnis commonly grows to 160-185 mm. Population differences. Populations of fishes belonging to G. brevipinnis are wide- spread in lakes and rivers throughout New Zealand, and examination of populations from too few and too isolated localities, together with the variability between these populations, led earlier workers to regard these series of populations as belonging to several species. The most variable character, and the one chiefly used to justify several species, is vertebral number. Arranging the data from lake populations in north-south order, there is a cline in the number of vertebrae, with no justifiable division of the populations into two or more groups (Fig. 11). There are some irregularities in the cline, but overlap of data from adjacent populations is usually substantial. The more than 250 mile geographic break between the popu- lations in Lake Taupo and the Nelson Lakes (Rotoiti and Rotoroa) coincides with the greatest break in the cline, but this is somewhat bridged by the more southern Lake Howard population. The vertebral cline appears to be related 366 Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 50 a 0 50 b Al 50 Cc 0 50 d 0 50 e of a 0 7 50 = ;>) C f Mm = a 0 =< 50) g 0 50 h 0 50 | 0 50 j 0 52 52 54 2) S) 56 By7/ 8 ay) 60 6l 62 VERTEBRAE Figure 11. Variation in vertebral number in lacustrine Galaxias brevipinnis populations, in north-south series. a, Lake Rotorua (8 examples); b, Lake Okataina (25 examples); c, Lake Waikaremoana (27 examples); d, Lake Kiriopukae (20 examples); e, Lake Taupo (31 examples); f, Lakes Rotoroa-Rotoiti, Nelson Lakes (20 examples); g, Lake Sumner (20 examples); h, Lake Howard (25 examples); i, Lake Wanaka (20 examples); j, Lake Mahinerangi (25 examples). New ZEALAND GALAXIIDAE * McDowall 50 ho AINANOAYS 58 59 60 61 367 63 62 64 VE Ree Binge Figure 12. Variation in vertebral number in diadromous Galaxias brevipinnis populations. a, Northwestern North Island —North Auckland District (30 examples); b, Southern North Island—Wellington District (26 examples); c, Midwestern South Island (22 examples); d, Southwestern South Island—Haast District (36 examples); e, Sub-Antarctic Islands of New Zealand —Auckland and Campbell Islands (34 examples). to temperatures, the number of vertebrae being lower in fishes from the more north- ern (warmer) lakes. Growth in these lakes is likely to be more rapid during critical developmental periods. Many workers (see Lindsey, 1961, for a recent summary) have noted the tendency for closely related species to have more parts (particularly vertebrae) towards the polar end of their range. If temperature is affecting vertebral number, then it is not a valid character for use in dividing the northern and southern population series into two species. Other meristic characters did not appear to ex- hibit clinal variation and varied rather ir- regularly, but Lindsay has noted that clinal variation in one character does not neces- sarily correlate with variation in another character. Sea-going specimens were found to have about the same number of vertebrae as lacustrine examples in the more southern lakes. They exhibited variation of similar extent to that seen in other species with marine whitebait. There is slight displace- ment towards greater vertebral number with increasingly southern location of populations (Fig. 12, cf. G. argenteus, Fig. 368 22 253 24 25 26 27 28 29 PREPELVIC LENGTH ct aN DARD LENGTH % Variation in body proportions in lacustrine Gal- A, Head length/standard length ratio; B, Prepelvic length/standard length ratio; a, Figure 13. axias brevipinnis populations. Lake Rotorua (7 examples); b, Lake Okataina (5 examples); c, Lake Waikaremoana (30 examples); d, Lake Kiriopukae (33 examples); e, Lake Taupo (18 examples); f, Lakes Rotoroa- Rotoiti, Nelson Lakes (20 examples); g, Lake Howard (21 examples); h, Lake Wanaka (20 examples). 5). The more disjunct sub-Antarctic island populations showed greater distinctness, as is predictable from their extremely southern position in the range of G. brevipinnis. The most variable morphometric char- acters proved to be head length and pre- pelvic length. In lacustrine populations both head length/standard length and pre- pelvic length/standard length ratios ex- hibited north-south clinal variation, similar to that of vertebral number (Fig. 13). Variation in head length in diadromous populations showed a slight trend towards increase in length with southern displace- ment, again with the sub-Antarctic island populations standing somewhat apart from mainland populations (Fig. 14). No other characters were found that dis- tinguished any group of lake populations from any other, or the lake populations from diadromous populations. Although the inclusion of all these populations in G. brevipinnis results in a somewhat more variable species than some other New Zea- land species of Galaxias, the alternative course results in two or more morpho- Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 20 21 22 23 24 25 26 HEAD LENGTH “STANDARD LENGTH % Figure 14. Variation in head length in diadromous Gal- axias brevipinnis populations. a, Northwestern North Island —North Auckland District (9 examples); b, Southern North Island—Wellington District (23 examples); c, Western South Island (14 examples); d, Sub-Antarctic Islands of New Zea- land—Auckland and Campbell Islands (21 examples). logically similar forms that differ only in clinal characters. If recognized as distinct, such species would be much more similar to each other than any other closely related species pairs in the fauna. This is similar to the case of the Northern Hemisphere salmonids in which euryhaline diadromous populations have become |lo- cally restricted to fresh water, either volun- tarily, as Atlantic salmon, Salmo salar, have in the Manapouri-Te Anau system in New Zealand, or by the development of down- stream barriers to migration. Bigelow and Schroeder (1963: 559), for instance, re- corded Osmerus eperlanus from both coastal-estuarine situations and landlocked situations. In some cases, as in S. trutta in New Zealand, the diadromous and fresh- water forms mingle and may form a single gene pool. This may also occur in G. brevipinnis, since few of the lacustrine populations are prevented from moving downstream and interbreeding with diad- romous populations. In other instances, in the northern salmonoids and in G. brevi- pinnis, populations are found to be truly landlocked and thus completely isolated geographically. Habitat. As presently defined, G. brevi- pinnis comprises forms inhabiting a variety of types of river and lake systems. Many diadromous populations are known in low- land streams, but they may migrate a great distance inland, e.g., into tributaries of the Wanganui River on the slopes of Mount Ruapehu at altitudes approaching 3,000 feet (900 m) and streams on Mount Egmont at more than 4,000 feet (1,200 m) (G. C. Kelly, pers. comm.). Diadromous populations occur in a few lakes, e.g., Lakes Mapourika and Kaniere, the juveniles migrating from the sea into the lakes and finally living as adults in the lake tribu- taries. The possibility that both lacustrine and diadromous populations occur in such lakes as these cannot at present be ex- cluded. Most lake populations appear to be re- stricted largely to the lakes and_ their tributaries; the juveniles shoal in the lakes and the adults inhabit the lake tributaries but are mostly absent from the lakes them- selves. A fluviatile habitat is to be expected for the adults, from their obviously de- pressed, benthic form, adapted to rapid waters, and from the very definite upstream migration of the juveniles. Lake Howard has no tributaries, and the adults are found in the lake amongst rocks near the shore, but this seems unusual. In a few cases, e.g., Lake Coleridge, it has been found that the lacustrine populations also invade the rivers below the lakes. Lacustrine populations vary greatly in altitude, from about 200 feet (60 m) in Lake Alice to in excess of 2,000 feet (610 m) in Lakes Waikaremoana, Rotoiti (Nelson Lakes), and Monk. However, most of the lakes lie between 600 and 1,500 feet (180-460 m). Whether diadromous or lacustrine, the adults of G. brevipinnis are characteristi- cally captured from small, cold, rapidly flowing, stable, rocky streams which are often heavily overgrown with bush. The fishes are very secretive and live hidden amongst boulders in the most swiftly flow- ing water. In streams unmodified by clearing of the forest and agricultural de- velopment, G. brevipinnis may form large and dense populations. Formation of landlocked populations. Many of the landlocked populations of G. brevipinnis must post-date the last Pleisto- New ZEALAND GALAXIIDAE * McDowall 369 cene glaciation. Fleming (1962: 89) showed that the lower limits of the ice cap in New Zealand during the last glaciation (about 15,000 years ago) would have com- pletely engulfed many of the South Island upland lakes, in which G. Dbrevipinnis is now present. Lake Mahinerangi is even more recent; Dollimore (1962: 345) re- ported that this lake was formed artifically as a hydro lake in 1911. Most of the lakes in the South Island occur in glacial valleys, and their formation resulted from the retreat of the ice and deposit of moraine (C. A. Fleming, pers. comm.); the now- resident fish populations must have entered the lakes since that time. If the lake populations are geographically isolated by landlocking from diadromous populations, their great morphological similarity to the diadromous form is due to the recency of their isolation. The thermal lakes are also recently formed, and their populations of G. brevipinnis are of very recent derivation, almost certainly post-glacial. Life history. Little has been reported on the breeding of G. brevipinnis. The oc- currence of a spring migration of juveniles suggests that, like other species with marine juveniles, spawning occurs pre- dominantly in the autumn and early winter. Ripe and mature adults were most com- mon in samples collected from March through May, although a single fully ripe female was found in a November sample. The eggs of G. brevipinnis are of moderate size and numerous, 1.3-1.6 mm diameter in a female 188 mm long and carrying about 7,500 eggs. The spawning habitat has not been de- scribed, but localities from which ripe, strippable adults were collected were not different from usual adult habitats, sug- gesting that there may be little or no breeding migration. On hatching, the larvae are apparently washed downstream into the sea (or lake) and develop there during the winter. The slender, transparent whitebait juveniles (Fig. 42) migrate upstream primarily dur- 370 Bulletin Museum of Comparative Zoology, Vol. 139, No.7 @ diadromous Galaxias_brevipinnis lacustrine Figure 15. Distribution of Galaxias brevipinnis (numbers in figure as in text, p. 371). ing the spring; in diadromous populations the migration occurs concurrently with that of other whitebait species, in huge, mixed- species shoals (McDowall, 1965a: 290), although probably early in the migration period. The transparent fishes become _ pig- mented soon after migration. Sub-adult coloration develops initially as an overall covering of melanophores. These become concentrated along the myotomes and de- velop into dark, vertical chevron-shaped bands, which subsequently become sub- divided to form an irregular blotching pat- tern. This bold blotching may persist in the adult, or may become progressively more and more fragmented to produce the vermiculations found in most adults. Distribution. G. brevipinnis is probably the most widely distributed species of Galaxias in the New Zealand region. If it is shown to be conspecific with G. weedoni, it has trans-Tasman distribution. Coastally it is widespread, though at present, few localities are known from the east coast between East Cape and Southland. It is very common on the west coast and also in upland lakes, especially east of the main divide in the South Island, and occurs on many islands, including the very remote Chatham, Auckland, and Campbell Islands. As with the other diadromous species, inland range is somewhat limited by physical barriers in the rivers up which the fishes migrate, but this limitation affects G. brevipinnis less than other galaxiids on account of its exceptional climbing ability. Populations believed to be diadromous are known from the following localities: Mangamuka Stream (Fig. 15: 1); Mero- whanara Stream, Waipoua System (2*); Waikato River (3); Te Puna Stream (4*); Whakatane River (5); Mokau River (6); Waiwakaiho River (7); Patea River (8); tributaries of the Wanganui River near Erua (9*); Ngaruroro River (10); Rangi- tikei River (12); Pohangina River (11, Phil- lipps, 1926a: 98) and Kahuterawa Stream (13), Manawatu System; Makahika River, New ZEeaLANpd GALAXIIDAE * McDowall 37] Ohau System (14*); tributary of Otaki River at Otaki Forks (15*); Negatiawa and Mangakotukutuku Streams, Waikanae System (16*); Horokiri Stream (17); Hutt River at Kaitoke (18); Kaiwharawhara Stream (20); Day’s Bay Stream (21* ); Lyall Bay and Owhiro Bay Streams (22); tribu- tary of Lake Onoke (19); Wairau River (23); Pokororo River, Motueka System (24*); Karamea River (25); Ngakawau River (26); Buller River (27); Grey River (28); Taramakau River (29); Lake Kaniere (30); Hokitika River (31); Waitaha River (32); Wanganui River (33); Whataroa River (34); Lake Mapourika (35*); Cook River (36); Moeraki River (37); Waita River (38); Haast River (39*); Okuru and Turnbull Rivers (40); Waiatoto River (41); Arawata River (42); Jackson Bay Stream (42a*); Awarua River (43); Ethne River (44); Waitaki River (45); Clutha River (46); Chatham Islands (47, Skrzyn- ski, 1967: 95); Campbell Island (48*); Auckland Islands (49*). Lacustrine populations occur in the fol- lowing lakes: Rotorua (50*); Rotoiti (51* ); Okataina (52*); Taupo (53*); Rotopou- namu (54); Rotoaira (55); Waikaremoana (56*); Kiriopukae (57*); Rotoroa (58*); Rotoiti (59*, Nelson Lakes); Bowscale Tarn (60); Sumner (61*); Taylor (62); Pearson (64, Stokell, 1949: 486); Coleridge (65*); Howard (66); Alexandrina (67); Ohau (68, Stokell, 1955: 23); Hawea (69); Wanaka (70*); Wakatipu (71); Hawdon (72, Stokell, 1949: 486); Alice and Mar- chant (73), and Katherine (74, all Cun- ningham, 1951: 74); Te Anau (75) and Manapouri (76, both Stokell, 1959a: 255); Mahinerangi (77*); Monk (78, Riney et al., 1959: 45). The absence of lacustrine populations in the region between the southern end of the volcanic plateau and the Nelson Lakes is almost certainly attributable to the ab- sence of upland lakes there. Otherwise, the distribution of lacustrine populations of G. brevipinnis is fairly continuous from the 372 Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 Figure 16. most northern lakes of the volcanic plateau to southernmost Fiordland. Galaxias vulgaris Stokell, 1949 Figure 16 Galaxias vulgaris Stokell, 1949: 491 (holotype: CMCNZ 72, seen; paratype: DMNZ 2069, not seen; locality: Rubicon River, Springfield, Canterbury ). Galaxias anomalus Stokell, 1959b: 265 (holotype: DMNZ 2776, seen; locality: the outlet of a spring which is drained by a ditch crossing the Ophir-Omakau Road a few chains north-east of the Ophir Hotel‘). Diagnosis. Differs from G. brevipinnis Giinther (Fig. 10) in characters noted in the diagnosis of that species (p. 363). Taxonomy. Stokell (1949: 491, 1959b: 256) has recognized two moderately large and slender species from upland, eastern South Island streams, in addition to G. brevipinnis. Populations of fishes of this type are present in most of the major river basins in the east of the South Island, from the Conway River south to the Waiau (Southland), and collection localities are numerous. G. vulgaris was recorded by Stokell from Canterbury, the Waiau River (Kaikoura ) to the Rakaia, and G. anomalus from streams in Central Otago. He did not 'Omakau is almost directly north of Ophir; the Ophir-Omakau Road at the Ophir Hotel runs in a northwest-southeast direction, so that I could find no locality agreeing with Stokell’s description. This area is a part of the Mahinurikia catchment, a Clutha River tributary, and samples were col- lected in this catchment, not far from Ophir. Galaxias vulgaris Stokell, 100 mm L.C.F., Maruvia River, Buller River System. discuss differences between G. anomalus and G. vulgaris, but the chief differences between his descriptions of the two species are head length—4.2-4.8 in standard length in G. vulgaris and 5.1-5.3 in G. anomalus— and in the length of the gill rakers—*long” and “very short” respectively. The two species are indistinguishable from meristic data published by Stokell. Examination of samples from 15 localities indicated that they form a single rather variable species such that the differences between G. vul- garis and G. anomalus, as defined by Stokell, are absorbed in inter-populational differences. The holotype of G. anomalus, though recorded from a drain, is typical of G. vulgaris as found in the shingly streams of the upper Clutha River system. Description. Trunk moderately slender, belly often deepened and rounded, some- what flattened dorsally with moderate development of a middorsal furrow, trunk profiles somewhat parallel; depressed an- teriorly on head, somewhat more slender posteriorly on caudal peduncle, which is moderately long, and somewhat longer than deep. Lateral line a moderately de- veloped midlateral groove, accessory lateral line not observed. Head long and blunt, rounded, much broader than deep. Lower jaw receding a little; jaws long in head, cleft moderately oblique, reaching to about middle of eye; lower jaw profile from ventral aspect moderately broad and shal- low. Eye of moderate size, towards upper head profile, interorbital convex to flat. Jaws with moderate development — of canines laterally; mesopterygoidal teeth strong; pyloric caeca long; gill rakers weak to moderate. Fins small and fleshy; median fins short- based, but fin extending well toward caudal base, well-rounded distally; anal origin well behind dorsal origin. Pelvic fin inserted somewhat behind midpoint of standard length, fin moderately long in pelvic-anal interval, which is long. Pec- toral fin inserted low lateroventrally, fin lamina usually directed ventrally; fin of moderate length and rounded in outline with middle rays longest. Caudal fin moderately long, emarginate, lobes of fin rounded, depth about equal to body depth; caudal peduncle flanges moderately de- veloped. Coloration. Basic body color brownish to olive, trunk covered dorsally and laterally with irregular and variably dense vermicu- lations, these disappearing ventrally, some- times bold blotches, regular chevron-shaped bands, or grading to an almost uniform darkening on the dorsal and dorsolateral trunk. Size. G. vulgaris has been found up to 150 mm long, and seems to commonly reach 100-115 mm. Variation. Meristic: dorsal 7 (4), 8 (74), 9 (106), 10 (24), 11 (1); caudal 14 (15), 5 ae a Coe Ma (A NSe(2)analls (i) oy), MOS) JOD Ce) I (4) pei : a ye rig 8 eae pectoral 10 (2), 1 ae a 12 (68), 13 (70), 14 (18), to (7) 62): riviera 5 (1D). 8 (M2) a 03), eee 9 A vertebrae 49 (3), 50 Ge La oe 2 (144), 53 (99), 54 (60), a S)) go Gaal) SI (bye eat rakers 2-7 B ), 2-8 (5), 2-9 (3), 2-10 (4), 3-6 (1), 3-7 (9), 3-8 (36), 3-9 (36), 3-10 (10), 4-7 (1), 4-8 (7), 4-9 (8). Morpho- metric: see Table 2, p. 365. Population differences. As 1 define the species, G. vulgaris is rather variable, comparable in variability to G. divergens and N. apoda. These three species are also the most wide-ranging species that are New ZEALAND GALAXIIDAE * McDowall 373 confined to fresh water and which are thus less able to disperse from one river basin to another, by marine routes. Gene flow tends therefore to be limited to population exchanges by means of stream capture and perhaps occasional extraordinary flood situ- ations, when waters of two neighboring catchments become confluent temporarily. In recent years, contact between river systems has been increased by the con- struction of irrigation canals that transfer water from one catchment to another, but nothing is known of the effect of these changes on the populations of G. vulgaris. The magnitude of variability in G. vul- garis can be seen in Figures 17-19, in which various morphological characters are shown with the populations listed in ap- proximately north-south order. It is not easy to arrange the populations in an order likely to express a temperature gradient, since the nature of the watersheds in which these populations occur and their altitude in the headwaters are very variable, even in cases of closely adjacent localities; e.g., the Hinds River drains coastal hills, whereas the nearby Ashburton River pene- trates deep into the Southern Alps, which reach well over 7,000 feet in the head- waters of the river. Similarly in the Clutha River System, the Poolburn and Cardrona Rivers are at similar altitudes, but the Poolburn derives its water from the low Rough Ridge, rising to less than 3,500 feet, whereas the Cardrona drains the higher Crown Range, reaching more than 6,000 feet. In comparing populations from the various rivers and trying to relate differ- ences to water temperatures, it is important to realize that even though populations may have occurred at similar altitudes, or be in close proximity to each other, temperatures may be very different, be- cause of the origin of the water. Water temperatures will be a function of altitude, latitude, and the nature of the watershed in the hinterland of the river, and since there is no way to relate these factors and predict water temperatures, it is not pos- 7 374 Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 50 a 0| 50 b 0) 50 Cc 0 50 d = 50 e 0 50 f ) 50 g os 0) 1 50 h 7A Mm 7) 0) ee 50 i = i?) 2 0 50 J 6) 50 k 6) 50 | 0 50 m 0 50 n 0 50 Oo (6) : 49 50 5| 52 53 54 55 56 5y7/ VERTEBRAE 7 AONANDAYS 0 = 14 15 16 17 18 CAUDAL FIN RAYS Variation in caudal fin ray number in Galaxias Other populations of Galaxias vulgaris (190 River System (15 Figure 18. vulgaris. a, examples); b, Linnburn Stream, Taieri examples). sible to relate character differences to water temperatures. Vertebral number, which re- sponds to temperature differences, was found to be very variable, but this variation was irregular; most populations were found to have a range of three to five vertebrae (Fig. 17). The most distinctive population in the series examined was from the Linnburm Stream, above the Waipori Falls in the Taieri River System. It is distinctive chiefly in the number of caudal fin rays, usually a stable character with 16 rays, but reduced to 14, or occasionally 15 in the Linnburn fishes (Fig. 18). In other meristic char- acters, this population is “normal.” Body depth at vent/standard length and depth of caudal peduncle/length of peduncle ratios for the Linnburn fishes showed that they are stouter than other populations in this species (Fig. 19B). Head length/ standard length ratio is also higher than in most, but several other populations, e.g., those in the Hurunui and Cardrona Rivers (Fig. 19C) were also found to differ con- New ZEALAND GALAXUDAE * McDowall 375 siderably from the bulk of the populations studied. In some of the more variable morpho- metric characters, somewhat clinal change is exhibited along a north-south axis, al- though various and different populations were found to be aberrant and not to relate to the general trends. Head length/stan- dard length (Fig. 19C), length of caudal peduncle/standard length and pre-dorsal/ pre-anal length (Fig. 19A) ratios were all found to exhibit this tendency to some ex- tent. These differences cannot justifiably be related to temperature or any other ecological parameter, with our present understanding of the species. Coloration was found to vary greatly. As in other characters, the Linnburn popu- lation was most unusual, being much darker, the trunk patterning almost black. In this species, color pattern seems to be related to habitat. The more northern Canterbury populations occurred in swift, shingly streams in wide, open valleys, with sometimes milky water derived from snow fields. These fishes tended to have oliva- ceous coloration and a diffuse color pat- tern. Further to the south, the fishes from Central Otago were much more _ boldly colored, the vermiculations being similar in form to those in the Canterbury fishes, but contrasting much more with the ground color. These fishes were generally col- lected from small, stable, clear-flowing streams, and these color differences appear to be related to differences in lighting con- ditions in the respective habitat types— diffuse, dim but rather constant lighting in the open but somewhat murky alpine Canterbury streams, but broken lighting, interrupted also by marginal stream cover, in the clear flowing Central Otago streams. j Figure 17. Variation in vertebral number in Galaxias vulgaris, localities in north-south order. a, Conway River (31 ex- amples); b, Waiau River (27 examples); c, Maruia River, Buller River System (26 examples); d, Hurunui River (18 examples); e, Rakaia River (19 examples); f, Ashburton River (29 examples); g, Hinds River (25 examples); h, Waitaki River (28 ex- amples); i, Cardrona River, Clutha River System (35 examples); j, Shag River (33 examples); k, Poolburn River, Clutha River System (41 examples); |, Totara Stream, Taieri River System (15 examples); m, Linnburn Stream, Taieri River System (15 examples); n, Aparima River (28 examples); 0, Wilanda Downs Stream, Waiau River System (Southland) (40 examples). 376 Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 PRE-DORSAL LENGTH ope ANAL LENGTH % Ean eae TANDARDADENGT AE - Figure 19. Variation in body proportions in Galaxias vulgaris, localities in north-south order. A, Predorsal length/preanal length ratio; B, Depth of caudal peduncle/length of caudal peduncle ratio; C, Head length/standard length ratio; a, Con- way River (15 examples); b, Waiau River (15 examples); c, Maruia River, Buller River System (15 examples); d, Hurunui River (15 examples); e, Rakaia River (13 examples); f, Ashburton River (30 examples); g, Hinds River (15 examples); h, Cardrona River, Clutha River System (15 examples); i, Shag River (15 examples); j, Poolburn River, Clutha River System (15 examples); k, Linnburn Stream, Taieri River System (14 examples); |, Aparima River (15 examples); m, Wilanda Downs Stream, Waiau River System (Southland) (15 examples). This is an interesting species, in which the study of inter-populational variation appears to be potentially profitable and to warrant further investigation. Since the populations occur in widely separated river basins, it is probable that at least some of the variation is simply a product of mosaic evolution. Further examination of the Linn- burn population may, however, show that subspecific or specific distinctness has been attained. Habitat. G. vulgaris lives normally in water type similar to that in which G. brevipinnis, G. prognathus, and particularly G. paucispondylus are found. In Canter- bury, most of the cold upland rivers thread their way back and forth over broad, un- stable flood plains. G. vulgaris occurs commonly in these rivers and their tribu- taries, mostly in the very fast and broken water. Further south, in Central Otago and Southland, the terrain is more stable, and the upland rivers are usually narrower and more strictly confined to their river courses. In these rivers and their tribu- taries, G. vulgaris also occurs in the rapid and broken water. It has generally not been found in streams entering lakes, al- though the Hurunui River, above Lake Sumner, is an exception. G. vulgaris is a typical, highly secretive galaxiid, and is found in the interstices of boulder rapids; it sometimes hides in marginal cover, where this is present. The claim has been made that G. anomalus can withstand droughts, like the mudfishes (Neochanna species), but I know of nothing to substantiate it, and I think it is doubtful that a species usually found in cold, swiftly flowing streams can aestivate. Life history. G. vulgaris is restricted throughout its life to flowing fresh waters, it has no whitebait juvenile, and probably has no migration of any magnitude. Larvae have often been collected with the adults, suggesting that spawning occurs in or near the customary adult habitat. Samples collected in December and January invariably contained only spent New ZEALAND GALAxUDAR * McDowall BT ry or rejuvenating fishes; some collected in April and May were showing considerable advance towards gonad maturity, while samples collected in October were mostly freshly spent, although a few individuals were ripe. Recently hatched larvae, 10-15 mm long, were collected in December. These data all suggest that spawning oc- curs in the early and middle spring, agree- ing partly with Stokell’s (1955: 25) ob- servation of spawning in winter and early spring. The larvae may be found swimming in small groups in backwaters and slack water at the edges of the streams. The eggs are moderately large, measur- ing about 1.5 mm diameter when ripe, and relatively few in number. The largest ripe female examined was 83 mm long and contained 865 eggs. Distribution. G. vulgaris occurs only in the South Island, chiefly on the east of the Southern Alps and the Kaikoura Ranges, but it has extended its range over the alps into the upper Buller River System. It is known from the following localities: Upper Buller River System near Maruia Springs (Fig. 20: 1*); Conway River (2*); Mason, Wandle, and Leeds Rivers, Waiau River System (3*); Hurunui River above Lake Sumner (4*); Cass River (5), Porter River (7), and Rubicon River (8) in the Wai- makariri River System (Stokell, 1949: 491); Ashley River and Selwyn River (Stokell, 1949: 491); Wilberforce River (6) and Harper and Avoca Rivers (6a*), Rakaia River System; North Branch (9*) and Taylors Stream (10*), Ashburton River System; Hinds River (12*); Rangitata River at Mesopotamia (11*); Orari River at Peel Forest (13); Haehaemoana River, Opihi River System (14); tributary of Lake Pukaki (Stokell, 1955: 25); Waitaki River at Otematata (15*); Shag River (17*); Swinburn (18*), Totara (20*), and Linn- bum (21*) Streams, Taieri River System; Cardrona (16*) and Poolburn (19*) Streams, Clutha River System; Mataura River (22); Aparima River (23*); Orawia 378 Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 @ Galaxias vulgaris Figure 20. Distribution of Galaxias vulgaris (numbers in figure as in text, p. 377). River and Waiau River at Wilanda Downs (245), Galaxias maculatus (Jenyns, 1842) Figure 21 Mesites maculatus Jenyns, 1842: 119 (holotype: BMNH 1917.7.14.6, not seen; paratypes (3): BMNH_ 1917.7.14.7-9, not seen; locality: fresh water brook, Hardy Peninsula, Tierra del Fuego). Mesites attenuatus Jenyns, 1842: 121 (holotype: BMNH_ 1917.7.14.11, not seen; locality: fresh water, Bay of Islands, New Zealand); Richard- son, 1843: 26. Galaxias attenuatus; Valenciennes, In Cuvier and New ZEALAND GALAXIIDAE * McDowall 379 Galaxias maculatus (Jenyns), 94 mm L.C.F., Figure 21. Valenciennes, 1846: 348; Ginther, 1866: 210; Hutton, 1872: 60, 1896: 317; Clarke, 1899: 78; Hutton, 1904: 51; McKenzie, D. H., 1904: 122; Regan, 1905: 368; Phillipps, 1919: 211, 1924a: 17, 1926b: 292; 1927a: 13; Hope, 1928: 389; Stokell, 1949: 479; McDowall, 1967b, 1968b. Galaxias forsteri: Kner, 1865: 320 (not G. forsteri Valenciennes, In Cuvier and _ Valenciennes, 1846: 351). Austrocobitis attenuatus: Ogilby, 1899: 158. Galaxias maculatus attenuatus: Stokell, 1966: 78. Diagnosis. Differs from G. usitatus Mc- Dowall (Fig. 23) in having more vertebrae and pelvic fin rays, shorter head with smaller eye, longer pelvic-anal interval. and the presence of a marine whitebait stage. Differs from G. gracilis McDowall (Fig. 24) in having amen higher vertebral count, more dorsal fin rays, more branchios- tegals and many fewer gill rakers; also in more anterior pelvic fin insertion, shorter pelvic-anal interval, much shorter head, broader interorbital, and smaller eye. Description. Slender bodied, trunk rounded, somewhat compressed and deeper than broad, much more slender on head and on caudal peduncle, which is short and very slender, depth much less than length. Lateral line a well-developed mid-lateral furrow; accessory lateral line not evident. Head small and slender, short; eye large, moderately deep in head, interorbital con- vex and moderately broad in head width, but head itself narrow; jaws short, about equal in length, cleft reaching to about anterior eye margin, slightly oblique, gape very narrow; profile of lower jaw from Ship Creek, South Westland. ventral aspect deep and rather narrow, U- shaped. Canine teeth lacking from jaws; mesopterygoidal teeth well developed; gill rakers well developed; pyloric caeca rudi- mentary or absent. Median fins rather small, membranous. Dorsal fin origin well back, fin short based, greatest length not much greater than basal length, distal margin of fin somewhat rounded. Anal origin more or less below dorsal origin, greatest fin length very little greater than basal length, distal margin of fin straight or concave, inclined to trunk axis, anterior rays much the longest. Pec- toral fin short in rather long pectoral-pelvic interval, fin inserted high laterally. Pelvic fins very short in long pelvic-anal interval, insertion somewhat behind midpoint of standard length. Caudal fin short, forked, depth sub-equal to body depth; caudal peduncle flanges weakly developed. age io dorsally Ont(7)a lO (58) 1iln( 59) 2 (4 as) -weaudallelis @l : 6 (80); mal WES (WO), Wy (sil, (5 p ils Ll ES Sas Oar Baa ectoral 11 (4), 12 (28), 13 ae 14 ee 15 (3); branchiostegals 5 (9), 6 (93), 7 (73), 8 (5); vertebrae 59 Me 60 (13), 61 (44), 62 (45), 63 (24), 64 (5); gill rakers Sa10062) ole (3) 4=10n10)5 Zaina); MAD (BS), SIO (29). soll (OD sae (al), Morphometric: see Table 3, p. 380. Coloration. Trunk pale creamish white, covered with greenish gray mottling dor- sally and laterally, mottling failing latero- ventrally and ventrally and varying from 380 Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 TABLE 3. MORPHOMETRIC VARIATION IN SHOALING SPECIES (FIGURES GIVEN AS PERCENTAGES OF DE- NOMINATOR OF RATIO). G. maculatus G. usitatus G. gracilis Min. Mean Max. Min. Mean Max. Min. Mean Max S.1../L..G.F. 87.7 90.9 91.7 88.5 89.3 90.1 Si 89S nolan B.D.V./S.L. N@s33- ML IRAE) 99 115 13.0 Os 1, IL L.C.P./S.L. 8.8 105 12.3 8.6 99 10.9 10.4 11.8 13.4 D.C.P./L.C.P. 49.0 59.5 68.5 59D 667 — 76:3 48.1 544 649 Pre D./S.L. 74.6 76.9 78.7 TAN 16:3 - 78.7 13.0 SA Gee oale Pre D./Pre A. 96.2 99.0 102.0 96.2 98.0 101.0 97.1 100.0 103.1 D.F.B./S.L. 7.8 9.1 11.3 8.5 10.1 12.5 UP 7.9 9.2 D.F.B./D.F.M. 58.5 70.4 82.0 61.0 68.5 77.5 58.5 65.8 72.5 A.F.B./S.L. 11.8 13.7 15.6 12.2 13.6 15.1 13.0 14.2 15.5 A.F.B./A.F.M. 78.1 84.0 90.9 (oe oles ond 75.2 848 90.1 Pre Pel./S.L. 48.8 50.9 52.9 495 52.9 55.6 53.1 54.6 56.2 Pec.Pel./S.L. 29:9 Se SAR Qe) S0lS) easel 98.4 31.4 35.1 Pec./Pec.Pel. 30:0 35:8) 4272, 35.6 39.0 43.2 32.3 369 42.4 Pel.An./S.L. DAA ii eo Oe2, DEQ) Dba Dio 20.0 21.6 24.0 Pel./Pel.An. 29.4 36.8 44.0 37.1 43.4 49.1 40.9 45.6 50.0 AL. / Sale: 18.5 20.0 21.6 220 23352 246 298 \QAR3 2515 H.D./H.L. 43.7 48.8 52.9 42.9 46.3 50.0 42.9 48.1 52.4 H.W./H.L. 45.1 516 56.2 43.7 48.8 53.5 46.7 49.3 53.5 Sn.L./H.L. 26D (28a) a0 25.9 28.7 32.6 25.9) 21- Oe oles P.O.H.L./H.L. ASI) 18:20) OOD 490 50.3 53.8 50.0 54.4 59.2 Io.W./H.L. 34.3 37.6 41.3 30.8 33.9 37.5 28.6 310 33.2 D.E./H.L. I) PAIS ABET 21.8 243 26.9 PD) GYAN | ONT B L.U.J./H.L. Weal Bilal SF Silo BH Bi 28.0 308 34.1 L.M./H.L. 25.6 28.7 32.5 28.6 32.9 36.9 26.1 28.5 33.3 W.G./H.L. 25.0 29.0 32.9 26.3 28.9 34.5 25.0 27.1 30.4 Fish examined 40 20 30 fine speckling to bold, irregular blotches. The belly, opercular covers, and eyes are silvery. The head is usually darker than the rest of the trunk, the fins almost colorless, except for a few melanophores along the fin rays and at the base of the caudal fin. Size. G. maculatus is known to grow to 169 mm, and commonly reaches 100-110 mm. Population differences. Examination of SO fishes from the Waikanae River and 51 from the Awarua River, localities about 400 miles apart, revealed no meristic differences between these populations. Samples from more distant Australian and South American populations have shown that there are clinal differences in some characters (McDowall, 1967b). Habitat. G. maculatus is found in di- verse habitat types, but appears to be most successful in small, stable, coastal and low- land streams, chiefly in gently flowing water, usually above tidal influence. Large shoals are often found in back-waters and similar areas where the water is slack. It is abundant in the darkly tannin-stained waters of bush and flax swamps and streams on the west coast of the South Island. However, G. maculatus inhabits a wide variety of water types, including quite swift, gravelly streams, where the shoals appear to break up, and the fish are usually found singly, or in twos and threes in cover at the stream margins. Upstream range of G. maculatus is usually very limited. Com- pared with other galaxiids, it has very poor climbing ability, and it is limited to streams below falls that other whitebait species are able to surmount. G. maculatus is probably the most prolific, open-living, and com- monly encountered species of Galaxias. SS SS ane, ce) ak : \ \ AN a S\ yo" SA \ Australia S 4) Distribution of Galaxias maculatus. Figure 22. Life history. Because of its economic importance in the New Zealand whitebait fishery, the life history of G. maculatus has been extensively studied (Hayes, in Hef- ford, 193la, b, 1932; McKenzie, M. K., n. d.; Benzie, 1961; Burnet, 1965; McDowall, 196Sb ). G. maculatus is peculiar in that it breeds amongst grasses on estuarine flats and that breeding occurs in synchrony with the high New ZEALAND GALAXxIIDAE * McDowall ne 351 cl zs e0\o" 40 ® Chatham Isiands Falkland spring tides. The ripe fish migrate down- stream into estuaries in large shoals and swim out over tidal flats covered by the exceptional tides at the full and new moons. The eggs are deposited amongst the bases of terrestrial plants, mostly grasses and sedges, and are left exposed when the tide recedes. They hatch at subsequent spring tide cycles and_ the larvae are washed out into the sea. The 382 Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 Figure 23. eggs will tolerate and hatch in water of salinities varying between pure fresh and pure sea water. Spawning is reported from September to June, but occurs mostly in the autumn, from March to May. The transparent whitebait juveniles (Fig. 40) migrate into fresh water during all months of the year, but most commonly in the spring, from August to November. Size at migration is mostly between 45 and 55 mm. The marine life lasts over the winter period between the autumn spawning and the subsequent spring migration. Most adults reach maturity at one year and are thought usually to breed only once. In exceptional cases, maturation is delayed for a further one or even two years, and some fish may survive an initial spawning and perhaps spawn again. The eggs are small, measuring about 1 mm diameter when ripe, but size is vari- able. Fecundity was found to vary from 137 to 13,000 in fishes between 47 and 135 mm long (McDowall, 1968b). Compared with other galaxiids examined, egg number is high. Distribution. G. maculatus is known from all over New Zealand, in coastal situ- ations, and also on the Chatham Islands. It occurs in southeastern Australia, Tas- mania, Lord Howe Island, Southern Chile, Patagonia, and the Falkland Islands. With Geotria australis, the southern lamprey, which has a similar range, it is probably the most widely dispersed species of fresh- water fish known (Fig. 22). Galaxias usitatus McDowall, 72 mm L.C.F., Lake Waiparera, North Auckland. Galaxias usitatus McDowall, 1967 Figure 23 Galaxias usitatus McDowall, 1967a: 7 (holotype: NZMD, seen; paratypes: DMNZ 4,500, seen; MCZ 45054, seen; USNM 201223, seen; lo- cality: Lake Waiparera, near Kaitaia, North Auckland). Diagnosis. Differs from G. maculatus (Jenyns) (Fig. 21) in characters noted in the diagnosis of that species (p. 379); differs from G. gracilis McDowall (Fig. 24) in having more vertebrae, more branchiostegals, fewer gill rakers, a shorter pelvic-anal interval, longer dorsal fin base, shorter caudal peduncle, the presence of serrations on the free margin of the oper- culum and the greater size attained. Description. Trunk cylindrical, slender, somewhat depressed on _ head, laterally compressed on caudal peduncle, which is slender and short. Lateral line an indistinct midlateral furrow; accessory lateral line not evident. Head long and slender, about as broad as deep. Eye large, close to upper head profile, interorbital more or less flat, broad relative to head width. Jaws equal and prominent, cleft slightly oblique, reach- ing to about anterior eye margin, gape rather narrow; profile of lower jaw from | ventral aspect a deep and rather narrow | U. Jaws without canines; mesopterygoidal teeth well developed; pyloric caeca lack- ing; gill rakers well developed; free margin | of opercular membrane finely serrate. Fins membranous and short, except anal, | which is long based. Dorsal fin origin well | New ZEALAND GALAXHDAE * McDowall 383 Figure 24. back, fin with moderately short base, maximum fin length somewhat longer than fin base, distal margin of fin slightly rounded. Anal fin origin more or less below dorsal origin; fin long based, but greatest fin length little greater than basal length, distal margin of fin about straight, inclined to trunk axis. Pelvic fin inserted at about midpoint of standard length, pelvic-anal interval short, pelvic fin mode- rate in pelvic-anal interval, short. Pectoral fin short, inserted rather high laterally. Caudal fin very short, forked, fin depth about equal to body depth; peduncle flanges poorly developed. Variation. Meristic: dorsal 9 (2), 10 (22) reli Gls) 210 (2)- caudal 14) 15 (CORG(G4) iia) lise ()= anal 12503); SCS) a4 (6) > (38). 167Cl):3 pelvic Gm(a0)). @ (LL)= pectoral Wil (1), 12 (5); 13 (21), 14 (12), 15 (2); branchiostegals Del) 6 (25), 7 (15)s vertebrae 54 (2), Dom (a) oon LO eroy (12). 58) (9), 59) (3,)E gill rakers 4-11 (2), 4-12 (6), 4-13 (1), d-ll (5), 5-12 (6). Morphometric: see Table 3, p. 380. Coloration. Trunk a dusky gray-brown and covered with irregular dark blotches, very similar to G. maculatus. In fresh specimens, the lateroventral and ventral abdomen are silvery, but in preserved ma- terial, colorless. Size. G. usitatus is known to grow to 681.5 mm. Examples from a sample _ col- lected from the type locality were com- monly 60-70 mm long. Galaxias gracilis McDowall, 59 mm L.C.F., Upper Lake Rototuna, North Auckland. Population differences. Only one popu- lation of G. usitatus is presently known. Habitat. G. usitatus was collected along the shores of Lake Waiparera, mostly amongst moderately open sedges growing in a few inches to a foot of water. It was also collected in a small, boggy, overgrown tributary that drains partly cleared manuka (Leptospermum sp.) scrublands. It is mostly a midwater swimming and shoaling species. Life history. Nothing is known of the breeding of G. usitatus, except that it must occur either in the lake or in the small tributary stream running into the lake. Population size in the tributary stream in March, when specimens were collected, was extremely low, so that if spawning does occur there, a definite spawning migration of some type must take place. Examination of the gonads showed that in March, the fish are approaching maturity, although breeding appeared to be some time away. The gonads were too immature for useful measurements of eggs or de- termination of egg number. The eggs appeared to be quite numerous, compa- rable in number with those of G. maculatus of similar size. From the stage of maturity, breeding appears likely to occur in late autumn or early winter. Distribution. G. usitatus is presently known only from Lake Waiparera, the type locality, and a small stream entering the lake from the south (Fig. 25). 384 Galaxias gracilis McDowall, 1967 Figure 24 Galaxias gracilis McDowall, 1967a: 6 (holotype: NZMD, seen; paratypes: DMNZ 4499, seen; MCZ 45053, seen; USNM 201224, seen; locality: Upper Lake Rototuna, Kaipara Harbour, North Auckland ). Diagnosis. Differs from G. maculatus (Jenyns) (Fig. 21) and G. usitatus Mc- Dowall (Fig. 23) in characters discussed in the diagnoses of these species (pp. 379 and 382 respectively). Description. Trunk cylindrical, slender, somewhat depressed on _ head, laterally compressed on caudal peduncle, somewhat deeper than broad. Caudal peduncle short and slender. Lateral line an_ indistinct lateral furrow; accessory lateral line not evident. Eye large; at upper head profile, interorbital flat, very narrow. Lower jaw protruding a little or equal in length to upper; lips thin, cleft of mouth slightly oblique, extending to about anterior eye margin; gape very narrow, lower jaw from ventral aspect deep and narrow, U-shaped. Jaws without canines; mesopterygoidal teeth moderately developed; gill rakers long; pyloric caeca lacking. Fins membranous and short, except anal, which is rather long based; anal origin more or less below dorsal origin. Distal margin of dorsal fin rounded to straight, anterior rays longest; margin of anal straight, anterior rays longest, maximum fin length little greater than basal length. Pelvic fins inserted rather posteriorly, pelvic-anal interval short, fin relatively short in pelvic-anal interval. Pectoral fin short, inserted high laterally. Caudal fin short, forked, depth about equal to body depth; caudal peduncle flanges poorly de- veloped. Variation. aa dorsal, “7 (2), 8 (23), 9 (22), 10 (3); caudal 15 (4), 16 (42), 17 (4); anal 12 (2), 13 (7 ) 14 (29), LoS) 16s) apelvicnG (ey 7ac4O)e (3); pectoral 12 (8), 13 Cx he (1 : branchiostegals 4 (2), SOAP es Tore (OAl)) (2); vertebrae 47 (1), 48 (16), 49 oa Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 50 (9); gill rakers 5-13 (2), 5-14 (2), 5-15 (1), 6-13 (1), 6-14 (12), 6-15 (7), 6-16 (3), 617 (1), 7-16 (1). Morpho- metric: see Table 3, p. 380. Coloration. Trunk densely covered with large melanophores that intensify on the head and dorsum of the trunk, failing lateroventrally and ventrally. Fresh speci- mens are silvery in these latter areas. Size. G. gracilis is known to grow only to 62.5 mm. Many examples in the very large sample collected were between 45 and 55 mm long, but few were larger. Population differences. Only one popu- lation of G. gracilis is known. Habitat. G. gracilis has been collected only from a small, coastal dune lake. Large numbers were collected from shallow water near the lake shore. Life history. The entire life history of G. gracilis occurs in fresh water, since the locality from where it is known is land- locked. Ripe males were present in the sample, collected in March, but no ripe or mature females. Breeding thus appears likely to occur some time in the autumn. In a female 47 mm long and approaching maturity, there were 604 eggs, 0.6-0.8 mm in diameter. Distribution. G. gracilis is presently known only from the type locality, Upper Lake Rototuna (Fig. 25). Galaxias divergens Stokell, 1959 Figure 26 Galaxias divergens Stokell, 1959b: 266 (holotype: DMNZ 2777, seen; locality: a rapid shingly stream flowing into the Maruia River about a mile west of the hot springs). Diagnosis. Very similar in form to G. paucispondylus Stokell (Fig. 29) but dif- fers in its stouter build, slightly shorter caudal peduncle, and more posterior pelvic fin insertion. The chief differences are the very reduced gill rakers and the lower number of fin rays in the pelvic and caudal fins. Differs from G. prognathus Stokell (Fig. 31) in its stouter build, the jaws being sub-equal with the upper jaw longer New ZEALAND GALAXIIDAE * McDowall 385 -_ > Galaxias usitatus N => G. gracilis * Gdivergens * 1 (+) G. paucispondylus e@ G. prognathus ¥D Figure 25. Distribution of Galaxias usitatus, G. gracilis, G. divergens, G. paucispondylus, and G. prognathus (numbers in figure as in text on pp. 390, 393, and 394). 386 Figure 26. in head length, and usually fewer caudal and pelvic fin rays. The difference in jaw form immediately separates the two species. Description. Trunk rounded to squarish in section, dorsally flattened with moderate development of a middorsal furrow, trunk profiles somewhat parallel, tapering an- teriorly to a small head and becoming slender posteriorly. Caudal peduncle very long and generally slender, depth about half length. Lateral line an inconspicuous midlateral groove; accessory lateral line not observed. Head short, broader than deep, head depth conspicuously less than body depth. Eye small, towards upper head profile, and interorbital flat. Jaws equal or lower a little shorter, short in head length; cleft moderately oblique and reach- ing to about anterior eye margin; jaw pro- file from ventral aspect broad and shallow, somewhat flattened anteriorly. Jaws with- out canines, mesopterygoidal teeth weak; pyloric caeca lacking; gill rakers reduced to indefinite stubs. Dorsal and anal fins small, somewhat fleshy at bases but membranous distally; short based and extending little beyond bases, distal margins rounded. Dorsal origin well forwards due to length of caudal peduncle, anal origin below or a little be- hind dorsal origin. Pelvic fin inserted at about midpoint of standard length; pelvic- anal interval moderately long, fin very short in interval. Pectoral fin inserted moderately high laterally; short in pectoral- pelvic interval, which is also rather short; distal margin of pectoral fin rounded, with middle rays longest. Caudal fin moderately Bulletin Museum of Comparative Zoologt SViole 39) Nowe Galaxias divergens Stokell, 68 mm L.C.F., Mountain Camp Stream, Pelorus River System. long, emarginate, depth about equal to body depth; flanges of peduncle weak to moderate. Variation. Meristic: dorsal 7 (9), 8 (58), 9 (34), 10 (10); caudal 13 (1), 14 (3 : 15 (83), 16°(6); anal 8i@3))r . 68), 1 ( (36), 11 (8); pelvic 6 (119), 7 (5); torall 9s Gio). Om (59))yaial el a branchiostegals 6 (11), 7 (64), § (4); vertebrae 47 (2), 48 (14), 49 Ts, 50 (69), 51 (81), 52 (44) ssn) eR ell rakers—these are so reduced and irregular in development that a satisfactory count was impossible. Morphometric: see Table 4, p. 387. Coloration. Basic body color usually a striking creamy-white, covered dorsally and laterally with irregular, darker, greenish brown to gray vermiculations, which fail lateroventrally and ventrally. Head usually dark, pigmentation extending down on to cheeks. Less commonly the coloration con- sists of more diffuse, dark speckling. Size. G. divergens is a small species, which is known to reach only 87 mm. Examples from the type locality were com- monly over 70 mm, but those from other localities were generally smaller, usually 60-70 mm long. Population differences. G. divergens has cS iS v a moderately broad range in the North | Island and the northwest of the South | Island, and considerable differences be- | tween populations were observed. Stokell (1959b: 266) described G. divergens from | a locality near Maruia Springs, noting that | “from shingly streams in the Well- | a form ington Province agrees with G. divergens New ZEALAND GALAXIIDAE * McDowall 387 TABLE 4. MORPHOMETRIC VARIATION IN SLENDER, ALPINE SPECIES ( FIGURES GIVEN AS PERCENTAGES OF DENOMINATOR OF RATIO). G. divergens Min. Mean Max. S.L./L.C.F. 85.5 87.7 90.9 B.D.V./S.L. OF leo 4a: (2 C.P:/S.E. 14.7 17.0 18.9 D.C.P./L.C.P. 40.5 52.6 66.7 Pre D./S.L. 67.1 69.9 74.6 Pre D./Pre A. 94.3 98.0 102.0 D.F.B./S.L. UO 93 12.4 D.F.B./D.F.M. 51.8 61.7 76.9 A.F.B./S.L. 9.1 10.4 13.5 A.F.B./A.F.M. 59.5 68.5 78.7 Pre Pel./S.L. 47.6 50. 52.9 Pec.Pel./S.L. YO) BRB Bro) Pec./Pec.Pel. 296 39.6 48.4 Pel.An./S.L. 19.3 22.4 25.8 Pel./Pel.An. 34.4 45.1 542 H.L./S.L. 16.9 19.7 21.9 H.D./H.L. 410 49.3 54.6 H.W./H.L. 50.0 59.5 70.9 Sn.L./H.L. D6 298) B33 P.O.H.L./H.L. 50.0 54.6 65.4 lo.W./H.L. 32.0 36.9 45.9 D.E./H.L. 16.0 182 21.3 L.U.J./H.L. 28.6 35.3 40.0 L.M./H.L. 26.2 31.2 36.0 W.G./H.L. 30.4 35.8 49.3 Fish examined 105 G. paucispondylus G. prognathus Min. Mean Max. Min. Mean Max. 84.8 86.2 88.5 86.2 88.5 90.1 10.0 11.1 13.6 8.3 9.5 11.9 16.8 19.0 21.4 15.3 16.7 18.0 41.3 45.7 51.8 35.7 42.6 48.8 65.8 69.0 72.5 67.1 70.9 73.5 94.3 97.1 101.0 93:0. Ob 2p O80 7.6 9.1 Teal UP 8.5 9.7 54.4 62.5 80.0 Died) 6429) ui 8.4 10.1 11.9 8.6 9.9 11.3 55.6 64.5 72.5 61.0 70.4 78.7 43.9 47.4 50.5 48.1 50.8 53.2 26s e294 oleG 30.4 33.4 35.0 37.9 46.6 54.4 eyo; ox ZRXIL 19.8 23.5 26.3 20.5 22.9 246 37.5 45.6 55.6 33.8 39.2 43.8 Wyfadl 19.1 20.8 Wee 18.5 19.8 39.2 48.1 59.5 40.0 448 52.6 57.1 61.7 63.3 50.5 546 63.3 26.5 294 34.6 29.08 28a. ole 51.0 56.8 61.4 48.8 51.0 544 QS 33.49 40:5 31.2 33.4 40.0 12.2 15.0 18.7 12.8 14.1 16.7 30.8 34.6 38.8 PKG). PAS) JB) BRBYB} 28.1 SA -Bioyll 30.6 32.9 35.6 30.6 35.1 40.3 Se3) Be) Bye 40 23 in the number of ventral rays and the ab- sence of pyloric caeca but has a head in length ratio of less than five, and a definitely curved mouth.” He expressed the view that “the characters concerned are rather more important than come within the author’s conception of subspecific distinc- tion.” Populations of fishes like these are now known to be quite widespread. They do not seem to fall into more than one taxon and certainly form an assemblage that stands apart from the other species of the upland-alpine, slender-species group. The differences between the populations are decidedly less than differences between these G. divergens-type populations and other species in the species group. Ac- cordingly, all these populations are in- cluded in a somewhat variable species, G. divergens Stokell. Meristic characters were found to be similar in all populations and all characters examined. Maxima and minima in dorsal, anal, caudal, and pelvic fin ray counts in no case differed by more than one element between populations, and pectoral ray, branchiostegal, and vertebral counts by no more than two elements. Overlap between populations was thus found to be broad. Vertebral number and pectoral ray number showed slight general increase along a north-south axis (Fig. 27), although in both characters, one population or another was found to interrupt the continuity of the variation. The body proportions were found to exhibit greater variation. Fishes from the Mangatarere population were con- siderably stouter in build than other popu- lations, this being evident in both the depth caudal peduncle/length of peduncle and body depth at vent/standard length ratios 388 fs AD Nan Os ta Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 50 47 48 49 50 ont 52 52 VERTEBRAE Jariation in vertebral number in Galaxias divergens, localities in north-south order. a, Tukuhou Stream at ga, Rangitaiki River System (54 examples); b, Hinaki Stream, Ruamahanga River System (74 examples); c, Mangako- eam Naikanae River System (54 examples); d, Hutt River (57 examples); e, Catchpool Stream, W ainuiomata 23 examples); f, Mountain Camp Stream, Pelorus River System (54 examples); g, Stream at Golden Downs, System (68 examples); h, Maruia River, Buller River System (21 examples). New ZEALAND GALAXUDAE * McDowall 17 18 19 20 21 22 HEAD LEN a STH oT ANDARD LENGTH % 40 50 60 70 DEPTH CAUDAL PEDUNCLE, - tu of pEDUNCLE 7 9 10 nl 12 13 14 BODY DEPTH AT VENT “STANDARD LENGTH % Figure 28. Variation in body proportions in Galaxias divergens, localities in north-south order. A, Head length/ standard length ratio; B, Depth of caudal peduncle/length of caudal peduncle ratio; C, Body depth at vent/standard length ratio; a, Tukuhou Stream at Horomanga, Rangitaiki River System (20 examples); b, Hinaki Stream, Ruamahanga River System (13 examples); c, Mangakotukutuku Stream, Waikanae River System (9 examples); d, Hutt River (14 ex- amples); e, Catchpool Stream, Wainuiomata River System (10 examples); f, Mountain Camp Stream, Pelorus River System (10 examples); g, Golden Downs Stream, Motueka River System (10 examples); h, Maruia River, Buller River System (17 examples). (Fig. 28B, C). The Catchpool and Manga- kotukutuku populations were also some- what more stout than the others. Head length is less variable, except that the sample from the type locality, at Maruia, stands distinctly apart from all other popu- lations (Fig. 28A). There is no obvious basis for this variability, and despite its extent, there appears to be only one taxon here, at the species level. Until the range of this species is thoroughly understood, it is not appropriate to name sub-species. 389 It seems likely that many more localities for G. divergens within the known range, and particularly between those in the southern Wairarapa and the disjunct popu- lation at Horomanga, will be discovered. Data from these may make the inter-popu- lation variation more comprehensible. Habitat. G. divergens is usually captured from small, moderately swiftly flowing headwater streams, which have gravel or boulder beds. Streams where G. divergens is abundant are usually stable, and often occur in narrow, steep gullies with little or no flood plain. The characteristic water type is turbulent but not broken; the fish characteristically live in the interstices of the stream substrate, and are almost al- ways hidden. Life history. G. divergens is restricted to fresh water and has been found only in flowing water, although a population is known in a tributary of Lake Rotoiti. Fishes collected from the Maruia, Golden Downs, Mountain Camp, and Catchpool populations in May were near maturity; others, taken in the Mangakotukutuku in September, the Catchpool and Hutt locali- ties in November, and the Horomanga in December appeared to be mature. Adults in a large sample from the Mangatarere, collected in late February, were found to be spent, and the sample contained many small juveniles, mostly between 20 and 25 mm long. Recently hatched juveniles 10-12 mm long were collected from the Hutt River at Kaitoke in early February. These data suggest a rather extended spawning period in the spring and summer. The eggs of G. divergens are of moderate size, 1.3-1.6 mm diameter, and very few in number, a female of 68 mm carrying 225 eggs. The breeding site is unknown to me. A search was made for the site when fishes were collected during December, when there were ripe fish in the population, and it was not discovered (G. A. Eldon, pers. comm.). Mature fish from the Horomanga locality exhibited a peculiar sex ratio, all the fish being females. The failure to find 390 Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 Figure 29. Galaxias paucispondylus Stokell, 70 mm L.C.F., Wilberforce River, Rakaia River System. the spawning site may be related to a spawning migration in association with breeding, perhaps explaining the absence of males from the Horomanga sample. Groups of juveniles may be found swim- ming freely in the still waters amongst rock piles at the edges of the pools and other places where there is very little flow. Distribution. G. divergens is presently known over a broad area of the southern North Island and the northwest of the South Island. A single locality in the Bay of Plenty is known. G. divergens has been recorded from the following localities: Tukuhou Stream at Horomanga, Rangi- taiki System (Fig. 25: 1*); a tributary of the Mangahao River at Mangamaire, Mana- watu River System (2); Bull Stag Creek, Kiriwhakapapa tributary (3*) and Hinaki Stream, a tributary of the Mangatarere Stream (5*), both in the Ruamahanga River System; Mangakotukutuku Stream, Waikanae River System (4*); Hutt River at Kaitoke (6*); Catchpool Stream, Wainuiomata River System (7*); tributary of the Wakamarina Stream, Pelorus River System (8*); tributary of the Motueka River at Golden Downs (9*), and another tributary, the Clarke River at Hope Saddle (10); tributary of Lake Rotoiti (11*, Nel- son Lakes); small tributary of Maruia River near Spring’s Junction, upper Buller River (12*, type locality). Galaxias paucispondylus Stokell, 1938 Figure 29 Galaxias paucispondylus Stokell, 1938: 203 (holo- type: CMCNZ 73, seen; paratypes (10): CMCNZ 74, not seen; locality: Acheron River, tributary of the Rakaia River, Canterbury), 1949: 480. Diagnosis. Differs from G. divergens Stokell (Fig. 26) in characters discussed in the diagnosis of that species (p. 384); differs from G. prognathus Stokell (Fig. 31) in having fewer anal fin rays, branchios- tegals, vertebrae, and gill rakers. It also has a slightly stouter build, longer pectoral fins, and more anterior pelvic fin insertion, but these morphometric differences are rather minor. As with G. divergens, G. paucispondylus differs from G. prognathus chiefly in having sub-equal jaws, this char- acter allowing immediate separation of the two species. Description. Very elongate and slender bodied, trunk almost square in section, middorsal groove present, indistinct, dorsal and ventral trunk profiles about parallel, with belly somewhat deepened and rounded anterior to the pelvic fins especially in ripe adults; depressed anteriorly on head, some- what compressed on caudal peduncle, which is very long and slender, much longer than deep. Lateral line a moderate midlateral furrow, accessory lateral line not observed. Head short and tapering, some- what depressed. Eye small, upper margin near upper head profile, interorbital flat or slightly concave. Lower jaw a _ little shorter than upper, lips well developed; cleft of mouth extends to about anterior eye margin; profile of lower jaw from ventral aspect broad and shallow, some- what flattened anteriorly. Canine teeth poorly developed or lacking in jaws; meso- New ZEALAND GALAXIIDAE °* pterygoidal teeth rather poorly developed; gill rakers variable, weakly to moderately developed, often irregularly spaced with large gaps suggesting loss of rakers; pyloric caeca short. Dorsal and anal fins somewhat fleshy at bases, short based; greatest fin length somewhat longer than base length, but fins not prominent. Dorsal fin insertion further forward in standard length than in most Galaxias (due to the great length of the caudal peduncle); anal origin usually a little behind dorsal origin. Pectoral fin quite small, rounded, inserted moderately high laterally. Pelvic fin short, inserted in front of midpoint of standard length. Caudal fin moderately long, truncated or slightly emarginate, fin depth about equal to greatest body depth, flanges of caudal peduncle moderately developed. Variation. Meristic: dorsal 7 (3), 8 (10), 9 (36), 10 (9), 11 (1); caudal 15 (3), 16 (oy liane) wamall 7. (ly) 8 (7). 9. (35), 10 (13), 11 (1); pelvic 6 (14), 7 (43), 8 (1); pectoral 10 (5), 11 (33), 12 (17), 13 (3); branchiostegals 5 (2), 6 (20), 7 (33), : foes 50) (2) ak GIO), SPAT Mos AW 3 (9), 54 (1), 55 (0), 56 (1); gill rakers 5 (dl i ), 1-6 (3), 1-7 (5), 1-8 (1), 2-6 ee 2-7 (11), 2-8 (6), 2-9 (1), 3-7 (4), S§ (1). Morphometric: see Table 4, p. a Coloration. The basic body color is a grayish cream, interrupted dorsally and laterally by usually bold greenish brown to gray vermiculations. These fail rather high laterally on the belly and caudal peduncle and do not extend much onto the fins except the caudal, which is often quite densely pigmented. The head is heavily pigmented dorsally and also laterally to just below the eyes, but the cheeks, ventral head, and belly are pale, virtually colorless. Quite commonly, this lack of pigmentation on the ventral trunk extends posteriorly beyond the anal fin, even as far as the caudal base. This is unusual in New Zea- land galaxiids, in which the lateral trunk pigmentation usually extends as far down McDowall 391 as the anal fin and covers the entire caudal peduncle. Size. Stokell (1949: 480) recorded G. paucispondylus growing to 4.4 inches (112 mm). The largest examined in this study was 104 mm; G. paucispondylus commonly grows to 80-85 mm. Population differences. Samples of G. paucispondylus studied came from a re- stricted area and only three distinct river systems. Samples from only two of these, the Rakaia and Ashburton, were of suffi- cient size to allow comparison, and the interpretation of inter-population differ- ences is difficult without a series of popu- lations. However, the more southern Stour (Ashburton) population was found to be generally more slender in form than that from the Harper-Avoca-Wilberforce Rivers (Rakaia). In the former, body depth at vent/standard length, head depth/head length, head width/head length, inter- orbital width/head length, and gape width/ head length ratios are all lower (Fig. 30). The meristic data did not show recogniz- able differences between populations. Habitat. G. paucispondylus occurs only in the swift, cold, snow-fed, boulder-gravel streams of sub-alpine and alpine Canter- bury. These rivers tend to be unstable, flood severely with heavy rains and rapid snow thaws, and the rivers wind across broad, flat, open, gravel plains, flanked on either side by steep and often denuded, unstable hills. G. paucispondylus is char- acteristically found in these rivers in the moderately deep, broken-water riffles where the flow is extremely rapid. Life history. G. paucispondylus is re- stricted to fresh water and has no juvenile whitebait stage. Ripe and spent adults were present together in samples collected in October, suggesting that spawning takes place in the southern spring. However, Stokell (1955: 32) reported that it occurs in March and April. It thus appears that G. paucispondylus may have prolonged breeding from spring through the summer to the autumn. The ovaries of fishes 392 Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 _ i) 10 11 12 13 14 BODY DEPTH “STANDARD LENGTH Z% AO 50 60 70 HEAD WIDTH 4 “HEAD LENGTH % INTER-ORBITAL WIDTH, weap LENGTH % 30 32 34 36 38 40 WIDTH OF GAPE “HEAD LENGTH % iriation in body proportions in Galaxias paucispondylus. A, Body depth at vent/standard length ratio; B, Head th ratio; S Head width/head length ratio; D, Interorbital width/head length ratio; E, Width of gape/ hear a, Rakaia River (15 examples); b, Stour River, Ashburton River System (20 examples). | NEw ZEALAND GALAXIIDAE * McDowall 393 Figure 31. collected in June were in moderately ad- vanced stages of maturation; others, col- lected in December, were invariably spent or had very immature ovaries. The eggs of G. paucispondylus are very large, about 2 mm diameter when ripe. They are very few in number, a female 88 mm long having only 269 eggs. The breed- ing site is unknown to me and is unde- scribed, but is probably close to the normal adult habitat. Distribution. G. paucispondylus is pres- ently known only from upland and alpine Canterbury, on the eastern side of the Southern Alps. Populations are known at the following localities: Waimakariri River and its tributaries, the Cass and Porter Rivers (Fig. 25: 14); the Harper, Avoca, Wilberforce (15*), and Acheron Rivers (16) in the Rakaia River System; Stour River, Ashburton River System (17*); Deep Creek, a tributary of the Rangitata River at Mesopotamia Station (18*). Galaxias prognathus Stokell, 1940 Figure 31 Galaxias prognathus Stokell, 1940: 422 (holotype: CMCNZ 75, seen; locality: Wilberforce River, Canterbury ), 1949: 480. Diagnosis. Differs from G. divergens Stokell (Fig. 26) and G. paucispondylus (Fig. 29) in characters discussed in the diagnoses of these species (pp. 384 and 390 respectively ). Description. Very elongate and slender in form, belly somewhat rounded and deep- ened in front of pelvic fins but tapering posteriorly to a slender caudal peduncle, Galaxias prognathus Stokell, 64 mm L.C.F., Avoca River, Rakaia River System. which is very long and much longer than deep. Trunk flattened dorsally with a well- developed middorsal furrow. Lateral line a well-defined midlateral crease; accessory lateral line not observed. Head small and short, tapering considerably, and dorsally flattened; shallow, much broader than deep. Eye small, towards upper head pro- file, interorbital flat. Lower jaw much longer than upper, upper jaw short, and mouth upturned, lips prominent. Cleft of mouth oblique, usually not reaching an- terior eye margin; lower jaw profile from ventral aspect broad and shallow, flattened anteriorly. Canine teeth lacking from jaws; mesopterygoidal teeth weak; gill rakers weak and irregular in development; pyloric caeca lacking. Dorsal and anal fins showing some basal fleshiness, very short based; greatest fin length somewhat greater than basal length but fins not prominent; distal margin of fins rounded. Dorsal fin insertion well for- ward, anal origin below or a little behind dorsal origin. Pectoral fin very small and short, somewhat triangular, with the long- est rays towards the upper edge of fin, inserted moderately high laterally. Pelvic fin also very small and short, inserted at about midpoint of standard length. Caudal fin moderately long, emarginate to slightly forked, fin depth about equal to body depth. Caudal peduncle flanges showing moderate development, extending about half-way along caudal peduncle to anal fin base. Variation. Meristic: dorsal 8 (2), 9 (16), ORGS) ella excaudal tone peliGeGlia)e L7 394 ()B enavell 10) (CES) 5 IL 7) 1 ee pelvic 6 (2), 7 (18); pectoral 11 (3), one 13 (1); branchiostegals 7 aoe 3) (G3) vertebrae 54 (3), 55 (10), oe (i) BY (CD) gill rakers 1-9 (1), 2-8 (4), 2-9 (9), 2-10 (1), 2-11 (1), 3-9 (1), 3-10 (3). Morpho- metric: see Table 4, p. 387. Coloration. Similar to the two preceding species, pale creamish gray with bold, dark, greenish gray vermiculations dorsally and laterally. In a similar manner to the color- ation of G. paucispondylus, the vermicu- lations fail rather high laterally, just below the eyes on the cheeks, and not far below the lateral line along the abdomen. The fins are largely colorless. Size. G. prognathus is known to reach 79 mm length. Fishes 60-70 mm_ long formed a substantial proportion of a large sample from the Wilberforce River. Population differences. Since adequate numbers of G. prognathus were collected from only one locality, and since all the samples were from the Rakaia River System, no meaningful comparisons of in- ter-population variation were possible. Habitat. Like G. paucispondylus, G. prognathus occurs in alpine boulder-gravel streams and rivers, and the two species are sometimes taken from the same water. Generally, G. prognathus occurs in shal- lower, turbulent but not broken water. Stokell (1949: 480) reported having col- lected it mostly in situations where “a side stream rejoins the main stream at such a gradient that the water percolates through the boulders, leaving their upper surfaces dry.” In the Wilberforce River, where G. prognathus was found to be quite common, the fishes were collected in “shal- low riffles up to four inches deep and not particularly fast, but not in the flats,” the stream bed composed of “boulders, stones and gravel with considerable silt and sand” (G. A. Eldon, pers. comm. ). Life history. Samples of G. prognathus all collected towards the These fish were mostly though a few were ripe, indicating examined were end of October. spent, Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 that spawning was about finished. The eggs are large, about 1.8 mm diameter, and very few in number, a ripe female 68 mm long, with full abdomen, containing only 93 eggs. It is almost certain that the life history is restricted to fresh water, prob- ably in the vicinity of the normal adult habitat. As was the case with G. paucispondylus, my observations are at variance with those of Stokell’s (1940: 424, 1955: 34); he re- ported that spawning occurs in the autumn. However, Stokell’s observations were based on a female taken in April with ova measur- ing only 1.16 mm diameter (1940: 424). My observation of ripe eggs measuring about 1.8 mm suggests that Stokell’s fish were not mature. Spawning is probably later in the year than “late autumn or early winter” as suggested by Stokell, and seems likely to occur in the early spring. Distribution. G. prognathus is known from alpine areas of the central South Island. I have seen samples from the Rakaia River System, the Harper, Avoca, and Wilberforce Rivers (Fig. 25: 15*), and the Maruia River, Upper Buller River System (13). Collections of fishes in the adjacent Waimakariri, Ashburton, and Rangitata Rivers have not contained G. prognathus. NEOCHANNA GUNTHER Neochanna Giinther, 1867: 306 (type species Neochanna apoda Giinther, 1867, by original designation ). Diagnosis. Characters generally those of © Galaxias, but with pelvic fins reduced or lacking; mesopterygoidal teeth reduced or lacking. Jaw teeth sometimes compressed and incisorlike, or conical, as in Galaxias. No supraethmoid or ventral ethmoid, vomer folded upwards in front of ethmoid | cartilage, ascending processes of premaxil- lae more or less meeting tips of frontals. No epipleural ribs. confluent anteriorly with dorsal and anal fins. Flanges of caudal | peduncle very strongly developed, usually | | New ZEALAND GALAXIIDAE * McDowall 395 Figure 32. Key to SpEcIES OF NEOCHANNA 1. Pelvic fins present _. N. burrowsius p. 395. Relvicwtinsmabsemt as eee ee OR, 2. Median fins long, 14-19 rays, D.C.P./ L.C.P. 129.8-200.0% _. N. apoda p. 398. Median fins shorter, 11-16 rays, D.C.P/ IU kCal?, CEPI NGG a N. diversus p. 402. Neochanna burrowsius (Phillipps, 1926) Figure 32 Galaxias burrowsius Phillipps, 1926c: 531 (holo- type: DMNZ 521, seen; paratype: DMNZ 4646, seen; locality: a drain on the farm of the late Mr. A. Burrows, West Oxford, Canterbury ); Stokell, 1949: 481. Galaxias burrowsii Phillipps, 1927a: 14, 1927b: 11; Stokell, 1938: 205. Saxilaga burrowsius: Scott, 1936: 110, 1966: 250. Paragalaxias burrowsii: Phillipps, 1940: 39. Diagnosis. Differs from N. apoda Giin- ther (Fig. 34) and N. diversus Stokell (Fig. 36) in the presence of pelvic fins, and of mesopterygoidal teeth in many examples; also in the shorter median fins and lower ray counts in these fins. Taxonomy. N. burrowsius has previously been included in the genus Galaxias, or in Saxilaga, which has characters intermedi- ate between Galaxias and Neochanna. In its general morphology, it is similar to N. apoda and N. diversus, but whereas these latter species have entirely lost the pelvic girdles and fins, they persist in N. bur- rowsius. It also often has a few weak meso- pterygoidal teeth, whereas Neochanna is usually described as having none. (In one specimen of N. diversus, I found a single tooth on each mesopterygoid.) These three species have the appearance of a radiation within the Galaxiidae, comprising Neochanna burrowsius (Phillipps), 108 mm T.L., Gawler Downs, Hinds River System. species adapted to temporary creeks and bogs, which are able to aestivate when these dry up. They look like a single phylo- genetic lineage, and their osteology sup- ports this. Osteological examination has revealed characters that indicate close relationship. In the ethmoid region of the skull, the three neochannoid species have lost the supra-ethmoid and ventral ethmoid bones. The ascending processes of the premaxillae have become pushed back over the ethmoid cartilage to meet the anterior tips of the frontals, and the vomer seems to have been folded upwards in front of the massive ethmoid cartilage, above the tip of the parasphenoid. In N. burrowsius and N. diversus, but not in N. apoda, the tips of the ethmoid cartilage, which diverge over the vomer, each have small tubular ossifi- cations. In none of the three species are there epipleural ribs, although all the New Zealand species of Galaxias have them. On the basis of these considerations, I include G. burrowsius Phillipps in the genus Neochanna Giinther. A minor nomenclatural problem exists in the spelling of the name burrowsius. In his original description, Phillipps (1926c: 531) named the species G. burrowsius, reporting that the fish was collected on the farm of a Mr. A. Burrows. In later papers (1927a, b, 1940), he has spelt the name burrowsii. Mr. Phillipps (pers. comm. ) has kindly advised me that he intended to name the species for Mr. Burrows, and not because of its habit of aestivating in small pockets in mud. The Zoological Code of 396 Nomenclature is not firm in the formation of patronyms, only recommending (1964: 33, recommendation 31 A) that a should be formed by the addition of “i” to the personal name, if masculine. Hoes rules for the emendation of name spelling (p. 35, art. 32a) are such that the original spelling must be maintained unless it con- travenes mandatory provisions on name formation, or there are obvious, inadvertent errors. Neither of these is the case, so the original spelling must stand. The specific patronym can be regarded as grammati- cally correct either as an adjective—the Burrows Galaxias—or as a noun in appo- sition. Numerous attempts to collect N. bur- rowsius from localities listed by earlier workers (Phillipps, 1926c: 532, Stokell, 1949: 482) have failed. The type locality appears to have disappeared as a habitat for N. burrowsius, since no creeks or drains could be found at the locality at West Oxford, from which Phillipps first obtained the species (K. F. Maynard, pers. comm. ). The present work is based on series col- lected from localities associated with the Hinds River. In his description of N. burrowsius, Phil- lipps (1926c: 531) pointed out that teeth were present “only on pre-maxillaries, lower jaw and tongue.” Stokell (1945: 129) noted that he had been unable to re- collect N. burrowsius from the original locality but (1938: 205, 1949: 482) re- described it from further, new localities. In the later of these papers, he pointed out that teeth may or may not occur on the mesopterygoids. Scott (1966: 250) ques- tioned the correctness of identifying the forms having toothed mesopterygoids with N. burrowsius. Since the condition is variable and Phillipps described the species from only two specimens, and since the subsequent collections of fishes included in this species have all been made from a restricted area of Canterbury, there seems little doubt that Stokell’s action is correct and that the “neochannoid” species present Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 in swamps and drains in Canterbury Province, in the vicinities of Ashburton and Christchurch, is N. burrowsius (Phillipps ). Description. Trunk much elongated and cylindrical, flattened dorsally with a deep middorsal furrow, dorsal and ventral trunk profiles parallel; little depressed on head, compressed on caudal peduncle, which is of moderate length and depth. Lateral line indistinct anteriorly, becoming a_ well- defined furrow posteriorly; accessory lateral line present, but weakly developed. Head rounded and blunt, cylindrical. Eye very small, deep set in head, with interorbital convex. Jaws very short, about equal or lower a little shorter, cleft of mouth ex- tends to about anterior eye margin; profile of lower jaw from ventral aspect somewhat deep and narrow, a broad U. Canine teeth lacking in jaws; mesopterygoidal teeth poorly developed and few in number, or lacking. Gill rakers short; pyloric caeca well developed. Median fins short based and low, with well-developed basal fleshiness. Overall fin length little greater than basal length, distal margins of fins more or less straight and tending to become parallel with trunk axis. Predorsal length moderate, anal origin usually below or a little behind dorsal origin. Pectoral fin very short, rounded, inserted moderately high laterally; pelvic fins much reduced, inserted at about mid- point of standard length. Caudal fin short and much rounded, fin depth about equal to body depth; caudal peduncle flanges very strongly developed and_ extending forwards to the insertions of the dorsal and anal fins, more or less confluent with the posterior ends of their bases. Variation. Na ae dorsal 7-C1) Sy@ ae 9 (14), 10 (5), 11 (3); caudal Gal (1); 13 (20), 14 (12): analeSiGae Ou Glane 10 (10), 11 (5), 12 (1); pelvic 4 (1), 5 (14); pectoral 10 (1), 15 sGlO) aloe branchiostegals 5 (1), 6 (9), 7 (3); verte- braccolh(2) 3526) 538 Ge 54 (7), 55 (1); gill rakers 2-9 (1), 2-10 (1), 3-8 (1), > New ZEALAND GALAXHDAE * McDowall 397 © Neochanna diversus @ N. apoda @ N. burrowsius Figure 33. Distribution of Neochanna burrowsius, N. apoda, and N. diversus (numbers in figure as in text pp. 398, 402, and 404). 398 Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 Figure 34. Soe) (OO) Cal) (05) SS (2), SSP 4-9 (2), 410 (2), 4-11 (3). Morpho- metric: see Table 5, p. 399. Coloration. Trunk a milky gray-brown, covered dorsally and laterally with fine, darker, greenish brown vermiculations, these extending well on to the fin bases. Belly paler, a milky brown. Size. An example from the Anama sample measured 146 mm total length. Amongst the few fish examined, a good proportion were 100-125 mm long. Population differences. Unfortunately, all the specimens examined were collected from the Hinds River System, in the same restricted area, so that no investigation of differences between populations was pos- sible. Habitat. N. burrowsius appears to be collected usually from small, muddy, or gravel-bed streams and creeks, often those draining swamps. Phillipps (1926c: 532) reported their ability to aestivate in small pockets of mud in the same manner as other Neochanna species are well known to do. Stokell (1949: 482) also reported specimens dug out of “damp earth and detritus at the bottom of a drain that had been dry for over a month.” Life history. N. burrowsius collected from the Hinds River System in early August were approaching maturity. All those taken in November were spent. Thus N. burrowsius probably spawns in the spring. No females from the August sample were sufficiently mature to permit egg counts or measurements, but egg ber appeared to be moderately high, mparable with that of G. macu- Neochanna apoda Gunther, 110 mm T.L., tributary of Mangatarere Stream, Ruamahanga River System latus. There is no whitebait stage and no obvious juvenile-adult metamorphosis, and it is almost certain that N. burrow- sius spends all its life in fresh water. Distribution. N. burrowsius is known only from the Canterbury District, South Island. It has been collected from the fol- lowing localities: West Oxford (Fig. 33: 27, type locality, Phillipps, 1926c: 532); Rangiora (28) and Tinwald (29, Stokell, 1949: 482); Gawler Downs, Anama_ Dis- trict, Hinds River (30*). Neochanna apoda Ginther, 1867 Figure 34 Neochanna apoda Ginther, 1867: 306 (holotype: BMNH_ 1965.11.5.8, not seen; locality: “near | Hokitika,” west coast, South Island); Hector, 1869: 402; Hutton, 1872: 61; Vollams, 1872: 456; Hutton, 1904: 51; Regan, 1905: 383; Phillipps, 1923: 62, 1926b: 297, 1927a: 14, 1940: 41; Stokell, 1949: 494. Diagnosis. Differs from N. burrowsius (Phillipps) (Fig. 32) in characters dis- cussed in the diagnosis of that species (p. 395); differs from and, particularly in the southern part of its range, fewer vertebrae. N. apoda has | much paler coloration than N. diversus. Description. Trunk elongated and some- | what rounded, middorsal furrow promi- nent; dorsal and ventral trunk profiles | about parallel; head little depressed an- | teriorly, caudal peduncle much _ com- pressed and thin posteriorly, short, much N. diversus Stokell | (Fig. 37) in having longer dorsal and anal | fin bases, a much shorter caudal peduncle, — a broader head, smaller eyes, and much — longer jaws. It also has more rays in the | dorsal and anal fins, more branchiostegals | New ZEALAND GALAXUDAE * McDowall 399 TABLE 5. MORPHOMETRIC VARIATION IN THE “NEOCHANNOID” SPECIES (FIGURES GIVEN AS PERCENTAGES OF DENOMINATOR OF RATIO). N. burrowsius N. apoda N. diversus Min Mean Max. Min Mean Max Min Mean Max. Sol Lig/1ally 87.0 89.3 91.7 85.5 88.5 90.9 88.5 90.9 92.6 B.D.V./S.L. 10.0 110 #£411.9 10.4 12.9 14.9 10.5 11.9 15.0 ete Py/ Sales HCE 2S lb. 2) 4.9 6.5 7.8 Oe lOO L2e9 D.C.P./L.C.P. 56.8 64.9 76.9 129.9 158.7 200.0 hy, ALLL yy) Pre D./S.L. 69.4 73.5 76.3 69:9) 7205) 75:8 A TAR GES Pre D./Pre A. 94.3 98.0 101.0 93.5 97.1 101.0 97.1 101.0 105.3 D.F.B./S.L. 97 116 14.0 IG), DAILAS) 5,8} 10.1 13.8 16.8 D.F.B./D.F.M. 59.9 71.9 84.0 75.2 83.3 90.9 69.4 78.7 92.6 A.F.B./S.L. 10.5 13.1 15.8 20 2224 a8 16.7 189 20.8 A.F.B./A.F.M. 69.9 80.7 87.0 78.1 86.2 96.2 84.8 90.9 98.0 Pre Pel./S.L. 48.9 51.3 53.8 — — — — — — Pec.Pel./S.L. 29.0 340 36.1 — — — — — — Pec./Pec.Pel. Mi Oss — = = = = = Pel.An./S.L. ONS) BIBS) BNI — — — — — — Pel./Pel.An. yl PAN Shays} — —— — — — — H.L./S.L. 16.7 184 20.2 ise, ALO GY) 186 20.1 21.8 H.D./H.L. 46.5 53.8 59.2 444 52.9 59.2 424 49.8 58.5 H.W./H.L. 52.1 59. 64.1 56.8 62.9 71.9 53.8 59.9 65.8 Sn.L./H.L. DAN Qo Sled 23:89 | 27.0)" 29:8 2516 Peeled P.O.H.L./H.L. 546 59. 63.3 57.1 63.7 70.4 58.8 62.9 68.5 To.W./H.L. 32.8 36.1 40.0 32.6 36.8 42.7 Bigs) esl Zi, il D.E./H.L. 10.3 12.5 15.3 8.3 11.2 15.5 ial 13.0 15.6 L.U.J./H.L. DU Bild BBB 36.2 39.7 44.1 27.0 30.4 32.3 L.M./H.L. 28.0 30.7 35.1 Sy exsey 0) ZEB 30.0 31.6 32.3 W.G./H.L. 34.5 38.0 42.4 36.1 41.7 50.0 30.8 35.2 38.3 Pec./H.L. — — — 424 53.3 62.1 45.2 50.6 56.8 Fish examined ilk 38 16 deeper than long. Lateral line a deep lateral groove; accessory lateral line pres- ent. Head moderately long, broader than deep, somewhat bulbous behind nape, and tapering abruptly from just behind eye forwards on to a rather slender snout. Eye very small, deep set in head, with interorbital convex. Jaws about equal, very long, cleft extending well below eye, about as far as posterior eye margin. Gape very broad, profile of jaw from ventral aspect quite deep but moderately broad, some- what flattened anteriorly. Canine teeth lacking from jaws; jaw teeth peculiarly compressed and _ incisorlike; mesoptery- goidal teeth lacking; gill rakers moderate to short; pyloric caeca strongly developed. Unpaired fins low, but very long based, with much basal fleshiness; greatest fin length little greater than basal length, dis- tal margin of fin straight, parallel to trunk axis; fin bases confluent posteriorly with caudal peduncle flanges. Anal origin usually a little behind dorsal origin. Pec- toral fin short, rounded, insertion high laterally. The fins all show coarse mar- ginal serration. Variation. Meristic: dorsal 14 ee 15 (GLO EGA ean (GLA rea Sen, ep aS 92) caudalise Gl) pl4e(S sls) Gillon @isg))e Le (S))s hare a (Ae MS Ee Gs (ear (10), : (4), 19° (6); pectoral 11 (2); 12 (15), 13 (20); branchiostegals 6 (12), 7 (19), 3 (5), 9 (1); vertebrae 52 (5), 53 (Oye 2 (8). ss) (iS), OG, ei (a), as (2), 59 (2); gill rakers 1-9 (1), 2-8 (18), 2-9 (12), 3-8 (7), 3-9 (4), 49 (1). Morphometric: see Table 5. Coloration. Trunk usually sandy col- ored, darker dorsally, with the usual 400 Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 50 a 0 so 20 2 b m OO mi 0 = (?) as 50) C (0) , 13 14 15 l6 IRs 8 19 DORSAL FIN RAYS 50 B x 0 - 50 fn fo) b S m 5 0 < 50 C 0 5/2 5S 54 5S) 56 Dy/, 58 59) VERTEBRAE Figure 35. Variation in meristics in Neochanna apoda. A, Dorsal fin rays; B, Vertebrae; a, Wellington District (8 examples); b, Wairarapa District (20 examples); c, Western South Island—Westland District (13 examples). irregular darker vermiculations of green- ish brown color. Coloration much browner than galaxiids in New Zealand, usually brownish gray to a purplish brown. Belly a pale cream- most 1 a 9re buff color. Fin bases well pigmented with vermiculations as on trunk. Samples from the west coast of the South Island were generally somewhat darker than northern ones, with the vermiculations tending to become resolved into rather bold, broad, dark, vertical bands. Size. Stokell (1949: 494) recorded N. apoda growing to 6.8 inches (173 mm). The largest example examined in the pres- ent study was 169 mm long. N. apoda appears commonly to reach 100-130 mm. Population differences. N. apoda_ ex- hibits wide variation in meristic and some morphometric characters. Unfortunately, only two populations were available of sufficient size to allow adequate com- parisons to be made. However, by group- ing samples it was possible to compare data from three major, discrete areas. These were the Wairarapa, the Welling- ton Province west of the main ranges, and the west coast of the South Island. The number of specimens from the Wellington area (five) is totally inadequate for definitive comparisons, but these few specimens showed substantial, interesting differences from the others. Vertebral number was found to be greatest in Well- ington fishes, somewhat lower in those from the Wairarapa, and minimal in those from the west coast of the South Island (Fig. 35B). Dorsal (Fig. 35A), anal, and caudal fin ray counts showed a similar trend; the greatest number of fin rays was found to occur in the more northerly populations. These data appear to be contrary to clines related to water tem- peratures, in which the populations in warmer areas would be expected to have fewer elements, unless there is something peculiar about the temperatures of water bodies from which these populations were collected. The variation in caudal fin rays is interesting in that this is otherwise by far the least variable of any of the meristic characters in the family. Branchiostegal number showed a different trend from other counts, being least in the Wairarapa samples and greatest in west coast samples, the Wellington fishes occupying an intermediate position. The head length/standard length ratio of the Wellington fishes was much lower New ZEALAND GALAXIIDAE * McDowall AQ] AD LENGTH : WE “STANDARD LENGTH % ———— ae B oe a eee ee LENGTH OF CAUDAL PEDUNCLE /.7 NIDARD LENGTH % Figure 36. Variation in body proportions in Neochanna apoda. A, Head length/standard length ratio; B, Eye diam- eter/head length ratio; C, Length of caudal peduncle/ standard length ratio; a, Wellington District (8 examples); b, Wairarapa District (23 examples); c, Western South Island—Westland District (8 examples). than that of those from the Wairarapa, in which the ratio was lower than those from the west coast. Eye diameter ex- hibited similar variability. The length of the caudal peduncle was similar in Wair- arapa and west coast material but greatly decreased in the few Wellington speci- mens examined (Fig. 36). This variability appears to generate no clear pattern and it is obviously necessary to await collection of more material, especially from the western part of the Wellington province, before these differ- ences can be understood. Eventually it may be necessary to divide what is here treated as a single species at either the species or sub-species level. Habitat. In the Wairarapa, N. apoda has been usually collected in the upper reaches of small, spring-fed creeks, often filled with bottom-rooted vegetation. On the west coast of the South Island it was found in the tannin-stained waters of bush swamps, under a heavy forest cover, and in streams flowing into flax swamps. N. apoda is typical of the New Zealand mud-fishes in 402 its ability to withstand desiccation of its habitat. G. A. Eldon (pers. comm.) re- ported collecting one specimen from be- neath a log in the middle of a partially cleared cow pasture, with no water near by at the time of capture. Phillipps (1923: 62) reported that it was collected from white pine swamps, and Reid (1886) reported that healthy N. apoda had been collected five or six feet down in clay, suggesting that the fishes follow moisture down holes left by rotted tree roots. Stokell (1955: 38) also mentioned its ability to bury itself in mud in times of drought. Life history. Little is known of the life history of N. apoda. Stokell (1949: 495) noted that males taken in October “had the milt almost fully developed and ap- peared to be within a week or two of spawning.” Davidson (n.d.) found that Wairarapa N. apoda spawn “probably . . . not before the end of November.” A specimen taken in June by Phillipps (1926b: 297) is reported to have been in spawning condition. Material collected from the west coast of the South Island in late October was all apparently re- cently spent, or in the early stages of gonad rejuvenation, whilst the sample from the Kaipaitangata System (Wair- arapa) collected in February contained a mixture of ripe and spent individuals. It appears that at present, no clear breed- ing period for N. apoda can be defined; it is possibly of long duration. The eggs of N. apoda are moderately large, about 1.75 mm diameter, and few in number. A female from the Wairarapa, 115 mm long, contained only 533 eggs. Growth of juveniles is almost certainly in or near the adult habitat, as examples only 27 1/2 mm long were collected with adults. They are, at this small size, similar in form and coloration to the adults. Distribution. N. apoda is found in the south of the North Island and on the west the South Island. It is known following localities: Opunake coast of from the Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 (Fig. 33: 8); Feilding (9, Davidson, n.d.); Rangitikei (10, Phillipps, 1923: 62); Palmerston North (11, Stokell, 1949: 494); Rongotea (12*); Otaki (13*); Waikanae (14*); Masterton (15, -Phillipps, 1926b: 297); tributary of the Kaipaitangata stream, Ruamahanga River System (16*); tributary of Lake Wairarapa at Pirinoa (17*); Oparara (18); Birchfield (19); Westport (20, Eldon, 1968); Greymouth (21, Stokell, 1949: 494); Kumara Junction (22*); Hokitika (23*); Ross (24, Eldon, 1968 ); Harihari (25*); Whataroa (26%). Neochanna diversus Stokell, 1949 Figure 37 Neochanna diversus Stokell, 1949: 495 (holotype: CMCNZ 76, seen; locality: Kaitaia, North Auckland ). Diagnosis. Differs from N. burrowsius (Phillipps) (Fig. 32) and N. apoda Giin- ther (Fig. 34) in characters noted in the diagnoses of these species (pp. 395 and 398 respectively ). Description. Rather slender bodied, trunk rounded in section without middorsal fur- row; dorsal and ventral trunk profiles parallel from about head to dorsal origin, although the belly deepens noticeably in ripe adults. Trunk much compressed pos- teriorly on caudal peduncle, but little depressed anteriorly on head. Caudal peduncle deep and relatively long. Lateral line indistinct, accessory lateral line pres- ent. Head short, very blunt, broader than deep; head profile smooth and rounded, snout profile very convex. Eye very small, set deep in head, interorbital very convex. Jaws short, about equal, or lower protrud- ing slightly; cleft of mouth reaching below anterior half of eye, oblique. Profile of jaw from ventral aspect rather broad and flattened anteriorly. Jaw teeth conical, lacking canines, usually no mesopterygoidal teeth, although one fish with one tooth on each mesopterygoid was observed. Gill rakers variable in length, from moderately to well developed; pyloric caeca also vari- able, usually moderately developed. New ZEALAND GALAXIIDAE °* McDowall 403 Figure 37. Neochanna diversus Stokell, 112 mm T.L., swamp at Waiharara, North Auckland District. Median fins moderately long based and expansive, but extending little along caudal peduncle beyond end of fin base, extremely fleshy basally, fleshiness failing on distal half of fin, which is quite membranous; margins of fins straight, more or less paral- lel to trunk axis. Anal origin at or a little in advance of dorsal origin. Dorsal and anal fins confluent with caudal peduncle flanges, which are very strongly developed. Caudal fin short, moderately fleshy at base, truncated to rounded. Pectoral fin short, membranous or slightly fleshy, rounded, inserted high laterally. Margins of fins serrate. Variation. Meristic: dorsal 10 (2), 11 (Co eel) else 21) 452) caudal 5 (GPG) 25) liad) anal i3n(4)s 145 15 (19), 16 (4); pectoral 11 (1), 12 (14), 13 (17), 14 (4); branchiostegals 6 (12), 7 (24); vertebrae 55 (2), 56 (9), 57 (7), 58 (12), 59 (10), 60 (2); gill rakers 2-8 (5), 2-9 (3), 3-8 (8), 3-9 (14), 3-10 (6), 4-8 (2), 4-9 (3), 4-10 (1). Morphometric: see Table 5, p. 399. Coloration. Dark colored; in populations examined from peat bogs, a dark smoky gray to almost black; profuse fine vermic- ulations cover the dorsal and lateral trunk and the fin bases. The belly is paler, smokey gray to somewhat rufous. Size. An example from the Waiharara series measured 122 mm total length. Samples contained few examples more than 90 mm long. Population differences. Noticeable dif- ferences in meristics were found for the two large samples of N. diversus examined. The Waiharara sample (the more northern locality ) was found to have generally fewer counted structures, especially vertebrae, anal fin rays, and gill rakers. The sample from the most southern locality, Mount Pirongia, included only four fish, but these appeared to be more similar to those of the Waiharara series than to that from the Hikurangi swamp (e.g., vertebral number, Fig. 38). In the Waiharara and Hikurangi samples, ranges for anal fin rays and gill raker counts showed decided displacement from each other, and vertebral number was almost disjunct in the two samples (although the Hikurangi distribution is noticeably skewed). Thus, as was the case with N. apoda, differences between samples from highly isolated localities appear to be considerable. Collections of samples from further, intermediate localities and at- 50 a a 2 50 pa) b = m 2 OQ “so 50 c 0 55 56 57 58 59 60 VERTEBRAE Figure 38. Variation in vertebral number in Neochanna Waiharara, North Auckland District (16 ex- amples); b, Hikurangi, North Auckland District (22 examples); diversus. a, c, Mt. Pirongia, Waikato District (3 examples). 404 tempts to relate water temperatures or other ecological parameters to morpho- logical characters may prove worthwhile. Habitat. Stokell (1949: 495) discussed the habits of N. diversus, reporting that one of his specimens came from the mud of a creek in the summer, and that six were taken, free-swimming, during the winter. In March, 1965, several days were spent in the vicinities of Mt. Pirongia, Kaitaia, and Whangarei searching for this species in muddy creeks and drains, without suc- cess except at Mt. Pirongia. Only three specimens were found (the fourth was ob- tained from a local farmer). At Waiharara, one of the creeks searched unsuccessfully was a narrow channel draining a substan- tial shallow lagoon. Near this lagoon is a large area of Kauri-gum swamp, in which there are many small depressions, a few inches to several feet across and up to 18 inches deep. When examined, most of the holes were heavily overgrown with sphagnum, and often filled with twigs to the extent that they were scarcely dis- tinguishable from the surrounding peat. It was here that N. diversus was found. By moving from hole to hole, clearing away the sphagnum and debris, waiting for the sediment to clear, and fishing in each with the electric fishing machine, it was found that each hole usually contained two or three fishes. Collections from the Hikurangi swamp, near Whangarei produced N. di- versus from similar habitats. It appears, then, that these holes, rather than the streams running through swamps or drain- ing them, are the characteristic habitat of N. diversus. Life history. Mature, adult N. diversus were present in the samples collected from the North Auckland area in March. A single example from the vicinity of Waihi, col- lected in January, contained ovaries at an early stage of maturation. Stokell (1949: 195) found that examples collected from Kaitaia at the end of July were fully ripe These data indicate that probably breeds during the or partly spent. Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 winter. This is reasonable for a species that lives in a habitat that may dry up in the summer and that is known to be able to aestivate when this occurs. The eggs of N. diversus are moderately large, those from the most mature female in the collection being about 1.6 mm diameter. These were not ripe; the fully ripe eggs may be somewhat larger. Cor- responding with their size, they were found to be rather few in number, only 940 being present in a female 119 mm long. Distribution. N. diversus was recorded by Stokell (1949: 495) from Waihopo (Fig. 33: 1), Kaitaia (3), and Mangawai (5). An example in the collection of the New Zealand Dominion Museum is from Waihi (6*); in the present study additional speci- mens were collected from Waiharara (2*); the Hikurangi Swamp at Whakapara (4*); Mt. Pirongia (7*). These localities sug- gest that it has a general distribution in swamplands in the Auckland Province, as far south as Waihi and Mt. Pirongia. SPECIES INCERTAE SEDIS Galaxias kaikorai Whitley Galaxias kaikorai Whitley, 1956c: 34 (holotype: GMUO 6330-1; locality: Fraser's Gully, Kai- korai, near Dunedin, late Pliocene diatomaceous shale ). Taxonomy. Whitley applied this name to a fossil collected from a Pliocene fresh- water deposit in southeastern New Zea- land. Stokell (1945) examined this fossil thoroughly, concluding that it belongs in | the genus Galaxias, but that it is probably | not conspecific with living species. Whitley (1956) supplied no diagnostic characters _ for the species, and its validity is undeter- | mined, but since the name was referred to a specimen and a partial description, the name has taxonomic standing. I have not examined the fossil and believe that further | similar fossils have recently been dis- covered at the same locality (P. M. Johns, | pers. comm.). Examination of these fossils | may help to clarify the situation. New ZEALAND GALAXIIDAE * McDowall A405 a 50 be A aul ZA\(0) re) fm b = 30 & m = 20 10 0 36 37 38 B39 4Q 4 AB ais dial ay ale aba aliey alc) 0) Gil Be STANDARD LENGTH mm B é m > io) P m = rom = =x 5 30, 27 Be NO iD) f— ss INS) CC), & = oO pH J Nh ATGIGNVIN HLONAT/HLONAT GYVGNVLS = AONANODAYS Figure 39. 0 BO BY BS BOT AON ANS 42°42 5 44 45) 46 47 48 49° 50 51 52 STANDARD LENGTH mm 39 40 41 42 43 44 45 46 47 48 49 50 51 52 STANDARD LENGTH mm 39 40 41 42 43 44 45 46 47 48 49 50 51 52 30 By BS) STANDARD LENGTH mm Identification of juveniles of diadromous New Zealand galaxiids (explanation of symbols in text, pp. 406-408). 406 Galaxias abbreviatus Clarke Galaxias abbreviatus Clarke, 1899: 80. Nomen nudum. This name was used by Clarke in a discussion of the Galaxias species of the west coast of the South Island. The origin of the name is unknown, it does not occur in earlier literature, and Clarke applied it to no description or type. DISCUSSION Identification of diadromous whitebait juveniles The juveniles of the diadromous species of Galaxias in New Zealand lack many of the diagnostic characters of their respective adults, e.g., definitive pigmentation and dentition, and body proportions are very different from those of the adults. Over- lapping meristic values for the five species also add to the difficulties of identification. Because of this, the species are difficult to distinguish (McDowall, 1964b: 142, 1965a ). In my 1964 study, G. fasciatus and G. postvectis were not properly separated, and Woods (1966: 177) succeeded in identify- ing only three of the five species occurring in the rivers of the west coast of the South Island. In New Zealand, G. argenteus, G. fasci- atus, G. postvectis, G. brevipinnis, and G. maculatus have whitebait juveniles (Mc- Dowall, 1966a: 13). In earlier studies, samples were collected from the Buller River (Westland) and rivers nearby, and the four species, in addition to G. macu- latus, were identified by rearing the fishes. The samples preserved at that time have been re-studied and the results are pre- sented here. The whitebait of G. maculatus (Fig. 40) is easily identified by its very bold pig- mentation. The lateral line is very clearly defined by a series of large melanophores (which are small or lacking in other spe- cies), and there are several to many very ielanophores on the dorsal trunk anteri the dorsal fin. In other species Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 these melanophores are lacking. More ob- jective characters that distinguish the whitebait of G. maculatus include its low pectoral fin ray number (11-15, usually 12-13), combined with high anal fin ray counts (14-18, usually 15-17). After eliminating G. maculatus white- bait from samples, a _ length-frequency histogram of the remaining fishes showed that there were two very different size categories (Fig. 39A, a and b); these sug- gest size difference for some species at migration. A plot of standard length against head length indicated the same division of the samples (Fig. 39B, a and b). Further study of the fishes in the smaller size range (Fig. 39A, a) showed them to comprise only one species, and from much experience in collecting and studying these fishes, and having studied series of de- velopmental stages from freshly migrated juveniles to fully pigmented sub-adults, I am quite confident that they are the white- bait of G. fasciatus (Fig. 41). By elimination, the fishes in the larger size group (Fig. 39A, b) comprised a mixture of G. argenteus, G. postvectis, and G. brevipinnis. The adults of the first two species are much stouter than that of the third, and this difference is also evident in the juveniles. A plot of standard length against body depth at vent enabled division of the fishes into two somewhat over- lapping groups, and the same division was produced in a plot of standard length against head length (Fig. 39B, b, 1-2). Examination of the fishes showed that the more slender ones (b, 2) also had a much shorter-based anal fin than the stout fishes, and that the anal fin origin was set further back from the dorsal origin. These two subjective characters enabled me to place the fishes in the overlap zone between the two major groupings, in the appropriate group. The slender fishes, from the slender- ness, the short anal fin set back from the dorsal fin, and again from examination of | developmental series, are the whitebait | juveniles of G. brevipinnis (Fig. 42). New ZEALAND GALAXIIDAE * McDowall 407 Figure 40. Whitebait juvenile of Galaxias maculatus (Jenyns), 53 mm L.C.F. Sree Dy pies ae ee econ NE Rear mace eet nog pe ERIM EERIE sk eR Te Figure 41. Whitebait juvenile of Galaxias fasciatus Gray, 48 mm L.C.F. Figure 42. Whitebait juvenile of Galaxias brevipinnis Gunther, 50 mm L.C.F. Figure 43. Whitebait juvenile of Galaxias argenteus (Gmelin), 50 mm L.C.F. 408 Bulletin Museum of Comparative Zoology, Wolk, Ist), INO: 7 Figure 44. Whitebait juvenile of Galaxias postvectis Clarke, 54 mm L.C.F. The remainder of the samples, containing the larger and stouter fishes, consisted of G. argenteus and G. postvectis (Fig. 39C), and these fishes could be separated on the basis of standard length (Fig. 39D). The adult of G. postvectis has a distinctly re- ceding lower jaw, whereas in G. argenteus the jaws are about equal; this character enabled the identification of those fishes with the receding lower jaw as G. post- vectis (Fig. 44), and the shorter ones with the lower jaw not receding, as G. argenteus (Fig. 43). I am fairly confident that the groups divided and identified by this pro- cedure are correct, but the only way to confirm this appears to be further rearing trials. This study has not isolated particular characters by which individual specimens of the four problematical species can be identified with any assurance. But it does show that it is possible to separate the species when they occur in mixed-species samples. In addition, the experience of carrying out this sorting procedure high- lights the subjective characteristics of the whitebait of each species, and with this experience, it becomes possible to sort the species more directly. If this procedure can be carried out on a large scale, it will be possible to analyze seasonal differences in migratory patterns and the various eco- logical parameters which, in differing ways, influence the migrations of each of the species. Such an analysis would constitute significant progress towards understanding and intelligently managing the fishery. In Table 6, the ratios for head length/ standard length and body depth at vent/ standard length of juveniles and adults of each of the five diadromous species in New Zealand are listed. These data show that in all the species, the head becomes proportionately much longer and the body TABLE 6. BODy PROPORTIONS OF DIADROMOUS GALAXIIDAE. Head length as % of standard length Body depth at vent as % of standard length Fish Min. Mean Max. Min. Mean Max. examined Gurr rontene {Juvenile 19.6 20.0 21.2 11.9 12.8 14.0 6 ae ) Adult 27.0 29.1 30.5 18.7 21.0 23.4 36 G. fasciatus { Juvenile 16.4 17.8 18.8 10.5 11.8 V3.2, 151 ) Adult 22.8 25.9 28.6 15.0 17.9 21.4 60 Gupostocets { Juvenile 18.0 19.4 21.1 Wile 12.6 14.9 16 ) Adult ee, 23.0 25.0 16.4 19.6 22,4 95 G. brevipinnis {Juvenile 16.3 178 18.8 OD Shia AO 186 ) Adult 20.6 23.6 28.7 11.0 HELO? 15.3 160 Cinna {Juvenile 124 145 16.1 8.8 94 11.2 40 ) Adult 21.6 10.3 11.6 12.9 40 18.5 20.0 deeper during growth. I have shown else- where (McDowall, 1968b ) that in G. macu- latus, the change in head length/standard length ratio is a result of trunk shrinkage just after migration, but I have insufficient material of the other species to determine whether this occurs in them also. Life history and distribution patterns Two distinct life history patterns are recognizable in the New Zealand Galaxii- dae. Five species, G. maculatus, G. brevi- pinnis, G. fasciatus, G. postvectis, and G. argenteus have numerous, small to moder- ate-sized eggs, they spawn mostly in the autumn or early winter, and the freshly hatched larvae are carried to sea and undergo juvenile development there. About six months later, in the following spring, the slender, transparent whitebait juveniles, 40-55 mm long, migrate upstream in large, mixed-species shoals, undergo a minor metamorphosis, become pigmented, and assume adult form. In four of these species, the sub-adults become solitary in habit and are usually found in stream cover of some variety. The caudal fin changes from forked to emarginate or truncated, and all the fins become thick and fleshy, espe- cially basally. In contrast, the adult of G. maculatus retains the shoaling habits, the forked caudal, and membranous fins of the juve- nile, and does not become secretive in habit. G. maculatus, which breeds in river estuaries on grass-covered, upper-tidal flats, differs from the other four diadromous species in its spawning habits. Although little is known about the spawning habits of these four species, they seem likely to breed in or near the adult habitats. In some populations of G. fasciatus, and many of G. brevipinnis, the life cycle is restricted to fresh water. What changes have occurred in the life history of G. fasciatus subsequent to becoming land- locked in the Kaihoka lakes is presently unknown, but the population in Lake Okataina, and the many populations of G. New ZEALAND GALAXIIDAE * McDowall 409 brevipinnis in upland lakes, have a life history pattern in the lakes similar to the diadromous populations in the sea, al- though the juvenile life is of course lacus- trine instead of marine. Two small lacustrine species, G. usitatus and G. gracilis, are landlocked derivatives of G. maculatus. What is known of the life histories of these species suggests that they continue to spawn in the autumn, but that the diadromous habits have been elimi- nated. An Australian landlocked derivative of G. maculatus shows peculiar modifi- cation of the life history of the parental species, but this is not known to have oc- curred in these New Zealand species (see Pollard, 1966: 14). The remaining seven New Zealand ga- laxiids have fewer, larger eggs, they spawn mostly in the winter and spring, complete their entire life histories in flowing, fresh water and do not appear to migrate. There is no whitebait stage in any of them, nor do they seem to exhibit any metamorphosis. How these species have been derived from the diadromous species is not obvious, but it seems likely that the easily dispersed diadromous species are ancestral to at least some of the entirely freshwater species. New Zealand was heavily glaciated dur- ing the Pleistocene (See Fleming, 1962: 89), and the rivers must have been much colder than now. Species which were prob- ably lowland or coastal during the glaci- ations must have become adapted to cold water and are now probably those found in upland-alpine areas, where temperatures are about as cold as they were coastally during the glaciations. By spawning in the spring, the young of these species develop and undergo early growth in somewhat less severe conditions than would be the case if they spawned in the autumn, as the diadromous species do. The possession of a marine migratory phase by the diadro- mous species may be an alternative strategy for surviving through the cold winter. De- velopment in the sea gives them access to the prolific marine plankton, and in the 410 winter, the oceanic water temperatures during the glaciations would have been less extreme than river temperatures, making it worthwhile for the young fishes to spend the winter in the sea. The occurrence of five diadromous species in New Zealand and only three in warmer Australia per- haps supports this argument, and analysis of the life history patterns of South Ameri- can Galaxiidae will be interesting, since Galaxias species occur in the far southern regions of Chile and Argentina. The migratory stages of the Northern Hemisphere salmonoids appear to behave in a similar fashion. Kendall (1935: 11) noted that Salvelinus species in the far north are essentially marine species, enter- ing fresh water occasionally for food and reproduction. Further south, “the marine forms gradually disappear, becoming al- most or quite exclusively freshwater in- habitants at the southern end of each range. The presence of a marine migra- tory stage in both the salmonids and galaxiids may be connected with attempts to avoid the rigors of the extremely low temperatures and icing of freshwater habi- tats. Egg number and egg size show good correlation with the alternative life history patterns. The species with marine life history stages have more numerous eggs, usually several thousand. High fecundity is perhaps related to high mortalities of larvae in the marine plankton and further mortality due to loss by dispersal in the sea and the hazards of making a migration from the sea into fresh water. Also related to increase in egg number, and the fact that the larvae enter the highly productive marine plankton, are the comparatively small eggs, with little yolk. The non- lacustrine species which are restricted to fresh water, in which mortality during the juvenile stages may be lower, and which don't suffer from dispersal by ocean cur- ‘ents (although they are carried down- tream by river currents), have much fewer eggs, usually numbering only a few hund- Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 red. Increased survival of the fewer eggs is favored by larger size; this may be advantageous in rapidly flowing water in giving the freshly-hatched larvae a better chance of resisting downstream dispersal and also to compensate for any paucity of food available to very tiny fishes in rapidly flowing water. Greater yolk volume is advantageous because the larvae can de- pend on this for food for a longer period, and also because the larvae, hatching into swiftly flowing water, are larger and thus better fitted to maintain their position in the stream and find shelter from the cur- rents. Subjective observations indicate that the same relationship between egg size and egg number exists in the New Zealand Eleotridae. The species with fluviatile adults but marine or lacustrine juveniles— Gobiomorphus huttoni (Ogilby), G. bas- alis (Gray), G. gobioides (Valenciennes ), and Philypnodon hubbsi (Stokell)—have much smaller and more numerous eggs than the single species which lives and breeds entirely in flowing, fresh water— P. breviceps Stokell. This suggests that for diadromous spe- cies, high larval mortality due to predation and dispersal plus high food availability in a productive marine plankton have favored selection for numerous, small eggs. In contrast, in the non-migratory fluviatile species, somewhat lower larval losses due to less predation and dispersal, and greater ability to resist dispersal as a result of in- creased larval size at hatching, have fa- vored selection for fewer, larger eggs, which carry more nutritive yolk. There is a very clear relationship between life history patterns and range in the New Zealand Galaxiidae. It was previously pointed out (McDowall, 1966a: 16) that the fact that “certain species of Galaxias have marine whitebait is reflected in their New Zealand distribution.” The present study, in altering the taxonomy of some problematical species, has further strength- ened this observation. Reference to the TABLE 7. DISTRIBUTION OF NEW ZEALAND GAL- AXIIDAE. Whitebait Juvenile North Island South of South Island Offshore North of South Island North Island West Coast East Coast Islands Localized argenteus . fasciatus postvuectis brevipinnis maculatus AK AM Xs ARK KK A KAKA KM A An! KM | py bd OM OM diversus - divergens - apoda - vulgaris - prognathus - paucispondylus — —-— — . burrowsius - - = Le K MK KKK | 1 xr i | ms 4 pS Pd I Ann AK | | | gracilis - - = . usitatus - - = AA ZAARAZAZ ARAAA X = Present, widespread. x = Present, restricted. — = Not present. species distribution maps (Figs. 7, 10, 20, 92, 25, 33) shows that the widespread species, and those which are found on the offshore islands, are the species which also have marine whitebait stages. G. macu- latus, G. brevipinnis, G. fasciatus, G. argen- teus, and G. postvectis have all been found over large areas, where suitable habitats have been searched, and various of these species have been recorded from the Cavalli, Chicken, Little Barrier, Mercury, Kapiti, Arapawa, D’Urville, Stewart, Chat- ham, Auckland, and Campbell Islands. G. maculatus is present in Australia, Tasmania, Lord Howe Island, New Zealand, and the Chatham Islands, Chile, Patagonia, and the Falkland Islands. There can be no ques- tion that the wide range of this species is entirely a result of the existence of the marine whitebait stage. In marked contrast, the non-migratory species have a much more restricted range. The two main islands of New Zealand are divisible into four somewhat dubious, but presently useful, faunal regions. The North New ZEALAND GALAXHDAE * McDowall Al] Island is divided centrally by the volcanic plateau, with large areas north and south of it. The areas east and west of the South- ern Alps-Kaikoura Mountains mountain chain are separated by these high moun- tains. In Table 7, the New Zealand Gal- axiidae are listed and their presence or absence in each of the four areas noted. This table illustrates dramatically the dif- ference between the diadromous and non- migratory species in breadth of range. It also shows that the species in the latter group are restricted to one or two, occas- ionally three, of the so-called faunal regions. Two species have very localized distri- butions, occurring in only one water body. Five species occur solely in one area or extend marginally into a second; e.g., G. prognathus and G. vulgaris are found primarily to the east of the Southern Alps, but both species appear to have crossed the divide once, in the vicinity of the Lewis Pass, and have entered the upper reaches of the Buller River System, flowing to the west (see p. 424). Two species occur widely in two faunal regions and one of these, G. divergens, is marginally present in a third. It should be noted that both these species are pres- ent primarily in the south of the North Island and the west of the South Island. Consideration of the range of some other animals, e.g., Paranephrops planifrons, the fresh water crayfish, suggests that the Buller-Nelson-West Coast area of the South Island has close biogeographic affin- ities with the south of the North Island. Fleming (1962) showed that for most of the Tertiary the two islands of New Zea- land were connected across Cook Strait, and his map for the Pleistocene shows that the river systems from the Wellington area in the North Island and the northwest of the South Island were confluent in the now submerged area. The distribution patterns of G. divergens and N. apoda are thus easily explained in terms of land bridging of Cook Strait and indicate that these two areas have close faunal affinities. 412 Data on species range make it quite clear that the presence of a marine stage has resulted in a broad geographical range. This same factor, the presence of a marine stage in the life history, has also imposed some restriction on inland penetration and altitudinal range. Species that migrate up rivers from the sea are excluded from areas that the migratory juveniles are unable to reach in their migration. The extent of restriction in altitudinal range varies from species to species and is, in part, related to the climbing ability of each species. Climb- ing ability is, in turn, related to certain morphological adaptations (see “Adaptive radiation,’ p. 414). G. maculatus appears to be completely confined to lowland and coastal streams and rivers, and has been found to be virtually incapable of climbing a low artificial weir about six feet high across the Waikanae River (McDowall, 1964b: 145). Other diadromous species present in the river (G. fasciatus, G. post- vectis, and G. brevipinnis) were found above the weir and were thus obviously climbing it. G. brevipinnis whitebait have also been observed climbing up the vertical concrete face at one end of an aqueduct beneath a road (K. F. Maynard, pers. comm.). G. brevipinnis, G. argenteus, G. fasciatus, and G. postvectis juveniles have been found to be extremely troublesome to keep in captivity because of their pro- pensity for climbing out of aquaria. This ability is important to species migrating upstream from the sea, especially in a mountainous country like New Zealand; amongst the diadromous species, those which penetrate far inland, especially G. brevipinnis, are also good climbers. The origin and age of the New Zealand galaxiid fauna It is now fairly generally believed that the Galaxiidae, together with the Retro- pinnidae and Aplochitonidae, are deriva- s of the primarily northern salmonoid tishes. The osteology of these three south- fami suggests that they are probably Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 derived from some very early osmeroid stock.! The three families are restricted to the Southern Temperate Zone, except for a single, more tropical species in New Cale- donia (this species is found only in cooler, mountain lakes). The northern salmonoid families which seem to have given rise to the southern families are found only in the Northern Temperate and sub-Arctic. Thus, this quite large and diverse group of fish families is confined to the temperate zones, exhibiting bipolarity at the subordinal level. They are cold water fishes and seem to be excluded from more tropical areas by temperature. If this is so, it is difficult to imagine how the southern families could have been derived from the northern fami- lies unless either there was a period in the early Tertiary when tempratures were con- siderably cooler than now, or southward dispersal took place by tropical sub- mergence. Until the Pliocene cooling which led to the glacial periods, temperatures in the Tertiary are thought to have been warmer than now. Fleming (1962: 66) suggested that New Zealand was warmer than now during the late Mesozoic and remained warm well into the Tertiary. If, as suggested, the salmonoid fishes are limited in range by temperature, tropical submergence seems to be the most likely means of traversing the tropics. Hubbs (1953: 325) conjectured that the three southern families are “pre-Tertiary relicts of groups that failed to persist in the tropics.” The basis for Hubbs’s conclusion is not clear. Certainly none of the living generalized salmonoids are tropical, or even subtropical. Existing temperature limitations and inferred temperatures since the end of the Mesozoic suggest that occur- rence of salmonoids in the tropics is un- likely at any time since the radiation that produced the southern salmonoid families occurred. ' The evidence on which this supposition is based is discussed elsewhere (McDowall, 1969; see also Weitzman, 1967). | Where the southern families entered the Southern Hemisphere is also not very clear. The existing predominance of the Gal- axiidae, as well as the Retropinnidae, in the Australasian region suggests that dis- persal occurred across the tropics in south- eastern Asia. More than 20 galaxiids are recorded from Australia, 14 from New Zealand, 4-5 from South America, but only one from South Africa. These details sug- gest that South Africa is a most unlikely source of origin and dispersal, and that it is, rather, the end of a chain of dispersal areas, under the influence of the west wind drift. If South America were taken as the center of dispersal, and if we use existing currents (see Fell, 1967), then the Aus- tralasian fauna must then have been de- rived by dispersal eastwards via South Africa, and again, the presence of only one species there suggests the improbability of this having taken place. G. maculatus is present in Australia, New Zealand and South America, but not South Africa, and this suggests that it originated in Australia and spread eastwards. For such reasons, I think that Australia is the most likely area for the origin and dispersal of the family Galaxiidae, and that the New Zealand fauna was derived from the Australian one. How the galaxiid fishes have dispersed is fairly clear, although the wide geographi- cal range of the Galaxiidae has puzzled many ichthyologists and zoogeographers. As recently as 1950, Stokell discussed means by which the family could have dispersed by land routes. Long ago, how- ever, Boulenger (1902: 84), noting that “most text books and papers discussing geographical distribution have made much of the range of a genus of small fishes, somewhat resembling trout, the Galaxias,” pointed out that some species are not re- stricted to fresh water. I have earlier (Mc- Dowall, 1964a) analyzed the derivation of the New Zealand freshwater fish fauna, and at that time I concluded that since all the families in the fauna (including the Galaxiidae) contain species that have New ZEALAND GALAXIIDAE * McDowall 413 marine stages, their presence in New Zea- land is simply and clearly explained by trans-oceanic dispersal. The existence of at least seven diadromous species of Galaxias (two in Australia, four in New Zealand, and one further species in both) strongly supports the hypothesis of oceanic dispersal, and the range of one of these species—G. maculatus—suggests that this dispersal continues (McDowall, 1966a). It appears that the East Australian current, which travels down the east coast of Aus- tralia and which may impinge on much of the west coast of New Zealand, supplies a suitable mechanism for dispersal from Aus- tralia to New Zealand (see Fell, 1962, 1967). However, it is uncertain how much this current affects the New Zealand region. Burling (1961: 51) suggested that the Tasman current, which flows on to the west coast of New Zealand, is derived from the East Australian current as it turns eastward towards the central Tasman Sea. The similarities between the marine faunas of southeast Australia and New Zealand (see Moreland, 1958, for fishes, and Fell, 1967, for evidence of recent dispersals) support the probability of water transport in this manner. However, Dr. B. V. Hamon (pers. comm.) suggested that eddies that break off from the East Australian current may be a mechanism by which water is carried across the southern Tasman Sea, but that at present there is no evidence to suggest that these eddies persist over such distances, and it is not known how long it would take such eddies to travel across. Fleming (1962: 105) concluded that “during the early Cretaceous, the geanti- cline west of the New Zealand geosyncline could have extended north to New Cale- donia,” but “at no time is there any evi- dence for direct trans-Tasman connection with Australia.”. There seems little likeli- hood of a suitable land bridge between New Zealand and Australia. In 1963 (p. 382) Fleming pointed out that if “the podocarps, Sphenodon, the frog Leiopelma, 414 many invertebrates, Nothofagus, and per- haps the ratite birds walked into New Zealand ... . across an Antarctic land bridge from South America lasting at least into the middle Cretaceous, we are left with problems almost as great as those solved. What kept the land dinosaurs, the early mammals and snakes from New Zealand?” On this basis, he gave “wavering support to the view that the dispersal of the Paleoaustral organisms, like that of the Neoaustral element, was across the sea.” The whole idea of land-bridge dispersal of the Galaxiidae to New Zealand must be discarded; or looking at the problem from a different perspective, the present range of galaxiid fish gives absolutely no support for any land connection between New Zealand and any other land area at any time. The age of the New Zealand galaxiid fauna is uncertain. The comparative sizes of the families of freshwater fishes in the New Zealand fauna (and none of them looks like a relict )—Geotridae, one species; Galaxiidae, 14 species; Retropinnidae, six species; Aplochitonidae, one species; An- guillidae, two species; Eleotridae, six species; Cheimarrichthyidae, one species —suggest that the Galaxiidae may form the oldest existing element in the freshwater fish fauna. The fossil record is of little help in dating the fauna, since the only galaxiid fossils are from the Upper Pliocene of Frasers Gully, Kaikorai near Dunedin. No other fossils of any freshwater fishes are described from New Zealand, and no galaxiids are recorded from other regions. Romer (1966: 356) listed a Galaxias from the Oligocene of New Zealand, but the original record of this fossil is unknown to me in any galaxiid literature. The Oligo- cene of New Zealand was a period of extreme marine transgression, and there was very little emergent land (see Fleming, 1962: 74); freshwater deposits would be very limited in extent, if present. The fossil record thus provides little information on the age of the family, or Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 on how long it has existed in New Zealand. The number of species in New Zealand, and their distribution pattern, suggests initial arrival in the early or mid-Tertiary, certainly considerably before the period of the Fraser’s Gully deposit. Romer (1966: 355) recorded salmonoid fishes back to the Upper Cretaceous, and one existing genus—Thymallus—in the Eocene. Osme- roid fishes from which the southern families seem to be derived are dated back to the Miocene. The prospect that the salmonoid fishes were present in the Australasian region as long ago as the beginning of the Tertiary must be recognized. Adaptive radiation Since the family Galaxiidae is by far the largest family of freshwater fishes in New Zealand, it probably represents an old ele- ment in the fauna. However, to what extent the present fauna represents the fauna of the Tertiary is not clear, and it may be that as a result of changes in land form, marine transgressions, and glaciations dur- ing and after the Tertiary, the present fauna is only a surviving remnant of a more extensive fauna. But there are no apparent relicts in the fauna. If the neo- channoid species represent a_ distinct radiation, independent of the Tasmanian mudfishes, they would seem to constitute an old element in the fauna, but the dis- tribution of the three New Zealand neo- channoid species seems to be a result of obvious geological events, like the rise of | the South Island mountain chains and the | central North Island volcanic plateau, or the marine transgressions of the earlier Tertiary. All the New Zealand freshwater fishes fall into species groups (or are the sole representatives of their family in New Zealand). And those species that have no | close relatives in the New Zealand fauna | have such relatives in Australia. The re- lationships and distribution patterns of the New Zealand freshwater fishes give no If these are old species, they | give no indication of significant extinction. | j evidence for the occurrence of much ex- tinction, and most appear to be rather recent in origin. Apart from man-caused extinction, I see no reason to suspect that the New Zealand freshwater fish fauna, including the Galaxiidae, has ever been any more diverse or speciose than it is at present. The galaxiid fauna has been built up in two ways—by invasion from Australia and by speciation in New Zealand. As I discuss in the section “Species groups and phylog- eny’ (p. 418), I think that invasion has played an extremely important role in the development of the fauna, and I think that evolution of the fauna in the New Zealand region has been slow and rather ineffective in populating New Zealand’s fresh waters. There is little question that the fauna is depauperate. Comparison of New Zealand’s freshwater fish fauna, comprising about 31 species, with that of Japan, 127 species (Okada, 1960), or the British Isles, 72 species (Regan, 1911), suggests that the freshwater habitats of New Zealand are far from saturated and that there is con- siderable scope for invasion. (This is not meant to imply that potentiality exists for the introduction of game fishes, since the past results of such introductions to New Zealand indicate that they may have serious effects on the existing fauna (McDowall, 1968a ).) It is not surprising that the fauna is depauperate. New Zealand has been geographically isolated from other land since at least the end of the Mesozoic (Fleming, 1962: 105), and this has com- pletely prevented the invasion of New Zealand by primary or secondary fresh- water fishes. What is interesting is the failure of the species arriving to radiate in the semi-vacant and highly productive freshwater habitats of New Zealand. I find it surprising that so much of the fauna is adventive and not a result of diversification of stocks in the New Zealand region. There are no herbivores in the New Zealand freshwater fish fauna. This is almost certainly a result of the scarcity of New ZEALAND GALAXxUDAE * McDowall 415 aquatic vegetation, which is itself at least partly related to the mountainous char- acter of many New Zealand rivers. Further- more, none of the species can be classed as piscivorous, data showing that some species may occasionally take fish, but that it is not customary (McDowall, 1965b, 1968b, Hopkins, 1965). Stream inverte- brates make up the bulk of the food of all species, except for very large examples of the long finned eel, Anguilla dieffenbachii (Cairns, 1942: 139). Galaxiids have invaded a wide variety of habitat types. Rapid-water species tend to be more slender in form, but are, in their general characters, similar to slack-water species. A few, e.g., G. brevipinnis, G. postvectis, have a receding lower jaw, which suggests adaptation to feeding off the stream bottom. Conversely, G. prog- nathus has a conspicuously protruding lower jaw, and this is a very obvious adaptation for picking invertebrates off the under sides of rocks. There is some vari- ation in dentition, but only in the degree of development of lateral canine teeth in the jaws and of mesopterygoidal teeth in the roof of the mouth. Most species are solitary and secretive. They are usually found in heavy cover and have great thickening of the fins, which provides resistance to fin damage when the fish swim amongst rocks and _ logs; these species also almost invariably have emarginate to truncated caudal fins. The pectoral fins are usually placed low latero- ventrally on the trunk, a modification prob- ably related to bottom resting and feeding habits. G. brevipinnis, in which this is most pronounced, also has deep corru- gations on the lower surfaces of the pec- toral fins, which may help the fish to grip the stream bottom and maintain its position in the very rapid flow of its habitat. In contrast, the few species that shoal and live openly in pools have more membranous fins, the pectorals are positioned much higher laterally on the trunk, and the caudal has a definite fork. It seems justi- 416 fiable to interpret these characters as being related to the shoaling or midwater habits. The functional nature of the differences is perhaps least obvious in the caudal fin, although in shoaling groups, e.g., clupeids, the caudal fin is invariably forked, whilst in secretive, cover-dwelling groups, e.g., umbrids, the caudal fin tends to be truncated or rounded. Furthermore, the juveniles of the diadromous species are shoaling in habit and have forked caudal fins, the fins only becoming emarginate as the fishes mature and become solitary and seek cover. A further character related to shoaling habits, and also characteristic of shoaling fishes in general, is silvery color- ation on the abdomen, present in the New Zealand Galaxiidae only in G. maculatus and its shoaling derivatives. The low position of the thick, fleshy pec- toral fins has served as a preadaptation for climbing. The diadromous species, which have the pectoral fins inserted low on the lateroventral trunk are much better climb- ers than G. maculatus, in which the fins are higher. The fact that the pectoral and pelvic fins are about flush with the ventral trunk and have the lamina facing down- wards seems very important in the great climbing ability of species like G. brevi- pinnis, G. fasciatus, and G. postvectis. The differences between the diadromous and freshwater species in egg size and number is discussed in the section on life history patterns (p. 409). A further dif- ference between these groups of species is the length of the gill rakers. The five diad- romous species and two lacustrine species (G. usitatus and G. gracilis) have long, well-developed gill rakers, whereas most other species have much shorter, sometimes irregularly-spaced rakers. The tendency towards gill raker reduction is especially evident in the slender alpine species, G. divergens and G. prognathus. These dif- ferences correlate with the existence of a shoaling, plankton-feeding phase in those with long rakers and the lack of . phase in the fluviatile species. Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 One of the most distinctive character divergences in the New Zealand Galaxiidae is the probably neotenic retention of shoal- ing habits and the associated morpho- logical peculiarities seen.in G. maculatus. It is notable, though, that this apparently did not originate in the New Zealand region, but its presence in New Zealand represents an independent invasion from Australia. The three neochannoid species have not hitherto been discussed in relationship to the radiation of the family. This is largely because their adaptations represent a radiation into the most peculiar niche oc- cupied by members of the family in New Zealand, though this group, too, might have originated in Australia and dispersed to New Zealand. If not, these species rep- resent the most striking example of inven- tiveness in the New Zealand galaxiid fauna. There are three species that occur in swamps, swampy creeks, and sometimes springs, and that have an ability to aesti- vate during droughts. Common morpho- logical characters, like loss or reduction of the pelvic fins, reduced eyes, long nostrils, low median fins that are more or less con- fluent with the caudal fin, rounded caudal fin, are interpreted as modifications for a semi-burrowing existence. The limited extent of the radiation of the New Zealand Galaxiidae suggests that they are a conservative group. In this they are similar to their Northern Hemisphere rel- atives—the Salmonidae and Osmeridae— which are principally cold water predators of rather uniform morphology, although their habits may vary. The differences that _ have developed in the Galaxiidae are | usually variations on the theme of solitary, | secretive invertebrate predators. The most | persistent variation involves adaptations to | different water types—placid pools in stable bush streams to rushing torrents in | rivers—and_ the | persistence or loss of the marine whitebait | juvenile stage. In the case of habitat dif- | open, unstable shingle ferences, morphology shows little variation —usually some elongation and depression in rapid-water species. The only obvious differences associated with the presence or absence of a whitebait stage are those related to survival of the progeny (egg size and number) and feeding (length of gill rakers ). These characters are those likely to be most affected by selection, since the low saturation of the New Zealand fresh- water habitat leads to little interspecific competition—selection is mostly intraspe- cific and related to feeding and reproduc- tive efficiency. The conservativeness of the group is seen in its failure to become open living and benthic, like the eleotrids, which are almost certainly more recent invaders. They have not become mid-water pool-living species like the species of Retropinna, but instead, when found in pools, generally skulk in marginal cover. They have mostly been unable to remain in streams that have lost their forest cover, and except in localities that may have been naturally alpine grass- land in Central Otago and alpine Canter- bury, streams without cover are mostly populated by eleotrids, eels, and sometimes Retropinna. Nor have they been particu- larly successful as lacustrine species. G. argenteus and G. fasciatus are known as adults from very few lakes; G. brevipinnis, although commonly found in lakes in the juvenile stages, is rarely lacustrine as an adult. The exception to all these is G. macu- latus and its derivatives. They are shoaling, open-living, pool-dwelling species, which have persisted in streams without cover and have become partly lacustrine. But apart from the landlocking in G. usitatus and G. gracilis, these characteristics do not represent evolution in the New Zealand region but are a product of dispersal of G. maculatus from Australia. The dominat- ing impression acquired from studying the New Zealand Galaxiidae is that they have adapted their basic structure and way of life to a variety of water conditions, and New ZEALAND GALAXIIDAE ° McDowall 417 have pursued their invertebrate-feeding habits in these. Analysis of sympatry in the family in New Zealand shows that there is likely to be little interspecific competition. G. usi- tatus, G. gracilis, Neochanna diversus, and N. apoda have never been collected from the same water bodies as other galaxiids. This is also true for most of the range of G. divergens, but this species is known to be sympatric with G. prognathus, G. vul- garis, and G. brevipinnis in two localities. G. vulgaris is found in some waters with G. brevipinnis below some of the South Island alpine lakes and with G. pauci- spondylus, G. prognathus, and N. burrow- sius in a few Canterbury rivers. The species that exhibit broad sympatry are the diadromous species—G. maculatus, G. brevipinnis, G. argenteus, G. fasciatus, and G. postvectis—plus G. prognathus and G. paucispondylus, both of which are known from very few rivers, and occur together and with G. vulgaris, G. brevipin- nis, or G. divergens. There is a likelihood here of competitive species interactions, and in these species there are some signs of significant niche differences; this is a profitable area for further study. The sympatry of the diadromous species is interesting in that these species appear to be a product of successive invasions of the New Zealand region. All five species probably occur together in mixed associ- ations as juveniles in the sea, and they migrate upstream from the sea together (McDowall, 1966a ). It is unusual, however, for the adults to be taken from precisely the same habitat type, even though several species may be present in the same pool- rapid series in just a few yards of stream. The habitats of G. fasciatus and G. argen- teus are especially similar, but the common association of these species in nature sug- gests that some habitat divergence is likely. A detailed study of microdistribution in river systems where various of the species are sympatric may provide interesting data 418 on niche and the effects of sympatry on niche breadth. Species groups and phylogeny The New Zealand Galaxiidae are easily divided into a series of small, distinct, species groups. These vary in their com- pactness, but they combine species that appear phylogenetically more closely re- lated to each other than any species in one group is related to those in other groups. Although it is possible to list characters that unify the species groups, many of these are adaptive. However, the general form and habits of the species in these groups show such similarities that they seem to be valid. The species groups are as well defined by the gaps between species groups as by the similarities be- tween the contained species. There is a group of small, slender species that are somewhat compressed in form and have membranous fins, a long-based anal (Fig. 45, B, f, g, h), a forked caudal, and shoaling habits. They vary somewhat in their reproductive cycle, but otherwise form a compact group. G. maculatus (Fig. 21) is one of the more distinctive New Zealand Galaxias species, and probably stands far apart from other New Zealand species, except its immediate derivatives. The fact that G. maculatus is present in Australia, Tasmania, New Zealand, the Chatham Islands, South America, and the Falkland Islands has important implications in an analysis of the phylogeny of the family in New Zealand. The only way this species could have attained its present distribution is by trans-oceanic dispersal. Existing ocean currents indicate that G. maculatus probably originated in Australia and dispersed eastwards, in the west wind drift. If this is so, then G. maculatus has no direct affinities with other New Zealand species groups. Since populations from these widely separated land areas are conspecific: (Stokell, 1966: 76, McDowall, 1967b) it seems that the original dispersal occurred quite recently. Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 Comparison of G. maculatus with G. usitatus and G. gracilis (see McDowall, 1967a: 3) shows that the latter two species, each of which is confined to one small lake, are landlocked derivatives of G. maculatus. Although the three species look very similar, there are marked differences in body proportions and meristic characters. The divergence which has taken place since G. usitatus and G. gracilis became landlocked indicates that the species are plastic in their general morphology. G. gracilis, for instance, appears to have traversed the full range of vertebral num- ber seen in the New Zealand Galaxiidae. G. maculatus represents a maximum, with 59-64, and G. gracilis a minimum for the family in New Zealand, with only 47-50. The argument that this is evidence that G. gracilis is not closely related to G. macu- latus conflicts with the evidence of their obvious similarity; and in view of the known relationship between vertebral number and developmental temperatures, basing affini- ties on vertebral number must be regarded as suspect. G. gracilis has diverged further from G. maculatus than has G. usitatus. This is in accord with the apparent ages of the small lakes in which these species are found. Lake Rototuna (G. gracilis) occurs in well-stabilized, rolling sand dunes, once covered with bush, and at an altitude of about 90 m. It is probably older than Lake Waiparera (G. usitatus), which occurs in still shifting sand dunes, no more than 36 m above sea level. Some workers, e.g., Whitley and Phillipps (1940: 229) and Phillipps (1940: 39), | have suggested that G. castlae Whitley | and Phillipps (=G. brevipinnis Gimnther) and G. paucispondylus are also landlocked derivatives of G. maculatus. Present under- standing of the nature of the fauna suggests | that this is most unlikely, and I think that | G. maculatus and its two derivatives belong | in a quite distinct species group. | A second species group includes three | very similar large, stout-bodied species, | New ZEALAND GALAXxUDAE * McDowall 9 8) 10 14 18 22 BODY DEPTH AT VENT CoANDARD LENGTH ° 50 60 70 80 90 100 BASAL LENGTH ANAL FIN AXIMUM LENGTH % 10 14 18 22 26 EMEMEVAMEIER eu ea Figure 45. Species groups in the New Zealand Galaxiidae, as indicated by body proportions. a, Galaxias argenteus; b, G. fasciatus; c, G. postvectis; d, G. brevipinnis; e, G. vulgaris; f, G. maculatus; g, G. gracilis; h, G. usitatus; i, G. paucispond- ylus; j, G. prognathus; k, G. divergens; |, Neochanna apoda; m, N. diversus; n, N. burrowsius. 420 which are mostly lowland dwelling. These species, G. argenteus, (Fig. 4), G. fasci- atus (Fig. 6), and G. postvectis (Fig. 9), form a compact group with their long, truncated to emarginate caudal fins, ex- pansive dorsal and anal fins, and very long pectoral and pelvic fins, deep bodies (Fig. 45A, a, b, c), short and very deep caudal peduncles, and rather long, broad heads. The accessory lateral line is developed, and in all three, a large, dark, blue-black blotch above and behind the pectoral fin base is present. They all spawn in the autumn, are diadromous, and are found in rather similar habitats. The three species have almost identical ranges, and_ their characters strongly suggest common an- cestry. But explaining their divergence within the New Zealand region is a prob- lem for which I see no solution. It is diffi- cult to imagine how allopatry could have developed, since all three species occur together in a common marine pool, migrate into fresh water in mixed species shoals, and are found together in the adult habi- tats. It is no simpler to see how speciation could have occurred had all three species remained sympatric. Even if a satisfactory model for sympatric speciation were to be constructed, this would necessarily involve habitat, reproductive, or behavioral dif- ferences of a nature and degree that seem to be lacking in these species. The only alternative is that these fishes invaded New Zealand from Australia several times, giv- ing rise to the three stout-bodied species. Their great predominance on the west coast of New Zealand (see Fig. 7) suggests that they have evolved in the swift, rocky bush streams there, rather than in the very different plains streams of the southeast, where they are largely absent. In the Australian galaxiid fauna, G. trut- taceus Valenciennes shares many of the morphological characteristics of the stout species and is a suitable ancestral type; as it has a marine whitebait juvenile (Lynch, 1965), it also has the necessary dispersal ability. Since the larval stages of the three Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 New Zealand stout-bodied species are be- lieved to spend many months in the sea, giving them plenty of time to disperse from Australia, and since the East Australian- Tasman currents may be favorable for such dispersal, multiple dispersal of fishes an- cestral to the stout species is quite com- prehensible. Castle (1963: 13) suggested that the New Zealand freshwater eels (Anguilla spp.) probably spawn some- where in the tropical Pacific, northeast of the New Hebrides; it would thus be the action of these currents which bring the leptocephali to New Zealand coasts—and they come in colossal numbers. The effi- cacy of these currents in dispersing fishes to New Zealand from the northeast seems demonstrated. Skrzynski (1967) and others have re- ported the following New Zealand species from the Chatham Islands, 420 miles east of New Zealand: G. fasciatus, G. argenteus, G. brevipinnis, and G. maculatus, as well as Gobiomorphus huttoni (Ogilby), Retro- pinna retropinna (Richardson), Anguilla australis schmidtii Phillipps, A. dieffen- bachii Gray, and Geotria australis Gray. Thus, unless these fishes have been on the islands since they were connected to main- land New Zealand, there must have been a great deal of dispersal to the islands from New Zealand, demonstrating considerable dispersal powers. If Galaxias argenteus, G. fasciatus, and G. postvectis have each reached New Zealand from Australia, with a common ancestry there, these dispersals must have been at sufficiently separated intervals to permit reproductive isolation to develop between successive invasions. In compari- son with the ideas of earlier workers on galaxiid fishes, that they must have reached New Zealand by means of land _ bridges, Gondwanaland, continental drift, etc., the idea of dispersal on several occasions is radical. But it is clear that there was a separate dispersal to bring G. maculatus; perhaps it is necessary to envisage inde- pendent dispersals for each of the three stout-bodied species. If so, then the minor habitat and morphological differences be- tween the three species are likely to be a product of interspecific competition forc- ing the three rather similar, sympatric species to specialize in some way. There is something inherently dissatisfying about the hypothesis of multiple invasions with- out any radiation of the group in New Zealand, but the formulation of a model allowing their speciation in New Zealand, either sympatrically or involving the de- velopment of geographic or other barriers to gene flow, eludes me. Looking at the entire New Zealand fish fauna, it appears that it is almost entirely an invasion of fauna, with little speciation in the New Zealand region. Moreland (1958) has shown that only 30 per cent of the fauna is endemic. Since 7 per cent of the fauna is freshwater, and about 94 per cent of the freshwater fishes are en- demic, it follows that only about 24 per cent of the marine fauna is endemic. Analysis of the fauna from _ Phillipps (1927a) shows that about 100 families, 200 genera, and 318 species occur in the fauna; thus there are two genera per family, and one and one half species per genus. Nearly one half of the genera have only one New Zealand species. Both the low endemism and the structure of the fauna suggest that the fauna is derived more by invasion from outside than from evolution within the New Zealand region (there is only one doubtful endemic marine fish family), and it is necessary to postulate fishes crossing the Tasman Sea a great number of times. Since, according to Moreland (1958: 28) 31 per cent of the species are presently shared with southeast Australia, recent trans-Tasman crossings must number in the vicinity of 100. The Galaxiidae, with their long-lasting oceanic, pelagic stages are admirably fitted for this dispersal, which may have occurred several times. G. brevipinnis has many characteristics also found in the stout-bodied species, but it is much more slender (Fig. 45A, d). In New ZEALAND GALAXIIDAE °* McDowall 42) addition to being widespread in New Zea- land, it is present on the Chatham Islands, 420 miles east of New Zealand, the Auck- land Islands, 290 miles south of New Zea- land, and Campbell Island, 150 miles southeast of the Auckland Islands. It has a marine whitebait and thus has potentiality for transoceanic dispersal, and its range in the New Zealand region shows that it is probably a more effective disperser than any other species of Galaxias, except G. maculatus. As indicated in discussing the taxonomy of G. brevipinnis, I think that G. weedoni should become a synonym; if these two species are not conspecific, they are certainly very closely related, and G. brevipinnis is derived from G. weedoni. Assuming that G. brevipinnis originated in the Australian region, it is necessary to explain its present distribution. As with the other species, it must have arrived in the New Zealand region in the East Austra- lian and Tasman currents. Dispersal of fishes from New Zealand to the Chatham Islands does not appear too difficult, judg- ing from the number of fishes that have succeeded in doing so. The difficult dis- persal seems to be that to the sub-Antarctic islands—Campbell Island and the Auck- land Islands. Dispersal from New Zealand is opposed by the west wind drift, which flows from west to east, south of New Zea- land, with a substantial northern displace- ment, i.e., the Ekman drift (Burling, 1961, see his chart 1), which flows northeast from the sub-Antarctic islands, towards the south and east of New Zealand. It is pos- sible that G. brevipinnis represents a direct dispersal from Australia to the islands; such a dispersal could have occurred if fishes in the East Australian current were picked up by the deeper west-east flowing west wind drift. The existence of a some- what more extensive land area to the south of New Zealand, perhaps until late in the Tertiary, makes dispersal directly to the southern islands more plausible. If G. brevipinnis was present in the sub- Antarctic islands prior to the Pleistocene 422 glaciations, its survival there during the glaciations becomes a critical question. If it became exterminated by the glacial ice cap, judging by existing current patterns, re-invasion of the islands from New Zealand would have been difficult, if possible. Gres- sitt (1964b: 11) reported that Campbell Island (the more southern of the two island groups) was “mildly glaciated during the Pleistocene .. . . there are cirques as well as glacial valleys but no proof of a con- tinuous ice sheet.” Gressitt (1964a: 548) thought that the fauna represented a de- pauperated fragment of an ancient subcon- tinental fauna, with over-sea colonization. Illies (1964: 215) examined the Plecoptera of Campbell Island, finding that “the exist- ence of running freshwater environments on the island must have lasted at least since the break-down of land connections,” i.e., for the present discussion, through the glacia- tions. This being the case, G. brevipinnis could probably have survived there too. G. vulgaris is very similar in general form and appearance to G. brevipinnis. None of its body proportions are greatly differ- ent, although it is a little less depressed in the head region and the jaws are nearer equal in length. In the area where the two species are sympatric, it has fewer verte- brae, fewer rays in some fins, and shorter gill rakers, but these are all minor differ- ences. The most significant difference is in the life history: G. vulgaris spawns in the spring and has no marine whitebait stage. This might seem a major difference, which precludes relationship between G. vulgaris and G. brevipinnis; yet Pollard (1966: 14) found that an apparently re- cent, landlocked derivative of G. maculatus in Victoria, Australia, has changed from autumn to spring spawning and from downstream to upstream migration prior to spawning. And this life history modifi- cation has been accompanied by very little morphological differentiation. G. vulgaris is virtually restricted to the eastern side of the Southern Alps in the South Island, but it is very widespread in Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 this area. This suggests that it must have evolved since the rise of the Southern Alps in the late Miocene and Pliocene. Its similarity to G. brevipinnis suggests that it is a derivative of this species, but how it diverged is unclear. G. brevipinnis and G. vulgaris are presently widely sympatric, since the former occurs in many lakes above the rivers in which the latter is found. The lacustrine populations of G. brevipinnis almost certainly post-date the Pleistocene glaciation. Apart from these lake populations, G. brevipinnis is very rare in the east of the South Island, and is prob- ably sufficiently rare for isolation of some population(s) in one or several of the river basins which lack lacustrine popu- lations of G. brevipinnis in the upper reaches. Such a happening may have al- lowed divergence of populations leading to G. vulgaris. Isolation may have been pos- sible in some Canterbury rivers that do not flow directly into the sea, but disappear into coastal gravel, but this seems unlikely to have been a long-term condition in any particular river, such as would allow speci- ation. There is a compact group of three small, slender, alpine species. They are confined to fresh, flowing water and do not have an obvious juvenile-adult metamorphosis. They are widespread in the southern half of the North Island and the northwest of the South Island, and extend over the Southern Alps into alpine Canterbury. This group comprises G. divergens, G. paucispondylus, and G. prognathus. Body proportions like slender trunk, very long and slender caudal peduncle, anterior dorsal fin insertion, short head with broad, shallow gape, short anal fin, bring these species together. Meristics are generally low. Of particular interest is the number of pelvic rays. G. divergens consistently has six rays, and al- though the two other species usually have seven, a high proportion of G. paucis- pondylus examined had six rays. In both G. prognathus and G. paucispondylus it was found that when there were seven rays present, the ray in each fin closest to the ventral midline was commonly much reduced and often unbranched. This is interpreted as illustrating the trend towards reduction in the number of pelvic fin rays, a process that has be- come absolute at six rays in G. divergens. In these species, the anteriormost of the interorbital pores on the head has migrated to a position very close to the posterior nostril. In G. divergens, and sometimes in the other two species, the opening of this pore has become confluent with that of the nostril. This is a trend not seen in any other New Zealand species; the pore and nostril are always well separated. A further significant character that unites these spe- cies is the loss of the postcleithrum in the pectoral girdle. Occasionally there is a barely staining splint, but usually the postcleithrum does not appear in alizarin- stained adult specimens. It is present, though, in all other New Zealand galaxiids. In addition, the supraethmoid has a char- acteristic irregular shape, the palatine lacks the process that in other galaxiids runs along the side of the face lateral to the mesopteryoid, the mesopterygoid and meta- pterygoid do not meet in the posterior corner of the orbit but are separated by a band of cartilage, and there are no proc- esses on the basioccipital. The species in this group exhibit some sympatry, G. di- vergens and G. paucispondylus being completely allopatric, but G. prognathus more or less bridging the geographic break between the other two. There is little question that these species have a common ancestry, probably in a species similar to G. paucispondylus or G. divergens, which are more generalized than G. prognathus. Our present knowledge of their ranges may not be sufficiently com- plete to allow valid conclusions about their evolution. In particular, the two disjunct localities known for G. prognathus strongly suggest that further localities remain to be discovered in the intervening area, especi- ally since the rivers there are superficially New ZEALAND GALAXUDAE * McDowall 423 similar to those in which G. prognathus has already been found. The present range of this species group suggests that the ancestral form had spread over the Wellington-Nelson-Buller region during or since the uplift of the Southern Alps. Their preference for cold alpine streams indicates that they may have evolved during the cold Pliocene-Pleisto- cene periods, and Fleming (1962: 81) con- sidered that the “climax of the Kaikoura orogeny occurred during the Pliocene. The existence of a single population (of G. divergens) in the east of the Volcanic Plateau far north of other known popu- lations is a problem. Fleming showed that the marine transgression of the Miocene persisted into the Pliocene, covering the southern half of the North Island, from Taranaki to northern Hawke’s Bay, except for the Wellington area, which was emer- gent and connected with the Nelson-Buller district. The northern population is con- specific with G. divergens, in which the pelvic fin ray number is only six, suggest- ing that this species must have spread north as the southern half of the North Island emerged towards and during the Pleistocene. It seems unlikely that this disjunction is a result of any earlier geologi- cal event, like the Miocene transgression; it is probably due either to destruction of the populations in the intermediate area— inland Hawke’s Bay and northern Wai- rarapa—by the known, recent volcanic activity, or to incomplete knowledge of the range of the species. G. divergens occurs in the south of the Bay of Plenty; the Wellington Province on both sides of the mountain ranges, and the northwest of the South Island as far south as the Buller River System at Maruia. G. prognathus is presently known only from the Buller River System at Maruia and the Rakaia System over on the eastern side of the Southern Alps. G. paucispondylus oc- curs in several river systems along the east- ern side of the Alps. This distribution pattern gives the im- 424 pression of a single, widespread species whose range became fragmented by the development of geographical barriers. If the ancestor of these species was wide- spread in the Nelson-Buller-northeast Canterbury regions during the rise of the Southern Alps, it would have been divided into two isolated series of populations by the mountains. Such a division would have sufficed to allow the divergence that has taken place between G. paucispondylus and G. divergens. G. prognathus, in its remarkably elon- gated lower jaw, is, in some ways, the most specialized species of Galaxias in the New Zealand fauna. Its specialization suggests that it may have been in competition with another, similar species, like G. paucis- pondylus or G. divergens. It is almost cer- tainly derived from one of them and the most recent species of the three. G. vulgaris is distributed east of the ranges in the South Island, but it has in- vaded the west, in the upper Buller River System, in the Maruia River. This seems likely to have occurred since the rise of the Alps, for two reasons. First, the Maruia population shows very little divergence from those on the east, certainly less than that seen between various populations in the east; in other words, its derivation from the east seems to be very recent. In ad- dition, the fact that G. vulgaris is present only in one western river system, and the fact that this river drains the lowest exist- ing pass in the northern alps, suggests that G. vulgaris has crossed the alps from east to west, and recently. The derivation of G. prognathus from either G. divergens or G. paucispondylus seems most simply accounted for if we presume that the parental species of G. prognathus was able to cross the alps. This would have brought G. divergens and G. paucispondylus into sympatry. If they had only recently become reproductively isolated, competition is likely to have been intense, forcing one of the species, prob- ably the invader, to specialize, leading to Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 G. prognathus. Since G. prognathus is now recorded from both sides of the Alps, it must also have recrossed the divide, and if it did evolve in competition with either G. paucispondylus. or G. divergens, it is not very obvious on which side it is likely to have evolved. If it evolved on the east and has reinvaded the west, there is nice agreement with the distribution pat- tern of G. vulgaris, since both species are known in the west only from the Maruia River, in the upper reaches of the Buller River System, near the Lewis Pass. It appears too much of a coincidence that both these species are known in the west only from this one river, and that the rivers headwaters drain the most easily crossed pass in the Alps. Stream capture seems quite feasible in the Lewis Pass area, and I am advised (P. M. Johns, pers. comm.) that there is a swampy area in the pass between the headwaters of the rivers on each side. I think that both G. vulgaris and G. prognathus have quite re- cently crossed from east to west and that G. prognathus probably evolved in the east. If so, I suspect that G. prognathus will be found in further alpine Canterbury rivers, particularly the upper tributaries of the Waiau, Hurunui, and Waimakariri Rivers. An alternative hypothesis, which is feas- ible but without obvious support from physiographic changes, is that the two species in the east evolved in separate river basins and have subsequently become sympatric, as the rivers wandered back and forth across the Canterbury Plains and various pairs of rivers became confluent for a time. These river changes must be the mechanism behind the present broad range of G. vulgaris, but, inasmuch as they have not sufficed to produce speciation in G. vulgaris, it seems unlikely that isolation of any basin has been of sufficient duration to allow the contained populations to evolve reproductive isolation. The final species group comprises the three species of Neochanna—N. burrow- sius, N. apoda and N. diversus—which pre- sent a distinctive facies. They have an elongate and rounded trunk, somewhat blunt and little depressed head, small eye (Fig. 45C, 1, m, n), and a prominent de- velopment of the tubular anterior nostril, which may extend forwards beyond the upper lip. The dorsal and anal fins (Fig. 45B, 1, m, n) are rather long and low, with extremely thick, fleshy bases, the caudal fin is much rounded, and the peduncle flanges are very strongly developed, more or less confluent with the dorsal and anal fins. In N. apoda and N. diversus the pel- vic fins and girdle have disappeared, and there are usually no mesopterygoidal teeth. The process of reduction is less complete in N. burrowsius, which has very small pelvic fins with only four or five rays, and only a few, small mesopterygoidal teeth, or none at all. Stokell (1945: 132) discussed the question of whether these three species form a natural phylogenetic unit, or whether their similarities are a case of convergence. He concluded that “if bur- rowsius is to be regarded as indicating the line of descent of Neochanna from Ga- laxias, it might be expected that a domi- nantly four rayed form, or a form lacking ventral fins but retaining vestiges of the pelvic bones would exist to indicate a fur- ther stage in the process of degeneration.” As Stokell found, there is no such inter- mediate stage, and this is what students of evolution have commonly, if not normally, found. In this instance, it seems to indicate that a widespread species, ancestral to all three, had reached a stage of modification about equivalent to that exhibited by N. burrowsius. After the populations making up N. burrowsius were isolated, the modi- fication continued, producing the other species of Neochanna, again probably from a common stock. A cursory examination of the three species shows that they are very similar to each other, certainly far more similar to each other than any one is to any New Zealand species of Galaxias. It seems likely that the common ances- New ZEALAND GALAXIIDAE * McDowall 495 tor of the neochannoid species was wide- spread in New Zealand in the Miocene as the marine transgression began. From this time onwards, the transgressions and great orogenies that followed would have frag- mented its range. The populations now known as N. burrowsius would have been isolated from populations in the north and west by the rise of the main South Island mountain ranges in the late Miocene and Pliocene, and were perhaps restricted to the present small pocket in the Canterbury Province by the cold of the glaciations. Following this, the populations in the north and west lost the already reduced meso- pterygoidal teeth and pelvic fins. During the Pliocene, the marine transgression cov- ered most of the southern half of the North Island, and the transgression would have isolated populations—in the north of the North Island, leading to N. diversus, and in the Nelson-Buller-West Coast area, lead- ing to N. apoda. Land connections between the North and South Islands lasting into the Pleistocene would have allowed N. apoda to reach the southern North Island, if it was not already present. The preceding discussion suggests some probable species groups and their phylo- genetic relationships, and thereby estab- lishes possible terminal branching points in the phylogeny of the species (Fig. 46). It seems almost certain that three of these branching systems, those containing (1) G. maculatus and its derivatives, (2) the stout-bodied species, and (3) G. brevipin- nis and G. vulgaris, originated independ- ently in Australia; G. maculatus occurs in Australia, the stout-bodied species seem to be derived from G. truttaceus, and G. brevipinnis is derived from G. weedoni or is conspecific with it. There are two other very closely knit and distinctive species groups in the New Zealand fauna. There is no evidence to suggest where the slender, alpine species originated. The neochannoid species have counterparts in the Tasmanian fauna, pres- ently treated by Scott (1936: 160) as two 426 Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 _ ORIGIN ay > Galaxias postvectis x< + ee S—°Falaxss fasciatus » iS iy ay © | Sa5 Saas pat) fat) & c Galaxias argenteus ie) I @ O » ae > 4 TG a : ae eo <= SS Calletiag brewiotnmis Si = TMS SS (9°) = pat) oO = Q Fn O° ie : 6 = = Galaxias vulgaris *~) lon) LE ww a > rm a B , co = Galaxias maculatus sat) set) ——— 2 m = Galaxias usitatus » St fos wn Galaxias gracilis Galaxias paucispondylus aS) nN Galaxias divergens Galaxias prognathus Neochanna_burrowsius é Neochanna diversus el /eajsny SUMO}! |[INbUe SBIXe]|ey iWaNed]O SEIXE]|eES Neochanna apoda Figure 46. Suggested phylogenetic relationships of the New Zealand species of Galaxias and Neochanna. species of Saxilaga. If, as workers like then it is possible that a neochannoid lemming (1963: 382) claim, the ratite birds, species, even though it does not possess a phenodon, Leiopelma, and other animals definite sea-going stage, could have rafted Vicars eal Pid T r ’ 6 cispersea to New Zealand across the sea, or otherwise dispersed to New Zealand. Presently, nothing is known about the euryhalinity of either New Zealand or Tas- manian species of this group, but it is certain to be much greater than that of Leiopelma, a small frog. If the Australian species, treated by Scott (1936) as the distinct genus Saxilaga (also said to include N. burrowsius from New Zealand and G. globiceps Eigenmann from Chile), have a common origin with Neochanna in New Zealand, it may be necessary to alter the generic arrangement of these species. The answer does not, however, seem to lie in whether G. cleaveri Scott and G. anguilli- formis Scott belong in the genus Savxilaga, but in whether or not it is preferable to regard them as species of Galaxias, showing evolutionary trends towards Neochanna, or to include them in Neochanna, as a distinct radiation of “mudfishes.” If these species do form a natural grouping, it seems natural to include them all in Neochanna. The inclusion of G. globiceps in this assemblage is supported neither by Eigen- mann’s description nor by his figure of the species. Before these problems can be clarified, the osteology of these species needs study. ACKNOWLEDGEMENTS Among many people to whom I am in- debted for assistance and advice during the preparation of this thesis, I mention the following: Dr. Giles W. Mead supervised the study, and his enthusiasm and frank and helpful advice have been greatly appreciated. I am grateful to Dr. Ernst Mayr and Dr. John F. Lawrence for advice on nomen- clatural problems; Dr. H. B. Fell for help- ful discussions; Dr. C. A. Fleming and Dr. B. V. Hamon for useful comments; Mrs. M. M. Dick for much practical help. The study material was largely from the collection of the Fisheries Research Di- vision of the New Zealand Marine Depart- ment, and I am grateful to the following members of the Division staff for assist- ance: Messrs. J. W. Brodie, G. D. Waugh, New ZEALAND GALAXIIDAE * McDowall 427 G. A. Eldon, K. F. Maynard, W. Skrzynski, C. L. Hopkins, H. J. Cranfield, and Dr. G. Re Fisht Dr Re AW Hordham, Mina Pollard, and Mr. D. D. Lynch also gener- ously supplied specimens, and Dr. P. H. Greenwood (British Museum, Natural His- tory), Dr. D. E. McAllister (Canadian Museum), Mr. J. M. Moreland (New Zea- land Dominion Museum), and the Curator of Fishes, Zoology Museum, University of British Columbia, kindly loaned specimens in their collections. Throughout the study, generous financial support was received from a New Zealand National Research Fellowship. LITERATURE CITED Barnarp, K. H. 1943. Revision of the indigenous fishes of the south-west Cape Region. Ann. South African Mus. 36(2): 101-263. BENz1E, V. 1961. A comparison of the life history and variation in two species of Galaxias, G. attenuatus and G. vulgaris. Unpublished thesis, University of Canterbury, Christ- church, New Zealand. 193 pp. Brerc, L. S. 1940. The classification of fishes, both recent and fossil. Trav. Inst. Zool. Acad. Sci. U.S.S.R. 5(2): 1-517. (Reprint Ed- wards, Ann Arbor, 1947.) BicELtow, H. B., ann W. C. ScHROEDER. 1963. Family Osmeridae. In: Fishes of the western North Atlantic. Mem. Sears Found. Mar. Res. 1(3): 553-597. Biocu, M. E., anv J. G. ScHNEmER. 1801. Systema Ichthyologicae Iconibus cx Illus- tratum. Schneider, Berlin, 584 pp., 110 pls. BrepDER, C. M., AND D. E. Rosen. 1966. Modes of Reproduction in Fishes. Natural History Press, New York, 941 pp. BouLeNncer, G. A. 1902. The explanation of a remarkable case of geographical distribution in fishes. Nature 67(1726): 84. Buruinc, R. W. 1961. Hydrology of circum- polar waters south of New Zealand. New Zealand Dept. Sci. Ind. Res. Bull. 143: 1-66. Burnet, A. M. R. 1965. Observations on the spawning migration of Galaxias attenuatus (Jenyns). New Zealand J. Sci. 8(1): 79-87. Cairns, D. 1942. Life history of the two species of fresh water eel in New Zealand. II. Food and interrelationships with trout. New Zea- land J. Sci. Tech. 23: 132-148. Caste, P. H. J. 1963. Anguillid leptocephali in the southwest Pacific. Zool. Pubs. Vict. Univ. Wellington (N. Z.) 33: 1-14. 428 CHAPMAN, W. M. 1944. The osteology and re- lationships of the South American fish Aplo- chiton zebra Jenyns. J. Morph. 75(1): 149- 165. CuarkE, F. E. 1899. Notes on Galaxidae, more especially those of the western slopes, with descriptions of new species. Trans. Proc. New Zealand Inst. 31: 78-91. CunnincHaM, B. T. 1951. Preliminary report on fishes. In: The New Zealand-American Fiord- land expedition (A. L. Poole, ed.). New Zealand Dept. Sci. Ind. Res. Bull. 103: 1-99. Cuvier, G. 1817. Les Galaxies. In: Le regne animal, Vol. 2, pp. 282-283. Deterville, Paris, 4 vols. Cuvier, G., AND M. A. VALENCIENNES. 1846. Hist. Nat. Poissons Vol. 18, pp. 508-511. Levrault, Paris, 22 vols, 8°. Davipson, M. M. n.d. Anatomy of Neochanna apoda Giimther, Galaxiidae. Typewritten MS, Library of Fisheries Research Division, New Zealand Marine Department, Wellington, New Zealand. DireFFENBACH, E. 1843. List of fishes hitherto detected on the coasts of New Zealand by John Richardson with descriptions by J. E. Gray and Dr. Richardson of the new species brought home by Dr. Dieffenbach. In: Travels in New Zealand. Murray, London, 2 vols. Dotumore, E. S. 1962. The New Zealand Guide. Wise, Dunedin, 981 pp. Epon, G. A. 1968. Notes on the presence of the brown mudfish (Neochanna apoda Ginther ) on the west coast of the South Island of New Zealand. New Zealand J. Mar. Freshw. Res. 2: 37-48. Fett, H. B. 1962. West wind drift dispersal of echinoderms in the Southern Hemisphere. Nature 193(4817): 759-761. . 1967. Cretaceous and Tertiary surface currents of the oceans. Oceanogr. Mar. Biol. Ann. Rev. 5: 317-341. Fieminc, C. A. 1962. New Zealand _ bio- geography; a palaeontologist’s approach. Tuatara 10(2): 53-108. 1963. Paleontology and southern bio- geography. In: Pacific Basin Biogeography, pp. 369-385 (J. L. Gressitt, ed.). Bishop Museum, Honolulu. Forster, J. R. 1778. Observations Made During a Voyage Round the World on Physical Geography, Natural History and Ethnic Phi- Josophy. Robinson, London, 649 pp. 1844. Descriptiones Animalium. stein, Berlin, 424 pp. J. F. 1789. Systema Naturae. Gmelin, Lipsiae, 3 vols., 13th ed. Lichen- (GMELID Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 GosuinE, W. A. 1960. Contributions towards a classification of modern isospondylous fishes. Bull. British Mus. (Nat. Hist.) Zool. 6(6): 327-365. Gray, J. E. 1842. Three hitherto unrecorded species of fresh water fish brought home from New Zealand and presented to the British Museum by Dr. Dieffenbach. Zool. Misc., p. 73. GrEENWOop, P. H., D. E. Rosen., S. H. Werrz- MAN, AND G. S. Myers. 1966. Phyletic studies of teleostean fishes with a provisional classi- fication of living forms. Bull. American Mus. Nat. Hist. 131(4): 339-456, 3 pls. Gressit, J. L. (ed.). 1963. Pacific Basin Bio- geography. Bishop Museum, Honolulu, 561 1964a. Insects of Campbell Island. Pac. Insects Monogr. 7: 1-663. . 1964b. Introduction. In: Insects of Campbell Island, pp. 1-33. Pac. Insects Monogr. 7: 1-663. Guntuer, A. 1866. Catalogue of the Fishes in the Collection of the British Museum, Vol. 6. British Museum, London, 8 vols. . 1867. On a new form of mudfish from New Zealand. Ann. Mag. Nat. Hist. (3)20: 305-309, 1 pl. Haast, J. von. 1872. Notes on some undescribed fishes from New Zealand. Trans. Proc. New Zealand Inst. 5: 272-278. Hector, J. 1869. Comment on Neochanna apoda. Trans. Proc. New Zealand Inst. 2: 402. . 1902. Notes on New Zealand whitebait. Trans. Proc. New Zealand Inst. 35: 312-319. Herrorp, A. E. 193la. Whitebait. Rep. Fish. New Zealand 1930: 15-16, 21-23. . 1931b. Whitebait. Rep. Fish. New Zea- land 1931: 14-17. . 1932. Whitebait investigation. Rep. Fish. New Zealand 1932: 14-15. Henperson, N. E. 1967. The urinary and genital systems of trout. J. Fish. Res. Bd. Canada 24 (2): 447-449, Hoar, W. S. 1957. Gonads and reproduction. In: The Physiology of Fishes (M. E. Brown, ed.). Academic Press, New York, 2 vols. Houuister, G. 1934. Clearing and staining fish for bone study. Zoologica 12(10): 89-101. Hope, D. 1928. Whitebait (Galaxias attenuatus ); growth and value as trout food. Trans. Proc. New Zealand Inst. 58: 389-391, 2 pls. Hopkins, C. L. 1965. Feeding relationships in a mixed population of fresh water fish. New Zealand J. Sci. 8: 149-157. Husss, C. L. 1953. Antitropical distribution of fishes and other organisms. Proc. Seventh Pacific Sci. Congr. 3: 324-329. Husss, C. L., anp K. F. Lacter. 1947. Fishes of the Great Lakes Region. Cranbrook Inst. Sci. Bull. 26: 1-186, pls. Hutton, F. W. 1872. Catalogue with diagnoses of species. In: Fishes of New Zealand (J. Hector and F. W. Hutton). Colonial Museum and Geological Survey, Wellington, 133 pp., 12s pls: . 1874. Notes on some New Zealand fishes. Trans. Proc. New Zealand Inst. 6: 104—107. . 1889. List of New Zealand fishes. Trans. Proc. New Zealand Inst. 22: 275-285. . 1896. Notes on some New Zealand fishes with description of a new species and a synopsis of the New Zealand species of Galaxias. Trans. Proc. New Zealand Inst. 28: 314-318. . 1901. On a marine Galaxias from the Auckland Islands. Trans. Proc. New Zealand Inst. 34: 198-199. . 1904. Index Faunae Novae Zealandiae. Dulau, London, 372 pp. Ixus, J. 1964. Insecta—Plecoptera. In: Insects of Campbell Island, pp. 208-215. (J. L. Gres- sitt, ed.). Pac. Insects Monogr. 7: 1-663. Jorpan, D. S. 1919. The genera of fishes, part II, from Agassiz to Bleeker, 1833-1858, twenty six years with the accepted type of each. A contribution to the stability of scientific nomenclature. Leland Stanford Jun- ior University Publication, University Series, pp. 163-284. Jenyns, L. 1842. The Zoology of H. M. S. Beagle During 1832-6. Part 4. Fish. Smith, Elder, London, 151 pp. KENDALL, W. C. 1922. Peritoneal membranes, ovaries and oviducts of salmonoid fishes and their significance in fish culture practices. Bull. United States Bur. Fish. 37: 185-208. . 1935. The fishes of New England. The Salmon family II. The salmons. Mem. Boston Soc. Nat. Hist. 9(1): 1-166, 11 pls. Kner, R. 1865. Reise der Osterreichischen Fre- gatte Novara um die Erde in den Jahren 1857-59. Zoologischer Theil, Fische. Wien, 433 pp., 16 pls. Lacuer, K. F., J. E. BARpAcH, AND R. R. MILuer. 1962. Ichthyology, the Study of Fishes. Wiley, New York, 545 pp. Linpsry, C. C. 1961. The bearing of experi- mental meristic studies on racial analyses of fish populations. Proc. Ninth Pacific Sci. Congr. 10: 54-58. Lyncu, D. D. 1965. Changes in Tasmanian fish- ery. Australian Fish. Newslett. 24(4): 13, ilfey, McDowa.i, R. M. 1964a. The affinities and derivation of the New Zealand fresh water fish fauna. Tuatara 12(2): 56-67. New ZEALAND GALAXIDAE * McDowall 429 . 1964b. A consideration of the question “What are whitebait?” Tuatara 12(3):; 134— 156. 1965a. The composition of the New Zealand whitebait catch, 1964. New Zealand J. Sci. 8(3): 285-306. . 1965b. Studies on the biology of the red finned bully, Gobiomorphus huttoni (Ogilby ), III, Food studies. Trans. Roy. Soc. New Zealand Zool. 5(17): 233-254. . 1966a. Further notes on Galaxias white- bait and their relation to the distribution of the Galaxiidae. Tuatara 14(1): 12-18. . 1966b. A guide to the identification of New Zealand fresh water fishes. Tuatara 14 (2): 89-104. . 1967a. New land-locked fish species of the genus Galaxias from North Auckland, New Zealand. Breviora 265: 1-11. 1967b. Some points of confusion in galaxiid nomenclature. Copeia 1967(4): 841-843. . 1968a. Interactions of the native and alien faunas of New Zealand and the problem of fish introductions. Trans. American Fish. Soc. 97: I-11. 1968b. Galaxias maculatus (Jenyns); the New Zealand whitebait. New Zealand Mar. Dep. Fish. Bull. (mew series) 2: 1-84. . 1969. Relationships of galaxioid fishes, with a further discussion of salmoniform classification. Copeia 1969(4): 796-824. McKenziz, D. H. 1904. Whitebait at the Anti- podes. New Zealand Illustrated Mag. 10: 122-1926. McKenzie, M. K. n.d. Embryonic and _ larval structures of Galaxias attenuatus (Jenyns). Unpublished thesis, Victoria University, Well- ington, New Zealand, 74 pp., 24 pls. McMixtitan, H. M. 1951. The osteology and relationships of the family Retropinnidae Gill. Unpublished thesis Canterbury University, Christchurch, New Zealand, 59 pp. MoreELAND, J. M. 1958. Composition, distribution and origin of the New Zealand fish fauna. Proc. New Zealand Ecol. Soc. 6: 28-30. Myers, G. S. 1949. Usage of anadromous, catad- romous and allied terms for migratory fishes. Copeia 1949(2): 89-97. Mutter, J. 1844. Uber den Bau und die Gren- zen der Ganoiden und iiber das natiirliche System der Fische. Abhandl. Akad. Wiss. Berlin 1844: 117-216. Ocitpy, J. D. 1899. Contributions to Australian ichthyology. Proc. Linn. Soc. New South Wales 24(1): 154-186. Oxapa, Y. 1960. Studies on the Freshwater Fishes of Japan. Mie Prefectural University, Mie, 860 pp., 61 pls. 430 Otiver, W. R. B. 1936. The tertiary flora of the Kaikorai Valley, Otago. Trans. Roy. Soc. New Zealand 66(3): 284-304. Puiiuirrs, W. J. 1919. Life history of the fish Galaxias attenuatus. Australian Zool. 1(7): 211-213. . 1923. Note on the occurrence of the New Zealand mudfish or hauhau, Neochanna apoda. New Zealand J. Sci. Tech. 6(1): 62-63. . 1924a. The New Zealand minnow. New Zealand J. Sci. Tech. 7(2): 117-119. . 1924b. The koaro, New Zealand’s sub- terranean fish. New Zealand J. Sci. Tech. 7 (3): 190-191. . 1926a. Second occurrence of Galaxias robinsonii in New Zealand. New Zealand J. Sci. Tech. 8(2): 98-99. . 1926b. Additional notes on fresh water fishes. New Zealand J. Sci. Tech 8(4): 289-298. . 1926c. New or rare fishes of New Zea- land. Trans. Proc. New Zealand Inst. 56: 529-537, 6 pls. 1927a. Bibliography of New Zealand fishes. New Zealand Mar. Dep. Fish. Bull. 1: 1-68. . 1927b. A checklist of the fishes of New Zealand. J. Pan-Pacific Res. Inst. 2(1): 9- 16. . 1940. The Fishes of New Zealand, vol. 1. Avery, New Plymouth, 87 pp. Potiarp, D. A. 1966. Land-locking in diadro- mous salmonoid fishes with special reference to the common jollytail (Galaxias attenuatus ). Australian Soc. Limn. Newslett. 5(1): 13-16. Powe x, L. 1869. On four fishes commonly found in the River Avon, with a consideration of the question “What is whitebait?” Trans. Proc. New Zealand Inst. 2: 84-87. Recan, C. T. 1905. A revision of the fishes of the family Galaxiidae. Proc. Zool. Soc. Lon- don 2: 363-384, 4 pls. 1909. The classification of teleostean fishes. Ann. Mag. Nat. Hist., ser. 8, 3: 75-86. . 1911. The Fresh Water Fishes of the Brit- ish Isles. Methuen, London, 287 pp., 37 pls. Rem, R. C. 1886. Rambles on the Golden Coast of the South Island, New Zealand. Colonial Printing and Publishing, London, 176 pp. Renpaunt, H. 1926. Fishes from New Zealand and the Auckland and Campbell Islands. Papers Dr. Mortensen’s Pacific Exped. 1915- 1916 30: 1-14. RicHarDsoNn, J. 1843. Report on the present state of the ichthyology of New Zealand. Rep. sritish Assn. 12: 12-30. ———. 1848. Ichthyology of the voyage of H. {. S. Erebus and Terror. In: The Zoology Bulletin Museum of Comparative Zoology, Vol. 139, No. 7 of the Erebus and Terror During the Years 1839-43. Longman, Brown, Green and Long- man, London. Riney, T., J. S. Watson, E. G. Tursotrr, anp W. E. Howarp. 1959. Lake Monk expedition; an ecological study in southern Fiordland. New Zealand Dep. Sci. Ind. Res. Bull. 135: 1eio: Romer, A. S. 1966. Vertebrate Paleontology, 3rd ed. University of Chicago, Chicago, 468 pp. SauvacE, H. E. 1880. Notice sur quelques pois- sons de Vile Campbell et ’Indo-Chine. Bull. Soc. Philom. ser. 7, 4: 228-233. SCHOENNHERR, C. J. 1838. Genera et Species Curculionidum, Vol. 4, No. 2, p. 1043. Roret, Paris, 8 vols. Scotr, E. O. G. 1934. Observations on some Tasmanian fishes with descriptions of new species. Pap. Roy. Soc. Tasmania 1933: 31- 53, 2 pls. . 1936. Observations on fishes of the family Galaxiidae I. Pap. Roy. Soc. Tasmania 1935: 85-112. . 1966. The genera of Galaxiidae. Aus- tralian Zool. 13(3): 244-258. SKRZYNSKI, W. 1967. Fresh water fishes of the Chatham Islands. New Zealand J. Mar. Freshw. Res. 1(2): 89-98. STOKELL, G. 1938. A new species of the genus Galaxias with a note on the occurrence of Galaxias burrowsii Phillipps. Rec. Canterbury Mus. (N.Z.) 4(4): 203-208, 1 pl. . 1940. A new species of Galaxias. Trans. Roy. Soc. New Zealand 69; 422-424. . 1945. The systematic arrangement of the New Zealand Galaxiidae. I. Generic and sub-generic classification. Trans. Roy. Soc. New Zealand 75(2): 124-137. ; 1949. The systematic arrangement of the New Zealand Galaxiidae. II. Specific classification. Trans. Roy. Soc. New Zealand 77(4): 472-496, 12 figs. . 1950. Fresh water fishes from the Auck- land and Campbell Islands. Cape Exped. Ser. Bull. (N.Z.) 9: 1-8. . 1954. Contributions to galaxiid taxonomy. Trans. Roy. Soc. New Zealand 82(2): 411- 418, 1 pl. . 1955. Fresh Water Fishes of New Zea- land. Simpson and Williams, Christchurch, 145 pp., 42 pls. . 1959a. Structural characters of Te Anau salmon. Trans. Roy. Soc. New Zealand 87 (3-4): 255-263, 2 pls. . 1959b. Notes on galaxiids and eleotrids with descriptions of new species. Trans. Roy. Soc. New Zealand 87(3-4): 265-269, 2 pls. 1960. The validity of Galaxias post- vectis Clarke, with notes on other species. Rec. Dom. Mus. New Zealand 3(3): 239. . 1966. A preliminary investigation of the systematics of some Tasmanian Galaxiidae. Pap. Roy. Soc. Tasmania 100: 73-79. SwiInNnERTON, H. H. 1903. The osteology of Cromeria nilotica and Galaxias attenuatus. Zool. Jahrb. 18: 58-70. Taytor, W. R. 1967. An enzyme method of clearing and staining small vertebrates. Proc. United States Nat. Mus. 122(3596): 1-7. VotuaMs, S. E. 1872. On the mudfish (Neo- channa apoda). Trans. Proc. New Zealand Inst. 5: 456-457. Waite, E. R. 1907. Basic list of New Zealand fishes. Rec. Canterbury Mus. (N.Z.) 1(1): 1-39. 1909. Vertebrata of the sub-antarctic islands of New Zealand. In: The Subantarctic Islands of New Zealand (C. Chilton, ed.). Philosophical Institute of Canterbury, Christ- church, 2 vols. 235—- New ZEALAND GALAXIIDAE * McDowall 431 WeirzMan, S. H. 1967. The origin of the sto- miatoid fishes with comments on the classifi- cation of salmoniform fishes. Copeia 1967 (3): 507-540. Wuittey, G. P. 1935. Whitebait. Victorian Nat. Melb. 52(3): 41-51, 1 pl. . 1956a. The story of Galaxias. Australian Mus. Mag. 12(1): 30-34. . 1956b. Name list of New Zealand fishes. In: Treasury of New Zealand Fishes (D. H. Graham), 2nd ed. Reed, Wellington, 424 pp. . 1956c. New fishes from Australia and New Zealand. Proc. R. Zool. Soc. New South Wales 1954-55: 34-38. WuitLey, G. P., anp W. J. Puiturrs. 1940. Descriptive notes on some New Zealand fishes. Trans. Roy. Soc. New Zealand 69(2): 228-236, 2 pls. Woops, C. S. 1966. Species composition of white- bait (Galaxias spp.). Rec. Canterbury Mus. (N.Z.) 8(2): 177-179. (Received 29 April 1968.) ADDENDA 1). Since this work was completed, an extensive re-analysis of the taxonomy of G. maculatus and its lake derivatives in 16 lakes in Australia, New Zealand, and South America, has been made. This analysis suggested that G. usitatus McDowall (p. 382) is best regarded as a local race of G. maculatus, although the validity of G. gracilis McDowall (p. 384) is confirmed (McDowall, In Prep. ). 2). Study of adult specimens of G. weedoni Johnston supported tentative sug- gestions (p. 364) that this species is a junior synonym of G. brevipinnis Giinther, which is thus shown to have trans-Tasman range (McDowall, In Press, Records of the Dominion Museum, New Zealand, vol. 7). 3). A recent checklist of the fishes of New Zealand by G. P. Whitley (Australian Zoologist, vol. 15, 1968, pp. 1-102) listed as valid several Galaxias species synony- mized by myself and earlier authors, and is inaccurate and misleading. Rulletin OF THE. Museum of. Comparative Loology — The Spider Genus Ariadna in the Americas (Araneae, Dysderidae) JOSEPH A. BEATTY HARVARD UNIVERSITY VOLUME 139, NUMBER 8 CAMBRIDGE, MASSACHUSETTS, U.S.A: JUNE 26, 1970 PUBLICATIONS ISSUED OR DISTRIBUTED BY THE MUSEUM OF COMPARATIVE ZOOLOGY HARVARD UNIVERSITY. BULLETIN 1863— Breviora 1952— Memotrrs 1864-1938 JoHnsontA, Department of Mollusks, 1941- OccASIONAL Papers ON Mo Lwusks, 1945— pene aa SS Geen RE ia iia ay a a ee aes SS Other Publications. Bigelow, H. B., and W. C. Schroeder, 1953. Fishes of the Gulf of Maine. Reprint, $6.50 cloth. — Brues, C. T., A. L. Melander, and F. M. Carpenter, 1954. Classification of In- sects. $9.00 cloth. Creighton, W. S., 1950. The Ants of North America. Reprint, $10.00 cloth. Lyman, C. P., and A. R. Dawe (eds.), 1960. Symposium on Natural Mam- malian Hibernation. $3.00 paper, $4.50 cloth. Peters’ Check-list of Birds of the World, vols. 2—7, 9, 10, 12, 14, 15. (Price list on request. ) Turner, R. D., 1966. A Survey and Illustrated Catalogue of the Teredinidae (Mollusca: Bivalvia). $8.00 cloth. Whittington, H. B., and W. D. I. Rolfe (eds.), 1963. Phylogeny and Evolution of Crustacea. $6.75 cloth. Proceedings of the New England Zoological Club 1899-1948. (Complete sets i | only. ) oe Publications of the Boston Society of Natural History. { Publications Office \ Museum of Comparative Zoology Harvard University Cambridge, Massachusetts 02138, U. S. A. © The President and Fellows of Harvard College 1970. eae = ee “Si SR EA ete pee ———— — — ane nasil THE SPIDER GENUS ARIADNA IN (ARANEAE, DYSDERIDAE) JOSEPH A. BEATTY INTRODUCTION The spider family Dysderidae, to which the genus Ariadna belongs, is one of a series of presumably primitive families called the haplogyne spiders, from the simplicity of their genitalia. Other haplogyne families are the Plectreuridae, Diguetidae, Sicariidae, Scytodidae, Caponiidae and Oonopidae. (The families Leptonetidae, Ochyrocerati- dae, and Telemidae, often included in the haplogyne series, are possibly more closely allied to the Araneoidea, and are omitted from consideration here. ) Dysderids are medium- to large-sized spiders (maximum body length about 25 mm) of either sedentary or wandering habits. Their web is of the tubular retreat type, sometimes with radiating trip lines extending from the mouth, but containing no viscid silk, and not functioning as a snare. The spiders have six eyes (or no eyes, as in a few cave-dwelling species ), two lungs, and a pair of tracheal spiracles close behind the lung apertures; they bear usually three, sometimes two, tarsal claws. The family is nearly cosmopolitan in dis- tribution, but is absent or rare in polar and cold northern temperate regions. The genus Ariadna is of interest because of its almost worldwide distribution, its occurrence on isolated islands, the many species that have been described, and the fact that it is haplogyne. The genitalic 1 Department of Zoology, Southern Ilinois Uni- versity, Carbondale, Illinois 62901. Bull. Mus. Comp. Zool., 139(8): 433-518, June, 1970 THE AMERICAS simplicity of haplogyne spiders makes iden- tification of specimens, especially females, difficult. As a consequence, the taxonomy of the haplogyne families has been in a confused state until recently. A revision of the American Ariadna was the principal objective of this study, in which, as far as possible, the males were distinguished by genitalia. A secondary objective, however, was a search for other taxonomic characters. Computer methods were used in this search and in the grouping of species, and the calculated results were compared with those obtained by a “clas- sical” taxonomic approach (Beatty and Bos- sert, in prep.). Although the secondary objective is probably of more general in- terest and application, all phases of the study are, of course, closely interrelated. ACKNOWLEDGMENTS This study was conducted under the direction of Dr. Herbert W. Levi of the Museum of Comparative Zoology. His advice, support, and encouragement have been of great assistance throughout the course of the research, and his patience with its protractedness is greatly appreciated. Professor F. M. Carpenter, in addition to serving as major professor, offered support in both academic and personal matters further than anyone has a right to expect. I am most sincerely grateful to him for many favors. The excellent library and the collections of the Museum of Comparative Zoology 433 434 have been indispensable. Especial thanks are due Dr. W. J. Gertsch, who made avail- able to me the collection of the American Museum of Natural History. This collection has formed the nucleus and major part of the material used in the present study, which could not have been made without it. Mrs. Alice Bliss Studebaker inked most of the drawings, and aided in checking the literature citations. Her assistance is sin- cerely appreciated. Travel grants provided by the Evolu- tionary Biology Committee of the Biology Department of Harvard University, and the Society of the Sigma Xi made possible two extensive collecting trips into the southern United States and Mexico. Public Health Service Research Grant AI-01944, from the Institute of Allergy and Infectious Diseases to Dr. Levi, partially assisted the research in various ways. Important collections were supplied by the following persons from collections under their care: Dr. G. Owen Evans and Mr. Douglas Clark, British Museum (Natural History); Dr. Gisela Rack, Zoologisches Staatsinstitut und Museum, Hamburg; Prof. M. Vachon, Muséum National d Histoire Naturelle, Paris; Prof. R. D. Schiapelli and Sra. B. Gerschman de Pikelin, Museo Argen- tino de Ciencias Naturales, Buenos Aires; Mr. Persio de Biasi and Dr. Hans Reichardt, Departamento de Zoologia, Secretaria da Agricultura, Sa0 Paulo; Dr. Ralph Crabill, U.S. National Museum, Washington; Dr. H. K. Wallace, University of Florida, Gaines- ville; and Dr. Charles A. Triplehorn, The Ohio State University, Columbus. To all of these I extend my sincerest thanks for their cooperation and assistance. Donations and loans from Mrs. D. L. Frizzell (Dr. H. Exline), Dr. Andrew A. Weaver, and Mrs. Stephanie Cannon were of considerable assistance, and I wish to express my appreciation to them. Others who have given or loaned speci- mens, assisted in their collection, or acted 4s companions on field trips include Drs. s3rady, James W. Berry, and An- 1 T ey Bi Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 drew A. Weaver, Mrs. Stephanie Cannon, Miss Allison MacDougall, Arnold Aschwan- den, Ivan Braun, J. E. Carico, and Fred- erick A. Coyle. Their cooperation is greatly appreciated. POSITION OF THE DYSDERIDAE AMONG THE HAPLOGYNE FAMILIES The haplogyne spiders are readily separa- ble into two groups that may deserve to be ranked as superfamilies. These groups are distinguished primarily by differences in chelicerae, posterior respiratory organs, and number of heart ostia. In the Plectreuridae, Diguetidae, Sicarii- dae, and Scytodidae, the chelicerae are chelate and are joined to each other basally over one-fourth or more their length; a single median tracheal spiracle is situated behind the middle of the abdomen but not immediately adjacent to the spinnerets, and the heart has three pairs of ostia. Within this group, the Plectreuridae and Diguetidae are similar to each other in some characters, but differ strongly from the Sicariidae and Scytodidae. The latter families are close to each other morpho- logically. In the Dysderidae, Oonopidae, and Caponiidae, the subchelate chelicerae are free from each other or are only barely con- nected by membrane proximal to the sclero- tized part of the appendages; a pair of large tracheal spiracles is situated just be- hind the lung openings, and the heart has two pairs of ostia. Additional data are summarized in Table 25. In the collection of the Museum of Com- parative Zoology, I have examined a few unidentified specimens that do not appear to fit well into any of the recognized fami- lies. Many new species and some new genera almost certainly remain to be dis- covered, especially in south temperate regions. The intermediate structural char- acters of some of these species will prob- ably discourage maintenance of seven separate families for the haplogyne spiders. Within the haplogyne series, the Dysderi- dae are distinguishable from the sicariioid families by the characters listed above. They differ from the other two families as follows: from all the Caponiidae and a few of the Oonopidae by their having six eyes; from most of the six-eyed oonopids by the arrangement of the eyes; from both the Caponiidae and Oonopidae by their well- developed book lungs, the presence of a tarsal claw on the palp of the female, their larger size and sedentary habits, and their use of silk for making tubular retreat webs. In habitus the larger dysderids seem primitive and similar to the plectreurids, filistatids, ctenizids, and liphistiids. Each of these other families is probably the most primitive living member of its par- ticular suprafamilial group. The body size, leg proportions, color, sedentary retreat- web building habit, and genitalic simplicity of these five families, although far from providing conclusive evidence, certainly suggest relationship. I believe that a more intensive study of this group of families than has thus far been published would yield important data pertaining to the phylogeny of the Araneae. NATURAL HISTORY OF ARIADNA Members of the genus Ariadna are sedentary nocturnal spiders. Their tubular webs of closely woven pure white silk are slightly widened at the mouth, from which single threads radiate in all directions. These threads are attached to the sub- stratum at distances of about two to ten millimeters from the tube. The spiders stand at the mouth of the tube at night or during periods of low light intensity during the day. They are highly sensitive to vibration of the radiating threads by poten- tial prey, reacting with a rapid dash from the tube and back again. The movement is too swift to be followed precisely with the naked eye, but it appears that the spiders do not emerge completely from the tube. The prey is seized in the two ARIADNA IN THE AMERICAS + Beatty 435 anterior pairs of legs. The double rows of heavy spines on the ventral surfaces of the legs probably function in preventing the escape of the prey. All species of Ariadna for which infor- mation is available place their webs in the same type of microhabitat: crevices and small holes. Depending upon the size of the spider, the diameter of the web ranges from one to about ten millimeters. The web is generally built in a crevice that has a width about equal to the diameter of the tube. On one occasion, however, I saw a web suspended vertically in the center of a pipe that had a diameter of about two inches. Levi (pers. comm.) reports that juvenile A. maxima in Chile often build their webs on top of each other, so that a series of tube webs spans a wide crevice. The spider is so abundant that this behavior may result from a shortage of crevices of a more appropriate size. In my experience, broken outcrops of rock are the sites most favored by Ariadna, but not all kinds of rock are equally acceptable. Unstable rocks such as shale are avoided; massive thick-bedded rocks usually do not have enough crevices of sufficient depth to support more than a scattered individual or two. One of the largest colonies I have seen was on Gibraltar Island in western Lake Erie. At one end of the island a weathered, highly fractured dolomite cliff rises 15-20 feet (5-7 m) from the edge of the water. The accessible portion of this outcrop, about 50 feet (17 m) long, is occupied by hun- dreds of A. bicolor webs. The amount of moisture present is also of considerable importance. I have never found Ariadna on a rock outcrop from which even a small amount of water was seeping, nor in highly insolated outcrops in the lowlands of the southwestern and Mexican deserts. At low elevations of 2700 to 4600 feet (820-1400 m) in southern Arizona, A. pilifera occurs sparingly in rocks along canyon bottoms. At higher elevations it gradually spreads to more 436 exposed outcrops. In the Santa Catalina Mountains north of Tucson, Arizona, I have taken it at a maximum elevation of 7500 feet (2280 m). Ariadna is also commonly collected in crevices on buildings and other man-made structures. I have collected A. bicolor in such places in Ohio, North Carolina, Florida, and Illinois, and A. pilifera in Arizona. Crevices in or under tree bark and palm fronds are frequently used, al- though less often than the above micro- habitats. Ariadna fidicina has been taken from beneath eucalyptus bark in California, and A. arthuri from under the bark of red mangrove in the Florida Keys. Additional microhabitats include crevices and small holes found within and about clumps of moss, beneath rocks or boards on the ground, among roots and stems of grass clumps (Barnes, 1953), in ground litter, and once in shipworm burrows in a hatch-cover thrown up on a Florida beach. All of these sites are used much less fre- quently than the three mentioned above, and the population densities are relatively low. Even where some member of the genus is abundant, suitable microhabitats are often so localized that the spider appears rare. In such cases it is usually found only in the more marginal kinds of microhabi- tats. A special search and special collecting methods are usually required to determine its actual frequency in a given area. A rock outcrop fitting the description given earlier will usually harbor at least a few Ariadna, and populations as high as five to ten per linear foot of crevice are not unusual. When the webs have been located, one can usually catch a large num- ber of specimens in a short time by using the technique described below. The only equipment needed is an ordi- nary dissecting needle and a vial, or a series of vials with cotton plugs if the spiders kept alive for a time. While ding the open mouth of a vial near, but touching, the mouth of the web, thrust are to be Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 the needle quickly into the crevice at an oblique angle to the web, and as far back from the mouth of the web as possible. The object is to close off the web behind the spider, so that it can not escape from the rear of the web into the crevice. Place the mouth of the vial against the web opening, and gently work the needle for- ward. The spider will be driven forward and will usually make a quick dash from the web into the vial. If the collecting is done at night with the aid of a head-lamp, the spiders will be found sitting at the mouths of the web. Under these conditions they may be captured easily and quickly by the above method. Two factors probably account for the colonial habit of Ariadna. The first is the relative scarcity and isolation of optimum microhabitats. The spiders are certainly present in many sub-optimum habitats, but in small numbers and _ thinly scattered. Second, no species of Ariadna, in fact no haplogyne spider of any kind, is known to balloon. The eggs of Ariadna are laid in the tube of the female and the young re- main in the tube for some time after hatch- ing. Judging from the size of known second-instar young and the smallest in- dividuals found inhabiting their own tubes, I would guess that the juveniles leave the female’s web between their second and third molts. Dispersal appears to vary from no more than a few millimeters to perhaps ten yards (10 m). The food habits of Ariadna are virtually unknown. In captivity I have fed them leafhoppers, small moths, and Drosophila, all of which they took readily. The insects available in largest numbers to the Gibraltar Island colony are caddis-flies, midges, and Mayflies, which emerge from the lake in thousands during the summer. Ariadna is probably like many spiders in eating other spiders readily, also. Courtship and mating in this genus have not been observed. Presumably it is similar to that of the related genera Segestria and Dysdera. Gerhardt (1921) described and illustrated the mating behavior of Seges- tria. Males of this genus lack the spurs present on the anterior legs of some male Ariadna. The male pushes back the fe- male’s legs with his, moves under the female’s carapace, and, holding the anterior part of her abdomen with his fangs, inserts both palps simultaneously into the genital groove. Mating behavior in Dysdera and in the mygalomorph spiders follows the same pattern. The eggs of Ariadna are laid in the pos- terior portion of the tube, without being enclosed in an egg sac. The eggs are not sticky and the female, when disturbed, may move through the egg mass with little difficulty. The very few threads which tie the eggs loosely together may, therefore, be the result of the movements of the female over a considerable period follow- ing the laying of the eggs. Egg masses and young spiders are rarely found with adult females in collections. The difficulty of extracting the entire web intact probably insures loss of some eggs and young even when attempts are made to collect them. Two collections of A. bicolor from the Emerton collection in the Museum of Comparative Zoology contain the following notes: “in closed bag with cocoon of about 25 eggs” (one female in this vial), and “with young in tube” (this vial contains two females and_ twelve young ). Two females of the same species that I collected on Gibraltar Island produced eggs in captivity. Females and offspring were preserved shortly after the eggs began to hatch. With one female were 26 eggs and hatchlings. There were a few more empty chorions than hatchling spiders; apparently a few young were lost even with careful removal from the web. The second female produced 71 eggs, a little more than half of which had hatched at the time of preservation. The remaining eggs, except one, contained well-developed embryos which probably would have emerged. Fresh egg masses, collected with three ARIADNA IN THE AMERICAS + Beatty 437 different females on Gibraltar Island con- tained respectively 46, 69, and 95 eggs. It is believed that no eggs were lost from these masses during collecting. Ariadna seemingly has an extended breeding season. There is a distinct tend- ency for males to be most abundant in late summer, but in Ohio and Pennsylvania, male A. bicolor may be collected from late May through September. The relative in- frequency of males in most collections suggests that they tend to mature mostly during a short period of time, and may live only a short time after mating. Mature females evidently live more than one sea- son. They may be collected, along with juveniles of all sizes, at almost any time of year. Winter, when the spiders retreat deep into their webs, is a possible ex- ception, but lack of collecting at this season prevents a definite statement. The chromosomes of one species of Ariadna, A. lateralis of Japan, have been reported by Suzuki (1952). The diploid number is eight, the smallest known for any species of spider. Sex determination is XO, the male being the heterogametic sex. In view of the tendency to regard high chromosome number as primitive and low numbers as derivative within a given taxon, the chromosome number of Ariadna_ is especially interesting. One species of the Liphistiidae has a 2N chromosome number of 94 or 96. Most araneomorph spiders range between 2N = 24 and 2N =36. PATTERNS OF SPINATION The cuticular structures referred to as spines both here and in araneological liter- ature in general, are certainly not spines in the entomological sense of the term. Ac- cording to the generally accepted definition (Snodgrass, 1935), a spine is an immovable outgrowth of the entire body wall, is lined by epidermal cells, and is not alveolate. The so-called spines of spiders, presumably even the largest of them, are alveolate, hence usually movable, and are secreted 438 by a single epidermal cell. Therefore they are setae. Setae of spiders fall into several inter- grading categories on the basis of diameter. These are usually referred to as spines, setae, and hairs. The distinctions among these three setal classes are useful and the terms are firmly fixed in the literature. I have continued to use all three words in their araneological senses in preference to causing confusion by the introduction of unfamiliar or newly-coined words. Most of the leg spines of Ariadna are borne on the metatarsi, tibiae, and femora. Occasionally there are one to a few on the patellae. In later instars, spination of juve- niles is like that of females. Second instar juveniles have a female pattern of spination, but the number of spines is much smaller than in adults. Femoral spines in mature specimens of both sexes are limited to the upper surface of the segment, except on the first, and rarely the fourth, leg. Females have, on the distal prolateral surface of the first femur, a single long slender spine. Two species, A. gracilis and A. multispinosa, have two or three such spines on each first femur (Figure 7). The constancy of these spines is almost one hundred percent in females and in juveniles after the first molt. In males the prolateral spines of the first femur are often suppressed. Usually the spines are still represented by setae that may be set in enlarged sockets or be some- what spiniform, indicating their homology with the spines of the female. In both sexes the spines of the upper surfaces of the femora are arranged in three longitudinal rows. One row runs along the dorso-median axis of the femur, the other two along the lateral margins of the upper surface. The cylindrical cross-section of the femur makes delimitation of dorsal and lateral surfaces somewhat ambiguous. All three spine rows have been arbitrarily termed dorsal, although the lateral ones show some tendency to “slip” down onto the lateral surfaces of the femora. Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 Dorsal spination of the femora is re- duced or absent in females but strongly developed in males. In females, retention of a single distal spine of the inner row on femora II and III (in about ten percent of the individuals) represents the lower ex- treme in spine reduction. A common pat- tern in females is a distal circlet of three dorsal spines, one in each row. At the observed upper limits in females are found two spines in each of the outer and middle rows, and three in the inner, in a few individuals. The spination of each of the four femora varies in an individual, usually by a reduction in the number of spines posteriad. Occasionally the fourth femur will bear a variable number of spines basally in the middle row. The number of dorsal femoral spines is highly variable in males both intra- and inter-specifically. The range in the speci- mens examined was zero to thirteen on a single femur. The majority of the spines are usually in the middle row. The lateral rows often equal each other in spine num- ber. On the tibia, the ventral spines occur in two rows, which on legs I and II are al- most always present in both sexes (Fig. 3); in males, however, the lateral spines occur in one or more rows and are almost always present; if present in females, the lateral spines occur in one row. The dorsal spines rarely occur in males; and a “super- numerary spine just outside the basal spine of one or both ventral rows in males. Variation in the number of tibial spines is considerable in the genus, but often very slight within a species. In females, lateral tibial spines on the first two legs are usually either present in numbers or al- most entirely absent in a given species. The number and arrangement may be variable in a species that has them. Par- ticular patterns are indicated in the species descriptions. The metatarsi bear a larger and more variable number of spines than the other podomeres. Only the main features of the variation can be dealt with here. In fe- males on legs I and II, there are two ven- tral rows of spines, usually with seven to thirteen spines per row (Fig. 13). Meta- tarsus III also bears two ventral rows, but rarely has more than three spines in the outer and two in the inner row. The fourth metatarsus has a distal comb of spines on the inner ventral surface. This comb is made up of a transverse row of two to eight modified spines set in contiguous sockets (Fig. 31). Elsewhere on the podo- mere, spines are few or absent. Lateral metatarsal spines occur sporadically in fe- males except on leg II, which usually has one or two prolateral spines. Metatarsal spination in males is very different from that in females, except on the fourth metatarsal comb, which is alike in the two sexes. The sexual differences in metatarsal spination consist of reduction in the number of ventral spines, and ad- dition of lateral and sometimes dorsal spines in the males. Male first leg. Of the American Ariadna, sixteen males are known. Eleven of these have the metatarsus of the first leg modi- fied in some way. The presence or ab- sence, and the particular form, of this modification furnish the best species char- acters for the males. The modified metatarsi are either trans- versely sinuous or sharply bent (Figs. 41, 51). If sinuous, they are also usually slender, and may bear one or two conical projections. If sharply bent, the metatarsi are thicker and bear one or two lateral projections. In either case, the projections may bear spines (Figs. 32, 33). In those males that have heavy first metatarsi with lateral projections (apophy- ses), modified spines occur on the first tibiae. These spines are very short, flat, and wide, and occupy the distal part of the inner ventral row of tibial spines. The number of spines so modified varies from one to three. ARIADNA IN THE AMERICAS + Beatty 439 DEVELOPMENTAL AND TRAUMATIC CHANGES IN SPINATION Spination in adults of a given species of Ariadna is slightly to moderately variable, depending upon which surface of which podomere is considered. The maximum variation noted in a single character (the number of spines in the ventral rows on metatarsus I) was plus or minus five from the range midpoint. The minimum was minus one spine (first femur, pro- lateral) where the modal number was one. In this case, variation was quite rare. Of the total number of spine characters, about twenty percent fall into the category of slightly variable; that is, their constancy is ninety percent or more. However, among several hundred adult Ariadna examined, not one specimen had a completely sym- metrical spination, and no two specimens had the same spination. The question of the basis of this vari- ation immediately arises, especially because at least some characters appear to be under rather strict genetic control. Comparison of young specimens of A. bicolor with adults of the species provides a partial answer to this question. Five broods of young were available produced from eggs laid in captivity. The spider- lings were known to be in the second instar. One brood was from Massachusetts, two were from Ohio, and two from Florida. Ariadna, as other spiders, hatches from the egg as an unpigmented spiderling without hair, setae, or spines. It molts al- most immediately upon freeing itself from the chorion. The second instar is provided with setae and spines, which may be less well differentiated from each other than they become later. In all five broods of A. bicolor, not only was every individual symmetrically spined, but each was spined exactly the same as the others, with one minor qualification: occasionally one of the spines was so slender that it resembled the irregularly arranged setae. However, its position indicated its nature clearly. 440 All spines at this state are relatively much slenderer than in the adults. The second instar leg spines of A. bicolor consist of two rows of three each on the ventral surface of tibia I, an outer row of three and a single inner spine on the ven- tral surface of tibia II, a single ventral spine on metatarsus III, two spines in the comb on metatarsus IV, and a single pro- lateral spine on femur I. The spination characters showing the greatest constancy in the adults are present in the second instar in almost their final number. The addition of only one spine to each of the ventral tibial rows and one to the metatarsal comb produces the spine number found in 90 percent or more of the adults. The prolateral spine on femur I remains single. The addition of these spines almost al- ways takes place at the second or third molt, after which, in these particular char- acters, there is usually no further change with subsequent molts. No statistically significant change in any feature of spi- nation occurred during the last molt in the few samples examined for such a change. The ventral spine rows of metatarsi I and II are the most variable of the spi- nation characters. There appears, however, to be a fairly regular pattern of addition of spines. The second instar, as noted above, has three spines in each of these rows. In the adult, the proximal and distal spines, and the one at the middle of each row, are of about equal length and are much longer than the others. These long spines are be- lieved to be the original three found in the second instar. Between these “primary” spines are others of varying lengths. In general, the distal spine in a series between two primaries is the longest, and the lengths of the others gradually diminish proximally until the next primary is reached. Sometimes there may be a long “secondary spine interrupting the series, breaking the series into two shorter ones that repeat the same pattern (Fig. 13). I | that the length of a spine is an Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 indication of the molt at which it ap- peared, and that rearing of a few species will reveal enough of the pattern of ad- dition of spines to allow aging of speci- mens on the basis of their spination. The uniformity of juvenile spination suggests that the genetic basis of spination in Ariadna is either subject to little vari- ation or is highly canalized, and that the variability in adult spination should be attributed largely to developmental “acci- dents.” Until large-scale rearing of Ariadna has been carried out, little can be said about the effect of environmental con- ditions on development and variation of spination. A few observations derived from preserved specimens are of interest, how- ever. Differential expression of a particular spine pattern is obvious in preserved material, and is of frequent occurrence. It is most noticeable in males. The number of spines in the ventral metatarsal rows of males is small, usually two or three spines per row, as compared with six to twelve or more in the females. In the wide spaces between the spines of the male, setae with slightly enlarged bases and sockets often occupy the spots where the spines of the female would be. The male pattern of spination on this segment may perhaps represent a suppression of the spines that have been added since the second instar, but the suppression is not necessarily com- plete. The same sort of effect is occasionally observed in females. A spine row that “should” contain four evenly spaced spines may have a conspicuous gap in it. Stand- ing in the gap may be a seta which corresponds to the missing spine. The phenomenon in both sexes suggests that, if we had any knowledge of the genetics of Ariadna, we might be speaking here of the expressivity of a gene. In cases such as the above the trichogen cells that produce the spines fail, either completely or partially, to function. There are also cases of overfunctioning or appar- ent duplication of these cells. In some males, the spines of the first metatarsi or tibiae are greatly thickened or widened as compared with the other leg spines (Fig. 40). Rarely an individual of either sex will be found in which a spine is slightly or deeply bifid at the tip. In a few of these cases the division extends to the base, so that two spines, closely appressed to each other, stand in the same socket. Too little is known of the histology of Ariadna epi- dermal cells to warrant suggestions of reasons for most cases of developmental duplication or multiplication of spines. Traumatic effects upon spine number are much more easily dealt with. Seizure of an appendage by another organism nearly the size of the spider or larger will usually result in the spider's sharply twist- ing itself free of the trapped appendage. No reflex autotomy occurs, but there are weak points in the appendages, and breaks at these points seal themselves off rapidly. If an appendage is severely mangled but not pulled off, the spider will itself later break off part or all of the appendage. Provided the animal has at least one more molt in the offing, no sooner than several days after the injury, the lost appendage will be regenerated. Regenerated appendages are at first smaller and paler in color than the original ones, and customarily have a reduced spination. With each further molt they be- come more like the unregenerated append- ages, but probably rarely catch up with them, except in female mygalomorphs. Al- though reduction in the number of spines is general in regenerated appendages, the particular pattern assumed by the spines is not very predictable; various irregulari- ties have been observed. An injury that does not result in autospasy of the appendage heals, leaving a localized scar. If a molt follows, various results may be observed. One specimen I examined had a gap in a spine row with an area of pale cuticle where the spines were missing. Examination of the exuviae ARIADNA IN THE AMERICAS + Beatty 44] of the previous instar of this individual showed a healed wound covering an arez which should have borne two to four spines. Perhaps a superficial injury might heal without effect upon spination, but such a result would be difficult to detect in field-collected material. Of especial interest is a specimen of A. bicolor, in which one of the metatarsal] rows contained 19 spines, the maximum number for the species otherwise being 12. Near the distal part of the segment. a cluster of half a dozen spines occupied a semicircular area lateral to the rest of the row. These extra spines were oriented in various directions, not extending diago- nally down and forward as did the other spines of the row. Probably a small wound lateral to the spine row had occurred in an earlier instar. The multiplication of trichogen and_tor- mogen cells, and their migration to the wound area during healing, would result in the production of supernumerary spines in an abnormal location. Wigglesworth (1954) describes the experimental pro- duction of this effect in the hemipteran Rhodnius. In summary, then, it appears that a large portion of the variation in spination in adult A. bicolor is produced by transient or local physiological changes, including trauma. The extent to which varying en- vironmental conditions might produce a harmonious variation of the spine pattern is unknown. SPECIATION IN THE GENUS ARIADNA To be able to point out lines of evolution among the American Ariadna that could be supported by considerable evidence would be gratifying. Unfortunately, this does not yet seem possible. Because of the paucity of material of many species and the morphological and ecological conserva- tism of the genus, differences are relatively few and slight in most cases. Morphologi- cally, any one of the American species 442 could probably have given rise to any of the others. Nevertheless, some possible relationships are apparent. It must, of course, be borne in mind that the following discussion is quite tentative and may re- quire extensive change when more material becomes available. Three groups of species are moderately well defined. Not all of the species fit into these groups, and the placement of some species in the groups is ambiguous. The bicolor group includes a series of species that occur from North America through Mexico, in the mountainous por- tions of western South America, and on several islands adjacent to the Mexican or South American mainlands. In this group are the widespread North American A. bicolor, A. pilifera of the U. S. and Mexico, A. pragmatica, A. weaveri, A. caerulea, A. cephalotes, A. murphyi, A. peruviana, and A. maxima of Chile. Two other species, A. isthmica from Central America and A. tovarensis from Venezuela, may belong here also. Five of these species are grouped _ to- gether by one of the more satisfactory of the Mahalanobis’ distance analyses, shown in diagram 7 of Beatty and Bossert (in prep.), and other methods of analysis associate several of them. Although A. bicolor, pragmatica, and caerulea are placed further from the above group than most other species, they decidedly belong in the group. Ariadna bicolor is actually quite similar in both sexes to A. pilifera. The distinctions between the species are primarily the increased numbers of spines on most appendages in the females of A. pilifera. The modifications of the male first metatarsi are very much alike. (Male characters were not included in the cal- culation of Mahalanobis’ distance. ) The elevational range inhabited by these species is almost totally unknown except for A. bicolor, recorded from near sea level to 7000 feet (0-2130 m), and A. pilifera, from about 2700-7500 feet (800- }m). It appears likely that this species taken 2306 Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 group represents a diversification of a stock that was distributed throughout temperate North America and the moun- tainous areas of South America. Indepen- dent migrations from various parts of this range are the probable origins of A. pragmatica on islands in the northern Gulf of California, A. weaveri on Clarion and Socorro Islands, A. murphyi on the Chin- chas Islands off Peru, and A. maxima on the Juan Fernandez Islands. The central American A. isthmica and Venezuelan A. tovarensis are somewhat similar to the members of the bicolor group, but apparently occur primarily in tropical lowlands, and are only doubtfully to be included in this group All eleven of the above species are characterized by the presence, in the fe- males, of two ventral rows of four spines each on the first tibiae, and an outer ven- tral row of four spines on the second tibiae. The inner ventral row on the second tibiae contains two spines in bi- color, pilifera, pragmatica, and weaveri, the northernmost members of the group; three spines in cephalotes of Colombia and Ecuador and tovarensis of Venezuela; and four spines in the other members of the group. The metatarsal comb contains two spines in murphyi and pragmatica (both island-inhabiting species); three in bicolor, weaveri and peruviana; and four in the remaining species. Only modal numbers are given for these characters. Except for pilifera and caerulea, however, there is relatively little intraspecific variation in the characters. Male characters, as far as they are known, show some agreement with this grouping. Of the four American species that have prominent spurs on the first metatarsi and enlarged spines on the first tibiae, three, bicolor, pilifera, and peruvi- ana, are in the bicolor group. The fourth species showing these modifications, gra- cilis, is morphologically far removed from most species of the genus. It is obvious, however, that the male leg modifications and the female spination are not strongly correlated with each other. The other eight members of the bicolor group in- clude males with sinuous first metatarsi without spines and others with unmodified metatarsi that are entirely spineless. Group- ing of the species on the basis of male characters produces a similar heterogeneity of female spine patterns within each group. The arthuri group includes A. arthuri, widespread on islands in the Caribbean, A. multispinosa in Hispaniola, and A. tar- salis in the Galapagos Islands. In these species, the first and second tarsi have more than four spines in each ventral row. The modal number for each row in arthuri is six. Too few specimens of the other species are known to establish a meaning- ful mode, but the observed range is four to nine in tarsalis (three specimens) and eight to eleven in multispinosa (two speci- mens). All but one of the specimens of the latter two species are immature. Only two other species show a similar tibial spination: tubicola from Venezuela and gracilis from the Amazon Basin. Both of these species are rather strongly divergent from the rest of the genus. The only affinities either of them show are weak ones with the multispinosa group. The multispinosa group appears to be distributed on the Caribbean and Gala- pagos Islands with further speciation on Hispaniola. Since, in spite of their seden- tary habits, Ariadna species are quite suc- cessful colonizers of new territory, one might expect a mainland member of the group to occur in Mexico or Central America. Prior occupation of this area by other species of the genus would, of course, reduce the possibility of such an occurrence. In fact, however, a_ single mature female collected by me in north- eastern Mexico near Monterrey appears to be a new species of the multispinosa group. Because only one mature and a few juve- nile specimens are on hand, the species has not been described. No other species ARIADNA IN THE AMERICAS + Beatty 443 from North or Central America resembles the multispinosa group in spination. In this group, arthuri and tarsalis each have two spines in the metatarsal comb, multispinosa has four. It is curious that, of the five American species in which the comb contains only two spines, all but obscura, from eastern Brazil, inhabit is- lands. Two other island species, weaveri and peruviana, have three comb spines, solitaria has four. The third species group centers around A. mollis of southeastern Brazil, Uruguay, and Argentina. Other species of the group are boesenbergii and four as yet unidenti- fied species known from the Sao Paulo and Diamantina areas in southeastern Brazil. The latter four species probably include A. crassipalpus Blackwall, A. conspersa Mello-Leitao, A. dubia Mello-Leitao, and A. spinifera Mello-Leitao. In the absence of type specimens, it has so far proved im- possible to assign the available specimens to any described species. They may not all belong to the mollis group, but certainly most of them do. According to the com- puter analysis, A. boesenbergii and, to a smaller extent, A. mollis link the mollis and bicolor groups. Ventral spination of the first two tibiae in the mollis group is generally like that of the bicolor group. The metatarsal comb contains four or more spines in all members of the mollis group. Both sexes of mollis, boesenbergii, and two of the unidentified species, are known. In all of these, the carapace and abdomen are predominantly light in color, and in mollis and boesen- bergii, there is a conspicuous pattern of light and dark bars on the abdominal dor- sum. It is impossible to be sure whether the unidentified species referred to above did or did not have a pattern in life. The pattern can be destroyed by poor preser- vation. The material of the other possible mem- bers of the group consists of two distinctly different males, one from Diamantina and one from Sao Paulo, and several females 444 from Sao Paulo. It is not possible to match either male with the females with any acceptable degree of probability. The fe- males are all very dark in color, in contrast with other members of the group. They may belong elsewhere. It appears possible that the La Plata River may have acted as a barrier per- mitting differentiation of an_ originally single stock into A. boesenbergii to the north and A. mollis to the south. Later migration would then result in the ob- served occurrence of both species on both sides of the river at the present time. Too little is known of the actual present dis- tribution of either species, however, to do more than point out the possibility. Mello- Leitao cited the occurrence of both species at several localities not shown on the dis- tribution map (Map 2) but, since he once incorrectly synonymized the two species, his identifications cannot be relied upon. This genus presents more interesting problems in the area occupied by the mollis group than it does anywhere else in the Americas. Of these problems, the nomen- clatural difficulty is the most obvious, but least interesting. Once the Mello-Leitao types become available, the proper names should be easily assignable to the speci- mens. Of much greater interest is the fact that, in all of the Americas, only in the area from southern Brazil to northern Argentina is there definite evidence of sympatry of two or more species of Ariadna. A collec- tion from Diamantina contained a male of A. boliviana and one of the unplaced males mentioned earlier. Another collection taken in Buenos Aires contained two female mollis and two female boesenbergii. Col- lections from the immediate vicinity of Sao Paulo include at least three and _ pos- sibly four species, none as yet identified. Field study of the microhabitat choice and general ecology of Ariadna, especially in the Sao Paulo region, should provide division of habitat and might suggest some reason for the allopatry of J data on Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 most American species. In general, a given species of Ariadna appears to be able to tolerate a wide range of climatic conditions as long as its preferred microhabitat is available. As a result, most species have an extensive geographic range, unless they occur on islands. Even one island species, A. arthuri, ranges over a large area (Map 4). How four species of these spiders, conservative as they appear to be in choice of microhabitat, can occupy a small area simultaneously is at present a seemingly insoluble puzzle. The occurrence of mollis and boesen- bergii together, apparently at a_ single collecting site, is of interest for a further reason. A single vial received from the Museo Nacional de Ciencias Naturales in Buenos Aires contained two females of each species. All four specimens had been identified as A. mollis. Although it is difficult to assess the de- gree of similarity or difference in such a generally uniform genus, to the naked eye, mollis and boesenbergii are certainly very similar in appearance. They are the only two species of the genus with a conspicuous abdominal color pattern, they are of about the same size, and they show no more than an average number of differences from each other in spination. Although they may not be the most similar pair of species in the Americas, they are not far from being so. This is exactly the reverse of the phenomenon of character displacement which has gained considerable attention recently. (I might add that I have yet to see any strong evidence for the occurrence of character displacement in any spider species. It must be admitted, however, that no one has made a careful search for it.) Whether any sort of behavioral or eco- logical displacement occurs in these species is not known. The remaining species, not placed in any of the above groups, are either di- vergent from the three described groups or are at the moment unplaceable. Three of them, boliviana, obscura, and _solitaria, i f if ; may be related to the bicolor group. How- ever, only a single female of obscura, one juvenile of solitaria, and one female and two males of boliviana have been seen. Without more material, I prefer to leave them unassigned. Three species, tubicola from Venezuela, gracilis from the Amazon Basin, and fidicina from the Pacific Coast of North America, are the most divergent of the American species. Except for some possible connection of gracilis with the multispinosa group, none of them is similar to any other American species (as close similarity goes in this genus). It is especially unfortunate that among these three species, both sexes are known only for A. gracilis. No further speculation on the evolution of Ariadna in the Americas appears worth- while at present. Until the Old World fauna of the genus is revised, and much more collecting has been done in the south- ern hemisphere where many undescribed species probably exist, no reasonable phylogeny can even be suggested, much less defended. Study of the ecology and be- havior may provide clues to the evolution- ary history of the genus, but if the animals are as uniform in these characters as they are in morphology, solution of the problem will be difficult. TAXONOMIC CHARACTERS IN HAPLOGYNE SPIDERS WITH PARTICULAR REFERENCE TO THE GENUS ARIADNA The haplogyne spiders (Haplogynae ) include the families Plectreuridae, Digueti- dae, Scytodidae, Sicariidae, Dysderidae, Caponiidae, and Oonopidae. They are two-lunged or lungless spiders with simple external genitalia. This simplicity is re- garded as primary and primitive. Palpi of the males lack hematodochae, the inflat- able membranes that expand the palpal organ, and often consist of a simple globose bulb that tapers to a spinelike embolus. In a few families (e.g. Diguetidae, Dysderi- dae), the bulb may be subdivided, with a ARIADNA IN THE AMERICAS + Beatty 445 conductor, but the palpal organ never approaches the complexity usual in higher araneomorph spider families. The female has a patch of slightly differentiated cuticle in the genital area, but other external genital structures are either absent or con- sist only of shallow depressions. For about 100 years, araneologists have relied heavily upon external genitalia for distinguishing among species of spiders. The genitalic simplicity of the haplogyne families sharply reduces the usefulness of the genitalia as specific characters. The taxonomy of haplogyne spiders, and of mygalomorphs (orthognaths), which also have simple genitalia, has therefore been considered difficult, and has been re- latively neglected until recently. In descriptions of haplogyne spiders, a wide variety of specific characters has been used. The palp of the male, sometimes the genital area of the female, the sper- mathecae of the female, the number and arrangement of the eyes, the number and dentition of tarsal claws and _ cheliceral teeth, the shape of the sternum, the length and proportion of the legs, and the spi- nation of the appendages have each been described by one or several authors as being distinctive of certain species. Heretofore, no attempt has been made to determine the range of variation of “diagnostic” characters within populations of a species, or, at least, the attempt has not been reported. The validity of many species is therefore questionable. METHODS Examination of the genus Ariadna for usable . taxonomic characters involved several steps. The literature was searched, and a list of diagnostic characters employed in the genus was drawn up. Next, speci- mens of several species were examined for the possible existence of additional char- acters that had not previously been used taxonomically. With an extensive list of characters prepared, numerical data for 446 each specimen were recorded, and _ the results were analyzed to determine whether they would provide statistically significant characters. In published descriptions of Ariadna species, almost every external feature has been described at least once. The selection of characters by a particular author seems, however, to have been based only on whatever happened to strike his eye. Only occasionally have supposedly diagnostic characters been pointed out, and = uni- formity, system, and completeness are absent. Often the characters are reported incorrectly. Bryant (1948), Chamberlin (1916), Mello-Leitao (1916, 1947), Pe- trunkevitch (1926), and Simon (1891, 1893a) all reported that the fourth legs, in species of Ariadna they described, were entirely spineless. Examination of hun- dreds of Ariadna specimens, including types of species described by the authors cited, reveals that the fourth leg is never spine- less in Ariadna, except in very smal] juve- niles, and even there only rarely. This mistake led to an erroneous statement in Chamberlin’s diagnosis of A. murphyi (1920) in which he states “metatarsus IV, armed at distal end of leg instead of leg TV being wholly unarmed.” Various authors (Blackwall, 1858, 1863; Mello-Leitao 1917; Petrunkevitch 1929) have described the chelicerae as being without teeth, or have given an incorrect number of teeth. Again, the descriptions were shown to be in error by examination of one of the specimens seen by these authors, and by the constancy of the num- ber of cheliceral teeth in many other speci- mens. It was also common practice to describe the ventral spination of the first and second tibiae and metatarsi as consisting of some number of pairs of spines. The spines are, in fact, not arranged in pairs, but in two distinct longitudinal rows (Fig. 13). The rows often do contain the same num- pines, but more often they do not. a nen the two rows are equal in spine LWO ber oat Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 number, the spines are not always opposite each other. The only notation of spine position that proved satisfactory in this study was one based on the potential presence of one to four longitudinal rows of spines on each of the leg surfaces (see p. 438). The examination of specimens revealed no new categories of characters, but it did show that many more characters were available than had previously been used in descriptions of new species. The in- accuracy of many such literature reports was clearly demonstrated. The list of char- acters ultimately settled on for intensive investigation and _ statistical analysis in- cluded the entire spination of the append- ages, the cheliceral teeth, and a series of measurements. Other characters were omitted as being unsuited for statistical] treatment. Data were recorded on IBM cards and analyzed by the IBM 7094 computer at the Harvard Computation Laboratory. Methods and results of the analysis are discussed in a separate paper (Beatty and Bossert, in prep. ). ASSESSMENT OF THE CHARACTERS For data on intraspecific variation in these characters, see Tables 2 to 23. The number of cheliceral teeth proved to be a generic character. All species of Ariadna normally have three teeth on the promargin and one on the retromargin of the fang furrow. Variation from these numbers is infrequent. Spination of the palpal patella, ordinarily the presence of one spine as opposed to the complete absence of spines, was a useful character in many instances. There is a small to moderate amount of variation, usually not bilateral. A small series of specimens would be sufficient to show the normal condition in all species. The spines of the tibia and tarsus of the palp are highly variable in number. Only occasion- ally was the difference between two species significant, and even then only unilaterally. Ranges in spine number on these segments usually overlap to such an extent that they are unusable as specific characters. The spines of the first two legs present most of the useful taxonomic characters to be found in the spination of the genus. The femur of most species bears one large prolateral spine. Variation of this char- acter in females and juveniles is virtually absent. (In males the spine is often sup- pressed.) Two species, A. gracilis and A. multispinosa, have two or three spines in this position. Dorsal femoral spines are normally either present or absent. Normally spineless species may have one dorsal spine, species normally having spines usually have two to four. Overlap of the range is rare. The presence or absence of prolateral and retrolateral spines on the tibiae is a very useful character. In species having lateral spines on these segments, difference in the number of spines is usually not helpful. Ventral spination of the tibiae is an excellent character, particularly that of the inner ventral row of the second tibia. With respect to the basic pattern of spination on this surface, there are two groups of species: in the first group the maximum number of spines in all rows is normally four, in the second the minimum is nor- mally five. In most species of the first group, the variation is only slight, but in A. pilifera and A. caerulea, it is consider- able. Among the species in the second group, the number of spines per row is much less constant. In all species, the inner ventral row on tibia II is least vari- able within a species, and it also shows considerable interspecific variation. The ventral rows of spines on the metatarsi usually have the widest range of variation of any spination character. Nevertheless, given a small series of specimens, they provide useful supplementary characters at least for species that fall near the ends of the range of intrageneric variation. Most spine arrangements of the third AAT ARIADNA IN THE AMERICAS * Beatty leg are not taxonomically useful. The ventral and lateral spines of the metatarsi are the most constant intraspecifically, but the range of interspecific variation is slight. Ventral spines of the tibiae are more variable intraspecifically, and also show a small range of variation from species to species. No character derived from the spination of the third leg can be considered often useful. The spination of the fourth leg, although denied existence by many authors, does provide one very useful character: the number of spines in a peculiar comb found on the inner ventral surface at the distal end of the metatarsus. This comb contains two spines in some species, three in a few others, four in many, and more than four in a few. Variation is slight, again with the exception of A. pilifera and A. caerulea. One or two spines are often present in an outer ventral row on metatarsus IV. Other spines of leg IV are usually dorsal or retrolateral on the femur, or ventral on the tibia; their occurrence is sporadic. The differences in measurements almost all resolve themselves into differences in overall size. The proportions of body and legs are remarkably constant throughout the genus. Here and there a species shows a distinctive feature, such as the relatively long narrow carapace of A. maxima or the short distal podomeres of A. tubicola, but, in general, a description of the proportions of one species would suffice for most of the others. A certain tendency exists for species that are closely similar in size to be quite distinct in spination and vice versa. The distance from posterior median to posterior lateral eyes, expressed in diam- eters of a posterior median eye, has been used rather extensively as a species char- acter by earlier authors. There certainly are interspecific differences in this char- acter, but so many variables are involved in assessing it that its use is very difficult. The eyes are so small that obtaining an accurate measurement is not easy at best, 448 and a slight difference in the angle at which the observed specimen is positioned under the microscope makes a great dif- ference in the distances being measured. Even with careful measurement, it was found that the intraspecific variation is considerable as compared with that men- tioned in published descriptions. I have felt that some use of this character should be possible, because the differences be- tween some pairs of species are striking to the eye. Thus far, however, no generally satisfactory method of describing the char- acter has been found. Characters not included in the statistical analysis are the following. Color was often described at length by early araneologists, and was used as a diagnostic specific character. But the hue, being likely to vary rather extensively within a species, is not especially reliable, whereas the pattern of coloration may be. In Ariadna, three groups of species are distinguishable on the basis of hue and pattern. One group is composed of species of light coloration, the abdomen ranging from whitish to yellow-orange, the car- apace from yellowish to deep red-orange. A second group is dark, with the abdomen purplish gray to dark brown, the carapace brownish or deep mahogany to almost black. Neither of these groups shows any distinctive color pattern on the abdominal dorsum. The third group contains a few species having a distinct dorsal abdominal pattern of yellow transverse bars on a purplish gray or brown background (Fig. I): Only the abdominal pattern is con- sistently usable as a diagnostic character. Color variation is extensive in many species, sometimes, although rarely, to the degree that an adult individual of a nor- mally dark species may be of a light color. The first instar young are, of course, un- pigmented, and darken gradually as they mature. Newly molted individuals are whitish, and darken over a period of days. In a few species of Ariadna, dark annuli Bulietin Museum of Comparative Zoology, Vol. 139, No. 8 are present on the legs. These seem nor- mally to be constant, although the number of specimens on hand is relatively small. The presence or absence of these annuli will quite likely prove tobe a useful diag- nostic character. The density, length, and color of the pilosity have been used in Ariadna taxon- omy. They can, under the proper circum- stances, be helpful, but are too variable to be diagnostic. Further, their proper use requires recently collected specimens, pref- erably ones which have molted only a short time prior to collection. In the normal activity of the spider some of the hair may be lost, and in preserved specimens the hair is often almost completely rubbed off, especially from the legs. In his description of Ariadna pilifera, O. P.-Cambridge (1898) said, “Behind the posterior extremity of the sternum, be- tween the coxae of the fourth pair of legs, is a small but distinct shining reddish brown chitinous plate of a truncate conical form, apparently beneath the connecting pedicle; on each side of it is a small oblique slit or orifice (perhaps spiracular open- ings?) The shape of this plate is probably a good specific character.” The plate re- ferred to is a ventral sclerite of the pedicel. It is of no taxonomic use, its shape being practically invariable from one species to another. Portions of the sclerite are often partially covered by folds of the thin ad- jacent cuticle. These folds may have pro- duced the appearance of lateral slits noted by Cambridge. I failed to find any such slits. Accurate observation of the shape of the sclerite is hindered by the cuticular folds, and efforts to uncover it are likely to damage the specimen. The structure of the male palp is quite useful in distinguishing species of Ariadna, even though it may not always be com- pletely diagnostic in itself. The differ- ences of greatest importance are the size of the organ relative to the size of the spider, and the length, thickness, and curvature of the embolic portion. The palp is more effectively described by drawings (Fig. 12) than it is in words; its generally small size and the helical curvature of the embolus make accurate and consistent measurement of a series of specimens al- most impossible. Proportions of the palpal organ do not present noticeable intra- specific variations, but the small number of males of most species in collections has prevented any significant study of vari- ation. The metatarsus of the first leg of the male is perhaps the best single diagnostic character found in Ariadna. It shows striking interspecific variation in diameter, curvature, spination, and possession of apophyses (Figs. 20-23, 50). The intra- specific variation shown by the only ade- quate sample of males (A. bicolor from Pennsylvania) does not affect the overall appearance of the metatarsus. Unfortu- nately, males of only about half the American species are known. The female has no epigynum, there being only a shiny, brownish, somewhat elevated patch of cuticle in the genital area. The internal genitalia present no usable taxonomic characters. There is a single median “seminal receptacle,” which probably does not actually receive sperm. Dorsal and posterior to the receptacle is a large membranous bursa copulatrix. Both receptacle and bursa have a _ uniform structure throughout all the American species. A TAXONOMIC REVISION OF AMERICAN ARIADNA In spite of the phylogenetic position accorded the haplogyne spiders, the taxo- nomic problems they present have not, in the past, been fully appreciated or ade- quately studied. Most araneologists have placed the haplogyne families at or near the base of the araneomorph line. The genitalic simplicity has been viewed as primitive, and other characters are be- lieved to provide supplementary evidence of primitiveness. ARIADNA IN THE AMERICAS + Beatty 449 If these families are really the most primitive araneomorphs, they offer, in the absence of an adequate fossil record, probably the best material for a study of spider evolution. If the lines of evolution leading from mygalomorph to advanced entelegyne spiders are distinguishable any- where in living animals, they should be found among the haplogynes. Until very recently the haplogynes have, unfortu- nately, been among the least known groups of spiders. During the past several years, revisions of some primarily North American hap- logyne families have been published (Gertsch 1958a, b, and c). Currently, Cooke in England and Alicata in Italy are studying the dysderid subfamily Dysderi- nae, and in America, Chickering is study- ing the family Oonopidae. But these studies constitute only a beginning of an understanding of haplogyne spiders. The fauna of the south temperate regions has been little examined. There is reason to believe that this fauna, when it becomes well known, may change our ideas of haplogyne classification drastically. Fur- ther, the studies mentioned above still rely heavily on genitalia for separating species, although females within a genus of haplogynes may be completely indis- tinguishable on this basis. Because of the absence of complex secondary genitalia, araneologists seem to have been at a loss for convenient species characters. The usefulness of characters other than genitalia, eye arrangement, and a few other obvious and traditional fea- tures, has been only slightly explored. It must be admitted that some haplogyne spiders actually have fewer external morphological structures than entelegyne spiders, and that identification of females in such genera may be extremely difficult. The excellent revisions of plectreurids, diguetids, and _ loxoscelids by Gertsch (1958a, b, and c; 1967) take into account leg length and proportions, spination, size, and eye relationships, in addition to geni- 450 talia. Even with these additional characters, the genus Loxosceles remains a difficult one to deal with. A paper by Cooke (1965b) is the first extensive investigation of non-genitalic morphological characters of haplogynes known to me. This present study, therefore, has been undertaken not only to provide a much needed revision of the genus Ariadna, but also to provide clues to kinds of characters that may prove useful in other genera or families of haplogyne spiders. METHODS Measurements. Specimens were mea- sured by the use of ocular grids in binocular dissecting microscopes. A variety of micro- scopes and grids was used for making measurements, so that neither the magni- fications nor the limits of accuracy of the figures are constant. In general, however, the measurements are accurate to about one-tenth unit of the micrometer grid, as is shown by repeated individual measure- ments. For the larger dimensions, the measurements are accurate to 0.1 mm, for the smaller dimensions to about 0.015 mm. Measurements were made with the speci- men in as nearly horizontal a plane as possible, along the lines shown in Figures 5 So. and: 5: A series of 24 measurements of various parts of each specimen was taken. Rel- atively few of these proved useful in species discrimination (see Beatty and Bossert, in prep.). The range and mean for total length, carapace length and width, and sternum length and width are given in the species descriptions. Figures and descriptions. The color de- scriptions are based upon alcoholic speci- mens, collected as recently as _ possible. Comparison of old museum specimens with living specimens of the same species shows that Ariadna generally retains its color well in alcohol, provided the initial preservation \ The carapace and appendages change color very slightly and properly done. Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 slowly, if at all. The abdomen is the most sensitive to improper preservation, and the most likely to change with increasing length of time in preservative. Well-pre- served specimens that have not been al- lowed to dry and have not undergone shrinkage of the soft parts away from the cuticle do not differ noticeably in color from living specimens unless they are quite old (fifty years or more). Illustrations were made with the aid of a camera lucida, usually at a magnification of 12x or 20X. In most cases only line drawings are presented. Color patterns are present in few species of Ariadna. A few drawings are presented to show the general appearance and structure of the genus (Figs. 1-7, 11-14). Structures of the cephalothorax and appendages of the females are so uniform that there is no reason to present drawings of these parts for each species. Similarly, the female genitalia have not been useful in dis- tinguishing species. The illustration given for one species (Fig. 14) serves equally well for all others. Two views of the male palpi are shown, from the prolateral and retrolateral aspects. The first metatarsus and tibia of the males are illustrated in dorsal view. Hair coverings are omitted from most drawings. Records. For most species few specimens were available and all records are given. For the United States, locality records are listed alphabetically by state, and counties of the states are also listed alphabetically. Specific localities are given for most spe- cies. For the common and widespread A. bicolor, only states and counties are re- corded. The major geographical areas are listed in geographic order, beginning in the north. Caribbean islands are listed alpha- betically under West Indies. States and territories of Latin American countries are listed alphabetically. The number of specimens collected at each locality is given. In addition to the é and 2 symbols used to represent mature specimens, “o” represents immature in- dividuals. Occasional collectors are named in the text. Family DYSDERIDAE Dysdérides C. L. Koch, 1837, Ubersicht des Arachnidensystems, vol. 1, p. 20. Type genus of family: Dysdera, Latreille, 1804. Subfamily SEGESTRIINAE Segestriinae Simon, 1893, Histoire Naturelle des Araignées, vol. 1, p. 319. Segestriidae: Petrunkevitch, 1933, Trans. Con- necticut Acad. Arts Sci., vol. 31, pp. 333, 365. In this paper, Petrunkevitch raised the subfamily Segestriinae to family rank. Type genus of subfamily: Segestria, Latreille, 1804. Ecribellate, haplogyne spiders of the sub- order Araneomorphae (=Labidognatha). Respiratory system consisting of a pair of book lungs, and tracheal tubes opening through a pair of spiracles just behind the lung slits. Heart with two pairs of ostia. Colulus small but conspicuous, short and wide, bearing several setae. Six spinnerets set close together, short, the anterior and posterior pairs two-segmented, the median pair one-segmented. Anal tubercle wide, and anteroposteriorly compressed. Chelic- erae barely united at base only, without apical lamina, normally bearing three pro- lateral and one or two retrolateral teeth. Labium longer than broad, not fused with sternum. Endites long, parallel, not con- verging in front of labium. Eyes six in three diads, the anterior median eyes lost. Tarsal claws three, the two superior claws pectinate, the inferior claw with a single minute tooth. Female pedipalp with a short claw. Third pair of legs directed forward rather than backward as with most other spiders. Legs and body covered by fine long hairs, appressed to nearly erect on the body, often erect and forming a conspicuous fringe on the anterior legs. Tarsi and metatarsi often scopulate, espe- cially in males. Rows of setae or heavy spines on the legs, especially on the two anterior pairs. Female copulatory organ ARIADNA IN THE AMERICAS + Beatty 45] much like that of the Diguetidae and Plectreuridae, with large membranous bursa copulatrix, and a single median sclerotized structure that is probably homologous with the seminal receptacle of other spiders, but does not appear to function as a site of sperm storage. Male palpal organ a sim- ple pyriform or long-conical bulb with a spinelike embolus, lacking any accessory structures. Kaston (1948, 1952) followed Gerhardt and Kastner's (1938) arrangement of spider families, with some modifications, mostly the splitting of various families. This increase in the number of families was largely due to the work of Petrunkevitch (1933, 1939), and included the separation of the dysderids into Dysderidae and Segestriidae. Cooke (1965a) apparently follows this scheme also. He states: “The family is divided into four tribes: Dysde- rini, Harpactini, Orsolobini, and Rhodini.” No segestriine genus is mentioned in his enumeration of genera of the family. Other araneologists, Bonnet (1955) for example, have continued to include the segestriines in the family Dysderidae. The subdivision of the family was dis- cussed by Petrunkevitch (1933) in the following words: “It seems to me now, however, more reasonable to elevate the subfamily Segestriinae to the rank of a family. They have many characters dif- ferentiating them from their nearest rel- atives, the Dysderinae, such as the position of the third pair of legs, the articulation of the coxae, the arrangement of the eyes. However, the tracheal system is alike in both Dysderidae and Segestriidae.” In his key in the same work Petrunke- vitch distinguishes the two families thus: “Third pair of legs directed forward. Sternum separated from carapace by soft membrane as usual. Eyes in 3 diads. Family Segestriidae. “Third pair of legs normally directed backwards. Sternum connected with cair- apace by hard chitin. Eyes in a transverse oval. 3 to 2 claws. Family Dysderidae.” 4 Ol bo The union of carapace and sternum, by sclerotization of the pleural membrane, occurs in a wide variety of spider families. It is a character common to all members of a family only in the Palpimanidae. From examination of a variety of specimens, it is apparent that whenever a spider becomes heavily sclerotized over much of the body (for whatever the adaptive reasons), one of the first accompanying morphological changes is that the carapace and sternum become fused to each other by scleroti- zation of the intervening membrane. This feature has been seen in the _loricate oonopids, some caponiids, several genera of theridiids, at least three genera of araneids, and some clubionids, in addition to the families mentioned above. These families are presently distributed among three superfamilies. In the oonopids, as in many families, heavy sclerotization is correlated with very small size, in the clubionids and theridiids also with ant- mimicry, in araneids with apparent pro- tective modifications of the abdomen, and in the caponiids and palpimanids possibly with xeric habitat, although this is a guess. The fusion of carapace and sternum is apparently not a character of much sig- nificance at the family level. The position of the third leg, held for- ward with the anterior two pairs instead of backward with the fourth, has been considered a unique character of the Segestriidae. Actually the araneid genus Micrathena shows this character also and, to judge by preserved specimens, so does the genus Plectophanes, variously placed in the Agelenidae or the Toxopidae. In- dividuals of Dysdera and related genera also occasionally rest with the third leg in a position midway between the forward and backward positions, almost perpen- dicular to the body axis. By examining Petrunkevitch’s own work, ind that of Buxton (1913) and Millot (1931), both of whom he quotes, one may many similarities between dysderine segestriine spiders. In both groups, { Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 in the species examined, the heart has two pairs of ostia, the cephalothorax is sup- plied with tracheae that originate from the second pair of spiracles, and the thora- centeron is of the simple type. Buxton (1913) described the coxal ex- cretory glands of various arachnids. He found that mygalomorph spiders have two pairs of glands, which open at the bases of leg coxae I and III. Each gland is made up of a saccule, a collecting tubule, a labyrinth, and an ectodermally lined ex- cretory canal. In the araneomorph spiders, one pair of glands and the collecting tubules are absent, and the labyrinth is considerably simplified. Variation in coxal glands within the araneomorphs is arranged by Buxton in a sequence from most primitive to most derivative. This sequence begins with the Dysderidae and Sicariidae, in which the labyrinth is a conspicuous loop and _ the saccule is functional. In most other fami- lies examined the labyrinth is present, but the saccule appears to be nonfunctional. Finally, in the Filistatidae, Pholcidae, Theridiidae, and Araneidae, the labyrinth has nearly disappeared so that the saccule and excretory canal are almost directly connected with each other. The haplogyne genera examined by Buxton included Scytodes, Loxosceles, Si- carius, and Dysdera. These genera repre- sent both subgroups of the Dysderoidea, so it is probable that the coxal gland structure is uniform throughout the super- family. The alleged difference in coxal articu- lation does not appear to exist. In both dysderines and segestriines the coxa is open distally across its entire cross-sectional area, and the trochanter articulates with the wall of this terminal opening. Proxi- mally, the articulation of coxa and body wall in both groups is on the dorsal surface of the base of the coxa. The only externally apparent difference between the dysde- rines and segestriines, with respect to the proximal part of the limb, is that the base of each coxa and trochanter is more con- stricted in the dysderines than in the segestriines. Considering the differences among other spider families, the similarities between dysderines and segestriines are so many, and the differences between them so few and minor that I can find no morphological grounds for separating them at the family level. The possibility that behavioral or distributional information may support their separation can not be denied at present. Until such information becomes available, both groups should remain in the single family Dysderidae. THE GENERA The Segestriinae include four named genera: Ariadna, Citharoceps, Segestria, and Segestriella. Citharoceps of the Pacific coast of North America, and Segestriella of South Africa are probably not valid genera, for reasons which will be dis- cussed below. Segestria, as presently known, is primarily a Holarctic genus consisting (Bonnet, 1958) of about 25 species. In the Nearctic it is limited to western North America. Seven species are cited as occurring respectively in Mada- gascar, India, Australia, New Zealand, southern South America, and the Gala- pagos Islands. The distribution of Seges- tria outside the Holarctic is very poorly known. Quite possibly a number of ad- ditional non-Holarctic species remain to be described. Ariadna is a very wide-spread, but not quite cosmopolitan genus. Approximately 100 species have been described but, of the 41 described from the Americas, only a few more than half are valid, and the same may be true of the Old World species. Few species of Ariadna occur in north temperate regions. Ariadna_ bicolor, A. fidicina, and A. pilifera occur in the United States and Mexico, A. lateralis in Japan, and A. insidiatrix in southern Europe and northern Africa. The number of species ARIADNA IN THE AMERICAS + Beatly 453 of the genus increases rapidly southward. The largest number of species will prob- ably ultimately be found in the temperate southern hemisphere. Except by introduction by man, no other genus of the Dysderidae has attained the wide distribution of Ariadna. It occurs on all continents except the Antarctic, and on such isolated islands as the Seychelles, New Zealand, the Galapagos, Hawaii, and the Juan Fernandez Islands. As presently known in the Americas, the species of this genus are allopatric, except for a region including southeastern Brazil, Uruguay, and east central Argentina, within which at least six species occur. With more in- tensive collecting this pattern may change drastically. Segestriella, described by Purcell (1904), is characterized as “Allied to Ariadna, Aud., but differing in having the body elongate cylindrical, the abdomen obtusely produced beyond the spinners, and the fourth pair of legs very short, not reaching hind end of abdomen when stretched out, and with the femur very short and stout, strongly swollen dorsally, the width of the femur between dorsal and ventral edges about 1/2 its dorsal length and almost twice the width of the first femur.” These distinctions are simply accentuations of a few characters common to the genus Ariadna, Purcell’s careful and detailed de- scription of Segestriella gryllotalpa, the only species of the genus, shows that it accords completely with Ariadna in spine pattern and cheliceral teeth. I have seen no specimens of Segestriella, but feel sure that the genus should be considered synonymous with Ariadna. The genus Citharoceps of Chamberlin (1924) was distinguished from Ariadna solely on the basis of its stridulating ap- paratus. The files are two patches of coarse transverse grooves which extend along the sloping sides of the cephalic region (Fig. 10). The picks are tubercles at the base of each first femur on the inner surface. In most respects Citharoceps 454 agrees with Ariadna, although it is rather divergent in spination and proportions, and is here synonymized with the latter genus. Two species of Citharoceps have been de- scribed, but they seem conspecific. Ariadna and Segestria have generally been distinguished principally by their eye arrangement (e. g. Comstock, rev. ed. 1948). Because placement of the eyes is often not a constant character within spider genera, Gertsch (pers. comm.) sug- gested that Ariadna and Segestria might not really be distinct from each other. However, Simon (1893a) used three other characters in his key to these genera: the shape of the labium, the cheliceral teeth, and the spination of the first pair of legs. Each of these appears to be a significant difference between the two genera. Further differentiating characters are found in the leg and pedipalp proportions, spination of appendages other than the first legs, the shape of the male palpal tarsus, the articu- lation of the bulb with the palpal tarsus, and the abdominal pattern. Although few Old World species of Ariadna have been examined, I consider Ariadna and Segestria separate and well-marked genera. Table 1 summarizes morphological char- acters of Ariadna (including Citharoceps ) and Segestria. This summary is based on examination of all the New World species of Ariadna, and four species of Segestria— S. florentina, S. pacifica, S. ruficeps, and S. senoculata. A few specimens of various Old World Ariadna and unidentified species of Segestria were also examined briefly. The characters cited show a high degree of constancy within each taxon. Genus Ariadna Audouin Ariadna Audouin, 1825, Explication Sommaire des Planches. in Savigny, Description de Egypte, p. 109. Type species by monotypy: Ariadna insidiatrix Audouin, op. cit., from \lexandria, [Egypt]. Pylarus Hentz, 1827, J. Boston Soc. Nat. Hist., , Type species by present designation bicolor Wentz, ibid., from northern Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 Macedonia Hogg, 1900, Proc. Roy. Soc. Victoria, 13:85. Type species by monotypy: Macedonia burchelli Hogg, ibid., from Victoria. Citharoceps Chamberlin, 1924, Proc. California Acad. Sci., ser. 4, 12:607. Type species by original designation and monotypy: Citharoceps fidicina Chamberlin, ibid., from Ensenada, Baja Cali- fornia. NEW SYNONYMY. Segestriella Purcell, 1904, Trans. South African Phil. Soc., 15(3):165. Type species by mono- typy: Segestriella gryllotalpa Purcell, ibid., from Stompneus, St. Helena Bay, Malmesbury Div., South Africa. NEW SYNONYMY. Description. Eyes: Six in number, the anterior medians missing. Posterior row of four eyes straight or slightly recurved. Anterior row of two narrower than pos- terior row. Lateral eyes of each side con- tiguous. Median eyes contiguous or very narrowly separated. The eyes are thus arranged in three closely spaced diads. Carapace subrectangular, rounded pos- teriorly, squarish anteriorly, slightly nar- rowed at about middle of cephalic region. Head only slightly elevated, thoracic region shallowly depressed laterad of head. Mar- gin of carapace horizontally flanged or slightly upturned, producing a thin, dark marginal line. Labium one and a half to two times as long as wide, sub-rectangular or sub-hexag- onal, distally truncate and usually shallowly notched. Endites spatulate, medial distal margins (anterior to labium) parallel, tips reach- ing fang, total length about 1.5 that of labium. Chelicerae short and moderately tapered. The short fangs are only slightly curved. Anterior cheliceral margin armed with three tiny conical black teeth, the posterior margin with one. The number of cheliceral teeth is probably as constant as a meristic character can be. Of 200 chelicerae ex- amined in one species, one had two teeth on the posterior margin instead of the usual one; there was no other variation. Anteriorly one of 200 had four teeth, and eight of 200 had two teeth. A few of the latter cases were probably ascribable to injury before or damage after collecting. TABLE |. ARIADNA IN THE AMERICAS * Beatty 45! Ariadna Cheliceral teeth Labium shape Spination (females ) Three anterior, one posterior, all tiny and conical. Short and wide, narrowed at each end. Greatest width near base. Tibiae and metatarsi of first two pairs of legs wth two ventral rows of heavy spines, mostly of short to me- dium length and flattened. Fourth UT UI STRUCTURAL CHARACTERS OF THE SEGESTRIINAE Segestria Three anterior, two posterior, all tri- angular. Long and narrow, nearly parallel-sided. Greatest width near middle. Tibiae and metatarsi of first two pairs of legs with two ventral rows of long, slender, nearly round spines. Total number of metatarsal spines legs with few spines. Legs and palps Male palpus tip. Bulb inserted at of Loxosceles. Abdominal pattern marked. Relatively short and stout. Tarsus short, of nearly uniform width throughout its length, notched at middle of tarsal length. Palpus similar to that Usually lacking. When present, con- sisting of transverse bars on a con- trasting background. Venter un- on these legs much smaller than in Ariadna. Fourth legs with many spines. Relatively long and slender. Tarsus long, the distal two-thirds much narrower than the basal third, not notched at tip. Bulb inserted over most of the basal third of tarsus. Palpus similar to that of Scytodes. Usually present, some species with a self-colored abdomen. Pattern con- sisting of a median dorsal longitu- dinal row of dark lozenges on a light background, plus many scattered small dark spots dorsally and ven- trally. Exposed portion of labrum white, bluntly rounded at tip, reaching to the ends of the endites. Sternum ovate to sub-rectangular, trun- cate anteriorly, bearing a narrow pointed articular process opposite the middle of each coxa. Abdomen longer than wide, usually con- siderably so, overhanging posterior part of carapace, extending slightly beyond base of spinnerets and anal tubercle, sub- cylindrical. Palps short and stout, bearing in females and immatures a pair of spines ventral to the claw, and prolateral spines on tarsus, tibia, and sometimes patella. Legs relatively short and stout in fe- males, long in males. Order of length 1-2-4-3, or occasionally in some individ- uals 2-143, the first and second legs always nearly equal in length. Tarsi obliquely truncate, the pretarsus, bearing the claws, set on the upper surface of the truncation. Superior claws pectinate in a single row, inferior claw short, with a single tiny tooth. Palp of male with tarsus short and ellip- tical, bulb pyriform, inserted at middle of length of tarsus. Embolus a simple spine variously curved, usually rather well set off from the bulb. Female genital area marked only by shiny brown cuticle externally. Internally, a tiny median sclerotized structure prob- ably represents the seminal receptacle. A large membranous sac, corresponding to the bursa copulatrix of plectreurids and dysderines, extends a short distance an- terior and far posterior to the epigastric groove. The seminal receptacle has a pos- terior opening in its somewhat triangular base. From the opening, a short folded blind tube extends forward. A short, curved, pointed tube extends anterodor- sally from the base of the receptacle, also ending blindly. The size of this receptacle 456 alone suggests that it does not actually function as a storage place for sperm. In several mature females, dissection revealed a yellowish white mass in the posterior part of the bursa copulatrix. Probably this mass was sperm. In a discussion of the female genitalia of Dysdera crocata, Cooke (1966) stated: spermatozoa are transferred, not directly to the ‘sperma- theca’ but into the bursa . The small proportion of spermatozoa that get into the ‘spermatheca’ makes it unlikely that the true function of this structure is sperm storage.” The female genitalia of haplogyne spiders are in need of further study with a view to elucidating the origin and evo- lution of the complex epigyna of higher spiders. Size. Total length varies from 4.0 to 16.0 mm in mature members of the genus, carapace length from 2.1 to 7.7 mm. Leg proportions are rather uniform in most of the American species. Leg I is usually longest, but leg II sometimes exceeds it. The difference between the two rarely ex- ceeds one millimeter and is usually much less. In females the first leg is about 2.5 times the length of the carapace, leg II very slightly shorter, leg III less than and leg IV slightly more than twice the carapace length. The longer-legged males have leg I about 3.5 times the carapace length, leg II about three and a quarter, leg III about two and three-quarters, and leg ITV about two and one-third times the carapace length. Coloration. Color in the genus Ariadna is generally dull, ranging from yellowish through reddish orange and mahogany to a deep brown that appears black to the naked eye. The extent of tanning of the cuticle is probably the chief determiner of color of the carapace and appendages. In only a few species does the abdomen bear a distinct color pattern. When present, the pattern ordinarily consists of transverse bars of yellowish on a purplish gray back- ground. By extension of the area of yellow- Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 ish, the pattern may come to be one of dark bars on a light background. The color pattern in Ariadna is never a median longi- tudinal series of dark lozenges, as is the rule in Segestria. The anterior legs are usually of about the same color as the carapace. The other legs become progressively lighter in color posteriad. Spination. The tarsi, trochanters, and coxae of all legs lack spines in both sexes of Ariadna. In only a few species are there patellar spines. The tibiae and metatarsi, especially the first two pairs, bear many spines. Most obvious of these are the two ventral rows of enlarged, flattened, for- ward-pointing spines. Femora also bear spines, on the dorsal and _ prolateral surfaces only. Key To AMERICAN Species oF ARIADNA Males la. First metatarsus armed with one or two strong apophyses, or with short thick spines set on low protuberances; meta- tarsus in dorsal view usually sinuous or strongly bent inward near base (nearly straight in A. peruviana). (Figs. 32- 33, 39). ee 2 lb. First metatarsus without strong apo- physes or short thick spines; in dorsal view either straight or slender and sinu- ous: (Figs: 21,51). eee 6 2a. First metatarsus without lateral or ven- © tral apophyses, with two or more short heavy spines arising from low protuber- ances (apophyses never bear heavy spines distally ); metatarsus sinuous or nearly Straight, (Bigs) 32) °39).\ =a 3 2b. First metatarsus with one or two large lateral or ventral apophyses, without short heavy spines; metatarsus curved or strongly bent inward at base. (Fig. 38). 4 3a. Heavy spines ventrolateral, opposite each other, one pair just distal to middle of metatarsus, usually a second pair at distal end of metatarsus. (Figs. 8, 26-27, 32). Caribbean from Curacao to Florida Keys EN te aes 3b. Heavy spines lateral, not opposite each other, a single pair placed proximal to middle of metatarsus . (Figs. 34-35, 39). 4a. With a single ventral apophysis which bears an unenlarged distal spine (this Ab. 5a. 5b. 6a, 6b. as Tb. 8a. 8b. 10a. 10b. spine is easily broken off). (Figs. 7, 28-29, 33). Amazon Basin gracilis With a pair of lateral apophyses which do not bear distal spines The prolateral apophysis slightly but distinctly proximal to middle of meta- tarsus, the retrolateral apophysis more proximal; middle pair of ventral spines inserted distal to the apophysis (rarely, the prolateral spine is at the base of the prolateral apophysis); transverse diam- eter of palpal bulb less than twice the maximum width of palpal tibia. (Figs. 38, 42-43). Southern Canada; Maine to Florida, southern California, and north- western Mexico. bicolor The prolateral apophysis distal to middle of metatarsus, retrolateral apophysis about at middle; middle pair of ventral spines inserted at bases of apophyses; transverse diameter of palpal bulb twice or more the maximum width of palpal_ tibia. (Figs. 20, 31, 36-37, 40-41). Arizona and Mexico. pilifera Metatarsus I straight in dorsal view 7 Metatarsus I slender and sinuous in GOrsall Ravi waa ae mk ean MST 2 10 Abdomen with a distinct dorsal pattern OfmtnAnISVerse sbaTsne=eemeens mT, Be 8 Abdomen unicolored, or with color mark- ing a broad longitudinal stripe. 9 Patellae of legs I and II each with a prolateral spine; metatarsal comb of leg IV with five to eight spines; midpiece of palp short, about equal to embolic por- tion in length. (Figs. 1, 16, 17, 22). Southeastern Brazil to Argentina. _ mollis Patellae of legs I and II without spines; metatarsal comb of leg IV with four spines; midpiece of palp long, much longer than embolic portion. (Figs. 3, 46-48). Uruguay and Argentina. scat sh Tag Bot OA ES ey ee eee boesenbergii Metatarsi I and II without spines (Figs. 50, 53, 56). Revilla Gigedo Islands ( Mexico ). weaveri Metatarsi I and II with spines. (Figs. 2, 18-19, 21). Chile, including Juan Fer- nandez Islands. maxima Embolic portion of palp much longer than midpiece, equal to or exceeding diameter of bulb; midpiece and embolic portion forming about a ninety degree angle with each other. (Fig. 55)... 11 Embolic portion of palp equal to or much shorter than midpiece, shorter than diameter of bulb; midpiece and embolic portion forming a_ relatively shallow angle with each other. (Figs. Wey Ilb. ae lb. 3a. 3b. 4a. Ab. Gb. ARIADNA IN THE AMERICAS + Beatly 457 44-45, 49). Colombia and Ecuador. eee. caerulea Without dorsal spines on femora I and II. (Figs. 23-25, 30). Central America. ok shan : isthmica With dorsal spines on femora I and II. (Figs. 51-52, 54-55). Bolivia and south- eastern Brazil. .............. boliviana Females Dorsal spines absent from femora I and II in 80% or more of individuals, remain- ing 20% usually with only one spine on each of these femora (rarely two or three ySpines i sok pei Cree cone eee ed Dorsal spines present on femora I anc II in 90% or more of individuals, usually four or more spines on each first femur, and two or more on each second femur. — 10 Comb of metatarsus IV with two or three spines; ventral tibial spines 4—4 on leg I, A= lor 4—=2\,on leg all; 2 ae ee 3 Comb of metatarsus IV with four or more spines; ventral tibial spines of leg I 4-4 or more, of leg II various com- binations|frome4—O0Mtor 5 —( ee 5 Comb of metatarsus IV with two spines; Patos Island and Cedros Island, upper Gulf of California, and coast of Sonora. MS Ba 2 Se a hn EE eee pragmatica Comb of metatarsus IV with three SPDUTVES 5k se itt BEDE Ae le an re ee +t Metatarsus III with one inner ventral spine (95%): third tibia with no pro- lateral spines (90%); posterior median eyes (PME) averaging 1.1 times their diameter from posterior lateral eyes (PLE). United States and northwestern IMIGxa CO, (ps eee ee ee ter ae ee bicolor Metatarsus III with two inner ventral spines (78%); third tibia with one or two prolateral spines (70% ); PME aver- aging 1.6 times their diameter from PLE. Revilla Gigedo Islands, Mexico. _. weaveri Ventral spines of tibia II 4-4 or more. — 6 Ventral spines of tibia I] fewer than 4-4, Ventral spines of tibiae I and II usually five to seven in each row; comb of metatarsus IV usually with five spines, sometimes with four; tarsi and metatarsi short; anterior legs with purplish gray annuli. Venezuela. tubicola Ventral spines of tibiae I and II usually four in each row; comb of metatarsus IV with four spines; tarsi and metatarsi of normal length; legs without annuli. _. 7 With prolateral and retrolateral spines on tibiae I and II. Central America. — isthmica 458 8a. 8b. 9a. Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 Without lateral spines on tibiae I and II (except occasionally on retrolateral sur- face of tibia I in cephalotes). ____-_-.-- 8 Abdomen with a dorsal pattern of trans- verse bars on a contrasting background. Uruguay and Argentina. boesenbergii Abdomen unicolored dorsally. Colombia and Ecuador. _.....- ae cephalotes With stridulating grooves on sides of cephalic region; abdomen occasionally with dorsal pattern of light bars on a dark background. California and Baja (Cellier: <(loiez, O10) fidicina Without stridulating grooves on cephalic region; abdomen purplish gray. Colom- bias sHicuadormeand, Perus esse = cephalotes With either two spines in the comb of metatarsus IV OR two to three prolateral spines on femur I. (Fig. 7). — 11 With more than two spines in the comb of metatarsus IV AND only one_ pro- lateral spine on femur I, _2 16 Comb of metatarsus IV with four spines; femur I with two or three prolateral spines. Comb of metatarsus IV with two spines; femur I with one prolateral spine. 13 With lateral spines on tibiae I and II; ventral spines of tibiae I and II five to eight in each row. Amazon Basin of Brazil and Peru. gracilis Without lateral spines on tibiae I and II; ventral spines of tibiae I and II seven to eleven in each row. Hispaniola. iota een eRe. UE EIEN BN, caer multispinosa Ventral spines of tibiae I and II four or fEWEr IM CAC li TO Wee = ie ee 14 Ventral spines of tibiae I and II five or TAN OTE WTA AC Me TO Wis ee 115) Palpal patella spineless; femur IV with one to three dorsal spines near base. Chinchas Islands off Peru. 222. -= murphyi Palpal patella with one spine; femur IV without dorsal spines. Eastern Brazil. a Peel EE SNR lien _ obscura Tibiae I and II each with one to two retrolateral spines; carapace length 2.0— 3.5 mm. Florida and West Indies. — arthuri Tibiae I with three retrolateral spines each, tibiae II with none; carapace length 3.7-4.8 mm (see text). Galapagos Islands. a Sees _ tarsalis Tibia II with one to three spines in inner veutral row (lOO Ze), te eaten ence ene i) Tibia IL with four spines in inner ventral row (95%), eae = 20 Dorsum of abdomen with a conspicuous vattern of transverse bars on a contrast- ng background: comb of metatarsus IV with five to eight spines. Southeastern Brazil to; Ane entinas ee ane mollis 17b. Dorsum of abdomen without such a pat- tern, usually unicolored; comb of meta- tarsus 1V with three to four spines. 18 18a. Metatarsi II and III usually each bearing a retrolateral spine; total number of spines on metatarsus III 10-11. Bolivia and! (Brazil )2t. 2 oe boliviana 18b. Metatarsi II and II spineless retrolater- ally; total number of spines on meta- tarsus III five to nine. _ 19 19a. Palpal patella without spines (907). Southwestern United States and Mexico. PU hh ln aa Ea pilifera 19b. Palpal patella with one spine (83%). Venezuela. (Js ee tovarensis 20a. Palpal patella without spines (100%). 21 20b. Palpal patella with one to three spines (95%). au. 22, 2la. Comb of metatarsus IV with three spines (75%); femur IV usually with one to four dorsal spines. (Peru. === peruviana 21b. Comb of metatarsus IV with four spines (75%): femur IV without dorsal spines. Colombia and) Ecuador. == caerulea 22a. Carapace length 4.2-7.7 mm, mean of 78 specimens 5.97 mm. Chile, including Juan Fernandez Islands. _._.......- maxima 22b. Carapace length 4.2 mm (single im- mature specimen). St. Vincent Island, West 2Indies, | 4... solitaria SPECIES DESCRIPTIONS Ariadna bicolor (Hentz) Figures 38, 42-43. Map 1. Pylarus bicolor Hentz, 1827, J. Boston Soc. Nat. Hist:, 4:225, ¢pl.22) figs, "Sa Oa live specimens from northern Alabama, lost. Pylarus pumilus Hentz, 1827, ibid., 4:226, pl. 2, fig. 5, juvenile. Type specimens from North Carolina and northern Alabama, lost. Ariadna pallida C. L. Koch, 1843, Die Arachniden, 10:90, pl. 350, fig. 817, @. Female holotype from Pennsylvania, not seen. Ariadne rubella Keyserling, 1877, Verhandl. der konig. kais. Zool. Bot. Gesell., Wien, 1877:229. Female holotype from Louisiana, New Orleans, in British Museum (Natural History ), examined. A. pennsylvanica: Simon, 1891, Proc. Zool. Soc. London, p. 556, nomen nudum. Simon attrib- uted this name to C. L. Koch, so A. pallida, from Pennsylvania, is evidently the species he intended to refer to. Ariadne mexicana Banks, 1898, Proc. California Acad. Sci., ser. 3, Zoology, 1(7):212. Syntypes from La Chuparosa, Baja California. One in California Acad. Sci. Collection, destroyed. The other in the Museum of Comparative Zoology, examined. NEW SYNONYMY. Ariadna philosopha Chamberlin, 1924, Proc. Cali- fornia Acad. Sci., ser. 4, Zoology, 12(28):606. Female holotype from Isla Partida, Gulf of California, in California Acad. Sci. Collection, examined. NEW SYNONYMY. Discussion. Although no types of either of Hentz’s species exist, the name A. bi- color may be assigned with certainty to this species. Specimens from both North Carolina and northern Alabama are avail- able, and are identical with the nearly continent-wide species to which the name A. bicolor has been applied for over 100 years. Other species of Ariadna are of very limited distribution in the United States: A. arthuri in the southern part of peninsular Florida and the Keys, A. pilifera in southern Arizona, and A. fidicina in the southern half of California west of the mountains. Map 1 ARIADNA IN THE AMERICAS * Beatty 459 BICOLOR A PILIFERA m@ FIDICINA @ PRAGMATICA OQ WEAVERI ve atlivtel G6: Au~-L~. G8 vy ISTHMICA In spite of Hentz’s denial of this fact, his Pylarus pumilus, described in half a dozen lines, can be nothing but the juve- nile of A. bicolor, as suggested by Emerton (1875). Koch’s description of A. pallida contains nothing distinctive of any par- ticular species of the genus. A large series of specimens from Pennsylvania, the type locality of A. pallida, differs in no signifi- cant way from other populations assigned to A. bicolor. Ariadne rubella Keyserling, from New Orleans, differs only by its reddish color, according to the description. Some Louisi- ana specimens are distinctly more reddish than most A. bicolor, but are otherwise indistinguishable. In the absence of freshly collected specimens from Louisiana, it is not even certain that the reddish color is natural. Banks’s A. mexicana was supposedly dis- tinguishable from A. bicolor by its more 460 slender build and smaller eyes. The former difference does not exist. The eyes are smaller, leaving a much wider space _ be- tween the posterior median and posterior lateral eyes than in U. S. specimens of bicolor, from which mexicana does not otherwise differ. This may be a geographic variation, but so few Mexican specimens are available that I do not feel justified in recognizing it even as a subspecies. Chamberlin’s A. philosopha from Isla Partida in the Gulf of California has small eyes also. It also occasionally has a pro- lateral spine on the second tibia, which is extremely rare in A. bicolor from other areas. Other differences are well within the range of individual variation of bicolor. The spine characters which supposedly distinguish philosopha from mexicana are much too variable to be so used. Ariadna bicolor is so remarkably uniform in most characters throughout the United States that, with additional material, it may be- come desirable to recognize a subspecies for the Mexican specimens. Color. Female. Described from a freshly collected specimen from Gibraltar Island, Ottawa County, Ohio. Carapace mahog- any, darker in head region. Abdomen purplish gray above and below, with a satiny luster, sometimes slightly iridescent, finely striate with yellow lines. Striations longitudinal on sides of abdomen, trans- verse ventrally and mid-dorsally. A nar- row yellowish lateral line on each side of abdomen, ending at the upper surface of the anal tubercle. First legs brown to very dark brown, sometimes with an olive green tinge, especially soon after molting. The other legs progressively paler posteriad, the fourth pair light mahogany or dark yellow- brown. Chelicerae very dark brown, al- black. Palps brown _ proximally, darker distally. most Labium and endites brown, endites with vhite tips, labrum white. Sternum bright Jany, margins Spinnerets whit vith pale brown transverse mark- darker. Bulletin Museum of Comparative Zoology, Vol. 139, No. § ings. Anal tubercle whitish. Epigastric plates yellowish, ventral surface of pedicel white or translucent. Genital area in the shape of a low wide triangle, slightly swollen, the cuticle brown and shining. Male. Overall pattern as in the female, but generally paler, the legs and carapace yellowish to medium brown in most cases. Structure. Dimensions of 80 females: total length 6.1-15.0 mm, mean 8.73 mm; carapace length 3.0-4.8 mm, mean 3.84 mm; carapace width 1.9-3.1 mm, mean 2.40 mm; sternum length 1.7-3.8 mm, mean 2.20 mm; sternum width 1.1-1.6 mm, mean 1.31 mm. Dimensions of 22 males from a single locality: total length 5.4-7.2 mm, mean 6.15 mm; carapace length 2.7-3.4 mm, mean 2.96 mm; carapace width 1.9-2.4 mm, mean 2.04 mm; sternum length 1.5- 2.0 mm, mean 1.75 mm; carapace width 0.9-1.2 mm, mean 1.04 mm. No other col- lection of males contains more than one to a few specimens. Some of these, how- ever, are larger than any specimen in- cluded in the above sample. Spination. See Table 4. Diagnosis. The comb of three spines on the fourth metatarsus distinguishes A. bi- color from all other New World Ariadna except A. peruviana, A. weaveri, and some specimens of A. pilifera. Females may be distinguished from peruviana by having only two, instead of four spines in the inner ventral row on tibia II, and from pilifera by the absence of lateral spines on the first two tibiae. Females of weaveri are quite similar to those of bicolor, but are readily distinguished by distribution, and by the characters given in the key. The pair of large lateral apophyses on the first metatarsus of the male bicolor separates it from all other known males except that of pilifera. The apophyses of pilifera are more distal, and the palpal bulb propor- tionately larger than in bicolor. Distribution. Maine to Florida and west to southern California, Baja California, and northwestern Mexico. Specimens are not available, however, from large parts of the central plains and the Northwest. The only Canadian records are from two islands in western Lake Erie. Records. County records only are given for the United States. CANADA. ONTARIO: Big Chicken Is- land, Lake Erie, under boards and stones; East Sister Island, Lake Erie, under boards and stones. UNITED STATES. ALABAMA: Col- bert, Jackson, Madison, Marshall. ARI- ZONA: Coconino. ARKANSAS: Benton, Washington. CALIFORNIA: San Diego. COLORADO: Chaffee, El Paso, Fremont. CONNECTICUT: Fairfield, New Haven. DISTRICT OF COLUMBIA. FLORIDA: Alachua, Gadsden, Hernando, Highlands, Indian River, Jackson, Lake, Levy, Lib- erty, Marion, Nassau, Orange, St. Johns. GEORGIA: Floyd, Fulton. ITLLINOIS: Bond, Jackson, Macoupin, Union, William- son. INDIANA: Brown, Crawford, Owen. KENTUCKY: Carter, Harding, Madison, Wolfe. LOUISIANA: Beauregard, Caddo, East Baton Rouge, Grant, Jefferson, Madison, Orleans, St. Charles. MAINE: Knox. MARYLAND: Montgomery, Prince Georges. MASSACHUSETTS: Barnstable, Essex, Middlesex, Norfolk, Plymouth. MISSISSIPPI: George, Hinds, Jackson, Oktibbeha. MISSOURI: Boone, St. Louis, Taney, Wayne. NEW JERSEY: Cape May, Mercer, Ocean. NEW MEXICO: county unknown. NEW YORK: Albany, Bronx, Nassau, Orange, Rockland, Suffolk. NORTH CAROLINA: Carteret, Durham, Guilford, Transylvania, Wake. OHIO: Ash- land, Cuyahoga, Hocking, Mercer, Ottawa, Wayne. OKLAHOMA: Comanche. PENN- SYLVANIA: Blair, Bucks, Franklin. TEN- NESSEE: county unknown. TEXAS: Bas- trop, Brazos, Cameron, Comal, Denton, Hays, Kerr, McLennan, Travis, Walker. VIRGINIA: Fairfax, Montgomery, Page. WEST VIRGINIA: Hancock. MEXICO. Isla Partida, Gulf of Califor- nia, 200, holotype and paratypes of A. philosopha. Santa Catalina Island. ARIADNA IN THE AMERICAS * Beatty 46] Ariadna pilifera O. P.-Cambridge Figures 20, 31, 36-37, 40-41. Map 1. Ariadne pilifera O. P.-Cambridge, 1898, Biol.- Cent. Amer., Arach. 1:235, plate 30, figure 9, 9 a-c, 2. Female holotype from Mexico, Du- rango, in British Museum (Natural History ), examined. Ariadne comata O. P.-Cambridge, 1898, Biol. Cent.-Amer., Arach., 1:235, plate 30, figure 8, 8a-c, 9. Female holotype from Mexico, Ori- zaba, in British Museum (Natural History), examined. NEW SYNONYMY. Ariadna acanthopus Simon, 1907, Ann. Soc. Ent. Belg., 51:263, figure 5, ¢. Male holotype from Mexico, Guanajuato, in Muséum National d'Histoire Naturelle, Paris, examined. Petrunke- vitch, 1911, Bull. Amer. Mus. Nat. Hist., 29: 131; Bonnet, 1955, Biblio. Aran., 2(1):730. NEW SYNONYMY. Ariadna jaliscoensis Chamberlin, 1925, Bull. Mus. Comp. Zool., 68(4):212. Female holotype from Mexico, Hacienda San Marcos, SW Jalisco, in Museum of Comparative Zoology, Cambridge, examined. Bonnet, 1955, Biblio. Aran., 2(1): 733. NEW SYNONYMY. Ariadna pilifera: F. O. P.-Cambridge, 1899, Biol. Cent.-Amer., Arach., 2:43; Petrunkevitch, 1911, Bull. Amer. Mus. Nat. Hist., 29:131; Bonnet, 1955, Biblio. Aran.,. 2(1):732. Ariadna comata: F. O. P.-Cambridge, 1899, Biol. Cent.-Amer., Arach., 2:43; Petrunkevitch, 1911, Bull. Amer. Mus. Nat. Hist., 21:131; Bonnet, 1955, Biblio. Axan;, 2(1)):732: Not Ariadna comata: Banks, 1929, Bull. Mus. Comp. Zool., 69(3):54. Discussion. This species is much more variable in some normally stable characters than any other American Ariadna. For a time it was believed that the specimens on hand might belong to as many as five species. The distributions of these appar- ent species made no_ sense, however. With long series of specimens from each of several scattered localities, the vari- ability of the species became evident. This variation is especially pronounced in the population found near Portal, Arizona. Types of all the species listed in the synonymy above have been examined. The holotype of A. acanthopus is a male, the other three are females. Chamberlin’s A. jaliscoensis differs most from the modal values for characters of other specimens assigned to pilifera. In its most divergent 462 characters, however, the jaliscoensis holo- type is usually still within the range of variation of other pilifera specimens. Only one character, the number of spines on the palpal tarsus, is entirely non-over- lapping. The name Ariadna comata was first applied to Central American specimens by Banks (1929), presumably because comata was the southernmost species recorded from Mexico. The type of comata, when examined, proved to be similar to that of pilifera, and quite different from the Cen- tral American form. The latter is described herein as a new species. Color. Male. Carapace dark reddish mahogany. Anterior legs a little lighter, remaining legs yellowish brown. Bulb of palp yellow. Remainder of palp, endites, labium, and coxae yellowish brown, ster- num darker. Abdomen purplish gray above and below. Spinnerets and anal tubercle brownish yellow. Female. Darker than male, the carapace and all legs deep mahogany, becoming almost black on distal podomeres of first leg. Underside of cephalothorax and legs scarcely lighter than upper side. Structure. Dimensions of eight males: Total length 7.2-10.6 mm, mean 9.15 mm; carapace length 3.8-5.0 mm, mean 4.36 mm; carapace width 2.5-3.1 mm, mean 2.76 mm; sternum length 2.3-3.1 mm, mean 2.74 mm; sternum width 1.30-1.63 mm, mean 1.502 mm. Dimensions of thirty females: Total length 9.7-15.0 mm, mean 12.02 mm; car- apace length 4.9-6.4 mm, mean 5.72 mm; carapace width 2.7-3.9 mm, mean 3.35 mm; sternum length 2.9-3.9 mm, mean 3.33 mm; sternum width 1.55-2.37 mm, mean 1.896 mm. Male palp. Bulb large, tibia not in- flated, the bulb 2.5 times the width of the tibia. Midpiece quite short and conical, embolic portion and midpiece together shorter than maximum depth of bulb. Em- bolic portion shallowly curved near distal end Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 Male first leg. Metatarsus strongly bent inward, bearing a large, acute retrolateral apophysis just proximal of the middle and a smaller prolateral one just distal of the middle. Tibia with two.to three of the outer ventral spines modified. Investiture. Male. Hair covering of abdo- men rather short. Tarsi of all legs scopu- late, metatarsi sometimes with a very short distal scopulate area. Female. Clothed throughout with con- spicuous long hair, normal in pattern of arrangement, but somewhat denser and longer than usual. Color of the hair varies from reddish or brown to black. Spination. See Table 17. Diagnosis. The variation in this species makes diagnosis or placement in a key somewhat difficult. The male is closely similar to that of A. bicolor. Only in these two species is the first metatarsus stout, strongly bent, and provided with a lateral apophysis on each side. The more distal placement of the apophyses and the pro- portionately larger palpal bulb distinguish the male of pilifera from that of bicolor. In the female of pilifera, lateral spines are present on tibiae I and II, separating the species from bicolor, which it generally resembles. The ventral spination of tibia II is normally 4—(1-2), and the metatarsal comb usually contains 3-4 spines (three in most populations examined); retrolateral spines are absent from metatarsi II and III and tibia TIT; metatarsi I and IT usually have 7-10 spines in each ventral row, and metatarsus IV usually has 1-2 spines in addition to the comb. Distribution. Southern Arizona to the Isthmus of Tehuantepec in southern Mexico. Records. ARIZONA: Cochise Co., Chiri- cahua Mtns., Portal and Southwestern Research Station, Chiricahua National Monument; Huachuca Mtns., Carr, Ram- sey, and Miller Canyons. Graham Co., Mt. Graham near Safford. Navajo Co., 12 mi. S of Show-low. Pima Co., Ajo Mtns., Organpipe Cactus National Monument, Alamo Canyon. Baboquivari Mtns., Brown Canyon, and Forestry Cabin; Santa Cata- lina Mtns., lower Sabino Canyon, Molino Basin, Peppersauce Canyon, Windy Point, Geology Vista, San Pedro Vista; Santa Rita Mtns., Madera Canyon, Roundup Camp. MEXICO. CHIHUAHUA: Primavera, 5000-6000 ft. COLIMA: 12 mi. E of Man- zanillo. DURANGO: Ojo de los Encinos. GUANAJUATO: no further data, 0. GUER- RERO: Ayotzinapa. HIDALGO: Jacala, Rancho Viejo near Jacala. MICHOACAN: Jiquilpan, Morelia, Tanci’taro, 6500 ft. MORELOS: Cuernavaca, 3 mi. E of Cuer- navaca, 9.6 mi. E of Cuernavaca. NAYARIT: Campostela. OAXACA: 67 mi. NW of Tehuantepec. PUEBLA: Tehuacan. SAN LUIS POTOSI: Tamazunchale, 3.8 mi. W of El Naranjo (Veracruz), E of Ciudad del Maiz. TAMAULIPAS: 5 mi. W of Altamira. VERACRUZ: Acultzingo, W of Orizaba, 15 mi. NW of Jalapa. Ariadna pragmatica Chamberlin Map 1. Ariadna pragmatica Chamberlin, 1924, Proc. Cali- fornia Acad. Sci., ser. 4, 12(28):606. Female holotype from Mexico, Sonora, Tepoca Bay, in California Academy of Science, examined. Bonnet, 1955, Biblio. Aran., 2(1):736. Ariadna scholastica Chamberlin, 1924, Proc. Cali- fornia Acad. Sci., ser. 4, 12(28):607. Female holotype from Gulf of California, Patos Island, in California Academy of Science, examined. Bonnet, 1955, Biblio. Aran., 2(1):737. NEW SYNONYMY. Discussion. Chamberlin separated his two species, scholastica and pragmatica, on the basis of the spacing of the posterior eyes. Only three specimens are available. In these the distance from PME to PLE is, respectively, 1.125, 1.667, and 1.882 times the diameter of a PME. This is a little more than the range of variation in A. bicolor (80 specimens). In other respects the specimens are quite similar. Recalling the variation of eye spacing shown by A. philosopha (=A. bicolor) in the same ARIADNA IN THE AMERICAS + Beatty 463 geographic region, it seems best to unite A. scholastica and A. pragmatica also. The treatment of all of Chamberlin’s species from the islands and shores of the Gulf of California has been difficult. The type localities are difficult of access, only a handful of specimens (some of which have been dried) are in collections, and the extreme climatic conditions of the region are seemingly operating on the animals to produce variation not usual in the genus. The presence of any Ariadna at all in such a hot arid region as coastal Sonora is surprising. More specimens are needed, but they may be difficult to find. Color. Carapace and chelicerae orange- brown. Anterior legs yellow-brown, darker distally. Succeeding legs lightening to yellow. Stermum, endites, and labium yellow-brown to orange-brown, coxae yel- low-brown to yellow. Abdomen purplish gray above, yellowish beneath. Structure. Dimensions of three females: Total length 8.6-12.3 mm, mean 10.33 mm; carapace length 3.94.6 mm, mean 4.31 mm; carapace width 2.24.1 mm, mean 3.03 mm; sternum length 2.24.0 mm, mean 2.94 mm, sternum width 1.33-2.57 mm, mean 1.823 mm. Investiture. Generally normal. The fringe on anterior legs is less conspicuous than usual. Spination. See Table 18. Diagnosis. Only the female is known. The absence of lateral spines from tibiae I and IJ, and the presence of only two spines in the metatarsal comb separate A. scholastica from the other American spe- cies. Distribution. Shore and islands of the Gulf of California. Records. MEXICO. SONORA: Tepoca Bay, 25 Apr. 1921, 2, (J. C. Chamberlin), holotype of A. pragmatica. GULF OF CALIFORNIA: Patos Island, 23 Apr. 1921, 2, (J. C. Chamberlin), type of A. scho- lastica; Cedros Island, 22 Feb. 1945, ¢, (B. F. Osorio Tafall). 464 Ariadna weaveri sp. n. Figures 50, 53, 56. Map 1. Holotype. A male from Mexico, Islas Revilla Gigedo, Clarion Island, in American Museum of Natural History. The species is named after my good friend and former professor, Dr. Andrew A. Weaver, who first introduced me to the study of spiders. Discussion. The sexual dimorphism in this species is much greater than is usual in Ariadna. The total body length of the males is only equal to or slightly more than the carapace length of some of the females. The male coloration is lighter than usual in relation to that of the females. Color. Male. Carapace orange-brown, narrowly flanged marginally, producing a thin dark marginal line in dorsal view. First legs slightly brownish yellow, re- maining legs yellow to yellowish white. Chelicerae yellow, endites and labium yellowish white. Sternum yellow suffused with purplish gray. Abdomen yellow suf- fused with purplish gray laterally and on the lateral parts of dorsum and _ venter. Dorsum with a broad median band of purplish gray finely mottled with yellow- ish, the band broadening posteriorly to the full width of the abdomen. Venter with a broad median band of clear yellowish white, the band narrowing posteriorly. Fine transverse yellowish lines break the dorsal band of the abdomen into sections that probably correspond with the (other- wise externally suppressed) segmentation of the abdomen. Seven of these sections are visible in the posterior third of the abdomen anterior to the anal tubercle. Female. Carapace, chelicerae, palps, and first legs distal to femur mahogany. The first femur and the remaining legs yellow to yellow-brown. Sternum, labium, and endites light mahogany. Abdomen uni- form purplish gray above and below in some beneath in specimens, yellowish other: Structure. Dirnensions of two males: Total length 4.0, 5.2 mm; carapace length Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 1.8, 2.1 mm; carapace width 1.3, 1.6 mm; sternum length 1.3, 1.5 mm; sternum width 0.78, 0.87 mm. Dimensions of seven females: Total length 7.8-12.2 mm, mean 10.10 mm; car- apace length 3.5-6.0 mm, mean 4.68 mm; carapace width 2.2-3.7 mm, mean 2.91 mm; sternum length 2.5-3.5 mm, mean 2.92 mm; sternum width 1.21-1.96 mm, mean 1.541 mm. Male palp. Bulb small, its greatest diameter about equal to diameter of tibia plus tarsus, and its transverse diameter about twice the width of tibia. Midpiece very short, conical, passing into the em- bolic portion at a curve. Embolic portion shorter than palpal tibia and not much longer than midpiece. Another shallow bend at tip of embolus. Palpal tibia thick- ened somewhat at base, but not inflated. Male first leg. Metatarsus and_ tarsus slender, almost straight, and entirely with- out spines or apophyses. Investiture. Male. Hair very short and sparse on the carapace, tending to be ar- ranged in rather regular longitudinal rows on the cephalic region. Elsewhere arranged as in A. isthmica, but lacking the scopulae of hooked hairs on tarsi and metatarsi. Female. Hair longer and denser than in male, especially on palps and legs. Spination. See Table 23. Diagnosis. Male. Metatarsus I nearly straight, slender, without spines or apo- physes. Metatarsal comb of three spines. Length of embolic portion of palp less than width of palpal bulb. Female. Tibiae I and I without lateral spines; ventral spines of tibia II usually 4-2 (the outer row, however, often con- tains one or two additional spines); meta- tarsal comb of three spines; usually with 1-2 prolateral spines on tibia HII. Carapace length 3.5-6.0 mm, mean 4.68 mm. Distribution. Mexico, the Revilla Gigedo Islands. Records. MEXICO. Clarion Island 7-S May 1955, 6 ¢ 2 200, from log, (McDonald and Blodget); Socorro Island, 1-5 May 1955, 2, (McDonald and Blodget). Ariadna caerulea Keyserling Figures 44-45, 49. Map 2. Ariadne caerulea Keyserling, 1877, Verhandl. der konig. kais. Zool. Bot. Gesell., Wien, 1877:227. Female and juvenile syntypes from Colombia, Bogota, in British Museum (Natural History ), examined. Ariadna caerulea: Petrunkevitch 1911, Bull. Amer. Mus. Nat. Hist., 29:131: Bonnet, 1955, Biblio. Aran., 2(1):732. Discussion. This species appears to be, as does A. pilifera, more variable in some characters than is usual. One of the six females had only three spines in the meta- tarsal comb, rather than four, and another had only two lateral tibial spines. It is possible that more than one species is included here but, since each of the fe- males was collected at a different locality, nothing is known of the range of intra- population variation. Until more specimens become available only one, presumably variable, species should be recognized. Color. Male and female. Carapace, an- terior legs, chelicerae, and palps reddish mahogany. Legs with longitudinal lighter stripes. Posterior legs paler. Sternum, endites, and labium orange-brown. Abdo- men uniform dark purplish gray above, only a little paler beneath, with a bluish surface sheen. Structure. Dimensions of one male: total length 3.9 mm; carapace length 2.2 mm; carapace width 1.6 mm; sternum length 1.4 mm; sternum width 0.84 mm. Dimensions of five females: total length 7.8-11.7 mm, mean 9.44 mm; carapace length 4.1-4.7 mm, mean 4.33 mm; car- apace width 2.5-2.9 mm, mean 2.64 mm; sternum length 2.0-2.7 mm, mean 2.31 mm; sternum width 1.28-1.63 mm, mean 1.424 mm. Investiture. Hair long and reddish, dis- tributed as usual. Fringes especially long on femora and tibiae [ and II. Spination. See Table 7. Diagnosis. TVibiae I and IL with lateral ARIADNA IN THE AMERICAS * Beatty 465 spines; tibia II with 4—(3-4) ventral spines. Metatarsus [TV with a comb of four spines and usually no additional ones. Metatarsi If and UL unarmed retrolaterally, II usually with a total of 5-9 spines. Male. Metatarsus of first leg slender and straight, without apophyses or modi- fied spines. Metatarsi I and II each with 2-4 spines. Palpal tibia conspicuously in- flated, bulb of palp small, length of embolic portion about equal to width of bulb. Distribution. Colombia and Ecuador, in mountains (Map 2). Records. COLOMBIA. CUNDINA- MARCA: Bogota (type locality ). MAGDA- LENA: S side of Sierra Nevada de Santa Marta, 8-11,000 ft (2440-3350 m); NA- RINO: 20 mi. E of Pasto; TOLIMA: 10 mi. W of Ibague; VALLE: 11 mi. W of Cale emis Waot Cali: ECUADOR. AZUAY: Lago Zurucuchu, 11 mi. W of Cuenca; PICHINCHA: 7 mi. S of Cayambe; TUNGURAHUA: Banos. Ariadna cephalotes Simon Figure 11. Map 2. Ariadna cephalotes Simon, 1907, Ann. Soc. Ent. Belg., 51:262. Female and juvenile syntypis from Bolivia, San Mateo, in Muséum National d'Histoire Naturelle, Paris, examined. The fe- male specimen is here designated as the lecto- type, and the juvenile as a _ lectoparatype. Petrunkevitch 1911, Bull. Amer. Mus. Nat. Hist., 29:131; Bonnet, 1955, Biblio. Aran., 2(1):732. Ariadna hotchkissi Chamberlin, 1916, Bull. Mus. Comp: Zool” 61(3): 216. Ply 10; shige § 45) eo: Immature holotype from Lucma, Cuzco, Peru, in Museum of Comparative Zoology, Cam- bridge, examined. Bonnet, 1955, Biblio. Aran., 2(1):732. NEW SYNONYMY. Discussion. Neither A. cephalotes nor A. hotchkissi has been reported in the primary literature since the original de- scriptions were published. No additional specimens are available from near the type localities of either. In 1965, however, Levi collected, at Tarma, Peru, a small series of Ariadna that agree well with the lectotype of A. cephalotes. The holotype of A. hotchkissi is an im- 466 mature specimen. It has a reduced spi- nation as compared with the lectotype of cephalotes, but is quite close to the im- nature lectoparatype of that species. The ‘ifferences in spination between the latter two specimens are in the ventral spination Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 CAERULEA oO CEPHALOTES Sd GRACILIS td MAXIMA a MOLLIS A BOESENBERGII Map 2 of the second tibia and the metatarsal comb. Although these are normally highly diagnostic characters, they do vary enough even in single demes to be of doubtful reliability when comparing singly collected specimens with each other. It seems better, in view of the usually extensive distributions of species in this genus, and the fact that all the specimens in question are from similar mountainous areas, to synonymize A. hotchkissi with A. cephalotes. If further collecting should demonstrate the distinctness of A. hotch- kissi, the Peruvian specimens described below should be retained in A. cephalotes. Color. Female. Carapace dark mahog- any, chelicerae darker. Legs yellowish brown, anterior legs and distal segments darker. Distal segments of palp almost as dark as carapace. Metatarsi I and II with distinct dark distal annuli, fainter annuli on tibiae I and II and metatarsi III and IV. Abdomen dark purplish gray. Male unknown. Structure. Dimensions of eight females: total length 9.2-11.4 mm, mean 9.78 mm; carapace length 4.4-5.0 mm, mean 4.54 mm; carapace width 2.8-3.4 mm, mean 2.93 mm; sternum length 2.6-3.0 mm, mean 2.64 mm; sternum width 1.43-1.70 mm, mean 1.491 mm. Investiture. As usual. Spination. See Table 8. Diagnosis. Generally lacking _ lateral spines on all tibiae (occasionally 1 or more retrolateral spines on tibia I), ventral spines 44 on tibia I, 4-3 on tibia II; abdo- men uniform purplish gray. Distribution. Bolivia and Peru. Records. BOLIVIA. San Mateo (female lectotype, juvenile lectoparatype). PERU. APURIMAC: 37 km S. of Anda- huaylas, 6 March 1951, 200 (E. S. Ross and A. E. Michelbacher); Cuzco: Lucma, 7000 ft (2130 m), Aug. 1911, Yale Peruvian Expedition (holotype of A. hotchkissi); JUNIN: Tarma, 3100 m (10175 ft), 11-12 ebm l96an09 2, (Hy We. Wevi), 14 Feb. 1965, “in ground” ? 200, (H. W. Levi). Ariadna murphyi (Chamberlin), new combination Map 3. Dysdera murphyi Chamberlin, 1920, Brooklyn Mus. Sci. Bull., 3(2):38. Female holotype ARIADNA IN THE AMERICAS * Beatty A467 from Peru, Chinchas Island, in Museum of Comparative Zoology, examined. Bonnet, 1956, Biblio. Aran., 2(2):1631. Discussion. This is an ordinary member of the genus Ariadna, without any morpho- logical characters more than usually sug- gestive of Dysdera. Chamberlin seems, for a brief time, to have considered Ariadna similar enough to Dysdera to warrant unit- ing the two genera. Later he described additional species in Ariadna. The correct placement of A. murphyi has not been made known previously. Only the female and juvenile are known. Color. Carapace and first legs orange- brown to rich reddish mahogany. The other legs paler. Abdomen light purplish gray to grayish yellow above, scarcely paler beneath. Structure. Dimensions of seven females: Total length 9.0-12.0 mm, mean 10.61 mm; carapace length 4.5-5.9 mm, mean 5.04 mm; carapace width 2.2-3.2 mm, mean 2.86 mm; sternum length 2.4-3.0 mm, mean 9.73 mm; sternum width 1.06-1.71 mm, mean 1.491 mm. Investiture. Fringes of hair on anterior legs straight, rather than curling. Lateral spines on tibiae I and II very short, half the tibial diameter or less. Spination. See table 15. Diagnosis. The presence of only two spines in the metatarsal comb distinguishes A. murphyi from most other species. The ventral tibial spination, 4—4 on tibiae I and II, separates it from the other species with two-spined comb. Distribution. Recorded only from islands off the coast of Peru (Map 3). Records. PERU. Chinchas Island, 12 Oct. 1919, 2 20, (R. C. Murphy), holotype and paratypes; South Chinchas Island, 23 Feb. 1935, 2 200. Ariadna peruviana sp. n. Figures 34-35, 39. Map 3. Holotype. Male from Lima, Lima, Peru, 1939 (W. K. Weyrauch) in the Museum of 468 BOLIVIANA M@ OBSCURA A CRASSIPALPUS AND DUBIA O MURPHYL O TARSALIS 4 SPINIFERA AND CONSPERSA © PERUVIANA The is an adjective referring to the country of origin of the species. Color. Male. Carapace mahogany. Chelic- erae yellowish brown. First pair of legs Comparative Zoology. name yellowish brown, the succeeding pairs lightening to yellow. Sternum, labium, en- dites, and coxae yellow-orange. Palps yellow. Abdomen yellowish white beneath and laterally, with a sooty median dorsal band about one-third the width of the abdomen anteriorly, widening to the full abdominal width posteriorly. Faint darker markings as in the female. Female. Carapace, chelicerae, palpal tibia and tarsus, and first legs distal to the ‘emur dark reddish mahogany. Succeeding lightening to yellow-brown. Labium ana endites mahogany. Sternum, coxae, Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 palpal femur and_ patella yellow-brown. Abdomen purplish gray above, darker in the axial third. Inconspicuous longitudinal lines resembling short dark brush strokes are visible on the dorsum of well-pre- served specimens. In some alcoholic specimens of both sexes the abdomen is almost white except for the sooty wash down the middle. The condition of these specimens suggests that the effect of preservation is chiefly re- sponsible for their paleness. Both the males are light-colored. Structure. Dimensions of two males: total length 6.8, 6.8 mm; carapace length 3.0, 3.6 mm; carapace width 2.3, 2.5 mm; sternum length 2.1, 2.3 mm; sternum width Wd RIE savant, Dimensions of six females: total length §.3-11.3 mm, mean 9.72 mm; carapace length 3.7-5.3 mm, mean 4.42 mm; car- apace width 2.3-3.4 mm, mean 2.81 mm; sternum length 2.3-3.2 mm, mean 2.72 mm; sternum width 1.2-1.8 mm, mean 1.44 mm. Male palp. Bulb small, twice width of palpal tibia or less. Midpiece of palp much longer than the embolus and about equal to palpal tibia in length. Embolus making about one quarter of a helical turn. Male first leg. First metatarsus moder- ately slender and bearing two lateral pro- tuberances or low apophyses, each of which ends in a short thick spine. The retrolateral protuberance is near the base of the metatarsus, the prolateral one is more distal, but still in the proximal half of the metatarsus. The distal spine in the outer ventral row on tibia I is somewhat shortened and thickened. Investiture. About as sexes. Spination. See Table 2. Diagnosis. Male. The heavy spines of the first metatarsus, placed in the proximal half of the podomere, and not opposite each other, are distinctive for the male of this species. Female. Lateral spines present on tibiae I and II. Ventral spines of tibiae I and II 4-4 (the proximal inner spine is generally more slender than the others). Metatarsal comb of three spines. Distribution. Known only from Peru (Map 3). Records. PI:RU. LIMA: La Molina, 250 m; Jan. 1961, 42 2 (R. Garcia ); Lima, 1939, in house, 662200 (W. K. Weyrauch); Lima, 9 Jan. 1955, 0, (E. I. Schlinger and E. S. Ross). LIBERTAD: Jequetepeque, 15 Feb. 1965, 0. (LL. Pefia). usual in both Ariadna maxima (Nicolet) Figures 2, 18-19, 21. Map 2. Dysdera maxima Nicolet, 1849, in Gay, Hist. fis. y polit. de Chile, 3:341, pl. 2, fig. 6, Ga-d, @. Holotype lost. Keyserling, 1877, Verhandl. der kénig. kais. Zool.-Bot. Gesell. Wien, 1877:230. ARIADNA IN THE AMERICAS + Beatty A469 Dysdera virens Nicolet, polit. Chile, 3:342. 1849, in Gay, Hist. fis. Holotype lost. Keyserling, 1877, Verhandl. der k6nig. kais. Zool.-Bot. Gesell. Wien, 1877:230; Petrunkevitch, 1911, Bull. Amer. Mus. Nat. Hist., 29:131; Bonnet, 1955, Biblio. Aran., 2(1):734. Dysdera incerta Nicolet, 1849, in Gay, Hist. fis. polit. Chile, 3:342. Holotype lost. Keyserling, 1877, Verhandl. der kénig. kais. Zool.-Bot. Gesell. Wien, 1877:230; Petrunkevitch, 1911, Bull. Amer. Mus. Nat. Hist., 29:131; Bonnet, 1955, Biblio. Aran., 2(1):734. Dysdera coarctata Nicolet, 1849, in Gay, Hist. fisy polit, Chile, 3: 344) plies tissanenva-ce lor Holotype lost. Simon, 1864, Hist. Nat. Araig- nées, p. 106; Keyserling, 1877, Verhandl. der konig. kais. Zool.-Bot. Gesell. Wien, 1877:239: Petrunkevitch, 1911, Bull. Amer. Mus. Nat. Hist., 29:131; Bonnet, 1956, Biblio. Aran., 2(2):1619. NEW SYNONYMY. Ariadna maxima: Simon, 1896, Actes Soc. Sci. Chili, 6:64; 1897, Actes Soc. Sci. Chili, 6:105, 107:1900, Rev. Chil. Hist. Nat. 4:49; 1902, Ergebn. Hamburger Magal. Sammilr., 6(4):11; 1905; Bull’ Socy Ent Er 19054) eialee Bae Cambridge, 1898, Journ. Linn. Soc. London, 27:17. Porter, 1914, Rev. Chil. Hist. Nat. 21 (6):180; 1917, Actes Soc. Sci., Chili, 25(2): 82. Petrunkevitch, 1911, Bull. Amer. Mus. Nat. Hist., 29:131. Berland, 1924, Nat. Hist. Juan Fern., Easter Isl., 3(3):423, 426; 1934, Publ. Soc. Biogeogr., 4:168. Bonnet, 1955, Biblio. Aran., 2(1):734. Discussion. Four of the five species de- scribed by Nicolet were collected in the vicinity of Santiago and evidently were based upon different ages, sexes, and color- ations of a single species. No types are available for any of the species. Fortu- nately hundreds of specimens from Chile are on hand, especially from the region of Santiago, but altogether covering most of the length of the country. All but four of the specimens belong to a single species. The fifth species described by Nicolet, Dysdera longipes collected in Valdivia, has also been synonymized with maxima (Petrunkevitch, 1911; Bonnet, 1955). Re- cently, however, Levi collected in Valdivia two male and two female Ariadna which appear to belong to two additional species. The name longipes is being reserved for one of these. Descriptions are deferred until more material is available. ATO For some reason Dysdera coarctata has not previously been synonymized with maxima, and has been listed in later works as a Dysdera (Porter, 1917; Bonnet, 1956). Nicolet’s illustration is obviously of an Ariadna, and there is no more reason to regard coarctata as a separate species than there is for any of the other synonyms. Color. In general like that of A. bicolor, but on the average darker. Male. Palps yellow-brown. Chelicerae and legs lighter in color than those of the female, mahog- any to yellow-brown. Color otherwise as in the female. Female. Carapace deep reddish mahog- any, sometimes with a tinge of maroon. Chelicerae almost black. Distal palpomeres and first legs reddish mahogany, remaining legs lightening to yellowish-brown. Labium and endites chestnut. Sternum and coxae yellow-brown, the sternum darker margin- ally. Abdomen light to dark purplish gray above and below. Large egg-filled females are paler than those not in reproductive condition, because of the stretching of the cuticle. Structure. Dimensions of 15 males: Total length 7.7-10.2 mm, mean 8.78 mm; carapace length 4.1-5.6 mm, mean 4.79 mm; carapace width 2.7-3.8 mm, mean 3.15 mm; sternum length 2.6-3.6 mm, mean 3.01 mm; sternum width 1.30-1.75 mm, mean 1.490 mm. Dimensions of 76 females: Total length 8.3-16.0 mm, mean 12.70 mm; carapace length 4.2-7.7 mm, mean 5.97 mm; car- apace width 2.5-4.4 mm, mean 3.42 mm; sternum length 2.5-4.5 mm, mean 3.46 mm; sternum width 1.30-2.30 mm, mean 1.723 mm. Male palp. Bulb small, tibia moderately inflated, the width of the bulb equal to or less than tibial width. Midpiece and em- bolic portion together shorter than tibia. Midpiece not clearly distinguishable from the embolic portion, the latter wider than in ler species. tirst leg. tarsus Metatarsus and Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 straight, the metatarsus without apophyses or modified spines. Investiture. Hair long, clothing entire body rather densely, in normal pattern. Fringes of anterior legs of female not very conspicuous. Spination. See Table 12. Diagnosis. Male. Metatarsus straight and not notably slender, lacking apophyses and modified spines. Metatarsal comb of four spines. Embolic portion and midpiece of palp together less than palpal tibia in length. Female. Lateral spines present on tibiae I and II, 44 ventral spines on tibiae I and II; four spines in metatarsal comb, plus 1-2 other ventral spines on the same podo- mere; metatarsus III with 5-9 spines, none retrolateral; carapace length 4.2-7.7 mm, mean 5.97 mm. Distribution. Chile, including Juan Fer- nandez Islands. Records. CHILE. ACONCAGUA: San Felipe. ANTOFAGASTA: Tal-tal. CAU- TIN: Temuco; Villarica, garden, and on buildings; NE of Villarica; 30 km NE of Villarica. COLCHAGUA: Chepica; Cun- aco, Fundo Millahue. CONCEPCION: Concepcion; Bosque de Ramuntcho; Tal- cahuana. COQUIMBO: La Serena; Pichi- dangui, Isla de los Locos. LINARES: Linares. LLANQUIHUE: 2-3 km NW of Ensenada; Peulla, 200 m, on buildings; Petrohué, buildings; Puerto Varas, 50 m, buildings; Parque Philippi. MAGEL- LANES: Punta Arenas. NUBLE: Chillan, Cordillera de Chillan. OSORNO: Osorno, city park; Purranqme; Termas de Puyehue, 240 m, buildings and gardens. SANTIAGO: Alhué; Cerro Santa Lucia; El Canelo; Quinta Normal; Tiltil; Yeso River, 1200 m, Cordilleras near Santiago. TALCA: Alto de Vilches, Andes; Talca. VALDIVIA: Valdivia, Isla Teja, farmland. WVAL- PARAISO: Bosque Relicto; Lagunillas; Limache; Los Horcones; Quintero, Playa Piratas; Ventana; Vina del Mar. JUAN FERNANDEZ ISLANDS: Mas a Tierra: Bahia Cumberland; Galpon, Valle Villagro; Puerto Ingles; Quebrada Pangal, Monte Oscuro, 100m; Mads Afuera: Cerro Innocentes, 1000 m; Chorro Dona Anna; Chorro de Varadero; Plano de Chosa, 800- 1000 m; Quebrada Casa; Quebrada Vaca. Ariadna isthmica sp. n. Figures 23—25, 30. Map 1. Ariadna comata: Banks, 1929, Bull. Mus. Comp. Zool., 69(3):54. Not A. comata O. P.-Cam- bridge. Holotype. A male from Panama, Barro Colorado Island, in the Museum of Com- parative Zoology. The name is an ad- jective meaning isthmian, in reference to the species’ Central American distribution. Discussion. There appears to have been no reason, other than a guess based on distribution, for Banks’ assignment of this species to Ariadna comata. As noted above, A. comata is here considered a synonym of A. pilifera. The present species is entirely Central American in distribution, while pilifera is recorded only from the south- western United States and Mexico. Color. Male. Carapace light to dark orange-brown with a narrow dark marginal line. Cephalic region lighter. Eyes nar- rowly rimmed with black. Chelicerae, labium, endites, and legs yellow to orange- brown, the anterior legs only slightly darker. Sternum yellow with a suffusion of purplish gray. Abdomen purplish gray over yellowish above, the yellowish show- ing through as many tiny light flecks. Venter of abdomen yellowish white with a dusting of purplish gray. Spinnerets and anal tubercle yellowish white. Female. Coloration generally as in male, but tending to be darker, the carapace and first legs often a rich reddish mahogany. Legs sometimes suffused with purplish gray at distal ends of podomeres. Structure. Dimensions of three males: Total length 4.6-6.9 mm, mean 6.00 mm; carapace length 2.4-3.3 mm, mean 2.93 mm; carapace width 1.6-2.2 mm, mean 1.92 mm; sternum length 1.5-2.0 mm, mean ARIADNA IN THE AMERICAS + Beatty 47] 1.78 mm; sternum width 0,92-1.12 mm. mean 1.020 mm. Dimensions of seven females: Total] length 7.5-11.7 mm, mean 9.04 mm: car- apace length 3.4-4.2 mm, mean 4.14 mm: carapace width 2.1-2.6 mm, mean 2.28 mm; sternum length 2.0-2.4 mm, mean 2.14 mm; sternum width 1.14-1.35 mm, mean 1.227 mm. Male palp. The bulb of the palp is small, its diameter less than twice that of the palpal tibia. The midpiece is short and narrows abruptly to the embolic portion, which joins the midpiece at a curve of almost ninety degrees. The embolic portion is equal to or slightly longer than the pal- pal tibia, and makes another nearly ninety degree bend at the tip. The palpal tibia is somewhat thickened at the base, but not inflated. Male first leg. Metatarsus slender and_ sinuous, spines or apophyses. Investiture. Male. Entire body with a sparse coating of fairly long dark, curving hair, most of it making about a 45 degree angle with the cuticular surface. A few scattered hairs are straight, stiff, and al- most erect. The chelicerae, tips of the endites, and tarsus of the palp are more densely clothed with hair than the rest of the body. On all the legs, most conspicu- ously the fourth, the tarsi and distal part of the metatarsi bear scopulae of stiff, short, translucent bristles bent into a minute hook at the tip. Female. Differing from the male only slightly. Hair denser on metatarsi and tibiae of the first two legs, usually forming a conspicuous fringe. The scopular hairs of the male are replaced by long curved recumbent hairs without hooked tips. Spination. See Table 11. Diagnosis. Male. First metatarsus slen- der and sinuous, without apophyses or modified spines. Embolic portion of palp equal to palpal tibia in length or slightly longer. Metatarsal comb of four spines. Carapace length 2.4-3.3 mm. First tibia and_ tarsus without thickened 472 usually with 3-5 retrolateral spines. Legs not annulate. Female. Lateral spines present on tibiae I and II. Ventral tibial spines 44 on legs I and II. Metatarsal comb of four spines. Metatarsi II and HI unarmed retrolaterally. A spine on palpal patella. Tibia IIT usually with one to two prolateral spines. Meta- tarsus III with seven to eight spines. Distribution. Central America (Map 1). Records. NICARAGUA. Musawas: Huas- puc River. PANAMA. Bella Vista; Porto Bello. CA- NAL ZONE: Barro Colorado Island; Fort Sherman; Canal Zone Forest . Preserve; Canal Zone Biological Area; Fort Davis; Gamboa. Ariadna tovarensis Simon Map 4. Ariadne tovarensis Simon, 1893, Ann. Soc. Ent. Fr., 61:448. Female and immature syntypes from Venezuela, Colonia Tovar, in Muséum National d’Histoire Naturelle, Paris, examined. Ariadna_ tovarensis: Petrunkevitch, 1911, Bull. Amer. Mus. Nat. Hist., 29:131; Bonnet, 1955, Biblio. Aran., 2(1):738. Discussion. As is the case with several other South American species, A. tovarensis is known from only a few specimens, so that variability of the diagnostic characters is poorly known. The geographically nearest species, A. tubicola, A. caerulea, and A. solitaria, are, however, distinguished by characters that seem to warrant recog- nition of tovarensis as a distinct species. Color. Carapace light mahogany, paler in cephalic area. Legs unmarked yellow to brownish yellow, the first leg a little darker than the others. Chelicerae, endites, labium and sternum brownish yellow. Abdomen shrunken from cuticle, but ap- parently was unmarked purplish gray above, scarcely lighter beneath. Structure. Dimensions of two females: Total length 7.3, 7.4 mm; carapace length 97 ‘ us 2.7, 3.1 mm; carapace width 2.1, 2.2 mm; sternum length 1.6, 2.0 mm; sternum width 0.98, 1.14 mm. Bulletin Museum of Comparative Zoology, Vol. 139, No. & Investiture. Much of the hair has been rubbed off. The pattern seems to be nor- mal, the density perhaps a little less than usual. Fringes on the anterior legs are straight or slightly curved. Spination. See Table 21. Diagnosis. Lateral spines present on tibiae I and II; ventral tibial spines 44 on leg I, 4-3 on leg I, with proximal inner spine reduced on tibia II; metatarsal comb of four spines, no other ventral spines on fourth metatarsus; metatarsi IT and HI un- armed retrolaterally. Distribution. Known only from north central Venezuela. Records. VENEZUELA. DISTRITO FEDERAL: ‘Colonia’ Tovar je o}owm enn Simon), the syntypes; La Silla, NE of Caracas, 21 Dec. 1930, ?, (J. G. Myers). Ariadna arthuri Petrunkevitch Figures 8, 26-27, 32. Map 4. Ariadna arthuri Petrunkevitch, 1926, Trans. Con- necticut Acad. Arts Sci., 28:48, fig. 8-10, “2”. Immature holotype from Sta. Maria Bay, St. Thomas, Virgin Islands, in Peabody Museum, Yale, New Haven, not seen; ibid., 1929, Trans. Connecticut Acad. Arts Sci., 30:59, figure 41, o. Ariadna bicolor: Lutz, 1915, Ann. New York Acad. Sci., 26:81; Franganillo, 1936, Los Aracnidos de Cuba hasta 1936, p. 38. Not A. bicolor (Hentz). Ariadna_ solitaria: Lutz, 1915, Ann. New York Acad. Sci., 26:81. Not A. solitaria Simon. Discussion. The holotype of A. arthuri could not be located by Petrunkevitch when an attempt was made to borrow it in 1962. Types of both the other West Indian species have been examined and are dis- tinctly different. The name A. arthuri is here applied to the common wide-ranging species found from southern Florida to Curacao—almost certainly the one to which it belongs. Petrunkevitch (1929) described and illus- trated a second immature specimen of A. arthuri, collected on Desecheo Island, 18— 20 February 1914. Presumably this spider is the one collected by Lutz and mentioned by him in his list of Greater Antillean spiders (1915) as A. solitaria. I have e. @ ARTHURI . XA O MULTISPINOSA B O SOLITARIA a 5 56 m™ = TOVARENSIS a TUBICOLA ont yy me examined the specimen and find that, while Petrunkevitch’s description is wrong in many details, the identification as A. arthuri is correct. The specimen, or specimens, from Cuba, listed by Lutz (1915) as A. bicolor, have not been seen. All available material from Cuba belongs to A. arthuri, however. Franganillo (1936) simply cites A. bicolor as occurring in Cuba with no further data. Probably his citation is based upon Lutz’s paper, to which he refers. Color. Male and female. Carapace, ster- num, and all appendages yellow to orange, slightly darker on first legs, chelicerae, and palpal tarsus (palps entirely light in male). Abdomen dusky yellow beneath and on sides, purplish gray above and around spinnerets ventrally. Structure. Dimensions of two males: total length 4.0, 4.5 mm; carapace length 2.1, 2.2 mm; carapace width 1.4, 1.5 mm; sternum length 1.36, 1.44 mm; sternum width 0.82, 0.82 mm. ARIADNA IN THE AMERICAS + Beatty 4 > Dimensions of four females and two last instar juveniles: total length 4.8-7.0 mm, mean 5.98 mm; carapace length 2.0- 3.9 mm, mean 2.67 mm; carapace width 1.2-2.2 mm, mean 1.64 mm; sternum Jength 1.2-2.1 mm, mean 1.55 mm, sternum width 0.73-1.14 mm, mean 0.922 mm. Male palp. The bulb is quite smal] and the tibia moderately inflated, the diameter of the bulb equalling 1.5 times the tibial diameter or less. The midpiece of the bulb is longer than either the spine or the depth of the bulb, and equals or slightly exceeds the length of the tibia. Male first leg. The tarsus and meta- tarsus are sinuous and slender. The middle third of the metatarsus is swollen. The inflation of the podomere is greatest at two-thirds of the distance from base to tip of the metatarsus. At this point, and opposite each other, are two ventrolateral protuberances; these are greatly enlarged spine sockets, each bearing a short thick spine. The distal pair of metatarsal spines and their sockets are somewhat enlarged also. The distal spine of the retrolateral ventral row on the tibia is slightly enlarged and flattened. Investiture. Male. Spines are conspicu- ous, mostly stout, and of medium length. On the first tibiae the longest spines equal or slightly exceed the tibial diameter. Hair mostly as usual. The tarsi of all legs except the first pair bear ventral scopulae. Female. Metatarsal and tibial spines on legs I and II lie at a very small angle to the podomeres, and the tips of the longer ones tend to curve inward toward the long axis of the leg. All the spines are slender, and the primary ones are very long, equalling or considerably exceeding half the length of the podomere. Spination. See Table 3. Diagnosis. Female. The presence of more than 4-4 ventral spines on tibiae I and II, and two spines in the metatarsal comb distinguish A. arthuri from all other American species except A. tarsalis, The latter may be separated by its ATA larger size, the lateral spination of the first two tibiae (see key or table), and distribution. Male. The structure of the first leg distinguishes this species from the other known males. Specific diagnostic char- acters are the heavy spines set opposite each other distal to the middle of the metatarsus. Distribution. Southern Florida and islands of the Gulf of Mexico and Carib- bean (Map 4). Records. FLORIDA: Lee Co., Boca Grande, under rocks. Monroe Co.: Big Pine Key, some taken from cracks and shipworm burrows in driftwood on beach; Bill Find’s Key, under bark of red man- grove; Rattlesnake Key, under bark of red mangrove; Squirrel Key, under bark of red mangrove. WEST INDIES. BAHAMA ISLANDS: South Bimini; Crooked Island. CUBA: 7 km N of Vinales, Trinidad Mtns., Buenos Aires; Soledad. CURACAO: Siberié, 3 km N of Savonet, “stones”; Piscadera Bay. PUERTO RICO: Aguas Buenas, “cave entrance”; Desecheo Island, “under fallen leaves in a sea-grape thicket.” VIRGIN ISLANDS: St. Thomas, Santa Maria Bay, “under bark of a log,” the holotype. LES- SER ANTILLES: no further data. Ariadna multispinosa Bryant Map 4. Ariadna multispinosa Bryant, 1948, Bull. Mus. Comp. Zool., 100(4):339. Female holotype from Dominican Republic, Loma Rucilla Mtns, in Museum of Comparative Zoology, examined. Discussion. Although known only from one female and one juvenile, A. multispi- nosa is quite distinct from most other species. Five species are similar to multi- spinosa in some characters of the spination, but are distinguished by other characters that have high diagnostic value. Only A. wthuri and A. tarsalis are very close to Until mature males and female of the latter two species ae ; ispinosa in structure. Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 are available, the relationships cannot be adequately assessed. Color. Carapace brown, legs yellowish brown, abdomen dark purplish gray above, dirty yellowish beneath. - Structure. Dimensions of female holo- type: Total length 9.4 mm; carapace length 4.1 mm; carapace width 2.6 mm, sternum length 2.0 mm; sternum width 1.33 mm. Investiture. Normal. Spination. See Table 14. Diagnosis. The absence of lateral spines from tibiae I and II, and presence of eight to eleven ventral spines in each row on tibiae I and II separate A. multispinosa from all other described American species. Distribution. Known only from Hispan- iola. Records. DOMINICAN REPUBLIC. Loma Rucilla Mtns, Pico del Yaque, 8- 19,000 ft (2440-3050 m), June 1938, 2, (P. J. Darlington, Jr.), holotype; Cordillera Central, near Valle Nuevo, rain forest, 6000 ft (1830 m), Aug. 1938, 0, (P. J. Darling- ton, Jr.), Paratype. Ariadna tarsalis Banks Map 3. Ariadne tarsalis Banks, 1902, Proc. Washington Acad. Sci., 4:57, plate 1, figure 9, immature. Immature holotype from Culpepper I., Gala- pagos Islands, in Museum of Comparative Zoology, Cambridge, examined. Banks, 1924. Zoologica, 5(9):95. Ariadna tarsalis: Petrunkevitch, 1911, Bull. Amer. Mus. Nat. Hist., 29:131; Bonnet, 1955, Biblio. Aran., 2(1):7387. Discussion. This species is quite similar to A. arthuri Petr. of the West Indies. Unfortunately adequate comparison of the two species is impossible, because A. tar- salis is known only from six immature specimens. The spination character that separates the two species is a minor one, but the difference in size is considerable. Color. Carapace and first legs orange- brown, the remaining legs paler orange to yellow-orange. Sternum and coxae yellow- orange, endites and labium brownish. Abdomen purplish gray above, paler be- neath. Structure. Dimensions of holotype: Total length 7.5 mm, carapace length 3.7 mm, carapace width 2.3 mm, sternum length 2.1 mm, sternum width 1.08 mm. Investiture. Normal. Spination. See Table 20. Diagnosis. Lateral spines usually present on tibiae I and II, ventral spines on these tibiae usually 6-6 or more; metatarsal comb of two spines; tibia I with 1-2 retro- lateral spines, tibia IT with none; carapace length 3.7-4.8 mm, mean of three speci- mens 4.15 mm. Distribution. The Galapagos Islands. Records. GALAPAGOS ISLANDS: Cul- pepper Island, immature holotype. South Seymour Island, April 1923, 0, (N. Banks). Duncan Island, 23 June 1929, 00, (H. H. Cleaves). Indefatigable Island, 20 June 1929, oo, (H. H. Cleaves). Ariadna mollis (Holmberg) Figures, 16, li7, 22, Map 2. Segestria mollis Holmberg, 1876, An. Agric. de la Repub. Argentina, 4:25, figure 6, 9. Holotype from Buenos Aires (?), lost. Segestria vulgarissima Holmberg, 1876, ibid., 4: 25, figure 7, 2. Holotype from Buenos Aires, lost. Ariadna mollis: Mello-Leitao, 1933, Arch. Escol. Sup. Agric. Med. Vet., 10(1):12; 1944, Rev. Museo La Plata, 3(24):312, 322, figure 1, 6; 1947, Arq. Museu Paranaense, 6(6):233, 234. Bonnet, 1955, Biblio. Aran., 2(1):735. Discussion. After examining specimens from Buenos Aires and Montevideo, Mello- Leitao (1933) concluded that only one Ariadna with a dorsal abdominal pattern occurred in these regions. Consequently he synonymized S. vulgarissima and A. boesenbergii with S. mollis, at the same time correctly transferring mollis to the genus Ariadna. Because the paper is likely to be in- accessible to many people, I here quote Mello-Leitao’s discussion in full: “Tendo examinado exemplares de Ariadna da Pro- vincia de Buenos Aires e de Montevideo ARIADNA IN THE AMERICAS + Beatty A7T5 (colhidos por mim no Cerro) e@ confront- ando-os com as descricoes de Holmberg e de Keyserling, conclui pela identidade das mesmas, tendo prioridade a designacao de Holmberg.” Working only from descrip- tions as he apparently was, (that is, with- out any type specimens ), it is not surprising that Mello-Leitao came to this conclusion. Keyserling’s description is reasonably de- tailed, but those of Holmberg are almost devoid of useful information. An examination of two specimens from the type series of A. boesenbergii (includ- ing both sexes) and a series of specimens from Buenos Aires reveals that mollis and boesenbergii are similar but distinct and apparently partly sympatric species. One vial from the Museo Argentino de Ciencias Naturales contained two female mollis and two female boesenbergii. The two species are almost identical in appearance, but are distinguished by several features of the spination. The ecological relationships of mollis and boesenbergii should be carefully in- vestigated. Not only are they different from most Ariadna in having an abdominal pattern, but they appear also to be sym- patric sibling species that would be of interest as a study of character displace- ment or its absence. At present there are too few specimens available from too limited an area to suggest whether or not character displacement has occurred. Mello-Leitao, in several papers, (1940, 1941, 1944, 1945, 1946), has given distri- bution records for A. mollis and A. boesen- bergii that include many localities not listed below. His identifications of neither species are trustworthy, so the actual dis- tribution of both species remains uncertain. Color. Male. Carapace and chelicerae orange-brown. First legs a little lighter than carapace, the remaining legs pro- gressively paler posteriorly. Sternum the color of the femora. Abdomen yellowish white dorsally with a purplish gray median longitudinal band in the anterior half and a series of short transverse bars posteriorly. A76 Female. Essentially as in male. The carapace and legs are darker, a rich reddish mahogany. The abdominal pattern on a female with enlarged abdomen consists of a lozenge anteriorly, back of which is a series of forward pointing chevrons. A narrow median band connects the first chevron to the lozenge, and continues to the anterior end of the abdomen (Fig. 1). The sides and venter of the abdomen are purplish gray. Structure. Dimensions of one male: Total length 7.1 mm, carapace length 3.6 mm, carapace width 2.3 mm, sternum length 2.2 mm, sternum width 1.22 mm. Dimensions of nine females: Total length 9.9-13.4 mm, mean 11.10 mm; carapace length 4.4-5.4 mm, mean 4.84 mm; carapace width 2.4-3.2 mm, mean 2.79 mm; sternum length 2.4-3.1 mm, mean 2.75 mm; sternum width 1.43-2.00 mm, mean 1.573 mm. Male palp. Bulb small, only slightly wider than tibia. Tibia scarcely inflated. Midpiece and embolic portion of palp equal to each other in length or midpiece a little shorter. Male first leg. Metatarsus and _ tarsus slender, sinuous, lacking apophyses or modified spines. Patella with a prolateral spine. Investiture. Male generally without un- usual features. Tarsi of legs I-IV scopu- late ventrally. Female as usual. Spination. See Table 13. Diagnosis. Male. First metatarsus sinu- ous, lacking apophyses and modified spines, metatarsal comb of five to seven spines; midpiece of palp short, about equal to embolic portion in length, abdomen with dorsal pattern. Female. Separated from most other American Ariadna by the pattern of bars on the abdominal dorsum. Presence of lateral spines on tibiae I and II, 4-3 ventral spines on tibia II, and 5-7 spines in the metatarsal comb distinguish it from A. voesenbergii. Two other South American reported to have abdominal pat- speci Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 terns, A. conspersa and A. crassipalpus, are not well enough known to be distinguished from A. mollis. (See discussion of these species under Ariadna Incertae Sedis be- low. ) Distribution. Brazil. Records. BRAZIL. PARANA: Caviana I., 1947, 2, (A. Maller). ARGENTINA. BUENOS AIRES: San Isidro, Oct. 1963, 2; Punta Lara, 6 Apr. 1950, 2200; Moreno, Oct. 1947, 22, (R: D. Schiapelli); Tigre; Nov: 1940) o3(@e Monros ); Gral. Madariaga, Jan. 1962, 2 2, (M. E. Galiano). Argentina and southern Ariadna boesenbergii Keyserling Figures 3, 46-48. Map 2. Ariadne Bésenbergii Keyserling, 1877, Verhandl. der konig. kais. Zool.-Bot. Gesell., Wien, 1877: 223, pl. 7, fig. 7, 7a-b, ¢. Syntypes from Monte- video, Uruguay, in the Zoologisches Staats-Mu- seum, Hamburg, examined. Ariadna_ bésenbergi: Petrunkevitch, Amer. Mus. Nat. Hist., 29:130. Ariadne bodsenbergi: Gerhardt, 1921, Arch. Natur., 87:92, 93, fig. 6, &. Ariadna mollis: Mello-Leitéio, 1933, Arch. Escol. super. agricul. med. vet., 10(1):12 (in part); ibid., 1947, Arq. Mus. Paranaense, 6:234. Bon- net, 1955, Bibliog. Aran., 2(1):730, 735. Not A. mollis (Holmberg). Discussion. Mello-Leitao (1933) mis- takenly synonymized A. boesenbergii with A. mollis, and later (1947) reiterated the synonymy in statements prefacing a key to Brazilian Ariadna. The two species are similar in size and appearance, but details of spination clearly separate them. Bonnet (1955) followed Mello-Leitao’s treatment, at that time the most recent taxonomic opinion. Among material borrowed from the Museo Argentino de Ciencias Naturales, I found a_ vial containing four female Ariadna collected at Moreno, Buenos Aires. Two of these specimens were A. mollis, the other two A. boesenbergii. The circum- stances of occurrence of the spiders are unknown, but certainly at least a strong 1911, Bull. presumption of sympatry is warranted. Only one other instance of collection of two species of Ariadna at one locality is known to me. (See under A. boliviana). Coloration. In both sexes the coloration is very similar to that of A. mollis, orange- brown carapace and legs, dorsum of abdo- men yellow with purplish gray transverse bars. Structure. Dimensions of male lectotype: total length 8.0 mm; carapace length 3.7 mm; carapace width 2.4 mm; sternum length 2.1 mm; sternum width 1.1 mm. Dimensions of three females: total length 7.9-9.5 mm, mean 8.63 mm; carapace length 3.64.2 mm, mean 3.78 mm; car- apace width 2.1-2.4 mm, mean 2.21 mm; sternum length 2.0-2.4 mm, mean 2.12 mm; sternum width 1.1-1.2 mm, mean 1.12 mm. Male palp. The bulb is small, the tibia short and much inflated, the diameter of the bulb equals less than 1.5 the tibial diameter. The mid-piece of the bulb is longer than either the depth of the bulb or the embolic portion, and slightly exceeds the length of the tibia. Male first leg. The metatarsus and tarsus are slender and slightly curved, but not sinuous. No apophyses, protuberances, or unusually heavy spines are present on the metatarsus. Investiture. Hair pattern presenting no unusual features. Spination. See Table 5. Diagnosis. Female. No prolateral or retrolateral spines on tibiae I or Hl; 44 ventral spines on tibiae I and II; dorsum of abdomen with a pattern of transverse bars on contrastingly colored background. Male. First metatarsus slender, not or only very slightly sinuous, without apo- physes or heavy spines. Metatarsal comb of 4 spines. Embolic spine shorter than palpal tibia, the latter short and inflated. Abdomen with dorsal pattern as in female. Distribution. Southern Brazil, Uruguay, east central Argentina (Map 2). ARIADNA IN THE AMERICAS + Beatty 477 iecords. BRAZIL. RIO GRANDE DO SUE: SRios Grande: URUGUAY. MONTEVIDEO (4 and 9? syntypes ). ARGENTINA. BUENOS AIRES: Mo- reno, Oct. 1947, 2? (R. D. Schiapelli). Ariadna boliviana Simon Figures 4, 6, 14, 51—52, 54—55. Map 3. Ariadna boliviana Simon, 1907, Ann. Soc. Ent. Belg., 51:262. Male and female syntypes from Espiritu Santo, Bolivia, in Muséum National (Histoire Naturelle, Paris, examined. Petrunke- vitch, 1911, Bull. Am. Mus. Nat. Hist., 29:130. Discussion. The odd distribution pattern of the two known collections of this species is probably a result of lack of thorough coverage of the area by collectors. Both collections are from upland areas, but a wide lowland, the Gran Chaco, lies be- tween them. In comparing males from the two localities, I can find no significant dif- ference between them. Since other species of the genus have a rather large elevational range (p. 442) it is probable that A. bolivi- ana occurs in suitable habitats in the Gran Chaco. Color. Female. Carapace light orange- brown, darker in cephalic region. Legs yellow-brown, anterior legs and _ distal podomeres darker. Faint darker distal annuli on tibiae and metatarsi I and II. Abdomen pale purplish gray above, yellow on sides and venter. Male. Carapace uniform mahogany, with very faint streaks radiating from thoracic groove. Abdomen purplish gray above, dirty yellow beneath. Legs yellow brown, first pair darker. Conspicuous purplish gray distal annuli on first tibia and meta- tarsus, fainter annuli on second tibia and metatarsus. Structure. Dimensions of two males: total length 6.5, 6.5 mm; carapace length 3.2, 3.4 mm; carapace width 2.1, 2.2 mm; sternum length 1.6, 2.0 mm; sternum width 1.10, 1.08 mm. 478 Dimensions of a single female: total length 7.8 mm; carapace length 3.8 mm; carapace width 2.4 mm; sternum length 2.3 mm; sternum width 1.37 mm. Male palp. Bulb of medium size, tibia somewhat inflated. Diameter of bulb more than 1.5 times that of tibia. Midpiece of palp short, less than diameter of bulb, about half the length of embolic portion. Embolic portion equalling tibia in length. Male first leg. Metatarsus and _ tarsus slender and sinuous, without apophyses or heavy spines. Tibia with ordinary spines only, none modified. Investiture. Male. Hair short and very sparse on carapace, otherwise normal. Tarsi and distal portion of metatarsi I-IV with ventral scopulae of short, translucent, minutely hooked bristles. Spines of rel- atively short to medium length, those on tibiae I and IT shorter than, to slightly longer than, diameter of podomere. Female. Normal. Spination. See Table 6. Diagnosis. Male. Metatarsus I slender and sinuous, lacking apophyses and modi- fied spines; bulb of palp small, embolic portion about equal to palpal tibia in length; metatarsal comb of four spines; metatarsi and tibiae I and II with purplish gray distal annuli; carapace length 3.2-3.4 mm. Female. Lateral spines present on tibiae I and IJ; 4-4 ventral spines on tibia I, 4-(2-3) on tibia II; metatarsal comb of four spines; metatarsi III bearing 9-12 spines of which 1-2 are retrolateral. Distribution. Bolivia, southeastern Brazil (Map 3). Records. BOLIVIA. ESPIRITU SANTO. and 2 syntypes. (Garlepp). BRAZIL. MINAS GERAIS: Diamantina, Minas de Serrinha, ¢, 1945 (Eliz. Cohn). a Ariadna fidicina (Chamberlin), new combination Figure 10. Map 1. Citharoceps fidicina Chamberlin, 1924, Proc. nia Acad. Sci. (4)12(28):608. Im- Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 mature holotype from Ensenada, Baja Cali- fornia del Norte, in California Academy of Science, examined. Citharoceps californica Chamberlin and Ivie, 1935, Bull’ Univ. Utah, 26(4): 85 Mics) 922=235mCr Female (?) holotype from Laguna Beach, Cali- fornia, in University of Utah collection, not seen. NEW SYNONYMY. Discussion. In its pattern of spination A. fidicina is rather divergent from most other Ariadna, but no more so than a few other undoubted Ariadna (e.g. A. gracilis). The genus Citharoceps was erected on the basis of the remarkable large coarse stridu- lating patches on the carapace. In the absence of males, the synonymy of Citharo- ceps with Ariadna may seem doubtful. The female, however, is clearly an Ariadna that happens to have stridulating grooves. Dis- covery of the male is expected to confirm the synonymy. The holotype of C. fidicina was listed (Chamberlin, 1924) as being a female. The specimen has been dried, and is in very poor condition at present. It appears to be immature. Citharoceps californica was described (Chamberlin and Ivie, 1935) in part as follows: “A larger and darker species than C. fidicina Chamberlin, which it otherwise resembles very closely. Known only from immature specimens, which range up to 9 mm in length.” The holotype of C. cali- fornica was not available, but I have ex- amined four of the paratypes (same collec- tion data as the holotype) and find they are all mature. The size and color differ- ences between the two described species are as one would expect as a consequence of the age difference. Several appendages are missing from the holotype of C. fidicina, so that comparison of spination is scarcely possible. A para- type, from the collection of the Museum of Comparative Zoology, was also ex- amined. There is no significant difference in any character between paratypes of C. fidicina and C. californica of similar size. There can be scarcely any question that the two are synonymous, Color. Carapace a dark mahogany. Chelicerae and palps darker than carapace, legs lighter than carapace. Anterior legs darkest, the others progressively lighter posteriorly. Sternum orange-brown. Abdo- men dark purplish gray above, sometimes with a series of yellow transverse markings producing an indistinctly barred pattern. Venter dirty yellowish mottled with pur- plish gray. Structure. Dimensions of ten females: Total length 7.8-10.8 mm, mean 8.97 mm; carapace length 3.6-4.5 mm, mean 4.00 mm; carapace width 2.1-2.7 mm, mean 2.35 mm; sternum length 2.0-2.4 mm, mean 2.23 mm; sternum width 1.16-1.43 mm, mean 1.302 mm. Investiture. Perhaps a little more densely clothed with hairs ventrally than other species, but the difference is scarcely noticeable in most preserved specimens. The spines are proportionately shorter than usual. The tibial spines are shorter than the diameter of the tibiae, and even the primary spines of metatarsi I and II exceed the metatarsal diameter only slightly. The extra pair of apical ventral spines on the metatarsi are usually between the distal spines of the two ventral rows. Spination. See Table 9. Diagnosis. The presence of a patch of coarse stridulating grooves on each side of the cephalic region immediately distin- guishes A. fidicina from all other American Ariadna. In addition, the first and second metatarsi have four apical ventral spines, also a unique character among the Ameri- can species. The metatarsal comb contains five to six spines. Distribution. Pacific Coastal region of North America from Pacific Grove, Cali- fornia, to Ensenada, Baja California. Records. CALIFORNIA. Monterey Co.: Pacific Grove; Los Angeles Co.: Glendale, Santa Monica Mtns, Saddle Peak, Agoura; Orange Co.: Laguna Beach, under bark of trees, holotype and paratypes of Citha- roceps californica; Santa Ana Canyon, 12 mi. E of San Juan Capistrano. ARIADNA IN THE AMERICAS + Beatty 479 MEXICO. BAJA CALIFORNIA DEL NORTE: Ensenada, holotype and_ para- types of Citharoceps fidicina. Ariadna gracilis Vellard Figures 7, 28-29, 33. Map 2. Ariadna gracilis Vellard, 1924, Arch. Inst. Vital Brazil, 2(2):160, figure 45, ¢. Male holotype from Caxias, Maranhao, originally deposited in collection of the Instituto Vital Brazil, not seen. Bonnet: 1955, Biblio. Aran., 2(1):732. Discussion. According to a letter from Dr. Roched A. Seba of the Institute, the holotype of A. gracilis is no longer in the collection of the Instituto Vital Brazil. Fortunately Vellard’s paper contains one of the few adequate descriptions of an Ariadna species, and a male in the collec- tion of the Museum of Comparative Zool- ogy matches the description closely. Assignment to this species of the females described below is based in part upon similarities in size, coloration, and meta- tarsal comb. The distribution of the female specimens, all unquestionably belonging to a single species, suggests further that A. gracilis is the common, if not the only, Ariadna throughout the Amazon Basin. Ariadna obscura’ Blackwall and A. taperae Mello-Leitao are described as having a single apophysis on the first meta- tarsus of the male, at least similar to that of A. gracilis. Either or both of these could conceivably be synonymous with gracilis, but their occurrence outside the Amazon Basin suggests otherwise. Mello-Leitao himself (1947) later synonymized A. taperae and another of his own species, A. campinensis, with obscura, without giv- ing reasons for doing so. No decisive information can be derived from the de- scriptions of any of these three species. Color. Carapace rich reddish brown to duller mahogany brown. The appendages and underside of cephalothorax show the usual pattern of variation with respect to carapace color. Abdomen purplish gray above and yellowish to entirely yellowish white beneath. Anterior legs of male 480 darker than those of female, otherwise the sexes are similar in coloration. Structure. Dimensions of one male: Total length 5.9 mm; carapace length 3.1 mm; carapace width 2.0 mm; sternum length 2.0 mm; sternum width 1.06 mm. Dimensions of thirteen females: Total length 7.1-9.8 mm, mean 8.59 mm; car- apace length 3.7-4.6 mm, mean 4.14 mm; carapace width 2.2-2.9 mm, mean 2.4 mm; sternum length 2.2-2.9 mm, mean 2.49 mm; sternum width 1.16-1.63 mm, mean 1.347 mm. Male palp. Bulb very small, in retro- lateral view its diameter not exceeding the maximum diameter of the tibia. Tibia somewhat inflated; midpiece of palp longer than spine or embolic portion, shorter than tibia. Midpiece and embolic portion to- gether slightly longer than tibia. An inner distal spine on the tibia. Male first leg. Metatarsus slender, sinuous, bearing a _ large ventrolateral apophysis at about the middle. The apophysis bears a forward-pointing spine distally. The right metatarsus has a thick, heavy, but very small prolateral spine distal to the apophysis. A slightly modified distal spine in the inner ventral row of the tibia. Investiture. Male. Hair largely rubbed off, but apparently normal in pattern. Tarsi of all but the first pair of legs with ventral scopulae at least distally. Female. No unusual features. Spination. See Table 10. Diagnosis. Male. The single ventral apophysis on the first metatarsus dis- tinguishes A. gracilis from all other Ameri- can species, except possibly A. obscura (see discussion above). Female. The presence of two or three prolateral spines on femur I separates the female from all other American species except A. multispinosa. From the latter, A. gracilis differs by having lateral tibial spines and fewer ventral tibial spines. Distribution. Northern Brazil and east- ern Peru, in the Amazon Basin and along Bulletin Museum of Comparative Zoology, Vol. 139, No. & river valleys in the higher regions (Map 2). Records. BRAZIL. AMAZONAS: Teffe. PARA: Belem. Caxias, ¢ holotype, St. André. BAHIA: Salvador. PERU. SAN MARTIN: Mishqui-Yacu, 20 km NE Moyobamba, 1200 m (3940 ft). Ariadna obscura (Blackwall) Map 3. Dysdera obscura Blackwall, 1858, Ann. Mag. Nat. Hist., 3(2):334. Immature holotype from Brazil, Pernambuco, destroyed. Blackwall, 1861, Ann. Mag. Nat. Hist., 3(48):446; Petrunkevitch, 1911, Bull. Amer. Mus. Nat. Hist., 29:132; Bonnet, 1956, Biblio. Aran., 2(2):1632. Ariadora [sic] campinensis Mello-Leitéo, 1916, Broteria, 15(1):13. Female holotype from Campina Grande, Paraiba do Norte, in Museu Nacional, Rio de Janeiro, not seen. Mello- Leitao, 1947, Arq. Mus. Paranaense, 6(6) :234; Bonnet, 1955, Biblio. Aran., 2(1):732. Ariadna taperae Mello-Leitao, 1926, Ann. Acad. Brasil. Sci., 1(2):93. Male and female syntypes from “Tapera,’ in Museu Nacional, Rio de Janeiro, not seen. Mello-Leitao, 1947, Arq. Mus. Paranaense, 6(6):234. Ariadna obscura: Mello-Leitio, 1947, Arq. Mus. Paranaense, 6(6):234. Ariadna taperana: Bonnet, 1955, Biblio. Aran., PX STK Discussion. The placement of these three species is made very difficult by the inac- cessibility of all three types and the fact that only one specimen is available from the general region where the types were collected, about 100 miles from the nearest type locality. Blackwall’s description, incomplete and incorrect as it is, suffices to place Dysdera obscura in the genus Ariadna. It contains nothing, however, which could possibly identify any given species of the genus. The type is not available, and Cooke (in litt.) states that the specimen was probably destroyed before Blackwall’s collection came into the Oxford Museum. In 1947, Mello-Leitao synonymized his own species campinensis and taperae with obscura, without giving any reasons for his action. It is unlikely that he had ever seen an authentic specimen of obscura. The synonymy is justifiable on the basis of dis- eee eee tribution, however. All three species, as well as the specimen assigned to A. ob- scura below, originate in a sector of east- ernmost Brazil about 350 miles (560 km) in diameter. Except for a few insular species, most frequently collected Ameri- can Ariadna have a range far larger than this. The probability that more than one species occupies this part of Brazil seems quite small. The discrepancies between the descrip- tions of Mello-Leitao’s two species and the specimen from Natal are more serious, al- though Mello-Leitaéo offers only a small fraction of the information available from his specimens. The disagreements in num- bers of metatarsal spines are relatively unimportant, lying, as they do, within the normal range of variation of a single deme in other species. The tibial spination, and to some extent that of the third metatarsus, offers problems, however. Both obscura and campinensis are described as having 5-5 or 6-6 ventral spines on tibiae I and II. The specimen from Natal has 4—4 on tibia I and 4-1 on tibia II. An unusually variable species, such as A. pilifera, might include all three variants of first tibial spination, but the difference between 4-1 and 6-6 on the second tibia, a leg segment normally showing a highly stable pattern of spination, is far too great for any single species known to me. Further inspection of Mello-Leitao’s de- scriptions, and comparison with many specimens of other species, suggests that his published data are in error, perhaps seriously so. Ariadna campinensis, for ex- ample, is described as having 6-6 ventral spines in four areas, the ventral surfaces of tibiae and metatarsi I and TH. Exami- nation of 158 specimens of A. maxima and A. bicolor reveals not a single specimen having such a degree of symmetry or uni- formity of spine numbers in all four areas. In fact, I can not locate in my records a single instance of complete symmetry in spination in any mature Ariadna. Almost certainly, therefore, the meager ARIADNA IN THE AMERICAS + Beally 15 | information on spination given by Mello- Leitao is inaccurate. With some reluctance, because of the magnitude of the discrepan- cies between descriptions and specimen, I conclude that the best course is to regard the three described species as synonymous, and to assume, until otherwise demon- strated, that the specimen on hand _ is Ariadna obscura. Color. Male. Blackwall (1861) describes the male as generally paler than the fe- male, but with the anterior legs browner, and the palpi yellowish white. Female. Carapace and legs I and II deep brown. The other legs paler. Ster- num, labium, and undersides of coxae only a little lighter than carapace. Abdomen dark gray above, slightly paler beneath. Structure. Dimensions of one female: total length 8.5 mm; carapace length 4.1 mm; carapace width 2.4 mm; sternum length 2.3 mm; sternum width 1.37 mm. Male palp. Details of structure unknown. Male first leg. Described by Blackwall (1861) as having a retrolateral or ventro- retrolateral apophysis ending in a_ short spine. Investiture. Female. Hair largely rubbed off, but apparently of normal pattern. Lateral spines of tibiae I and II, and pro- lateral spines of patellae quite short, ap- pressed, and almost invisible against the dark brown leg. Spination. See Table 16. Diagnosis. Male. Blackwall’s descrip- tion of the male indicates a similarity to A. gracilis, but to no other known male. The data given are insufficient to diagnose A. obscura more precisely. Female. The metatarsal comb of two spines distinguishes A. obscura from most other species. From the others with two- spined comb, it may be separated by the presence of spines on the patellae of legs I and II, a unique character in females of American Ariadna. Distribution. Eastern Brazil, in the states of Rio Grande do Norte, Paraiba, Pernam- buco, and possibly Bahia. 482 Records. BRAZIL. RIO GRANDE DO NORTE: Natal, June 1911, Stanford Exp., 2, (W. M. Mann); PARAIBA: Campina Grande (type locality of A. campinensis ); PERNAMBUCO: no further data (type locality of A. obscura); Tapera (Pernam- buco or Bahia?), no further data (type locality of A. taperae ). Ariadna solitaria Simon Map 4. Ariadne solitaria Simon, 1891, Proc. Zool. Soc. London, 1891:556. Immature holotype from Lesser Antilles, St. Vincent Island, in British Museum (Natural History), examined. Ariadna solitaria Petrunkevitch, 1911, Bull. Amer. Mus. Nat. Hist., 29:131; Bonnet, 1955, Biblio. Aran., 2(1):737. Not A. solitaria: Lutz, 1915, Ann. New York Acad. Sci., 26:81 (=A. arthuri). Discussion. Although known only from the immature holotype, this species is clearly distinct from the two other West Indian Ariadna, both of which have more ventral tibial spines. No other Ariadna has been taken in the Windward Islands or the nearby coast of northern South America. Possibly A. solitaria will ultimately prove to have a more extensive range in this region. Color. Carapace and chelicerae orange- brown, lighter in cephalic region. Legs and palps yellow, tarsus of palp, tarsi of legs I and II, metatarsus of leg I darker. Endites, labium, and sternum _ yellow. Abdomen grayish yellow above, yellow beneath. Structure. Dimensions of holotype: Total length 9.0 mm, carapace length 4.2 mm, carapace width 2.0 mm, sternum length 2.1 mm; sternum width 0.979 mm. Investiture. Largely rubbed off, but ap- parently plentiful and of usual arrange- ment. Spination. See Table 19. Lateral spines present on two tibiae; ventral spines of tibiae I ind IT 4-4; metatarsal comb of 4 spines. Spination of third leg: 4 ventral, 2 pro- Diagnosis. first Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 lateral, 0 retrolateral spines on metatarsus; 2 ventral, no lateral spines on tibia. Distribution. Thus far found only on St. Vincent, Windward Islands. Record. ST. VINCENT: Baronallie, near sea-level, open valley, under rubbish, 0, (H. H. Smith). Ariadna tubicola Simon Map 4. Ariadne tubicola Simon, 1893, Ann. Soc. Ent. Fr., 61:448. Immature syntypes from Vene- zuela, Caracas, in Muséum National d Histoire Naturelle, Paris, examined. Ariadna tubicola: Petrunkevitch, 1911, Bull. Amer. Mus. Nat. Hist., 29:131; Bonnet, 1955, Biblio. Aran., 2(1): 738. Discussion. The two immature syntypes are the only specimens presently known. They seem to me sufficiently distinct from the other Ariadna of northwestern South America to be considered a_ separate species. Again, many more specimens will be required to establish the status of the form firmly. Color. Carapace orange, legs yellow. Metatarsi I and II with distal and sub- basal dark annuli. Chelicerae, endites, labium, and sternum brownish yellow. Abdomen in poor condition, yellow with a suggestion of a median series of purplish gray markings of the usual form. Structure. Dimensions of immature lectotype: Total length 7.4 mm, carapace length 3.0 mm, carapace width 1.8 mm, sternum length 1.7 mm, sternum width 1.08 mm. Investiture. Hair reddish brown, of usual distribution. Spination. See Table 22 Diagnosis. Lateral spines present on tibiae I and IH; ventral tibial spines more than 4-4 on I and II; metatarsal comb of 4—5 spines; a single prolateral spine on femur I; palpal patella with one prolateral spine; tarsi very short. Distribution. North (Map 4). Record. VENEZUELA. central Venezuela DISTRITO FE- DERAL: Caracas, 00, types. (E. Simon), syn- Ariadna Incertae Sedis Map 3. Ariadna conspersa Mello-Leitao, 1940, Arq. Inst. Biol. Sao Paulo, 11(30):256. Dysdera crassipalpus Blackwall, 1863, Ann. Mag. Nat. Hist:, ser. 3; 11(61)):43. Ariadna dubia Mello-Leitao, 1917, Broteria, 15:82. Ariadna spinifera Mello-Leitaio, 1947, Arq. Museu Paranaense, 6(6):233, figure 1, ¢@. No holotypes or specimens from the type localities of any of these species have been seen. Blackwall’s description is worth- less, those of Mello-Leitao are so frag- mentary, and probably inaccurate, as to be unusable. Two of the species, crassipalpus and dubia, were described from specimens taken in Rio de Janeiro, a male and a fe- male respectively. The holotypes of A. spinifera and A. conspersa are a male and a female from Curitiba, Parana. Similar situations involving A. bicolor and A. maxima were readily solved, and it is tempting to apply the same procedure to the present species. Four species de- scribed from the United States and four from Chile were reduced to one in each case when study of many specimens from widely scattered localities revealed that the populations in each area were quite uniform. My original inclination, therefore, was to unite all the southeastern Brazilian species under one or, at most, two names. Certainly the description of a male and female from each of the two Brazilian sites suggested each pair of names referred to a single species. The relatively short airline distance of 400 miles separates Rio de Janeiro and Curitiba. Eight other American Ariadna, all the species that are fairly well-known, range over distances much greater than 400 miles. Furthermore, in North America, Central America, and the West Indies the pattern of distribution of Ariadna is one of allopatry of all species so far as presently known. This statement is true for most of South America, also, but so few specimens ARIADNA IN THE Americas + Beatty 453 of most species have been collected in South America that the known distribution there is of little significance. Unfortunately the simple treatment that was appropriate for A. bicolor and A. maxima can not be justified for the Bra- zilian species. A small collection of speci- mens from Sao Paulo (almost exactly half- way between Rio de Janeiro and Curitiba ) has been examined. This collection con- tains three, or possibly four, species. Ariadna mollis has also been collected in southeastern Brazil, and may occur in the Sao Paulo region. At present I find the task of matching the available specimens with the published names impossible. Examination of type material would be a step toward solving the problem, but in itself might be in- sufficient. Series of specimens from several localities, certainly including Rio de Ja- neiro and Curitiba, will be required. These series must include a number of mature females (five at the minimum ), and should also include at least one male from each locality. The material already on hand suggests that some of the species are not very different from each other morpho- logically. The region from Buenos Aires, Argen- tina, to Diamantina in the Brazilian state of Minas Gerais presents more serious taxonomic problems and more interesting biological ones in the genus Ariadna than any other part of the Americas. The only indications of sympatry of two or more species of Ariadna are in this area and in southern Chile. Besides the situation de- scribed above, A. mollis and A. boesen- bergii have apparently been taken together in Buenos Aires, and A. boliviana was found with another (undetermined) spe- cies at Diamantina. Plainly, on-the-spot investigations at least in southeastern Brazil are needed. LITERATURE CITED Banks, N. 1929. Spiders from Panama. Mus. Comp. Zool., 69(3):53-96. Bull. 484 Barnes, R. D. 1953. The ecological distribution ot spiders in nonforest maritime communi- ties at Beaufort, North Carolina. Ecol. Mono- graphs, 23:315-337. Beatty, J. A., AND W. H. Bossert. (in prep). A computerized study of American Ariadna (Araneae: Dysderidae). BLACKWALL, J. 1858. Characters of a new genus and descriptions of three recently discovered species of Araneidea. Ann. Mag. Nat. Hist., ger, B Bassil—selsy 1861. Descriptions of several recently discovered spiders. Ann. Mag. Nat. Hist., ser. 3, 8:441-446. . 1863. Descriptions of newly discovered spiders captured in Rio Janeiro by John Gray, Esq., and the Rev. Hamlet Clark. Ann. Mag. Nat. Hist., ser. 3, 11:29—45. Bonnet, P. 1955. Bibliographia Araneorum. Toulouse, 2(1):1-918; , 1956) "Op. cit., 2( 2): 919-1925: . 1958. Op. cit., 2(4):3027—4230. Bryant, E. B. 1948. Spiders of Hispaniola. Bull. Mus. Comp. Zool., 100(4):332—447. Buxton, B. H. 1913. Coxal glands of the arach- nids. Zool. Jahrb. Suppl., 14:231-282. CHAMBERLIN, R. V. 1916. Results of the Yale Peruvian Expedition of 1911. The Arachnida. Bull. Mus. Comp. Zool., 60(6):177-299. . 1920. South American Arachnida, chiefly from the Guano Islands of Peru. Bull. Brook- lyn Inst. Arts Sci., 3(2):35-44. 1924. The spider fauna of the shores and islands of the Gulf of California. Proc. California Acad. Sci., ser. 4, 12:561-694. CHAMBERLIN, R. V., AND W. Iviz. 1935. Mis- cellaneous new American spiders. Bull. Univ. Wtah; Biolk Sen, 2.8) 179: Comstock, J. H. 1948. The spider book. 2d ed., rev. W. J. Gertsch. Ithaca, New York: Com- stock Publ. Co., p. 306. Cooke, J. A. L. 1965a. Spider genus Dysdera (Araneae, Dysderidae). Nature (London), 205 ( 4975 ) : 1027-1028. 1965b. Systematic aspects of the ex- ternal morphology of Dysdera crocata and Dysdera_ erythrina (Araneae, Dysderidae ). Acta Zool., 46:41-65. 1966. Synopsis of the structure and function of the genitalia in Dysdera crocata (Araneae: Dysderidae). Senckenbergiana Biologica, 47:35-43. EmMertON, J. H. 1875. Notes and descriptions. In N. M. Hentz, Descriptions and figures of the Araneides of the United States. Occ. Pap. Boston Soc. Nat. Hist., 2:15-164. FRANGANILLO, B. P. 1936. Los Aracnidos de Cuba hasta 1936. La Habana, p. 38. GernmarptT, U. 1921. Vergleichende Studien iber die Morphologie des minnlichen Tasters Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 und die Biologie der Kopulation der Spinnen. Arch. Naturg., 87:78—-247. GeruHarptr, U., AND A. Kastner. 1938. Araneae —Echte Spinnen— Webspinnen. In W. Kikenthal and T. Krumbach, Handbuch der Zoologie, 3(2):497—656. ° GertscoH, W. J. 1958a. The spider family Diguetidae. Amer. Mus. Novitates, 1904: 1-24. . 1958b. The spider genus Loxosceles in North America, Central America, and the West Indies. Amer. Mus. Novitates, 1907: 1-46. . 1958c. The spider family Plectreuridae. Amer. Mus. Novitates, 1920:1-53. 1967. The spider genus Loxosceles in South America. Bull. Amer. Mus. Nat Hist., 136(3):121-173. Kaston, B. J. 1948. Spiders of Connecticut. State Geo. Nat. Hist. Survey, Connecticut, 70:48. . 1952. How to Know the Spiders. Dubu- que, Iowa: Wm. C. Brown Co., p. 25. Lutz, F. E. 1915. List of Greater Antillean spiders with notes on their distribution. Ann. New York Acad. Sci., 26:71-148. Meuio-Lerrao, C. pe. 1916. Notas Arachno- logicas. IV. Novas especies do Brasil. Bro- teria, 14:12-13. . 1917. Notas Arachnologicas. V. Especies novas ou pouco conhecidas do Brasil. Broteria, 15:74-102. . 1933. Catalogo das aranhas argentinas. Arch. Esc. Sup. Agric. Med. Vet., 10(1): 3-63. . 1940. Aranhas do Parana. Arq. Inst. Biol. (Sao Paulo), 11(30):235—257. . 1941. Las arafias de Cérdoba, La Rioja, Catamarca, Tucuman, Salta y Jujuy. Rev. Mus. La Plata, N. S., Secc. Zool., 2:99-198. 1944. Catalogo das aranhas do Rio Grande do Sul. Arq. Mus. Nac., (Rio de Janeiro), 37:149-245. . 1945. Arahas de Misiones, Corrientes y Entre Rios. Rev. Mus. La Plata, N. S., Secc. Zool., 4:213-302. . 1946. Aranhas del Paraguay. Not. Mus. La Plata, N. S., Secc.\Zool., 1 317—50: . 1947. Aranhas do Parana e Santa Cata- rina, das Colecédes do Museu Paranaense. Arq. Mus. Paranaense, 6(6):231—304. Minor, J. 1931. Anatomie comparée de Vintestin moyen céphalo-thoracique chez les Araignées Zeits. Morph. Okol. Tiere, 21:740- vraies. 764. Petrunkevitcu, A. 1911. A synonymic index- catalogue of spiders of North, Central and South America with all adjacent islands. Bull. Amer. Mus. Nat. Hist., 29:130-132. 1926. Spiders from the Virgin Islands. Trans. Connecticut Acad. Arts Sci., 28:21-78. . 1929. The spiders of Porto Rico. Part one. ‘Trans. Connecticut Acad. Arts Sci., 30:1-158. . 1933. An inquiry into the natural classi- fication of spiders, based on a study of their internal anatomy. Trans. Connecticut Acad. Arts Sci., 31:303-389. . 1939. Classification of the Araneae, with key to suborders and families. Trans. Con- necticut Acad. Arts Sci., 33:133-190. Prckarp-CampBrincE, O. 1898. Biologia Centrali- Americana, Arachnida, Araneidea. London, 1:235. Porter, C. EF. 1917. Apuntes sobre aracnologia chilena. I. Sinopsis de los Disdéridos. Rev. Chilena Hist. Nat., 21(6):172-182. Purceti, W. F. 1904. Descriptions of new genera and species of South African spiders. Trans. South African Phil. Soc., 15(3):115-173. Srmon, E. 1891. On the spiders of the island of St. Vincent. Part 1. Proc. Zool. Soc. London, 1891 :549-575. . 1893a. Histoire naturelle des Araignées. Paris, 2:308-322. . 1893b. Arachnides. In Voyage de M. E. Simon au Venezuela (Décembre 1887-Avril 1888). 2le Mémoire. Ann. Soc. Ent. Fr., 61 :423-462. SnNoperaAss, R. E. 1935. Principles of Insect Morphology. New York: McGraw-Hill Co., pp. 56-57. Suzuki, S. 1952. Cytological studies in spiders, III. Jour. Sci. Hiroshima Univ., (Zool.), 15 :23-136. WiccLEsworTH, V. B. 1954. The Physiology of Insect Metamorphosis. Cambridge: Cam- bridge University Press, pp. 52-54. (Received 21 August 1968.) ARIADNA IN THE AMERICAS Beally 485 ADDENDUM Since this work was completed, ad- ditional specimens have become available through the cooperation of Dr. W.. J. Gertsch and Mr. Vincent Roth, of the American Museum of Natural History. One of these specimens is the first known male of Ariadna fidicina (originally Citharoceps fidicina). It has the stridulating patches on the carapace, as in the female. Other structural characters agree with the defi- nition of Ariadna, confirming the synonymy of Citharoceps with Ariadna, as expected. The remaining new material consists of one male, many mature females, and a few juveniles of A. tarsalis, collected on several of the Galapagos Islands. Mature Ariadna from the Galapagos have not previously been available to me. These specimens agree in all respects with those described above as Ariadna peruviana, new species. Therefore A. peruviana is hereby synony- mized with A. tarsalis Banks. The presence of A. tarsalis on the mainland of South America is unexpected and _ surprising. However, a coastal species at Lima is admirably placed for rafting to the Gala- pagos on the Humboldt Current. A later paper will give details on the above specimens, with illustrations, and with modifications of the descriptions and keys for A. tarsalis and A. fidicina. 486 Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 TABLE 2. SpINATION OF ARIADNA PERUVIANA Male N = 4 Female N = 11-12 Range Mode Sa Range Mode n Palp Patella 0 0 4 Oa 0 12 Tibia 0) 0) 4 3=5> 3 6 Tarsus 0 0 4 6-8 6 6 Leg: 1—Meta.—Ventral ou. 1 il 4 8-10 9 U Ventral in. Y 2 4 8-10 9 5 Prolateral 0-1 0 3 0 0) 11 Tibia—Ventral ou. 4-6 4 3 4-6 4 14 Ventral in. 4 4 4 4-5 4 10 Ventral su. 0-1 0,1 2 ea. 0 0 11 Dorsal 1 1 4 0 0 11 Femur—Dorsal ou. il 1 4 il 1 ll Dorsal mi. 2 0} 4 2 1 10 Dorsal in. =o) 2 3 0-2 I 8 Prolateral i 1 4 il 1 ial 2—Meta.—Ventral ou. 3 3 4 7-10 9 oh Ventral in. 3 3 4 9-11 9 6 Prolateral ]-2 it 3 0 0 11 Retrolateral 3 3 4 0 0 itil Tibia—Ventral ou. 4 4 4 325 4 6 Ventral in. 1-2 lee?) 2 ea. 4 4 11 Prolateral 3 3 4 8} 3 10 Retrolateral 4 4 4 O23) 3 8 Dorsal 0-1 il 3 0) 0 iil Femur—Dorsal ou. 2 2 4 0-1 1 10 Dorsal mi. 4 4 4 |-4 1 6 Dorsal in. ]-2 9) 3 =o) 1 a 3—Meta.—Ventral ou. 3 3 4 3 3 2, Ventral in. 2 2 4 9 2 12 Prolateral 4 4 4 BoA! 4 8 Retrolateral 8} D8 2ea 0-3 il 6 Tibia—Ventral ou. 5) 3 3 3 3 12 Ventral in. 1 1 3 0-2 0) 10 Prolateral 2 2 4 eo) 9) ila Retrolateral 1E3 1,3 2 ea. 0 0) 12 Femur—Dorsal ou. 0-2 1 2 0) 0 19 Dorsal mi. 3—5 3 2 0-3 0 6 Dorsal in. 2 9 4 il? 9) 9 4—Meta.—Ventral ou. 2-3 2,3 2 ea 1-2 1 8 Ventral in. 3-4 3 3 Bal 3 9 Ventral su. il 4 0-1 1 Ul Tibia—Ventral ou. 0-1 0, 1 2 ea 0-1 0) 10 Ventral in. 0-1 0 8 0 0 12 Retrolateral |-2 132) 2Qea 0-1 0 10 Femur—Dorsal ou. 0 0) 4 0 0 12 Dorsal mi. 7-11 lea 0-4 2 6 9} 0-1 0 Hal Dorsal in. ARIADNA IN THE AMERICAS * Beatty 487 TABLE 3. SPINATION OF ARIADNA ARTHURI Male IN" Female N 9-10 Range Mode n : Range Mode n Palp Patella 0 0 4 0 0 s) Tibia 0 0 4 4-9 4 5 Tarsus 0) 0 4. 5-8 (6}3'7/ 3 ea. Leg: 1—Meta.—Ventral ou. 2 2 4 6-9 ilk i Ventral in. 3 3 4 6-8 7,8 4A ea Tibia—Ventral ou. 7 ih 4 5-8 6 4 Ventral in. 1 1Be2) Qea ARG 6 ii Prolateral 3 3 4 2-3 3 G6 Retrolateral 3 3 4 9) 9) 10 Dorsal 1 i 4 0 0 10 Femur—Dorsal ou. 1 1 4 il il 10 Dorsal mi. 1? iL, 2. ea. 1 il 10 Dorsal in. o; 9} 4 1&9, 2 6 Prolateral 0-1 0,1 2 ea. 1 il 10 2—Meta.—Ventral ou. 4 4 4 6-9 7 5 Ventral in. 3 3 4 7-8 8 8 Retrolateral 9) 9} 4 0 0 10 Tibia—Ventral ou. 7 7 4 6-8 rif 5 Ventral in. 1 1 4 Ae 6 ih Ventral su. 0-1 0, 1 2 ea. 0 0) 10 Prolateral 3 3 4 123 3 6 Retrolateral 4 4 4 IY 9 9 Dorsal 0-1 0 3 0 0) 10 Femur—Dorsal ou. 1 1 4 iL 1 10 Dorsal mi. 2) ) 3 0-1 1 qf Dorsal in. 2, 2 a 1-2 2 6 3—Meta.—Ventral ou. 3 3 4 3 3 10 Ventral in. il I 4 1 il 10 Prolateral 2 2 4 0-3 0 4 Retrolateral 2) 9) 4 0) 0 10 Tibia—Ventral ou. 3 3 4 2-3 3 9 Ventral in. 0 0 4 0 0 10 Retrolateral o=3 DB 2 ea. j-2 iL 9 Femur—Dorsal ou. Il 1 4 0-1 il 6 Dorsal mi. 0-3 lea. 0) 0) 10 Dorsal in. 1 1 4 0-1 0 6 4—Meta.—Ventral ou. i 1 As 1 1 10 Ventral in. 2 2 4 2 2; 10 Ventral su. 1 1 as 0 0 10 Femur—Dorsal ou. 0 0 4 0 0 10 Dorsal mi. 025 5 2 0 0 10 iL 0, 1 2. ea. 0) 0) 10 Dorsal in. 0- 488 Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 TABLE 4. SpINATION OF ARIADNA BICOLOR Male N = 100 Female N = 200 Range Mode n Range Mode Palp Patella 0 0 100 O=1 = 0 Tibia 0 0 100 0-6 3 Tarsus 0 0 100 5-16 9 Leg: 1—Meta.— Ventral ou. 2-3 Do) 97 6-11 8 Ventral in. 3-4 3 96 ill 8 Tibia—Ventral ou. 3-5 4 97 3-6 4 Ventral in. 5 4 86 2-6 4 Prolateral 0-5 4 46 0) 0 Retrolateral O=5 4 79 0 0 Femur—Dorsal ou. 0-2 1 90 0-1 0 Dorsal mi. 1-6 3 24 0-2 0 Dorsal in. 0-2 2 55 0-2 0 2—Meta.—Ventral ou. 9-4 3 97 b=, 8 Ventral in. ei 3 95 TAD 9 Prolateral 0-3 i 45 0 0) Retrolateral 0-2 2 82 0) 0 Tibia—Ventral ou. 3-6 4 90 3-5 4 Ventral in. 19 9, 97 0-4 ® Prolateral 0-5 3 46 0 0 Retrolateral 0-5 3 Bi 0 0 Femur—Dorsal ou. 0-2 il 66 0 0 Dorsal mi. 0-5 4 42 0-2 0 Dorsal in. 0-3 il 68 0-2 1 3—Meta.—Ventral ou. al 3 96 2-6 3 Ventral in. 0-2 IL 94 0-3 l Prolateral 0-5 2 54 1-4 2 Retrolateral 0-2 0 92, 0 0) Tibia—Ventral ou. 1-6 4 44 0-4 2 Ventral in. 0-3 1 52 0 0 Prolateral 0-4 ] 44 0-1 0 Femur—Dorsal ou. 0 0 100 0 0) Dorsal mi. 0-5 3 48 0-1 0 Dorsal in. (9) 1 85 0-2 il 4—Meta.—Ventral ou. 13 9) 81 022 1 Ventral in. oo 3 91 Ol 3 Prolateral 0-3 9} 64. 0 0 Tibia—Ventral ou. 0-4 2 28 0) 0 Femur—Dorsal ou. 0) 0 100 0 0 Dorsal mi. =} 3 AO 0 0 0-2 1 0 0 Dorsal in. ARIADNA IN THE AMERICAS + Beatty 489 TABLE 5, SPINATION OF ARIADNA BOESENBERGII Male N = on = j Female N 7-8 Range Mode n Range Mode n Palp Patella 0 0 2 0 0) Tibia 0 0 2 9-4 3,4 3ea Tarsus 0 0 2 4—6 5 Leg: 1—Meta.—Ventral ou. 3,4 lea 5-9 8 3 Ventral in. 4 4 2 6-8 A 3 Prolateral il 1 9) 0) 0 7 Retrolateral 2 9} 2 0 (0) iT Tibia—Ventral ou. 4,6 1 ea. 4 4 7 Ventral in. 4 4 2 4 4 Wf Prolateral 3,4 lea. 0 0 fl Retrolateral 6 6 2 0 (0) 7 Femur—Dorsal ou. 1 1 2 0 0 7 Dorsal mi. 8} lea. 0-1 0 6 Dorsal in. 2 2 2 0-1 0) 6 2—Meta.—Ventral ou. 3 3 D} 6-9 7 4 Ventral in. 4 4 9) 6-8 ah 4 Prolateral 1 1 2 0 0 8 Retrolateral 2 2 2; (0) 0 8 Tibia—Ventral ou. 4 4 2 4-5 4 vik Ventral in. 4 4 2) 3-4 4 6 Ventral su. 1 il 2 0 0 8 Prolateral 4 4 2 0 0 8 Retrolateral 5,6 lea. @) 0 8 Femur—Dorsal ou. 2 9, 9 0 0 8 Dorsal mi. ORS lea. 0-1 0 6 Dorsal in. 2 2, 2 0-2 2 4 3—Meta.—Ventral ou. 3 3 2 923 3 6 Ventral in. 2) 9} 2 1 1 8 Prolateral 3 3 2 ]-2 I 5 Tibia—Ventral ou. 4 4 2, 1-4 3 5 Prolateral 2 2 y 0 0 8 Retrolateral 1 1 2 0 0 8 Femur—Dorsal ou. 0 0 2, 0 0 8 Dorsal mi. 3 3 2 0) 0 8 Dorsal in. 1D 1 ea. 0-1 1 6 4—Meta.—Ventral ou. 2 ® 2 1 1 d Ventral in. 4 4 2} 4 4 il Tibia—Ventral ou. OS} lea 0 0 7 Femur—Dorsal ou. 0 0 2 0 0) af Dorsal mi. 8,9 lea. 0 0 7 Dorsal in. 1 1 2 0-1 0 5 490 Bulletin Museum of Comparative Zoology, Vol. 139, No. § TABLE 6. SPINATION OF ARIADNA BOLIVIANA Male N = 4 Female N = 2 Range Mode n Range Mode Palp Patella 0 0 4 it 2 2 Tibia 0 0 4 6 6 2 Tarsus 0) 0 4 10 10 2; Leg: 1—Meta.—Ventral ou. 1-2 Il 2 ea. 7,8 it Ventral in. 2-3 2,3 2 ea. 8 8 2 Tibia—Ventral ou. 4 4 4 4 2 Ventral in. ]-2 12, 2, ea. 4,5 1 Ventral su. 0-1 0,1 2 ea. 0 0 2 Prolateral 3-4 3,4 2ea 3 3 2 Retrolateral 7-8 eS Qea 3 3 2 Dorsal ]-2 ] 3 0 0 2 Femur—Dorsal ou. il l 4 1 ] 2 Dorsal mi. 3-4 4 3 1 1 2 Dorsal in. ee 9) 3 2 2 2 Prolateral ile IP 4 1 1 2 2—Meta.—Ventral ou. 3-4 3 3 Gs 1 Ventral in. 9) 2 4 8,9 1 Retrolateral 2 2 4 i 1 2 Tibia—Ventral ou. 4 4 4 4,5 1 Ventral in. 1-2 1b, 2 ea. 3 3 2, Ventral su. 1 1 4 0) 0 9 Prolateral 3 3 4 3 3 2 Retrolateral 4-5 4,5 2ea 3,4 le: Dorsal 0-1 1 3 0 0 2 Femur—Dorsal ou. i 1 4 1 1 9} Dorsal mi. 3-4 4 3 1 il 2 Dorsal in. =2; 1,2 2 ea, 2 2 2 3—Meta.—-Ventral ou. 3 3 4 3 3 9) Ventral in. ]-2 iL 2, ea. 2.) le Prolateral 9-8 ae 2ea 3,4 le: Retrolateral 2} 2 4 1) le Tibia—Ventral ou. 3-4 3 3 3 3 2 Prolateral 2 9) 4 ik le Retrolateral 33 2 3 1 1 2 Femur—Dorsal ou. 0-1 1 3 1 I 9) Dorsal mi. 2-4 3 2 1 1 2 Dorsal in. 19) 3 2 1 1 2 4—Meta.—Ventral ou. O-1 0, 1 2 ea. 0) 0 2, Ventral in. 4-5 4 3 3,4 1 Femur—Dorsal ou. 0) 4 0 0 2 Dorsal mi. 225 lea 0 0 2 Dorsal in. 0-1 0 0 ® 0,1 2 ea. ARIADNA IN THE AMERICAS + Beatty 49] TABLE 7. SPINATION OF ARIADNA CAERULEA Gas N = 9} Female N = 11-12 Range Mode n Range Mode n Palp Patella 0 0 2 0 0 12 Tibia 0) 0 2 Bab) 3 6 Tarsus 0 0 2 4—] J 5 5 Leg: 1—Meta.—Ventral ou. i 1 1 8-11 8, 10 5 ea. Ventral in. il il 1 811 8,9 4 ea. Tibia—Ventral ou. 6 6 1 4-6 4 7 Ventral in. 1 Il il 4-7 7 8 Prolateral 4 4 1 0-3 2 6 Retrolateral 4 4 il 0-2 2 5 Femur—Dorsal ou. 9) » 1 0-1 1 10 Dorsal mi. 4 4 1 0-1 il 11 Dorsal in. 1 1 1 1-2 9» 10 Prolateral 1 1 1 1 1 12 2—Meta.—Ventral ou. 2 2 9) 7-10 7-9 3 ea Ventral in. 3} 3 2 8-11 ) 5 Retrolateral 0-1 0, 1 lea. 0 0) 11 Tibia—Ventral ou. 6 6 2, 4-7 4 6 Ventral in. 2-3 9.3 lea. 3-4 4 9 Ventral su. 0-1 0, 1 lea. 0 0 11 Prolateral 3—4 3,4 lea. 0-3 3 4 Retrolateral 4 4 ® 0-3 0 8 Dorsal 1-2 1 lea 0 0 11 Femur—Dorsal ou. 2 ) 2 0-1 il 8 Dorsal mi. 3-4 3), al lea. 1 1 1l Dorsal in. =? 2) lea I 1 11 3—Meta.—Ventral ou. 3 3 1 3 3 12 Ventral in. ik 1 1 0-2 i W Prolateral 4 4 1 0-3 2. 6 Retrolateral 9; oO) 1 0 0 12 Tibia—Ventral ou. 4 A 1 2-3 3 i Ventral in. 1 1 1 0 0) 12 Prolateral 2 2, 1 0-1 0 9 Retrolateral 3 3 1 0-2 0 10 Femur—Dorsal ou. il al I 0-1 0 1l Dorsal mi. 4 4 1 0 0 12 Dorsal in. i it 1 0-1 It 10 4—Meta.—Ventral ou. 8} OS lea 0-1 0 ah Ventral in. 3 3 2; 3-4 4 9 Prolateral 0-1 0,1 lea. 0 0 1l Tibia—Ventral ou. ] 2} 0 0 il Femur—Dorsal ou. 0 0 2 0 0 11 Dorsal mi. 3-5 3,5 lea 0 0 11 Dorsal in. 1 1 2 0 0 11 492 Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 TABLE 8. SPINATION OF ARIADNA CEPHALOTES TABLE 9. SPINATION OF ARIADNA FIDICINA Female N = 15-18 Female N = 14-20 Range Mode n Range Mode n Palp Patella 0 0 17 Palp Patella 0. 0 20 Tibia 2-5 3 13 Tibia 3-5 5 s0 Tarsus 5-15 8 5 Tarsus 7-13 #11 5 Leg: Leg: 1—Meta.—Ventral ou. 7-11 10 fi [= Met VentralvoueeG 5 16 Ventral in. 6-11 9 8 Wentenleint ary 5 10 Tibia—Ventral ou. 4-7 4 14 Ventral su. 2 2 18 Ventral in. 4-5 4 16 Tibia—Ventral ou. 4 4 18 Retrolateral 0-4 0 11 Wentrileint = oe 4 17 Femur—Prolateral 1 1 17 Femur—Prolateral 1 1 18 2—Meta.— Ventral ou. (Mal 9 5 2—Meta.—Ventral ou. 3-5 4,5 8ea Ventral in. 6-10 8 7 Ventral in. 4-6 4 10 Tibia—Ventral ou. 4-6 4 14 Ventral su. 9) 2 17 Ventral in. 3-4 3 1 Tibia—Ventral ou. 4 4 ilk Femur—Dorsal ou. 0 0 16 Ventral in. 0-1 0 12 Dorsal mi. 0 0 16 Femur—Dorsal ou. 0 0 7 Dorsal in. 0-2 0 14 Dorsal mi. 0 0 iL/ 3—Meta.—Ventral ou. 3 3 18 Dorsal in. 0-1 0 16 Ventral in. 1 1 18 3—Meta.—Ventral ou. 3 3 19 Prolateral 1-2 1 12 Ventral in. 1-2 1 16 Retrolateral 0-2 0 15 Wesntall Gn 12o iL 18 Tibia—Ventral ou. 2-3 3 13 Prolateral 1-2 1 18 Femur—Dorsal ou. 0 0 18 Tibia—Ventral ou. 1-3 3 14 Dorsal mi. 0 0 18 Femur—Dorsal ou. 0 0 19 Dorsal in. 0-2 0 16 Dorsal mi. 0 0 19 4—Meta.—Ventral ou. i 1 18 Dorsalin. 0-1 0 18 Ventral in. 3-4 A 16 4—Meta.—Ventral ou. 1-2 1 12 Ventral in. 4-6 5 15 Tibia—Ventral ou. 0-2 0 17 Ventral in. 0-2 0 10 ARIADNA IN THE AMERICAS + Beatty 493 TABLE 10. SpPINATION OF ARIADNA GRACILIS Male N = 2 Female N 23-26 Range a ENioiel TE a Range Mode n Palp Patella 0 0) 2 0-2 0) 19 Tibia 1 1 9) 6-10 i 9 Tarsus 0 0 2 TM 9 § Leg: I—Meta.—Ventral ou. 1 2 7-10 8 1] Ventral in. 9} 2 2) 7-10 g 1] Prolateral 0-1 0,1 lea. 0 0) 23 Tibia—Ventral ou. 6 6 2 6-7 6 21 Ventral in. 2 2 2 6-7 6 19 Prolateral 3 3 2 = 8) 3 19 Retrolateral 4 4 9) 9-3 2 22, Femur—Dorsal ou. 1 1 2 1 1 23 Dorsal mi. 4 4 2 1-2 2 13 Dorsal in. 9 2) 9 1-2 2 99. Prolateral 0 0 2 DB 2 17 2—Meta.—Ventral ou. 3 3 9» 7-9 8 16 Ventral in. 3 3 2 ql 8,9 10 ea. Retrolateral I 1 2) 0 0 24 Tibia—Ventral ou. 4 4 2} 5-8 6 11 Ventral in. O} 2 2 7/ 5 16 Prolateral 3 3 2 9-3 3 19 Retrolateral 4 4 2 2 2 24 Femur—Dorsal ou. 1 1 o) 1 1 24 Dorsal mi. 3-4 ont lea. 1-2 1 13 Dorsal in. 2 2 2 Ie® 2 23 3—Meta.—Ventral ou. 3 3 2 3-4 3 24 Ventral in. 1 1 2 1 1 26 Prolateral 3 3 2 2-3 3 25 Tibia—Ventral ou. 2-3 8) lea. 1-2 2 22 Prolateral 2 2 9 0-1 0 19 Retrolateral 1-2 ile 1 ea. 0-1 0 25 Femur—Dorsal ou. =) 12, lea. 2 1 22, Dorsal mi. 3-4 oa lea. 0-1 0 23 Dorsal in. 2 2 2 ]-2 1 23 4—Meta.—Ventral ou. 0 0) 2 0 0 26 Ventral in. 4 4 2 3-4 4 21 Femur—Dorsal ou. 0 0) 2 0 0 26 Dorsal mi. 8 8 2 0 0 26 Dorsal in. il il » 0 0 26 494 Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 TABLE 11. SprnATION OF ARIADNA ISTHMICA Dorsal in. Male N = 5-6 Female N = 11-14 Range Mode n Range Mode Palp Patella 0 0 0 0-1 a 13 Tibia 0 0) 0 4-7 6,7 5ea Tarsus 0 0 0 7-11 8 6 Leg: 1—Meta.— Ventral ou. 9} 2 5 6-8 vi 8 Ventral in. = 3 4 6-8 0 10 Prolateral 0-1 0 3 0 0 12 Tibia—Ventral ou. 4 4 5 4-5 4 Il Ventral in. T=? 1 3} 3-4 4 11 Ventral su. 0-1 1 3 0 0 12 Prolateral 3-4 3 4 3 3 12 Retrolateral Bay/ Sao 2 ea. 3 3 2, Dorsal 0-1 0 3 0 0 12 Femur—Dorsal ou. 1 il 5 1 1 12 Dorsal mi. =o 9) 3 il 1 2, Dorsal in. 2 2 5 1 1 12 Prolateral 0 0 5 1 1 12 2—Meta.—Ventral ou. 3 3 5 6-8 7 ) Ventral in. 3 3 5 6-9 7,8 5 ea. Prolateral 0-1 0 3 0 0) 112, Retrolateral 0-1 1 3 0 0 12 Tibia—Ventral ou. 4-5 4 4 4-5 4 9 Ventral in. 2, 2 3 4-5 4 11 Ventral su. 1 l 5 0 0) 12 Prolateral 3 3 5 2-3 3 ll Retrolateral 3-4 4 4 De 3 8 Dorsal 0-1 0 4 0 0 12 Femur—Dorsal ou. 1 1 5 il 1 12 Dorsal mi. 3 3 3 0-1 l Il Dorsal in. =o 2 3 |-2 1 7 3—Meta.—Ventral ou. 3 3 5 3 3 ll Ventral in. 2 2 5 ]-2 2 9 Prolateral D8 3 4 2-3 3 10 Retrolateral 0-2 0 3 0 0 ll Tibia—Ventral ou. 3 3 5 9-4 3 9 Ventral in. 0-1 0 4 0 0 ll Prolateral 2) 2 4 0-2 2 8 Retrolateral 3} 33 4 0-1 0 9 Femur—Dorsal ou. 0-2 il 3 0-1 il 10 Dorsal mi. =A 3 3 0 0 13 Dorsal in. ]-2 il 3 0-2 l 10 4—Meta.—Ventral ou. 0-1 il 4 1 1 14 Ventral in. 4 4 5 3-4 4 13 Femur—Dorsal ou. (0) 0 5 0 0 14 Dorsal mi. 13} iL 3 0 0 14 0 5 0) 0 14 => | _— ARIADNA IN THE AMERICAS + Beatty 495 TABLE 12. SPINATION OF ARIADNA MAXIMA Male N = 27-30 Female N 200 Range Mode n Range Mode n Palp Patella 0 0 30 0-3 | 173 Tibia 0 0) 30 1-6 4 138 Tarsus 0) 0) 30 4-9 6 11] Leg: 1—Meta.—Ventral ou. Dl 3 15 6-16 10 53 Ventral in. 2-8 5 12 8-16 i 55 Prolateral 0-2 I 13 0 0 200 Retrolateral 0-9 3 ll 0 0 2.00 Dorsal 0-5 0 22, 0 0 200 Tibia—Ventral ou. 3-5 4 23 3-6 4 184 Ventral in. 3-4 4 26 3=5 4 185 Ventral su. 0-3 il 19 0 0 2.00 Prolateral Bil 4,5 Sea 0-3 3 95 Retrolateral 9-16 11 8 0-4 3 100 Dorsal 0-5 0 20 0 0 200 Femur—Dorsal ou. 13} 1 18 0-2 1 182 Dorsal mi. 0-4 iL 16 0-2 i 170 Dorsal in. 1-3 9} 21 1-3 2 179 Prolateral 0-1 1 19 0-1 1 197 2—Meta.—Ventral ou. 5-5 3 18 6-15 9 49 Ventral in. G7 5 10 7-15 11 48 Prolateral 0-2 1 23 0 (0) 200 Retrolateral 1-10 5 i 0 0 2.00 Dorsal 0-2 0 SAL 0) 0 200 Tibia—Ventral ou. 4-6 4 23 4-7 4 189 Ventral in. 4 4 Oi 3-4 4 195 Prolateral 2-6 3 12 0-3 2 85 Retrolateral 8-14 11 7 0-3 1 90 Dorsal 0-3 0 29, 0 0 200 Femur—Dorsal ou. 8} 2 7 0-2 I 7s} Dorsal mi. 1-7 3,4 Sea 0-2 1 158 Dorsal in. =p) 2) 25 1-3 il 186 3—Meta.—Ventral ou. 3-4 3 24 9-4 3 194 Ventral in. 1 2 25 =o: 2 148 Prolateral 3-4 3 26 |-4 3 184 Retrolateral 1-4 3 12 0 0 200 Tibia—Ventral ou. 3-6 3 20 j-4 2) 189 Ventral in. 1-3 1 24 0 0 2.00 Prolateral 9-4 2, 23 0-3 9) 149 Retrolateral 3-10 5 8 0-2 0) 134 Femur—Dorsal ou. 1-3 @ 13 0-1 0 189 Dorsal mi. 3-6 5 13 0-1 0 197 Dorsal in. 1-2 2 26 0-2 1 120 4—Meta.—Ventral ou. 1-3 9 23 0-3 1 174 Ventral in. 3-4 4 28 4-5 4 199 Ventral su. 0-1 0 25 0 0 200 Prolateral 1-6 5 10 0 0 200 Retrolateral 1-3 2 24 0 0 200 Dorsal 0-2 0 26 0 0 2.00 Tibia—Ventral ou. 0-3 1 17 0 0 200 Ventral in. 0-2 il 26 0 0 200 Prolateral 0-1 0 26 0 0 200 Retrolateral 1-3 1 14 0 0 200 Femur—Dorsal ou. 0-1 0 28 0 0 200 Dorsal mi. 0 0 3 0-2 0 187 1 19 0) 0 2.00 Dorsal in. i ; 496 Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 TABLE 13. SPINATION OF ARIADNA MOLLIS Male N = 2 Female N = 16-18 Range Mode n Range Mode n Palp Patella 0 0 2 0-1 1 13 Tibia 0 0 2 3-6 5 12 Tarsus 0 0 9 6-9 6 6 Leg: 1—Meta.—Ventral ou. 2 2 2 7-11 8 5 Ventral in. 2-3 253 1 ea. 8-11 8 7 Tibia—Ventral ou. 4-5 4,5 lea. 3-4 A 15 Ventral in. 0-2 0, 2 lea. 2-4 4 15 Ventral su. 1 ] 9} 0-1 0 13 Prolateral 2-4 2,4 lea. 2-3 2 11 Retrolateral 4 4 9 0-2 0 7 Dorsal 3 3 2 0 0 17 Pat’l—Prolateral 1 1 2 0 0 7 Femur—Dorsal ou. ] 1 o) 0-1 1 16 Dorsal mi. 1 1 2 1 1 17 Dorsal in. 1-2 ty ® lea. 2 2 17 Prolateral 0 0 2 1 l 17 2—Meta.—Ventral ou. 3 3 2 7-11 8 5 Ventral in. 3 3 2 8-11 9 a Retrolateral 1-2 ile? lea. 0 0 Ie Tibia—Ventral ou. 4-5 4,5 lea. 4 4 18 Ventral in. 3 3 2) 2-3 3 17 Ventral su. 1 1 2 0 0 18 Prolateral 3-4 3,4 1 ea. 1-3 3 i) Retrolateral 4-5 4,5 lea. 0 0 18 Dorsal 1 1 9 0 0 18 Pat 1—Prolateral 0-1 0, 1 lea. 0 0 18 Femur—Dorsal ou. 1 1 2 0-1 1 12 Dorsal mi. 3 3 2 |-2 1 17 Dorsal in. 1 il 2 ]-2 2 15 3—Meta.—Ventral ou. 3 3 2 3 3 17 Ventral in. 1 1 2 ]-2 2 9 Prolateral 9) 9} 2 9-4 3 12 Retrolateral 2 2 2 0 0 17 Tibia—Ventral ou. 3 3 2 1-3 3 10 Prolateral 1 1 2 0-3 2 10 Retrolateral 0 0 ® 0-1 0 16 Femur—Dorsal ou. 0 0 D, 0-1 0) 15 Dorsal mi. 4=5 4,5 lea. 0-2 1 8 Dorsel in. 1 1 2 0-1 1 16 4—Meta.—Ventral ou. 1 1 OH 1 1 17 Ventral in. vil 7 2 5-8 6 13 Femur—Dorsal ou. 0 0 2 0 0 iY Dorsal mi. 5 5 2 0-2 0 8 0-1 0, 1 1 ea. 0 0 hy Dorsal in. ARIADNA IN THE AMERICAS * TABLE 14. SPINATION OF ARIADNA MULTISPINOSA TaBLe 15. SpminaATION oF Female N = 2 7 Range a Palp Patella 0 2) Palp Patella Tibia 3 2 Tibia Tarsus 4 9} Tarsus Leg: Leg: 1—Meta.—Ventral ou. 9-10 lea. 1—Meta.— Ventral ou. Ventral in. 10 2 Ventral in. Tibia—Ventral ou. Il 2 Tibia—Ventral ou. Ventral in. 8-10 lea. Ventral in. Femur—Dorsal ou. 1 y) Prolateral Dorsal mi. 0-1 lea, Retrolateral Dorsal in. 0 2 Femur—Dorsal ou. Prolateral 923 1 ea. Dorsal mi. 2—Meta.— Ventral ou. 7-8 lea. Roe i Ventral in. 8-10 lea. rolatera Tibia—Ventral ou. 10-11 lea. 2—Meta.— Ventral ou. Ventral in. 8 9 Ventral in. Femur—Dorsal ou. 0-1 lea. Tibia—Ventral ou. Dorsal mi. ] Dy} Ventral in. Dorsal in. 0-1 2 Prolateral Retrolateral 3—Meta.—Ventral ou. 3 2 F i = a ore Nenteallec 9 2 emur— oe ou. Prolateral i 2} ee na Tibia—Ventral ou. 3 2 an : ; ‘ 3—Meta.— Ventral ou. 4—Meta.— Ventral ou. 0 - Vena Ventral in. 4 2 Sa Mrvawall Tibia—Ventral ou. Prolateral Retrolateral Femur—Dorsal ou. Dorsal mi. Dorsal in. 4—-Meta.—Ventral ou. Ventral in. Femur—Dorsal mi. Beatty ARIADNA Female N 12-14 Range Mode 0) 0) 0-3 2 3-8 4 My 6-9 8 4 4 4 4 1-5 3 0-5 4 0-1 Il 0-3 1 0-2 0 1 1 4-9 6,8 6-10 7 4 4 3-4 4 1-4 2; 0-3 0 0-1 1 1-4 3 1 1 3 3 1-2 1 9=3 2 2-4 3 0-2 0 0-1 0 0 0 0-3 0 1-2 1 1-2 1 2-3 2 13} 2 497 VMURPHYI ea. 498 Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 TABLE 16. SPINATION OF ARIADNA OBSCURA Female N = 2 Prolateral Retrolateral Tibia—Ventral ou. Ventral in. Prolateral Retrolateral Pat’ l—Prolateral Femur—Dorsal ou. Dorsal mi. Range n Palp Patella 1 2 Tibia 5 2 Tarsus 8-9 lea. Leg: 1—Meta.— Ventral ou. 6-7 lea. Ventral in. 5-6 Lea. Prolateral 9) 2, Retrolateral 2) 2 Tibia—Ventral ou. 4 2 Ventral in. 4 2 Prolateral 4 2 Retrolateral 4 ® Pat l—Prolateral 2 2, Retrolateral 1 9} Femur—Dorsal ou. 1 2 Dorsal mi. il ® Dorsal in. ® 2 Prolateral il 9) 2—Meta.—Ventral ou. 5-6 lea. Ventral in. vit lea. 9, 2, 9 2; 2 2, 2 2 2, 9, Dorsal in. Ventral ou. Ventral in. Prolateral Retrolateral 3—Meta. Ventral ou. Prolateral Retrolateral Tibia Femur—Dorsal ou. Dorsal mi. Dorsal in. Ventral ou. Ventral in. 4— Meta. lop) SEW NWrwW NrRrFbwDP Wr K ND ! owl Nore So re OO WNNMrFRrRNHNWNNNWW wo wo ea. ea. ARIADNA IN THE AMERICAS + Beatty 499 TABLE 17. SPINATION OF ARIADNA PILIFERA "nN 20-22 Female N 71-50 Range Mode 7 n : Range Mode n Palp Patella 0 0 22 0-1] () 72 Tibia 0 0 22, 2-6 4 1] Tarsus 0 0 22, 5-14 9 25 Leg: 1—Meta.—Ventral ou. 9} 2 16 6-10 8 35 Ventral in. Bet) 5 10 6-10 8 36 Prolateral 0-1 il 13 0 0) rial Tibia—Ventral ou. 4-6 5 12 Be 4 4] Ventral in. 0-2 1 8 B7/ 4 46 Ventral su. 0-1 iL 18 0 0 (i Prolateral 3-5 4 16 0-4 2 Dil Retrolateral B55 4 17 0-4 1 34 Femur—Dorsal ou. =} l 15 0-2 1 63 Dorsal mi. 1-4 ip 8ea 0-3 1 52 Dorsal in. 2 2 20 0-2 iy) 66 Prolateral 0-1 IP 10 il 1 72 2—Meta.—Ventral ou. 3-4 3 19 ali 8 34 Ventral in. 3-4 3 20 Teal 9 27 Prolateral 0-1 0 15 0 0 7 Retrolateral 0-2 1 12 0 (0) Tl Tibia—Ventral ou. AL eri 5) 10 We 4 48 Ventral in. 0-2 2 12 1-3 2 40 Ventral su. 0-1 iL 20 0 0 72 Prolateral ats) 4 10 0-5 3 38 Retrolateral Gad 4. 17 0-1 0 69 Femur—Dorsal ou. 1p) 1 12 =o 1 29 Dorsal mi. 1-4 1 8 0-3 1 55 Dorsal in. 1-3 2 17 1-3 2 63 3—Meta.—Ventral ou. 3 3 21 3-4 3 75 Ventral in. 0-2 2, 18 |-2 py) 71 Prolateral 2-4 3 16 1-5 3 58 Tibia—Ventral ou. oo 3 13 1-5 3 44 Prolateral 9-4 3 12 0-4 2 34 Femur—Dorsal ou. 0-1 0 19 0-1 0 69 Dorsal mi. 0-4 3 8 0-2 0 32 Dorsal in. 123 ® 17/ 1-3 2 66 4—Meta.—Ventral ou. 0-2 2, 18 1-3 ® 63 Ventral in. 2-4 4 2} 3-5 3 49 Tibia—Ventral ou. 0-1 1 18 0-1 il 52 Retrolateral 0 0 22, 0-3 0 58 Femur—Dorsal ou. 0 0 2) 0 0 rill Dorsal mi. 2-6 4 A 0-2 0 39 Dorsal in. 0-3 1 10 0-2, 0) 55 Retrolateral 0 0 22, 0-4 0 32 500 Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 TABLE 18. SPINATION OF ARIADNA PRAGMATICA TABLE 19, SpINATION OF ARIADNA SOLITARIA Female N = 4-6 Juvenile N = 1-2 Range Mode n Range n Palp Patella 0 0 6 Palp Patella 1 2 Tibia 2-4 3 3 Tibia 3-5 lea. Tarsus 7-9 9 3 Tarsus 6 2 Leg: Leg: 1—Meta.—Ventral ou. 7-8 The 3 ea 1—Meta.—Ventral ou. rit 1 Ventral in. JAl®O. 2 ea. Ventral in. ri I Tibia—Ventral ou. 4 4 6 Tibia—Ventral ou. 4 il Ventral in. 4-5 4 4 Ventral in. 4 1 Femur—Dorsal ou. 0 0 6 Ventral su. 1 J Dorsal mi. 0-1 0 4 Prolateral 3 1 Dorsalin. ~ 0-1 0,1 3ea Retrolateral 3 1 Prolateral 1 1 6 Femur—Dorsal ou. l il 2—Meta.—Ventral ou. 7-9 7,8 2 ea. ae ou 5 : Ventral in. 8-10 8 3 Been I 1 Tibia—Ventral ou. 4 4 6 ; Ventral in. 129) ile veal 2—Meta.—Ventrol ou. 6 2 Prolateral 0-1 0, 3 ea. Ventral in. u 2 Femur—Dorsal in. if 1 6 et noe ou. ; 5 entral in. 3—Meta.— Ventral ou. 2-3 3 5 Pralateral 9-3 een Ventr al in. 1-2 1 5 Retrolateral l 9 Prolateral i 2 5 af Femur—Dorsal ou. 1 2 Tibia—Ventral ou. |-4 3 3 . ee Dorsal mi. l 2 p Pro on 0-1 0 4 Dorsal in. I-® lea. —D in. il 1 6 cas Ce We 3—Meta.—Ventral ou. 3 2 4— Meta.— Ventral ou. 1-3 I 4 Ventral in. l 99, Ventral in. 2 2, 6 Prolateral 2 9 Femur—Dorsal ou. 0 0 6 Sania eeNentraleour 9 2 Dorsal mi. 0-1 0 4 ; DYyaeel a 0 0 6 Femur—Dorsal ou. 0-1 lea. E ; Dorsal mi. 0 2 Dorsal in. 9) 2, 4—_Meta.— Ventral ou. 9) 2 Ventral in. 4 2 TABLE 20. SPINATION OF ARIADNA TARSALIS ARIADNA IN THE AMERICAS © TABLE 21. Juvenile N = 6 Female N = 4 Range Mode — KBE NWA AB CLO KF NHFE RWKH KB AO Beatty 50) NWR WWKRKER WD PWKEK WHE KNW Be DONHNY ee SPINATION OF ARIADNA TOVARENSIS ea. ea. ea. Range Mode n Palp Patella 0 0 6 Palp Patella Tibia 3-6 3 4 Tibia Tarsus 6-9 6 3 Tarsus Leg: Leg: I1—Meta.—Ventral ou. 8-9 9 4 1—Meta.— Ventral ou. Ventral in. 8-9 8 5 Ventral in. Tibia—Ventral ou. 5-9 5,9 2 ea. Tibia—Ventral ou. Ventral in. 4-7 Wf 2 Ventral in. Prolateral 0-3 2 3 Prolateral Retrolateral 0-3 3 4 Retrolateral Femur—Dorsal ou. 0-1 il 4 Femur—Dorsal ou. Dorsal mi. 0-2 il 4 Dorsal mi. Dorsal in. 0-2 1 4 Dorsal in. Prolateral il il 6 Prolateral 2—Meta.—Ventral ou. 8-9 8 5 2—Meta.—Ventral ou. Ventral in. 7-10 9 3 Ventral in. Tibia—Ventral ou. 6-11 7 2 Tibia—Ventral ou. Ventral in. 4-6 4 3 Ventral in. Prolateral 0-2 0 3 Prolateral Femur—Dorsal ou. 0-1 0,1 3 ea. Retrolateral Dorsal mi. 12, [eo 3 ea. Femur—Dorsal ou. Dorsal in. 1-2 2 3 ea. Dorsal mi. 3—Meta.—Ventral ou. 3 3 6 Dorsal in. Ventral in. 2) 1 4 3—Meta.—Ventral ou. Prolaterai D8) 3 4 Ventral in. Tibia—Ventral ou. 3 3 6 Prolateral Prolateral 0-1 0 5 Tibia—Ventral ou. Femur—Dorsal ou. 0 0 6 Prolateral Dorsal mi. 0-2 0 4 Femur—Dorsal in. Dorsal in. I : : 4—Meta.— Ventral ou. 4—Meta.—Ventral ou. 1 1 6 Ventral in. Ventral in. 9-3 2 5 502 Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 TABLE 22. SPINATION OF ARIADNA TUBICOLA Juvenile N = 4 Range Mode Palp Patella Tibia Tarsus Leg: 1—Meta. Ventral ou. Ventral in. Ventral ou. Ventral in. Prolateral Retrolateral Tibia Femur—Dorsal ou. Dorsal mi. Dorsal in. Prolateral Ventral ou. Ventral in. 2—Meta. Ventral ou. Ventral in. Prolateral Tibia Femur—Dorsal ou. Dorsal mi. Dorsal in. Ventral ou. Ventral in. Prolateral 3—Meta. Ventral ou. Prolateral Tibia Femur—Dorsal mi. 4—Meta.—Ventral ou. Ventral in. 1 7 9 Ol BD FOoW OFONRA DH KF ORrFOHEDN &-” v or eb ro © fon) WON eB! s Or WNONnN We WK KNW WY BKWKR WWND BW 23. SPINATION OF ARIADNA WEAVERI Malo Ni aad Range Noten Palp Patella 0 0) Tibia 0 0 Tarsus 0) (0) Leg: 1—Meta.—Ventral ou. 0 0 Ventral in. 0 0 Tibia—Ventral ou. 4-5 5 Ventral in. 2} 2 Ventral su. 0-1 0) Prolateral 1,3,4 Retrolateral 3-4 3 Femur—Dorsal ou. 0 0 Dorsal mi. 0-1 il Dorsal in. 2 O} Prolateral 1 il 2—Meta.—Ventral ou. 0 0 Ventral in. 0 0 Tibia—Ventral ou. 4 4 Ventral in. 1=2 il Prolateral 1-4 il Retrolateral 0-1 0,1 Femur—Dorsal ou. 0 0 Dorsal mi. 0-1 0,1 Dorsal in. 1-2 1 3—Meta.—Ventral ou. 3 3 Ventral in. 1 1 Prolateral 0) 0 Tibia—Ventral ou. 12 9) Prolateral 0-1 0 Femur—Dorsal ou. 0 0) Dorsal mi. 0 0 Dorsal in. 0-1 0,1 4—Meta.—Ventral ou. 19) il Ventral in. 3 3 Ventral su. 0-1 0, 1 Femur—Dorsal ou. 0 0 Dorsal mi. 8 i} Dorsal in. 0 0 ARIADNA IN THE AMERICAS * NDNreNMNWN WO NNWE PBR WWNYW G2 bo Wii oG Bw bo ee He G2 He ea. ea, ea. ea. Female N Range Mode 0 0 3-9 3 6-15 12 7-10 8,9 —9 8 4-5 4 to) 4 0 0 0 0 0 0 0 0 0-1 0 O-1 0 1 iL 7-11 7,9 9-16 10 4-6 4 1-4 2 0-1 0 0 0 0 0 0 0 1-2 1 3-4 3} 1-2 2, 0-3 2 13 2. 0-2 1, 0) 0 0 0) 1-2 l 1-2 1 3 3 0 0 0 0 0) 0) ) 0 Beatly 13-14 503 ea. Ca. 504 Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 TABLE 24, DraAcnostic FEATURES OF HAPLOGYNE SPIDER FAMILIES Plectreuridae Diguetidae Sicariidae Scytodidae Chelicerae Colulus Male palp Female genitalia Heart ostia Anterior respira- tory organs Posterior respira- tory organs Joined basally; laminate, chelate. Minute plate bear- ing 2 setae. Short tarsus, large bulb; embolus slender, simple, or flat, two-parted; no conductor Joined basally; laminate, chelate. Tiny, conical, with 2 lateral setae. Short tarsus, large bulb; slender simple embolus; large scoop-like conductor. With bursa copula- With bursa copu- trix; no sclerotized latrix; a single seminal receptacles, median seminal Lungs Median tracheal spiracle behind middle of abdo- men; tracheae probably rudi- mentary. receptacle. Lungs Median tracheal spiracle behind middle of abdo- men; tracheae simple, restricted to abdomen. Joined basally; laminate, chelate. Conspicuous, con- ical, with about 12 setae. Tarsus and bulb small; embolus conical basally, slender and sim- ple distally; no conductor. iP 3 pairs Lungs Median tracheal spiracle behind middle of abdo- men; tracheae lost. Joined basally; laminate, chelate. Conspicuous, con- ical or rounded, with 11-20 setae. Tarsus variable, bulb large or small; embolus con- ical basally, slender and simple distally; no conductor. No bursa copula- trix; a pair of seminal receptacles, these sometimes united. 3 pairs Lungs Median tracheal spiracle behind middle of abdo- men; tracheae simple, restricted to abdomen. TABLE 25. Dracnostic FEATURES OF HAPLOGYNE ARIADNA IN THE AMERICAS + Beatty 505 SPER FAminres ( Continued ) Dysderinae Segestriinae Oonopidae Chelicerae Colulus Male palp Female genitalia Heart ostia Anterior respira- tory organs Posterior respira- tory organs Free, subchelate, not laminate. Absent, or a tiny plate with 2—4 setae. Variable, similar to segestriines or partly sub- divided apically into several projections With copulatory bursa; T-shaped median seminal receptacle. 2 pairs Lungs Pair of tracheal spiracles just behind epigastric groove; tracheae well-developed, entering cephalo- thorax. Free, subchelate, not laminate. Large, rounded, with several setae. Tarsus short or long, bulb large; embolus conical basally, slender and simple distally; no conductor. With copulatory bursa; tiny median seminal receptacle. 2. pairs Lungs Pair of tracheal spiracles just behind epigastric groove; tracheae well-developed, entering cephalo- thorax. Free, subchelate, not laminate. Absent, or a tiny plate with 2 setae. Tarsus small; bulb large and glob- ular with variable embolus, or no separate bulb; no conductor. 2 pairs Greatly reduced lungs, or tracheae only. Pair of tracheal spiracles just behind epigastric groove; tracheae well-developed, entering cephalo- thorax. Caponiidae Free, subchelate, not laminate. Absent. Tarsus short, bulb globular, embolus short to very long, curved, bifurcate at tip; no conductor. 2 pairs Tracheae. Pair of tracheal spiracles just behind epigastric groove; tracheae well-developed, entering cephalo- thorax. 506 Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 Plate |. Fig. 1. Ariadna mollis (Holmberg). Dorsal view of female from Cavinna, Parana, Brazil. Fig. 2. Ariadna maxima (Nicolet). Sternum, endites, labium, and labrum of female from Mas Afuera Island, Juan Fernandez Islands, Chile. Fig. 3. Ariadna boesenbergii Keyserling. Ventral view of tibia Il of female lectoparatype from Montevideo, Uruguay. Figs. 4, 6. Ariadna boliviana Simon. 4. Lateral view of carapace of male lectotype from Espiritu Santo, Bolivia. 6. Lateral view of carapace of female lectoparatype from Espiritu Santo, Bolivia. Fig. 5. Dorsal view of eye region of female Ariadna sp., showing lines along which measurements were made. Fig. 7. Ariadna gracilis Vellard. Mesal view of femur | of female from St. André, Marajao, Brazil. ARIADNA IN THE AMERICAS + Beatty 507 3 a 4 7 508 Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 Plate Il. Figs. 8-9. Ariadna arthuri Petrunkevitch. Female from South Bimini, Bahama Islands, showing lines along which measurements were made. 8. Dorsal view of carapace. 9. Ventral surface of cephalothorax. Fig. 10. Ariadna fidicina (Chamberlin). Dorsal view of carapace of female from Laguna Beach, California, showing stridulating grooves. Fig. 11. Ariadna cephalotes Simon. Ventral view of abdomen of female lectotype from San Mateo, Bolivia. Fig. 12. Left palp of male Ariadna, showing regions of palpal organ. A, bulb; B, midpiece; C, embolic portion. Fig. 13. Ventro-lateral view of metatarsus | of female Ariadna, showing one of the rows of ventral spines. Fig. 14. Ariadna boliviana Simon. Ventral view of genital region of female lectoparatype from Espiritu Santo, Bolivia. Overlying tissue removed to expose seminal receptacle. Fig. 15. Lateral view of leg | of Ariadna sp., showing lines along which measurements were made. Hl (( ( (| PNA Sophos 3 Se 510 Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 Plate Ill. Figs. 16-17, 22. Ariadna mollis (Holmberg). Male from Tigre, Buenos Aires, Argentina. 16. Left palp, mesal view. 17. Left palp, lateral view. 22. Tibia and metatarsus |, dorsal view. Figs. 18-19, 21. Ariadna maxima (Nicolet). Male from Mas Afuera Island, Juan Fernandez Islands, Chile 18. Left palp, lateral view. 19. Left palp, mesal view. 21. Tibia and metatarsus |, dorsal view. Fig. 20. Ariadna pilifera O. P. Cambridge. Tibia and metatarsus | of male, dorsal view. (Holotype of Ariadna acanthopus Simon from Guanajuato, Mexico.) Figs. 23-25. Ariadna isthmica sp. n. Male holo- type from Barro Colorado Island, Canal Zone, Panama. 23. Tibia and metatarsus |, dorsal view. 24. Left palp, mesal view. 25. Left palp, lateral view. ARIADNA IN THE AMERICAS + Beatty 51] By) Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 Plate IV. Figs. 26-27, 32. Ariadna arthuri Petrunkevitch. Male from South Bimini, Bahama Islands. 26. Left palp, mesal view. 27. Left palp, lateral view. 32. Tibia, metatarsus, and tarsus |, dorsal view. Figs. 28-29, 33. Ariadna gracilis Vel- lard. Male from Téfé, Amazonas, Brazil. 28. Left palp, lateral view. 29. Left palp, mesal view. 33. Left tibia and meta- tarsus |, dorsal view. Fig. 30. Ariadna isthmica sp. n. Male from Barro Colorado Island, Canal Zone, Panama. Left meta- tarsus and tarsus IV showing scopulae. Fig. 31. Ariadna pilifera O. P.-Cambridge. Female from Southwestern Research Station, 5 mi W of Portal, Cochise Co., Arizona. Ventral view of metatarsus IV showing comb and outer row of ventral spines. 514 Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 Plate V. Figs. 34-35, 39. Ariadna peruviana sp. n. Male from Lima, Lima, Peru. 34. Left palp, mesal view. 35. Left palp, lateral view. 39. Left tibia and metatarsus |, dorsal view. Figs. 36-37, 40-41. Ariadna pilifera O. P. Cambridge. Male from Southwestern Research Station, 5 mi W of Portal, Cochise Co., Arizona. 36. Left palp, lateral view. 37. Left palp, mesal view. 40. Left tibia |, lateral view. 41. Left tibia, metatarsus, and tarsus |, dorsal view. Figs. 38, 42-43. Ariadna bicolor (Hentz). Male from Mohican State Park, Ashland Co., Ohio. 38. Left tibia and metatarsus |, dorsal view. 42. Left palp, mesal view. 43. Left palp, lateral view. 40 Git 42 516 ~=— Bulletin Museum of Comparative Zoology, Vol. 139, No. 8 Plate VI. Figs. 44~— 45, 49. Ariadna caerulea Keyserling. Male from Sierra Nevada de Santa Marta, Magdalena, Colombia. 44. Left palp, mesal view. 45. Left palp, lateral view. 49. Left tibia and metatarsus |, dorsal view. Figs. 46-48. Ariadna boesenbergii Keyserling. Male lectotype from Montevideo, Uruguay. 46. Left palp, lateral view. 47. Left palp, mesal view. 48. Left tibia and metatarsus |, dorsal view. Figs. 50, 53, 56. Ariadna weaveri sp. n. Male from Clarion Island, Revilla Gigedo Islands group, Mexico. 50. Left tibia and metatarsus |, dorsal view. 53. Left palp, mesal view. 56. Left palp, lateral view. Figs. 51-52, 54-55. Ariadna boliviana Simon. Male lectotype from Espiritu Santo, Bolivia. 51. Left tibia, metatarsus, and tarsus |, dorsal view. 52. Right metatarsus and tarsus IV, showing comb. 54. Left palp, anterior view. 55. Left palp, mesal view. _ a hel aS An 10) Bk iy igen ae a eT Bt ea i me Hd Oki ns Mh ‘\o Cor iu BARUAVIND NM ih ; HON ‘ ve } DATE DUE ee a — — | H — DEMCO, INC. 38-2931 lary MONT 3 2044 : - aah nh BOO ent tre now enn nsammeas ae . pian amenencnn in ee euamie at he comes —_* Tarere cert = piprenhintne mw tant Te Ne aipapesed: dameiea oD “ : Siacpapreree : . nag ee eAP ATCT i ta