HARVARD UNIVERSITY Library of the Museum of Comparative Zoology US ISSN 0047-4100 etin OF THE seum o oology The Carpolestidae Early Tertiary Primates from North America KENNETH D. ROSE HARVARD UNIVERSITY CAMBRIDGE, MASSACHUSETTS, U.S.A. VOLUME 147, NUMBER 1 21 FEBRUARY 1975 PUBLICATIONS ISSUED OR DISTRIBUTED BY THE MUSEUM OF COMPARATIVE ZOOLOGY HARVARD UNIVERSITY Breviora 1952- Bulletin 1863- Memoirs 1864-1938 Johnsonia, Department of Mollusks, 1941- Occasional Papers on Mollusks, 1945- SPECIAL PUBLICATIONS. 1. Whittington, H. B., and E. D. I. Rolfe (eds.), 1963. Phylogeny and Evolution of Crustacea. 192 pp. 2. Turner, R. D., 1966. A Survey and Illustrated Catalogue of the Teredini- dae (Mollusca: Bivalvia). 265 pp. 3. Sprinkle, J., 1973. Morphology and Evolution of Blastozoan Echinoderms. 284 pp. 4. Eaton, R. J. E, 1974. A Flora of Concord. 236 pp. Other Publications. Bigelow, H. B., and W. C. Schroeder, 1953. Fishes of the Gulf of Maine. Reprint. Braes, C. T., A. L. Melander, and F. M. Carpenter, 1954. Classification of Insects. Creighton, W. S., 1950. The Ants of North America. Reprint. Lyman, C. P., and A. R. Dawe (eds.), 1960. Symposium on Natural Mammalian Hibernation. Peters' Check-list of Birds of the World, vols. 2-7, 9, 10, 12-15. Proceedings of the New England Zoological Club 1899-1948. (Complete sets only.) Publications of the Boston Society of Natural History. Price list and catalog of MCZ publications may be obtained from Publications Office, Museum of Comparative Zoology, Harvard University, Cambridge, Massa- chusetts, 02138, U.S.A. © The President and Fellows of Harvard College 1975. THE CARPOLESTIDAE EARLY TERTIARY PRIMATES FROM NORTH AMERICA KENNETH D. ROSE1 Dedicated to the memory of Professor Glenn Lowell Jepsen 1903-1974 CONTENTS Introduction 2 History of Study 3 Stratigraphic and Geographic Occurrence _ 6 Systematic Revision 10 The Phyletic Position of Saxonella 50 Morphology and Function 51 The "Plagiaulacoid" Dentition 51 Major Features and Trends in die Carpolestid Dentition and Skull 52 Occlusion and Dental Function 55 Dietary Inferences 61 Origin, Affinities, and Interrelationships 62 Acknowledgments 67 Literature Cited 67 Appendix 70 Abstract. The Carpolestidae were archaic pri- mates of the superfamily Plesiadapoidea. They have been recovered from strata of Middle Paleo- cene (Torrejonian) to early Eocene ( Wasatchian ) age in western North America. Although known only from jaws, teeth, and a few cranial fragments, carpolestids have very characteristic dentitions by which they are easily recognized. Most diagnostic are the enlarged, serrate Pj, and the enlarged, poly- cuspate P3 and P*. 1 Museum of Comparative Zoology, Harvard Uni- versity, Cambridge, Massachusetts. The systematic revision includes emended diag- noses for all taxa. Three genera, Elphidotarsius, Carpodaptes, and Carpolestes, are recognized. There are nine valid species: two in Elphido- tarsius, five in Carpodaptes (including one new species), and two in Carpolestes. Saxonella, a Middle Paleocene primate from Europe originally described as a carpolestid, is an aberrant plesia- dapoid, here placed in a family separate from carpolestids and plesiadapids. New material described provides the first pub- lished information on the upper teeth of Elphido- tarsius, the lower incisors of Elphidotarsius and Carpolestes, the anterior teeth of Carpolestes, the posterior part of the mandible of Carpodaptes, and the snout of Carpolestes, as well as additional evidence bearing on dental formulae and intra- specific variability. Analysis of wear facets indicates that the molars of carpolestids resemble those of plesiadapids not only morphologically, but also functionally; they were used in the shearing (Phase I) and grinding ( Phase II ) stages of mastication. The specialized premolars were probably most important during the "puncture-crushing" stage, when the blade! ike Pi was used to tear or cut food. They were less important during Phase I and probably ineffective during Phase II. The three genera, Elphidotarsius, Carpodaptes, and Carpolestes, form a natural sequence both morphologically and stratigraphically, indicating that they are representatives of a single generic lineage. Each is rather restricted temporally: Elphidotarsius has been found only in Torrejonian beds, Carpodaptes is known chiefly from Tiffanian strata, and Carpolestes occurs in "Clark forkian" and earliest Wasatchian deposits. The nature and extent of similarities between Bull. Mus. Comp. Zool, 147(1): 1-74, February, 1975 2 Bulletin Museum of Comparative Zoology, Vol. 147, No. 1 carpolestids and plesiadapids indicates that the two families are closely related through a common ancestor. INTRODUCTION The radiation of primates in the Paleo- cene was one of the most successful episodes in the early evolution of euthe- rians. In North America, members of this initial deployment, particularly plesiada- pids, became common or even dominant forms in many local faunas. Other "archaic prosimians" included in this first major radiation of lower primates were the Paromomyidae (including Phenacolemur), the Carpolestidae, and, according to some authors, the Picrodontidae and the Micro- syopidae. (Recent attempts [e.g. Cartmill, 1972; Martin, 1968] to banish the "archaic prosimians" from the Primates or to trans- fer them to the Insectivora are unjustified when available evidence is considered in toto.) Renewed interest has focused on these early primates in recent years, resulting in a few broad reviews (e.g. Simons, 1963, 1972; 4 McKenna, 1966; Szalay, 1968a, 1972b). There have also appeared several more specific works that have contributed significantly to our understanding of Paleocene primates (e.g. Simpson, 1955; D. E. Russell, 1959, 1964; Szalay, 1968b, 1969a, 1969b, 1972a; Gazin, 1971; Butler, 1973; Gingerich, in press; and Bown and Ginge- rich, 1972, 1973). One outcome of the current crescendo of interest in early pri- mates has been the realization of their great diversity, and also of the presence of features in common which unite them more or less closely as members of the Primates. The relationships of these early forms to primates of Eocene or later time have been a subject of investigation. It has long been known that Paleocene Plesiadapis and Pheiuicolemur gave rise to Eocene species of the same genera. Recent evidence has strengthened the possibility that some other Paleocene primates may also have had Eocene descendants, e.g. Plesiolestes is a plausible ancestor for some Eocene micro- syopids ( Bown and Gingerich, 1972, 1973 ) . The interrelationships of the Paleocene primate families and genera are still not fully understood, but a relatively clear picture of affinities and evolutionary trends can be reconstructed in some cases. One family for which this is now possible, the Carpolestidae, forms the subject of this paper. This aberrant group comprises three genera that have been found in deposits of Middle Paleocene to earliest Eocene age in western North America. The three genera, Elphidotarsius, Carpodaptes, and Carpo- lestes, constitute a well-documented time- transgressive structural sequence. The Carpolestidae are known solely from dentitions and gnathic and cranial frag- ments; unfortunately, no postcranial ele- ments can be confidently referred to the family. Such evidence as has been available has suggested close ties between the Plesia- dapidae and the Carpolestidae, but a number of peculiar features clearly segre- gate the latter at the family level from all other early primates. The most conspicuous feature ( and also the most significant taxo- nomically) is the specialized lower fourth premolar. Through time this tooth became progressively enlarged, multicusped, and bladelike. Concomitant with this develop- ment was the reduction of the anterior dentition, except for the large, procumbent medial incisor, which remained prominent in size throughout the lineage. This par- ticular combination of features in the mandibular dentition has evolved inde- pendently in several unrelated mammalian groups and has been termed the "plagiau- lacoid dentition" (Abel, 1931; Simpson, 1933; also see below, p. 51). As hyper- trophy of P4 proceeded in carpolestids, there was corresponding enlargement of the upper third and fourth premolars (which occlude with P4). P3 and P4 lengthened anteroposteriorly and developed three longitudinal crests bearing cuspules, some- what resembling upper molars in multi- tuberculates. (In fact, the first discovered Carpolestidae • Rose upper premolar of a carpolestid was mis- taken for a tooth of a multituberculate. ) These features have been and still are re- garded as diagnostic of the family Carpo- lestidae. Since the discover)' of the first carpo- lestid specimen about half a century ago, only a small number of articles on carpo- lestids have appeared, and most have been descriptions of new taxa. By 1970, only about twenty specimens had been described or figured in the literature, but large num- bers of specimens had been recovered which remained unpublished. Persistent field work in Paleocene deposits has pro- duced sizable samples of these specialized early primates (the largest single species sample from one site is greater than 60 specimens). More than 300 specimens are now known, many of which reveal in- formation previously unknown. The most extensive collections have been recovered for Princeton University by field operations under the direction of Professor G. L. Jepsen; these investigations have been undertaken in the Bighorn Basin for more than forty years. Only three carpolestid specimens discovered by the earliest of these expeditions have been described pre- viously (Jepsen, 1930). The extensive Princeton collection, as well as numerous other new specimens, provide ample carpo- lestid material to serve as the basis for a systematic revision. In addition, better pre- served specimens and samples of larger size can now contribute significantly to our understanding of carpolestids and their role in the initial radiation of the Primates. ABBREVIATIONS AMNH CM MCZ PU — American Museum of Natural History, New York — Carnegie Museum, Pittsburgh — Museum of Comparative Zool- ogy, Harvard University, Cam- bridge, Massachusetts — Princeton University Museum, Princeton, New Jersey ROM — Royal Ontario Museum, To- ronto, Ontario UA — University of Alberta, Edmon- ton UKMNH — University of Kansas Museum of Natural History, Lawrence UMMP — University of Michigan Mu- seum of Paleontology, Ann Arbor UMVP — University of Minnesota, Min- neapolis USNM — United States National Mu- seum, Washington, D.C. UW — University of Wyoming, Lara- mie YPM — Peabody Museum of Natural History, Yale University, New Haven, Connecticut B — Breadth L — Length MD — Mandibular depth, measured on the buccal side beneath the anterior root of Mi. N — Number of observations (speci- mens) OR — Observed range s — Standard deviation V — Coefficient of variation X — Mean Dental terminology used in this study is that of Szalay ( 1969a ) . Dimensions of the cheek teeth and depth of the mandible have been measured in the manner illustrated in Figure 1, using an ocular micrometer fitted to a lens of a binocular microscope. HISTORY OF STUDY The first carpolestid described was Carpo- claptes aulacodon Matthew and Granger ( 1921 ) . The generic name means "fruit eater"; the specific name alludes to the vertical grooves on the lower fourth pre- molar, a condition similar to that of P4 in plagiaulacoid multituberculates and certain marsupials. The species was based solely on the type specimen and, in the absence of any comparable species, Matthew and Granger were duly reluctant to suggest its 4 Bulletin Museum of Comparative Zoology, Vol. 147, No. 1 B Figure 1. Schematic drawings to demonstrate method of taking measurements. All dimensions are in milli- meters. A) crown view of P4-Ma maximum mesiodistal length, as shown. Breadths (B) are maximum dimensions measured perpendicular to length. B) lateral view of mandible. MD = mandibular depth beneath anterior root of M,. C) crown view of PJ-M3. Carpolestidae • Rose affinities. They speculated only that "it may be a primate, a menotyphlan insecti- vore, or neither" (Matthew and Granger, 1921:6). J. W. Gidley (1923) proposed Elphido- tarsius florencae for a mandibular fragment preserving P4 through M3. Again the holo- type was the only known specimen. He noted resemblances of the molars to those of fossil tarsioids such as Tetonius, therefore placing ElpJiidotorsius provisionally in the Tarsiidae, the name then applied to many Paleocene and Eocene primates. He did not compare the specimen to Carpoclaptcs. Carpolestes ("fruit stealer") was named by Simpson (1928) to accommodate the new species Carpolestes nigridens from the Eagle Coal Mine at Bear Creek, Montana. Simpson regarded Carpolestes as a close relative of Carpodaptes, classifying both as aberrant members of the Tarsiidae. The following year he described a second spe- cies from Bear Creek, Carpolestes aquilae (Simpson, 1929). Jepsen ( 1930 ) named a third species of Carpolestes, C. dubius. Included in his hypodigm was a maxilla with four teeth, the first upper dentition of a carpolcstid to be identified. Jepsen recognized that the type and only known specimen of the sup- posed multituberculate Litotherhun compli- catum (Simpson, 1929), an isolated upper premolar from Bear Creek, was actually P3 of Carpolestes. ( This eliminated the Multi- tuberculata from the known fauna of Bear Creek.) Like Matthew and Granger, he deferred definite ordinal assignment for Carpolestes, stating, "It is possible to select suites of characters which, taken by them- selves, would place Carpolestes in any one of several orders," (Jepsen, 1930:523). The relationship of Elphidotarsius to Carpodaptes and Carpolestes was recog- nized by Simpson ( 1935b ) , who proposed the family Carpolestidae for the reception of the three genera. He also noted adaptive features shared by this family, the Plesi- adapidae, and the Apatemyidae. As diag- nostic characters of the Carpolestidae, Simpson listed the enlarged lower fourth premolar and the procumbent enlarged lower incisor. The dental formula remained controversial, however. Jepsen (1930) had proposed the mandibular dental formula 1.0.4.3, regarding the tooth immediately behind the enlarged incisor as Pa. Simpson ( 1935b: 10 ), although conceding that it was "impossible to determine whether the fol- lowing tooth is a canine or Pi", considered it "slightly more probable that it is the canine". Discussing the degree of relation- ship among the three genera, Simpson (1937b: 161) remarked that the morphologic sequence Elphidotarsius-Carpodaptes-Car- polestes "may be a direct phylogeny, al- though the possible age difference between the last two genera seems too small to per- mit such a marked structural advance in a direct descendant, and it is more likely that some collateral evolution is involved". In 1936, Simpson described Carpodaptes hazelae from the Scarritt Quarry in the Crazy Mountain Field of Montana. The next year, he described the most complete known upper dentition of a carpolcstid (referable also to C. hazelae), preserving three molars and four antemolar teeth. The most anterior tooth unfortunately has been lost since Simpson's description, and it is not preserved in any caq^olestid found sub- sequently. In the same paper, Simpson (1937a) reasserted his view that carpolestids represent early aberrant primates and again stressed the resemblances to Plesiadapis, concluding that these similarities could only indicate close affinity and common an- cestry. Dorr's field work in the Hoback Basin of western Wyoming yielded a Tiffanian fauna which included a previously un- known carpolcstid, Carpodaptes hobacken- sis (Dorr, 1952). Dorr suggested that this species might be more nearly intermediate between Elphidotarsius and Carpolestes than either other named species of Carpo- daptes. Gazin (1956b) reported the occurrence of Carpolestes, cf. C. dubius, in latest Paleo- 6 Bulletin Museum of Comparative Zoology, Vol. 147, No. 1 ccne beds near Buckman Hollow, south- western Wyoming, the only record of a carpolestid in the southern part of the Green River Basin. The Carpolestidae were regarded as strictly North American until the descrip- tion of Saxonella, from Walbeck, Germany, by D. E. Russell (1964). Due to several significant differences between the North American carpolestids and the new Euro- pean form, Russell was obliged to redefine the family in order to include Saxonella, and to create two subfamilies, the Carpo- lestinae (for the three North American genera) and the Saxonellinae (for Saxo- nella). This classification was adopted by some subsequent authors (e.g. Romer, 1966; McKenna, 1967). Nevertheless, Saxonella seems to have as much or more in common with plesiadapids. Some similarities were mentioned by Russell (1964). Szalay (1968a, 1969a) reiterated the resemblances and intimated a relationship between Saxonella and the Plesiadapidae. Van Valen (1969) went a step further and formally placed the Saxonellinae in the Plesiadapidae. He grouped Carpolestidae with the Paromo- myidae, Microsyopidae, Plesiadapidae, and Picrodontidae in a new superfamily, the Microsyopoidea. The first carpolestids from the Paleocene of Canada were reported by L. S. Russell (1967). He referred three fragmentary specimens from Alberta to a new species, Carpolestes cygneus. As noted by Krish- talka (1973), however, this species is clearly referable to Carpodaptes. Krishtalka de- scribed additional specimens of Carpo- daptes from Canada, this time from the Cypress Hills of southeastern Alberta. Both Carpodaptes and a new species of Elphidotarsius, E. shotgunensis, were re- corded in the Shotgun Local Fauna of western Wyoming (Gazin, 1971). The as- sociation of two different genera of carpo- lestids is otherwise unknown. Szalay (1972b) suggested that the genus Carpolestes should probably be regarded as a synonym of Carpodaptes (a view depicted in a chart by Szalay earlier [1969a: fig. 27], but not mentioned in the text of that paper ) . He also proposed a new inter- pretation of the dental formulae of carpo- lestids. In analogy with his interpretation of the formula in Pronothodectes and in Middle Paleocene paromomyids, he sug- 2.1.3.3 gested the formula ' ' ' or Elphido- 2.1.3.3 H tarsius, and the same formula, except for the loss of Po, for Carpodaptes and Carpo- lestes. He also figured a tooth as the first known upper incisor of a carpolestid (Szalay, 1972b: fig. 1-9). Other notable discussions of the Carpo- lestidae not cited above include those of Abel (1931), Simpson (1940), Hill (1953), Saban (1961), and Simons (1963, 1972). STRATI GRAPHIC AND GEOGRAPHIC OCCURRENCE Carpolestids have been discovered in beds of Middle Paleocene (Torrejonian) to earliest Eocene (Wasatchian) age in the Rocky Mountain region of western North America (see Fig. 2). Where different genera have been found in the same depositional basin, they occur in strati- graphic succession, Elphidotarsius below Carpodaptes (except in the Shotgun Local Fauna), and Carpodaptes below Carpo- lestes. The temporal range of each genus appears to be rather strictly limited, sug- gesting that carpolestids may be of value in determining the approximate ages of faunas. Elphidotarsius is known only in beds of Torrejonian age. Its latest known occurrence is in the late Torrejonian Shot- gun Local Fauna, where it is associated with the earliest known Carpodaptes. All other records of Carpodaptes are restricted to the Tiffanian (Late Paleocene). Carpo- lestes first appears in the Silver Coulee beds of the northern Bighorn Basin (Princeton Quarry level) and is known from several sites regarded as "Clarkforkian" (latest Paleocene) in age, as well as a few of Early Eocene age. Carpolestidae • Rose Only at Shotgun do two carpolestid spe- cies occur together. The Shotgun Local Fauna is diverse, and the sample is very large but consists predominantly of isolated teeth. Included in the fauna are other pairs of related genera rarely or never found together elsewhere. There has been some question as to the age of the Shotgun fauna and, in fact, whether the entire assemblage is the same age. Some forms suggest a late Torrejonian age while others are more indicative of the Tiffanian. Pat- terson and McGrew ( 1962 ) believed that the fauna indicated an early Tiffanian age, but more recently they (personal com- munication, 1973) and C. B. Wood (per- sonal communication) regard the age as late Torrejonian. The primates, described by Gazin (1971), include Palaechthon, Palenochtha, PJesio- lestes, Paromomys, and Elphidotarsius (all recorded only from Torrejonian deposits), Pronothodectes (usually from the Torre- jonian but known also from the early Tiffanian [Gazin, 1956a]), and Phena- colemur, Plesiadapis, and Carpodaptes (all typically Tiffanian or later in age). From the progressive nature of some species of the Torrejonian genera, Gazin ( 1971 ) in- ferred a late Torrejonian age for the Shot- gun assemblage (as he had earlier, [Gazin, 1961]), but he conceded that the presence of the three genera otherwise unknown in pre-Tiffanian faunas1 "might suggest a mixture of materials from different levels" (Gazin, 1971:15). The latter hypothesis is more compatible with the primate evidence, particularly the carpolestids. Specimens of Carpodaptes from Shotgun bear closest resemblance to those from Cedar Point Quarry (definitely of Tiffanian age) and, as noted above, Carpodaptes has not been found at any 1 The only purported Torrejonian record of Plesiadapis is in the Battle Mountain Local Fauna of the Hoback Basin, first considered of Torre- jonian age (Dorr, 1958), but now regarded as early Tiffanian (Gingerich, personal communi- cation ) . other pre-Tiffanian site. But aside from the primates, the fauna is overwhelmingly Torrejonian in aspect, with few Tiffanian forms; the reverse would be expected in a Tiffanian deposit containing reworked Tor- rejonian fossils. Resolution of this problem is, however, beyond the scope of this paper, and herein the age of the Shotgun Local Fauna is accepted as late Torrejonian. The uniqueness of the Shotgun fauna may be due in part to its unusual paleo- environment. In late Paleocene time, the site was flanked on the east by the extensive, probably saline Waltman Lake, which was connected at least intermittently to the Cannonball Sea (McGrew, 1963). Similar paleoecological conditions have not been sampled elsewhere in the North American Paleocene. Mention should also be made of the "Clarkforkian" problem. The use of this term, proposed by H. . E. Wood, et al., (1941), has met some opposition in recent years. For example, R. C. Wood (1967), who reviewed the Clark Fork fauna and stratigraphy, concluded that available evi- dence "scarcely warrants recognition of the Clark Fork as a provincial age, faunal zone, or member of the Polecat Bench Forma- tion" (Wood, 1967:28). Nevertheless, field work in the type Clark Fork area and in other Late Paleocene deposits in recent years has yielded considerable new evidence suggesting that the term "Clarkforkian" may be valid and useful as a North American Land Mammal "Age" (D. C. Parris, R. E. Sloan, personal communication). The genus Carpolestes has been regarded as character- istic of the "Clarkforkian" (Sloan, 1969: fig. 5) and is one of several forms whose overlapping temporal ranges may be used in a redefinition of the "Clarkforkian" (Sloan, in litt., 5/22/73). Sloan (1969) placed Princeton Quarry just below the Tiffanian-"Clarkforkian" boundary; more recently he suggests that it is approximately at the boundary ( Sloan, personal communi- cation, 6/28/73 ) . He regarded Olive, Bear 8 Bulletin Museum of Comparative Zoology, Vol. 147, No. 1 Saskatchewan • Wyoming Colorado Creek, and Buckman Hollow to be within the "Clarkforkian" (Sloan, 1969). Except for the Olive Local Fauna, here considered Tiffanian, I have adopted Sloan's view and regard the Princeton Quarry level as earliest "Clarkforkian". Pending a revised definition of the "age" (Sloan and Parris, in prepara- tion), the name "Clarkforkian" appears herein in quotation marks. To date, more than 300 earpolestid speci- mens (mostly jaws) have been collected (Elphidotarsius: 30+; Carpodaptes: 200+; Carpolestidae • Rose Carpolestes: 80+). They are among the Bench sequence, more than 20 jaws of commonest mammals at many Middle and Elphidotarsius are included in the collec- Late Paleocene sites, and one may infer tion from Rock Bench Quarry, and more that they were not uncommon members of than 60 jaws of Carpodaptes have been re- the biocoenose. For example, in the Polecat covered at Cedar Point Quarry, making it Figure 2. Map of localities where carpolestids have been found. Locality Age Formation Occurrence Reference to locality or to carpolestid record 1) Swan Hills 2) Cypress Hills 3) near Roche Percee 4) Crazy Mountain Field a) Gidley Quarry b) Scarritt Quarry c) Princeton Loc. 11 5) Eagle Coal Mine at Bear Creek 6) Circle 7) Olive 8) Medicine Rocks Site 1 near Ekalaka 9) Judson 10) Polecat Bench region a) Rock Bench Quarry b) Long Draw Quarry (Carbon Co., Mon- tana) c) Silver Coulee beds: Princeton, Schaff, Fritz, and Storm Quarries d) Paint Creek e) Twisty-turn Hollow 11) Big Horn County a) Cedar Point Quarry b) Divide Quarry c) Cleopatra Reservoir site 12) Togwotee Pass 13) Dell Creek 14) Shotgun 15) Badwater, "Malcolm's Locality" 16) Buckman Hollow 17) Mason Pocket at Tiffany Late Paleocene (Tiffanian) Late Paleocene (Tiffanian) Late Paleocene (Tiffanian) Middle Paleocene (Torrejonian) Late Paleocene (Tiffanian) Late Paleocene (Tiffanian) Late Paleocene ("Clarkforkian") Late Paleocene (Tiffanian) Late Paleocene (Tiffanian) Middle Paleocene (Torrejonian) Late Paleocene (Tiffanian) Paskapoo Ravenscrag Ravenscrag Fort Union (Lebo Member) Fort Union (Melville Member) Fort Union Fort Union Tongue River Tongue River Tongue River Tongue River Middle Paleocene Polecat Bench (Torrejonian) Late Paleocene (latest Tif- fanian) Late Paleocene (Tiffanian- "Clarkforkian" boundary) Paleocene-Eo- cene boundary earliest Eocene (Wasatchian) Late Paleocene (Tiffanian) Late Paleocene (latest Tif- fanian) Late Paleocene ("Clarkforkian") Late Paleocene ("Clarkforkian" or Paleocene- Eocene bound- ary) Late Paleocene (Tiffanian) Middle Paleocene (latest Torre- jonian) Late Paleocene (Tiffanian) Polecat Bench Polecat Bench Willwood Willwood Polecat Bench Polecat Bench Polecat Bench "lower variegated sequence" Hoback Fort Union (Shotgun Member) Fort Union (Shotgun Member) Carpodaptes cygneus Carpodaptes, cf. C. cygneus Carpodaptes cygneus Elphidotarsius florencae Carpodaptes hazelae Carpodaptes sp. Carpolestes nigridens Carpodaptes, cf. C. hazelae Carpodaptes sp. Elphidotarsius, cf. E. florencae Carpodaptes, poss. C. cygneus Elphidotarsius, cf. E. florencae Carpodaptes jepseni ? Carpolestes dubius Carpolestes nigridens Carpolestes sp. Carpodaptes hazelae Carpodaptes jepseni Carpolestes sp. Carpolestes sp. L. S. Russell, 1967 Krishtalka, 1973 Krause, personal commu- nication Gidley, 1923; Simpson, 1937b Simpson, 1936 Simpson, 1937b Simpson, 1928, 1929 Sloan, in D. E. Russell, 1967 Sloan, in D. E. Russell, 1967 Princeton University site, unpublished Holtzman and Sloan, per- sonal communication Princeton University sites, in Jepsen, 1930, or un- published Princeton University sites Late Paleocene Wasatch (Chappo ("Clarkforkian") Member) Late Paleocene Tiffany (Tiffanian) Carpodaptes hobackensis Elphidotarsius shotgunensis and Carpodaptes, cf. C. hazelae Carpodaptes sp. Carpolestes sp. Carpodaptes aulacodon McKenna, 1972 Dorr, 1952 Gazin, 1971 Krishtalka, personal com- munication; Black and Dawson, 1966 Gazin, 1956b Matthew and Granger, 1921; Simpson, 1935b 10 Bulletin Museum of Comparative Zoology, Vol. 147, No. 1 second in abundance to Plesiadapis at this productive site. On the other hand, carpo- lestids arc decidedly less common at some localities, such as the Crazy Mountain Field, where only 11 specimens of Carpo- daptes arc known from Scarritt Quarry, and only a single jaw of Elpliidotarsius has been recorded among 382 identifiable mammalian jaws from Gidley Quarry (Simpson, 1937b:34). Nevertheless, the wide geographic distribution and often common occurrence of carpolestids con- tribute to their utility in correlating Paleo- cene faunas. SYSTEMATIC REVISION Maglio (1971:372) recently stated: "... the goal of the paleo-biologist ... is not the recognition of fossil "taxa" and the establishment of a formal terminology for ever smaller segments of a phyletic continuum. On the contrary, it is the establishment of evolutionary units that can be traced through long periods of time and with which broad evolutionary phenomena can be studied." I concur with this view. Systematica should be a means for better understanding animals and their evolution; it should not be an end in itself. The systematic revision presented here has been approached with this in mind. Synonymy of named species has been proposed only when there can be little doubt that the taxa involved are conspecific. Closely similar established spe- cies have been retained when any con- sistent distinctions could be discerned. New species have not been proposed unless specimens were demonstrably different from existing species. I believe that this somewhat conservative approach will pro- mote a clearer understanding of the evolution and interrelationships of the Carpolestidae. Order PRIMATES Linnaeus, 1758 Infraorder PLESIADAPIFORMES Simons, 19721 Superfamily PLESIADAPOIDEA Trouessart, 1897 Family CARPOLESTIDAE Simpson, 1935 Carpolestidae Simpson, 1935b: 9. Type Genus: Carpolestes Simpson, 1928. Included Genera: Carpodaptcs, Elpliidotarsius, and Carpolestes. Distribution: Middle Paleocene ( Torrejonian ) to Early Eocene (early Wasatchian) of western North America (Alberta, Saskatchewan, Mon- tana, North Dakota, Wyoming, and Colorado). Emended diagnosis: Very small aberrant ^ , ?2.?1.?3.3 , primates. Dental formula . Man- F 2.1.2-3.3 dibular dentition characterized by enlarged, rooted, anteriorly inclined medial incisor, followed by greatly reduced tooth (lateral incisor). Root of enlarged incisor implanted at about 45° angle to vertical and extend- ing back no further than to a point beneath P.i. Canine present but also much reduced. Pi absent; P2 either very small and button- like (Elpliidotarsius) or lost (Carpodaptes and Carpolestes). P3 small, double-rooted (Elphidotarsius), or single-rooted (Carpo- daptes and Carpolestes). P4 enlarged (relatively more so in later forms), special- ized into polycuspate, trenchant blade; talonid heel small, consisting of one cusp, distinct in Elpliidotarsius and Carpodaptes, becoming merged with blade in Carpo- lestes. Proliferation of apical cusps on P4 occurring in later forms. Molars, except M], deviating little from generalized plesi- adapid pattern. Trigonid of Mi longer anteroposterior^ than in M2 or M3; widely splayed in Carpodaptcs and Carpolestes, with paraconid directly anterior to proto- conid, forming continuation of P4 blade. Metaconid of Mi always posterolingual to 1 This name first appeared in Simons ( 1972 ) but was credited to "Simons and Tattersall, 1972", a work which has not been published. Carpolestidae • Rose 11 protoconid. Molar talonids broad, basined, with distinct hypoconid and entoconid, small or indistinct hypoconulid (except on M3). Talonid of M3 smaller, more com- pressed anteroposteriorly than in M2. M3 not reduced, with pronounced third lobe bearing two cusps (twinned hypoconulid). P4 (particularly) and molars exodaenodont. Mandible shallowest in Elphidotarsius, deepest in Carpolcstes. At least 2 incisors and small canine (?) present in upper jaw. P2 small, single- rooted, bearing one main cusp. P34 either of plesiadapid type, smaller than molars, relatively unspecialized ( Elph idotarsius ) , or greatly modified, larger than molars, polycuspate with 3 longitudinal rows of cusps (Carpodaptes and Carpolestes). Upper molars with primitive plesiadapid morphology; hypocone small but distinct on M1 2, less distinct or represented by shelf on M3. Discussion: The Carpolestidae were a rapidly evolving group of archaic primates in which extreme specializations of P4 and P3-4 were achieved. The three known genera constitute a sequence that, in general, shows enlargement and special- ization through time; they are almost surely in a single lineage. The earliest genus, Elphidotarsius, is relatively unspecialized and easily derivable from a form morpho- logically close to the earliest plesiadapids; it contrasts with Carpolestes, in which the family traits are fully manifested. Teeth in which the base of the crown extends laterally well beyond the roots have been described as "exodaenodont". The condition was first observed in dimylid insectivores (Hiirzeler, 1944; Saban, 1958) and later noted in picrodontids and some bats (McGrew and Patterson, 1962). It is especially conspicuous in the lower cheek teeth, particularly P4, of carpolestids. Both Hiirzeler and Saban suggested that it cor- relates with malacophagy, but this seems to have little basis. A molluscan diet is verv unlikely for picrodontids and bats (as pointed out by McGrew and Patterson, 1962:6), and equally improbable for carpo- lestids. The dental formula in carpolestids has been controversial. The interpretation I present differs from that of most previous authors (e.g., Matthew and Granger, 1921; Jepsen, 1930; Simpson, 1935b, 1937b; Dorr, 1952; Hill, 1953; Saban, 1961; and Simons, 1972) and is in accord with that recently proposed by Szalay (1972b). (Justification of the new formula is presented below. ) The upper dental formula is necessarily uncertain, since anterior teeth are known only in the most advanced form, Carpo- lestes, and even in that genus only roots are preserved. If the antemolar teeth are cor- rectly interpreted, there were two incisors, a canine, and three premolars. Elphidotarsius Gidley, 1923 Elphidotarsius Gidlev, 1923: 10; Simpson, 1937b: 162. Tvpe Species: Elphidotarsius florencae Gidley, 1923. Included Species: E. florencae and E. shot- gunensis. Distribution: Middle Paleocene ( Torrejonian ) of Wyoming and Montana. Emended Diagnosis: Small, relatively unspecialized carpolestids. Lower dental formula 2.1.3.3. Medial incisor (Ii?) en- larged and procumbent; crown lanceolate in outline; root extending back to a point approximately below canine. Lateral incisor (I2?) and canine known only from alveoli. Alveolus of P2 very small, smaller than those of PL and P3, and situated slightly lingual to them. P3 premolariform, larger relatively and absolutely than in Carpo- daptes and Carpolestes; crown with promi- nent apical cusp preceded by lower, less distinct cuspule, and followed by low but distinct heel; two roots, partially fused, occupying a single alveolus. P4 enlarged (larger than Mi), bladelike, bearing 4 longi- tudinally arranged apical cusps followed by lower but well defined talonid cusp. Mi 12 Bulletin Museum of Comparative Zoology, Vol. 147, No. with anteroposteriorly extended trigonid, paraconid and metaconid lingual to proto- conid, the metaconid the more lingual of the two. Talonid of Mi with distinct hypo- conid and entoconid. Trigonids of M2 and M3 anteroposteriorly compressed. Upper dentition anterior to P3 unknown. P3-M3 all broader (buccolingually) than long. P3 distinctly smaller than P4, with prominent lingual cusp and major buccal cusp, the latter followed by a smaller buccal cusp. P4 about same size as M1, more triangular, longer buccally, shorter lingually; moder- ately specialized with buccal row of 4 cusps, median anteroposterior crest bearing large central cusp, and pronounced lingual cusp directly internal to the latter; incipient hypocone at internal end of posterolingual cingulum. Hypocone of molars formed as in P4, small, but more distinct than in P4, connected to protocone by "nannopithex fold". Pronounced ectocingulum and an- terolingual and posterolingual cingula on M1"3, weaker on P3'4; cingula not con- tinuous onto lingual face of teeth. Elphidotarsius florencae Gidley, 1923 Figures 3, 4 Elphidotarsius florencae Gidley, 1923: 10; Simp- son, 1937b: 163. Holotype: USNM 9411, left mandible with P4-M3. Hypodigm: type specimen only. Horizon and Locality: Middle Paleocene (Tor- rejonian), Lebo Member, Fort Union For- mation: Gidley Quarry, Crazy Mountain Field, Montana. Emended Diagnosis: Less progressive and slightly smaller than E. shotgunensis. P4 and Mi shorter anteroposteriorly but broader buccolingually than in E. shot- gunensis. Lower molars as long as, or longer than, broad. Trigonid of Mi more anteroposteriorly extended (i.e., paraconid and metaconid more widely separated) than in M2 or M3, but less so than in E. shotgunensis. Discussion: Simpson's (1937b) thorough Figure 3. Elphidotarsius florencae Gidley, holotype, USNM 9411, left P4-M3. Crown view (above) and lat- eral view (below). X 6. discussion precludes the need for redescrip- tion of the holotype. The recently described species E. shotgunensis (Gazin, 1971), permits a more precise definition of E. florencae. Unfortunately, since only one specimen is known, information on intra- specific variability is unavailable, unless the sample described below, Elphidotarsius sp., cf. E. florencae, is definitely referable to it. Elphidotarsius sp., cf. E. florencae Gidley Figures 4B, 5A, 6-8, 34A, 34B The largest sample of a population of Elphidotarsius is from the Middle Paleo- Carpolestidae • Rose 13 lm m A B Figure 4. Crown view of right P4-M, of Elphidotar- sius, to same scale. A) E. florencae, holotype, USNM 9411, from Gidley Quarry. B) £., cf. E. florencae, PU 14792, from Rock Bench Quarry. C) E. shotgunensis, holotype, AMNH 88311. cene ( Torrejonian ) Rock Bench Quarry, Polecat Bench Formation of Park County, Wyoming. The more than twenty speci- mens are closely comparable to the holo- type of E. florencae. There are no significant morphologic differences, but the individuals in the Rock Bench sample are, in general, slightly larger than the holotype. A Student's t-test comparing the dimensions of the type with those of the Rock Bench specimens yielded inconclusive results, neither favoring erection of a new species nor declaring the two samples positively conspecific. Description: The Rock Bench Quarry sample includes many specimens that pro- 1 MM Figure 5. Lower left medial incisors of carpolestids. A) Elphidotarsius, cf. E. florencae, PL) 14282. B) ICar- podaptes sp., UW 6530 (reversed). C) Carpolestes dubius, PU 14235. vide new information about Elphidotarsius. A few reveal the anterior part of the man- dible and suggest the dental formula 2.1.3.3. The Torrejonian plesiadapid Prono- thodectes, which is close to Elphidotarsius in many features ( see below ) , has a similar dental pattern and the formula 2.1.3.3. (e.g., PU 14783; Szalay, 1972b; Gingerich, personal communication). Analogy with Pronothodectes strengthens the interpreta- tion of the lower dental formula of Elphi- dotarsius adopted here. The enlarged medial incisor is preserved in one specimen, PU 14282 (Figs. 5A and 6). It is lanceolate in outline and quite broad at the base, much broader than in either of the other carpolestid genera. The dorsal (lingual) surface is broad and slightly convex, bounded laterally by a ridge running from the tip to the base of the crown, where it merges with a promi- nent internal cingulum. The incisor, al- though shorter, somewhat resembles that of Plesiolestes. A small basal cusp arising from the internal cingulum is found in Pronothodectes and Plesiadapis, but is con- spicuously absent from the incisor of Elphidotarsius. An interstitial facet is present in PU 14282, but there is no occlusal facet such as observed in Plesia- dapis by Gingerich (in press). 14 Bulletin Museum of Comparative Zoology, Vol. 147, No. 1 Figure 6. Elphidotarsius, cf. E. florencae Gidley, PU 14282, left dentary with medial incisor, P4-M3. Crown view (above) and lateral view (below). X 6. Rock Bench Quarry. Three alveoli are present between the mesiodistally compressed and slightly larger enlarged incisor and P;{. They are inter- than that for the canine. Its slight anterior preted to have held 1^, C, and P^. The inclination suggests that I2 was somewhat alveolus just behind the medial incisor is procumbent. The second alveolus is slightly Carpolestidae • Rose 15 Figure 7. Elphidotarsius, cf. E. florencae Gidley, PL) 14791, right dentary with P3-M3. Crown view (above) and lateral view (below). X 6. Rock Bench Quarry. larger than that of P2, supporting the view position that P2 has been lost in Carpo- that it held the canine. The socket for P2 daptes and Carpolestes, both of which have is situated just lingual to those of the canine one less tooth in the mandible, and P3. Its diminutive size favors the sup- P3 in the Rock Bench sample is pre- 16 Bulletin Museum of Comparative Zoology, Vol. 147, No. 1 molariform and relatively unreduced. Its two roots are partially joined, occupying a single alveolus. The crown bears a well developed central cusp followed by a low, well differentiated talonid cusp. A short anterior shelf rises in a tiny, low cusp, an- terior to the main cusp. P4 and the lower molars are very similar to those in the type of E. florencae. In some individuals, the cheek teeth (P4-M3) are relatively broader buccolingually than in the type. M2, particularly, is almost al- ways as broad as it is long (as in the type), and often is broader than long. Neverthe- less, the extent of variation observed in the Rock Bench sample and in other large samples of carpolestids indicates that this feature is not of taxonomic importance. P4 in the Rock Bench sample bears four apical cusps and a lower, pronounced talo- nid cusp, but it is comparable in length to E. florencae and shorter than in E. shot- gunensis. The third apical cusp is the largest and highest and is probably homol- ogous with the protoconid. (This cusp in the type of E. florencae shows apical wear, thus appearing lower than the cusp anterior to it.) Just behind, lower, and slightly lingual to the third cusp is a cusp probably homologous with the metaconid. The second cusp appears to be the homologue of the paraconid and is nearly as prominent as the protoconid. Anterior to it is an ac- cessory cusp, the lowest and least developed of the four. Butler ( 1973 ) has also sug- gested these homologies. As in the holotype of E. florencae, Mi is relatively generalized. The trigonid is only slightly splayed (less anteroposteriorly extended than in E. shotgunensis) but is more buccolingually compressed than in M2 and M3. The paraconid of Mi is usually the weakest trigonid cusp, while the protoconid is the largest. In M2 and M3, the metaconid is usually the highest trigonid cusp and the paraconid remains small and low. The mandible in Elphidotarsius is shal- lower than in Carpoclaptes or Carpolestes. Two specimens in the Rock Bench sample (PU Nos. 17439 and 17736) reveal the upper cheek teeth of Elphidotarsius. The better of these, PU 17439, is a right maxil- lary fragment with P3-M3 and two alveoli anterior to P3 (Figs. 8, 34B). Each of the anterior alveoli held a single-rooted tooth, P2, and C or possibly P1. The latter was the smallest of the known upper teeth. P3 and P4 reveal new evidence pertinent to the origin and affinities of the Carpolestidae. They are unspecialized relative to their development in Carpoclaptes and Carpo- lestes, but compare very closely in structure with P:; 4 of Pronothodectes. The resem- blance is striking and is indeed closer be- tween Elphidotarsius and Pronothodectes than between the former and either of the other carpolestid genera. The similarities involve much of the dentition and are surely indicative of true relationship. They support the conclusion that the Carpolesti- dae and the Plesiadapidae shared a close common ancestor. It is emphasized, none- theless, that certain features are present in P:i~4 of Elphidotarsius that clearly fore- shadow developments more fully mani- fested in Carpodaptes and Carpolestes. P8 is substantially smaller than P4, and the latter and M1 are subequal. Both P3 and P4 are pyriform in outline, considerably longer on the buccal side than lingually. Both are much shorter antero- posteriorly than in the two later genera. P3 has two buccal cusps, a prominent one about in the center of the buccal crest, and a smaller cusp behind it. A longitudinal crest lies just lingual to the middle of the tooth and bears one central cusp. Internal to this is a lingual shelf, unfortunately frac- tured anteriorly in the only known speci- men, so the possible presence of a cusp there can be neither verified nor ruled out. There is a faint ectocingulum. P4 is larger and has ( as in later carpolestids and Prono- thodectes) three longitudinal sections (rows of cusps, or crests with one or more cusps). There are four buccal cusps. The most anterior is small, possibly homologous with the parastyle. Behind it, located at approxi- Carpolestidae • Rose 17 Table 1. Metrical data for lower cheek teeth of Elphwotarsius, cf. E. florencae, from Rock Bench Quarry, Polecat Bench Formation. N OR ( mm ) X s V length 16 1.7-1.9 1.79 ± .02 .085 ± .015 4.8 ± .8 Pi breadth 15 1.3-1.6 1.40 ± .03 .107 ± .020 7.6 ± 1.4 length 15 1.4-1.6 1.50 ± .01 .038 ± .007 2.5 ± .5 Mi breadth 14 1.3-1.6 1.44 ± .02 .093 ± .018 6.5 ± 1.2 length 15 1.3-1.6 1.41 ± .02 .083 ± .015 5.9 ± 1.1 M2 breadth 15 1.4-1.7 1.50 ± .02 .093 ± .017 6.2 ± 1.1 length 11 1.8-2.1 1.95 ± .02 .082 ± .017 4.2 ± .9 M3 breadth 10 1.2-1.4 1.29 ± .02 .074 ± .017 5.7 ± 1.3 Mandibulai • depth 14 2.8-3.5 3.12 ± .05 .197 ± .037 6.3 ± 1.2 mately the center of the crest, is a promi- nent cusp (paraeone?), followed by another well-developed cusp ( metacone? ) . The last cusp on the buccal crest is much smaller and may be homologous with the metastyle. A central, longitudinal, creseentie ridge bears one large cusp anteriorly. Internal to this, a well-expressed cusp rises anteriorly from the lingual cingulum and is followed by a slight rise of enamel, probably an incipient hypocone. There is an ectocingu- lum, and anterolingual and posterolingual cingula are joined internally by the lingual cingulum. The upper molars deviate little from the primitive plesiadapid pattern. They are very similar morphologically to the molars in other carpolestids, which have been thoroughly described by previous authors (Jepsen, 1930; Simpson, 1937a). The three trigon cusps are high and prominent; the paraconule and metaconule are smaller but distinct. A cingular hypocone, connected to the protoeone by a "nannopithex fold", is well expressed on M1 and M2 but is virtually absent from M3. Pronounced cingula are present on the external, an- terior, and posterior borders, except on M3, where the postcingulum is restricted to the lingual half of the tooth. The molars de- crease in size from M1 to M3, and in M3 the posterobuccal corner (metacone area) is greatly reduced. The molars in PU 17439 are more buccolingually distended than in many specimens of other carpolestids. If the allocation of the Rock Bench Quarry sample to E. florencae is correct, speculations can be offered regarding the apparent size discrepancy between speci- mens in this sample and the holotype. The holotype may represent an unusually small individual, or, more probably, the Gidley Quarry population was, on the whole, slightly smaller than that of Rock Bench. Such differences in local populations are not uncommon in Recent species and should be expected in extinct species as well. Furthermore, although Gidley Quarry and Rock Bench are spatially and tempo- rally close, there may be a slight age differ- ence. The Elphidotarsius specimens now known suggest that Gidley Quarry is slightly older, and plesiadapid fossils strengthen this view (P. D. Gingerich, personal communication), although Sloan (1969; fig. 5) considered Rock Bench to be the earlier of the two. To summarize, the Rock Bench Quarry sample of Elphidotarsius is best referred to E. florencae. The possibility cannot be dis- missed, however, that additional specimens from Gidley Quarry or an equivalent level nearby could reveal sufficient distinctions from the Rock Bench population to warrant specific separation for the latter. Three specimens ( PU Nos. 16916, 19764, and 19780) from the Medicine Rocks (Tongue River Formation) near Ekalaka, Montana, are also tentatively referred to 18 Bulletin Museum of Comparative Zoology, Vol. 147, No. 1 Figure 8. Above: Elphidotarsius, cf. E. florencae Gidley, PU 17439, right maxilla with P3-M3. Crown view. X 6. Rock Bench Quarry. Below: Elphidotarsius shotgunensis Gazin, holotype, AMNH 88311, right P4-M,. Crown view. X 6. E. florencae. All are fragmentary mandibles with partial dentition, but only one speci- men, PU 19764, preserves all of P4. In P4 of this specimen, the presumed homo- logues of the molar trigonid cusps are prominent, with both paraconid and meta- conid situated slightly lingual to the proto- conid. The fourth, anteriormost cusp is much lower and very small, smaller than in other specimens referred to E. florencae. Whether this feature is significant taxo- nomically is unclear at present; PU 19764 is Carpolestidae • Rose 19 otherwise like the holotype and the Rock Bench specimens. Elphidotarsius shotgunensis Gazin, 1971 Figures 4C, 8 Elphidotarsius shotgunensis Gazin, 1971: 33. Holotype: AMNH 88311, right mandibular frag- ment with Pi-Mi and anterior alveoli. Hypodigm: Type and MCZ 18775, * a right Mi. Horizon and Locality: Late Middle Paleocene (late Torrejonian ) , Shotgun Member, Fort Union Formation: near Cottonwood Creek, Sect. 30-31, T. 6 N., R. 3 E., Wind River Rasin, Wyoming. Diagnosis (modified after Gazin): P4 longer and narrower than in E. florencae; Mi narrower than in E. florencae. Primary portion of P4 with 4 apical cusps, possibly less well defined than in E. florencae. Trigonid of Mi more elongate than in E. florencae, i.e., paraconid and metaconid more widely separated, but both lingually placed with respect to protoconid. Discussion: Gazin described this form thoroughly, and there is no new material since his study to contribute further in- formation. A re-examination of the type mandible, however, results in a different interpretation of the anterior dentition. The three alveoli immediately anterior to P4 probably held P3, P2, and the canine. An- terior to the canine, Gazin suggests the presence of three incisors. This is very unlikely, for no known Paleocene primate had three lower incisors. The most anterior alveolus is enlarged and anteriorly in- clined, as in other carpolestids. It is fol- lowed by a smaller alveolus, for I2. Gazin interpreted a groove behind this as possibly the alveolus of a third incisor, but the groove is small and poorly defined, and it appears to be situated at a break in the mandible. It is most improbable that it represents an alveolus. (It should be noted that Gazin was duly hesitant in suggesting the presence of three incisors.) Specimens 1.5 U 03 CD m 0? * 1.0 -I 1 1 This specimen has apparently been lost. 1.5 ^ 2.0 P4 Length Figure 9. Scatter diagram of P4 dimensions of Elphi- dotarsius. Black circles = £., cf. E. florencae from Rock Bench Quarry. F = £. florencae, holotype, from Gidley Quarry. S = E. shotgunensis. holotype. of Elphidotarsius from Rock Bench Quarry, discussed above, indicate the presence of only two lower incisors. Gazin referred an upper premolar, MCZ 18774, to this species. Inspection of the specimen and comparison with other carpo- lestids reveals that it is a very water-worn premolar of Carpodaptes, also known from this locality. With the elimination of this specimen, the known teeth of E. shotgunen- sis are reduced to P4 and Mi. Nevertheless, the morphology of these two is clearly more advanced than in E. florencae and suggests a trend toward Carpodaptes. Carpodaptes Matthew and Granger, 1921 Carpodaptes Matthew and Granger, 1921 : 6; Simp- son, 1935: 10; Dorr, 1952: 82. Type Species: Carpodaptes aulacodon Matthew and Granger, 1921. Included Species: C. aulacodon, C. hazelae, C. hobackensis, C. cygneus, and C. jepseni (new species ) . 20 Bulletin Museum of Comparative Zoology, Vol. 147, No. 1 Distribution: Late Middle Paleocene (late Torrejonian) of Wyoming, Late Paleocene (Tif- fanian) of western North America: Alberta, Saskatchewan, Montana, North Dakota, Wyo- ming, and Colorado. Emended Diagnosis: Lower dental formula 2.1.2.3. Enlarged medial incisor longer and more slender than in Elphido- tarsius. Lateral incisor (I^?), C, and P.s reduced to tiny, single-rooted vestiges with buttonlike crowns. P- absent. Pt larger, both relatively and absolutely, than in Elphidotarsius, but smaller in both respects than in Carpolestes; apical cusps number- ing 5 or 6, followed by talonid heel less distinct than in Elphidotarsius, but gener- ally more distinct than in Carpolestes; heel decidedly lower than trigonid of Mi; slight constriction of enamel at crown separating talonid from primary blade. Trigonid of Mi elongate and open, paraconid immedi- ately anterior or slightly anterolingual to protoconid; protoconid the most prominent trigonid cusp. M2-.s not significantly differ- ent from those of Elphidotarsius and Carpo- lestes, but third lobe of M3 usually larger than in Elphidotarsius; incipient mesoconid sometimes present on cristid obliqua of M3. Upper dentition with at least two reduced teeth anterior to P3 (presumably P2 and C); both with one major cusp. P3-4 pyriform, polycuspate, greatly enlarged over those of Elphidotarsius, but not so large as in Carpolestes. P3 and P4 subequal, or P3 slightly smaller than P4, both larger than molars; each with 3 well-developed longi- tudinal rows of cuspules, buccal row long- est. P3 without pronounced anteroexternal spur typical of Carpolestes; external row with 4 cusps, the most anterior separated somewhat from the 3 posterior; median row a crescentic crest, with one or more cuspules, situated slightly buccal to middle of tooth; lingual section with two subequal cusps. P4 with buccal row of 5 or 6 cusps, the 2 anterior ones separated slightly from the 3 or 4 posterior ones; median crest similar to that of P:i; lingual region with 3 cusps, the central cusp largest. Molars very similar morphologically to those of other earpolestid genera. Carpodaptes aulacodon Matthew and Granger, 1921 Figures 10, 11A Carpodaptes aulacodon Matthew and Granger, 1921: 6; Simpson, 1935b: 12. Holotype: AMNH 17367, left mandible with P,-M3. Hypodigm: Type specimen only. Horizon and Locality: Late Paleocene (Tif- fanian), Tiffany Formation: Mason Pocket, Tiffany, Colorado. Emended Diagnosis: P4 bearing 5 poorly- defined apical cusps, followed by distinctly separated, low talonid heel; anteriormost cusp almost indistinct, as low as talonid and separated somewhat from following 4 closely appressed cusps; lateral profile roughly triangular with apex almost pointed, in contrast to more gently rounded profiles of P4 in other species of Carpodap- tes. P4 longer than in C. hazelae, C. hobackensis, or C. cygneus, shorter than in C. jepscni or species of Carpolestes. Molars not significantly different from those of other species. Discussion: This revised diagnosis of Carpodaptes aulacodon is the first pub- lished since other species were added to the genus. The type mandible clearly re- veals the great reduction of the teeth be- tween the enlarged incisor and P4, a characteristic of carpolestids which is mani- fested to a greater degree in Carpolestes and Carpodaptes than in Elphidotarsius. The structure of P4, the most diagnostic mandibular tooth in carpolestids, differs from that of all other species of Carpo- daptes, and we may be confident (despite the existence of only one specimen) in recognizing AMNH 17367 as a representa- tive of a distinct species. Carpolestidae • Rose 21 f £. h -* r- Figure 10. Above: Carpodaptes aulacodon Matthew and Granger, holotype, AMNH 17367, left P3-M3. Crown view. :■ 6. Below: Carpodaptes sp., PU 14639. Left maxilla with P4-M2, crown view. X 6. Carpodaptes hazelae Simpson, 1936 Figures 11C, D, 12-18 Carpodaptes hazelae Simpson, 1936: 21; Simpson, 1937a: 5. Holotype: AMNH 33854, right mandible with P4-M3. Hypodigm: Type and AMNH Nos. 33853, 33855, 33887, 33979-33985, from Scarritt Quarry; PU Nos. 19558, 19572, 19574, 19596, 19601, 19936, 19939, 19953-19955, 19958, 19968, 19969, 19985, 20007, 20010, 20011, 20034, 20060, 20064, 20065, 20068, 20084, 20087, 20610, 20615, 20630, 20634, 20656, 20719, 20808, 20812, 20820, 20839, 20840, 22 Bulletin Museum of Comparative Zoology, Vol. 147, No. 1 1mm Figure 11. Outlines of right P4 of Carpodaptes (lateral aspect), to same scale. A) C. aulacodon, holotype, AMNH 17367 (reversed). B) C. jepseni, holotype, PL) 20716. C) C. hazelae, holotype, AMNH 33854 (Scarritt Quarry). D) C. hazelae, PL) 20084 (Cedar Point Quarry). E) C. hobackensis, UMMP V55124. F) C. cygneus, holotype, ROM 05622. 20886, 20889, 20898, 20900, 21213, 21266, 21279, 21297, 21299, 21316, 21317, 21330, 21341, 21350, 21351, 21357, 21384, 21393, 21399, 21413, 21419, 21437, 21438, from Cedar Point Quarry. Horizon and Locality: Late Paleocene (Tif- fanian) of Montana and Wyoming: Scarritt Quarry, Melville Member, Fort Union Forma- tion, Crazy Mountain Field, Montana; Cedar Point Quarry, Polecat Bench Formation, Big Horn County, Wyoming. Emended Diagnosis: Close to C. aula- codon in size, but mean length of P4 less than in C. aulacodon. P4 with 5 or 6 vari- ably defined apical cusps followed by a low but distinct talonid heel; more rounded in profile and more quadrate in occlusal view (at base) than in C. aulacodon. Molars not significantly different morpho- logically from other species, but larger than in C. hobackensis and C. cygneus, smaller than in C. jepseni. Discussion: Carpodaptes hazelae was originally described and defined from a small sample collected at the Scarritt Quarry in the Crazy Mountain Field. The much larger sample from Cedar Point Quarry is here referred to Carpodaptes hazelae and necessitates the revised diag- nosis presented above. C. hazelae is now represented by more specimens than any other species of the family. There can be little doubt that the carpolestids from Cedar Point Quarry all represent the same species. Although con- siderable variation is observed, graphs of tooth dimensions reveal normal, unimodal distribution; there is no indication of sexual dimorphism and no evidence of the pres- ence of more than one taxon. Since the different variations do not fall into con- sistently distinct groups, there is no practi- cable method of distinguishing more than one species, and indeed, no justification for so doing. On the other hand, a casual examination of some of the specimens here referred to C. hazelae might suggest that the Cedar Point Quarry sample represents a species distinct from that of Scarritt Quarry. But, as noted above, separate populations of a species not uncommonly show minor differences, and this should not be unexpected in C. hazelae. The large sample of this species now known provides us with a good cross-section of the variation which may occur intraspecifically in carpo- lestids, and it is substantial. But almost all variation is limited to the specialized pre- molars, P4 and P3-4. Although some speci- mens in the Cedar Point sample differ in size or in minor morphologic details from the type specimen (from Scarritt Quarry), others are virtually identical to the type. The Scarritt Quarry specimens tend to be slightly larger and their lower cheek teeth (P4-M3) relatively broader than the mean Carpolestidae • Rose 23 i;w Figure 12. Carpodaptes hazelae Simpson, AMNH 33980, left dentary with P4-M3. Crown view (above) and lateral view (below). X 6. Scarritt Quarry. values for the Cedar Point sample, but several Cedar Point specimens exceed in size even the largest Scarritt specimen. P4 in the best preserved individuals from Scar- ritt Quarry bears 5 sharp, relatively large apical cuspules followed by a lower, distinct talonid heel (Fig. 12). In the Cedar Point specimens, the apical cuspules, 5 or 6 in number, are usually (but not always) less well defined (Fig. 16). The sixth cuspule arises low at the anterior of the blade in about 75% of the known specimens (44 of 24 Bulletin Museum of Comparative Zoology, Vol. 147, No. 1 Figure 13. Above: Carpodaptes hazelae Simpson, AMNH 33980, left maxilla with P2-M3. Crown view. X 6. Scarritt Quarry. Below: Carpodaptes hazelae Simpson, PU 19939, right maxilla with P3-M3, crown view. X 6. Cedar Point Quarry. Carpolestidae • Rose 25 * Figure 14. Carpodaptes hazelae Simpson, PU 20839, right dentary with P4-M3. view (below). • 6. Cedar Point Quarry. Crown view (above) and lateral 59 specimens); it is variably developed, ranging from an incipient cuspule to one as well defined as any of the other apical cusps. The talonid of P4 in the Cedar Point specimens is usually not so distinctly set off from the blade as in the Scarritt indi- viduals, but this feature, too, is variable. Only a few upper dentitions are known for Carpodaptes hazelae. The Scarritt Quarry and Cedar Point Quarry specimens are similar except for the position of one cusp on P4 ( Fig. 18 ) . In the lingual section of P4 in three Scarritt specimens, the central cusp (the largest of the three lingual cusps) is situated approximately between the other two lingual cusps and contributes to the lingual face of the tooth. This same cusp in the two known specimens from Cedar Point is situated posterolabial to the anteriormost lingual cusp, thus it con- tributes almost nothing to the lingual face of the tooth. The fourth Scarritt specimen 26 Bulletin Museum of Comparative Zoology, Vol. 147, No. 1 «*'t*t ST*;-*4* 'A #^ * ■ • ♦ J Figure 15. Carpodaptes hazelae Simpson. Above: PL) 21350, right dentary with P4-M3, crown view. Below: PU 21351, right dentary with P4-M3, crown view. X 6. Cedar Point Quarry. Carpolestidae • Rose 27 Figure 16. Carpodaptes hazelae Simpson. Above: PU 21351, right dentary with P4-M3, medial view. Note pres- ence of 6 apical cusps. Below: PU 21350, right dentary with P4 cusps. Both ■ 6. Both from Cedar Point Quarry. M3, medial view. Note presence of 5 apical preserving P4 (AMNH 33981) is nearly intermediate between these two morphol- ogies, suggesting that the difference is not taxonomically significant. In addition to the Scarritt Quarry and Cedar Point Quarry remains referred to this species, there are a few specimens from the Late Torrejonian Shotgun Local Fauna (MCZ Nos. 18763, 18776, 18777, 19683, 19684) that are closely comparable to C. hazelae. They are the earliest recorded representatives of the genus Carpodaptes, and the only ones known from pre-Tif- fanian sediments. An isolated P4 (UMVP 5007) from the Circle Local Fauna, Tongue River For- mation, near Circle, Montana, is probably referable to C. hazelae. Specimens of Carpodaptes from the Ravenscrag Formation in Alberta were referred to Carpodaptes, cf. C. hazelae, by Krishtalka (1973), although (as he noted) they are smaller than typical C. hazelae. They are closer in size and morphology to Carpodaptes cygneus and are here tenta- tively referred to the latter. The dental morphology of C. hazelae has been described in detail by Simpson ( 1936, 1937a). No new material has been re- covered from Scarritt Quarry since Simp- son's work, and the material he described in 1937 still includes some of the most complete carpolestid specimens known. Among the Cedar Point specimens are several which preserve the anterior portion of the mandible, but only alveoli are pre- served anterior to P3. From these speci- mens it is clear that there were three diminutive, single-rooted teeth between the large medial incisor and P4. I interpret these as a lateral incisor, the canine, and P3. P3 is preserved in only one specimen, PU 28 Bulletin Museum of Comparative Zoology, Vol. 147, No. 1 'Ml f^ Figure 17. Carpodaptes hazelae Simpson. Above: PL) 21297, left dentary with P4-M2, crown view. Below: PU 20084, right dentary with P4-M3, crown view. Both X 6. Both from Cedar Point Quarry. Carpolestidae • Rose 29 1mm B Figure 18. Crown view of right P4 of Carpodaptes hazelae, to same scale. A) AMNH 33980 (reversed), from Scarritt Quarry. B) AMNH 33981, from Scarritt Quarry. C) PU 19939, from Cedar Point Quarry. 20630, and is much reduced compared to that of Elphidotarsius. It is a vestigial peg- like tooth, with the same morphology as P3 in C. aulacodon (and similar, in fact, to P3 in Carpolestes). The morphology of P4 has been discussed above. It should be emphasized again that although P4 is unquestionably the most diagnostic mandib- ular tooth in carpolestids, it is also the most variable. The extent of its variability is especially well demonstrated in the large sample of C. hazelae. The molars (lower and upper) of the Cedar Point specimens are virtually in- distinguishable from those of the Scarritt Quarry specimens. There is some variation, of course, in relative proportions (e.g., relative breadth of the lower molars). A maxillary dentition (AMNH 33980) from Scarritt Quarry, described by Simpson ( 1937a ) , included two small, single-rooted teeth anterior to P3. Unfortunately, the anteriormost of these (C?), unknown in any other carpolestid specimen, has been Table 2. Metrical data for lower cheek teeth of Carpodaptes hazelae, combined samples from Scarritt and Cedar Point Quarries. N OR (mm) X s V p length breadth 56 2.1-2.8 2.46 ± .02 .142 ± .013 5.8 ± .5 55 1.2-1.8 1.54 ± .02 .134 ± .013 8.7 ± .8 M length Ml breadth 53 1.3-1.6 1.47 ± .01 .078 ± .008 5.3 ± .5 52 1.2-1.6 1.34 ± .01 .097 ± .010 7.2 ± .7 length M2 breadth 39 1.1-1.4 1.29 ± .01 .070 ± .008 5.4 ± .6 39 1.2-1.6 1.36 ± .01 .091 ± .010 6.7 ± .8 M length Ms breadth 17 1.7-2.0 1.86 ± .02 .093 ± .016 5.0 ± .9 17 1.0-1.4 1.18 ± .02 .103 ± .018 8.7 ± 1.5 Mandibular depth 36 3.1-4.0 3.49 ± .04 .228 ± .027 6.5 ± .8 30 Bulletin Museum of Comparative Zoology, Vol. 147, No. 1 Table 3. Metrical data for lower cheek teeth of Carpodaptes hazelae from Cedar Point Quarry only. N OR (mm) X s V p* length 53 2.1-2.8 2.46 ± .02 .143 ± .014 5.8 ± .6 breadth 52 1.2-1.8 1.53 ± .02 .132 ± .013 8.6 ± .8 Ma length 47 1.3-1.6 1.46 ± .01 .077 ± .008 5.3 ± .5 breadth 46 1.2-1.5 1.32 ± .01 .087 ± .009 6.6 ± .7 M2 length 33 1.1-1.4 1.29 ± .01 .070 ± .009 5.4 ± .7 breadth 33 1.2-1.5 1.34 ± .01 .079 ± .010 5.9 ± .7 M3 length 14 1.7-2.0 1.86 ± .03 .101 ± .019 5.4 ± 1.0 breadth 14 1.0-1.3 1.16 ± .03 .094 ± .018 8.1 ± 1.5 Mandibular depth 33 3.1-4.0 3.46 ± .04 .221 ± .027 6.4 ± .8 lost since Simpson's description. Both P3 and P4 are hypertrophied and specialized in C. hazelae. P4 has 5 or 6 buccal cuspules, the two anterior ones closely associated and separated slightly from the three or four posterior cuspules. The specialization of P3 is especially dramatic in comparison to the small, generalized P3 of Elphidotarsius. Morphological details of P3 and P4 in C. hazelae have been listed above in the diag- nosis of Carpodaptes and in Simpson's (1937a) description. Carpodaptes hobackensis Dorr, 1952 Figures 11E, 19, 20A Carpodaptes hobackensis Dorr, 1952: 82. Holotype: UMMP 27233, left mandible with L, Pi-M3, and roots of I2, C, Pa. Hypodigm: Type and UMMP Nos. 27234, 27253, 55124. Horizon and Locality: Late Paleocene (Tif- fanian), Dell Creek Local Fauna, Hoback Formation: UM-Sub-Wy Locality No. 1, Dell Creek Quarry, Sublette County, Wyoming. Diagnosis (modified after Dorr, 1952): About the same size as C. cijgneus but smaller than all other known species of Carpodaptes. P4 lower and more gently rounded in lateral profile than in other species of Carpodaptes, except C. cijgneus; crest S-shaped in occlusal view due to deep posterointernal excavation of tooth, bearing 5 small but distinct apical cusps and a talonid cusp; talonid not so clearly sepa- rated from main blade as in C. hazelae or C. aulacodon; incipient lower anterior Table 4. Metrical data for lower cheek teeth of Carpodaptes hazelae from Scarritt Quarry only. N OR (mm) X s V p* length breadth 4 4 2.5-2.6 1.7 2.58 ± .03 1.70 .050 ± .018 0 1.9 ± .7 0 Mj lengdi breadth 7 7 1.4-1.6 1.4-1.6 1.51 ± .03 1.47 ± .03 .069 ± .018 .076 ± .020 4.6 ± 1.2 5.1 ± 1.4 M2 length breadth 7 7 1.2-1.5 1.4-1.6 1.34 ± .04 1.47 ± .03 .098 ± .026 .076 ± .020 7.3 ± 2.0 5.1 ± 1.4 Ma length breadth 4 4 1.8-2.0 1.2-1.4 1.90 ± .04 1.28 ± .05 .082 ± .029 .096 ± .034 4.3 ± 1.5 7.5 ± 2.7 Mandibular depth 4 3.7-3.8 3.75 ± .03 .058 ± .021 1.5 ± .5 Carpolestidae • Rose 31 seventh cusp variable in occurrence; tooth very similar to that of C. cygneus, differing in having less pronounced vertical ribs developed beneath apical cusps on lingual side, less developed posterior apical cusp (cusp just anterior to talonid heel), and, consequently, a deeper posterolingual de- pression than in C. cygneus; weakly ex- pressed posterointernal cingulum present, extending forward beyond midpoint of tooth. Weak internal cingulum at base of trigonid of Mi. Discussion: This species is very close to the subsequently described species C. cygneus (L. S. Russell). Specimens now known reveal minor but apparently con- sistent differences between the two, and both are accordingly retained as valid taxa in this review. They are so similar, how- ever, that they must be very closely related, and it is not inconceivable that future evi- dence could provide justification for placing C. cygneus in synonymy. Although only four specimens are known (no upper teeth have been recovered) the entire lower dentition is represented. The type mandible preserves parts or all of the lower teeth. The enlarged incisor un- fortunately was damaged in preparation, but as restored it is a relatively longer and more slender tooth than in Elphidotarsius, much closer to the enlarged incisor of Carpolestes. This is the only incisor of Carpodaptes known, except for one isolated tooth from the Shotgun Local Fauna (UW 6530), possibly from Carpodaptes sp. (see Fig. 5B). Though only roots of the I2, C, and P3 remain, their crown shapes are clearly indicated by impressions in the matrix. All were diminutive, single-rooted, peglike teeth with bulbous crowns. P4, with its posterointernal depression, S-shaped crest, and talonid less distinctly separated from the main blade than in other species of Carpodaptes, bears a closer resemblance to P4 of Carpolestes than does P4 of any other species of Carpodaptes, except possibly C. jepseni (which in other ways foreshadows Carpolestes). For these Figure 19. Carpodaptes hobackensis Dorr, UMMP V55124, right dentary with P4-M3. Crown view (above) and medial view (below). X 6. reasons Dorr (1952: 84) suggested in- directly that C. hobackensis might be closer to the ancestry of Carpolestes than any other species of Carpodaptes then known. The diminutive size of C. hobackensis relative to other species of Carpodaptes casts doubt on such a possible relationship, however. No additional specimens have been re- covered since Dorr's report. His thorough description obviates the need for further detail here. 32 Bulletin Museum of Comparative Zoology, Vol. 147, No. 1 A 1 MM Figure 20. Lingual view of right P., of A) Carpodaptes hobackensis, and B) Carpodaptes cygneus. Carpodaptes cygneus (L S. Russell), 1967 Figures 11 F, 20B Carpolestes cygneus L. S. Russell, 1967: 19. Holotype: ROM 05622, right mandibular frag- ment with Pi. Hypodigm: Type and paratypes, ROM Nos. 0562,3 and 05624, from Swan Hills, Alberta; numerous specimens in the University of Alberta collection from near Roche Percee, Saskatche- wan, including UA Nos. 8733, 8742, 8799, 8800. Horizon and Locality: Late Paleocene (early [?] Tiffanian) of Alberta and Saskatchewan: Paskapoo Formation of the Swan Hills, north- central Alberta, and Ravenscrag Formation of southern Saskatchewan (and Alberta?). Emended Diagnosis: Small earpolestids near Carpodaptes hobackensis in size but smaller than other known species. P4 with 5 or 6 small apical cusps; posterior cusp (anterior to talonid cusp) well expressed; vertical ribs beneath apical cusps on lingual side pronounced, noticeably better de- veloped than in C. hobackensis; postero- internal depression not as deep as in C. hobackensis. Discussion: Russell named this species on the basis of three jaw fragments each holding a single tooth: two lower fourth premolars and one upper third premolar. Fortunately, these are the most diagnostic teeth. Russell regarded the species as closest to Carpolestes dubius and cited the resemblances between the two as justifi- cation for allocating the new species to the genus Carpolestes. As Krishtalka (1973) pointed out, all the features noted by Russell are characteristic of Carpodaptes; there is no special resemblance to Carpo- lestes. In fact, C. cygneus is remarkably close to C. hobackensis, a form described fifteen years earlier. The only noticeable differences in the dentition as known occur in P.,; these have been detailed above under C. hobackensis (See also Fig. 20). A large sample recently recovered from the Ravenscrag Formation of Saskatchewan by D. Krause (personal communication) is referable to C. cygneus. Many specimens are considerably more complete than Russell's original lot and show the mandib- ular dentition, other than P4, to be virtu- ally indistinguishable from that of C. hobackensis. A thorough description of these will be published by Krause. P3, known from one of the paratypes ( ROM 05624 ) and a few of Krause's speci- mens, shows some distinctions from that of C. hazelae (the only other species of the genus in which upper teeth are known). ROM 05624 is smaller and more square in Carpolestidae • Rose 33 occlusal view than P8 of C. hazelae. In the posterior third of the tooth there is a short invagination of the posterior wall in the form of a crest roughly parallel to, and between, the median and lingual cusp rows. This crest is present also in UA 8799, from the Ravenscrag of Saskatchewan (but this tooth is closer in shape to that of C. hazelae than to ROM 05624). Krishtalka ( 1973 ) referred a sample from the Ravenscrag of Alberta to Carpodaptes, ef. C. hazelae. In size and P4 morphology, the specimens are more like C. cygneus than C. hazelae, and are here designated Carpodaptes sp., cf. C. cygneus. One of these is a P3 (UA 5857) which lacks the crest observed in ROM 05624 and UA 8799. Also allocated tentatively to C. cygneus are a few isolated teeth (UMVP collection, uncatalogued ) from Tiffanian deposits of the upper Tongue River Formation at Olive, Montana. R. C. Holtzman (personal communica- tion) has recently collected specimens of Carpodaptes from the lower part of the Tongue River Formation near Judson, North Dakota (UMVP Nos. 6062, 6063, 6142, 6190, 6217, 6355, 6517, and 6518). They show resemblances to both C. cygneus and C. hohackensis. UMVP Nos. 6062 and 6142 lack the prominent lingual ridge be- neath the last apical cusp of P4, a char- acteristic of C. cygneus, consequently showing the posterolingual hollowing typi- cal of C. hohackensis. On the other hand, UMVP Nos. 6063 and 6190, in which the ridge is well developed in P4, are like C. cygneus. The Judson sample may represent the sympatric occurrence of the two spe- cies, but suggests that they may be synony- mous. Description of the specimens is in preparation by Holtzman. Carpodaptes jepseni, new species Figures 11B, 21 Etymology: Named for the late Professor Glenn Lowell Jepsen, in recognition of his outstanding contributions to our knowledge of Paleocene mammals, and in particular for his efforts in the recovery of the majority of known carpo- lestid specimens. Holotype: PU 20716, right mandible with P4-M2. Hypodignv Type specimen only. Horizon and Locality: Late Paleocene (Tif- fanian), Polecat Bench Formation: Divide Quarry, NE 1/4, SW 1/4, Sect. 16, T. 54 N, R. 95 W., Big Horn County, Wyoming. Diagnosis: Largest known species of Carpodaptes, about the size of Carpolestes nigridens and duhius. P4 very high crowned, bearing 6 well-developed apical cusps fol- lowed by lower, separate talonid heel; last apical cusp lower but better developed than those anterior to it, as in C. cygneus; talonid lower than trigonid of Mi, and distinctly separated from main blade as in C. hazelae and C. aulacodon; lateral profile close to that of C. aulacodon, approaching a more pointed form than in other species. Para- conid of Mi only slightly subordinate to protoconid. Molars longer and broader than in other species. Discussion: This problematical speci- men shows features which resemble both Carpodaptes and Carpolestes, but it does not fit very conveniently into either. The balance of features, however, seems to weigh in favor of Carpodaptes, to which it is here referred. At present, this is pref- erable to a proposal of a new genus. The number and clarity of the apical cusps of P4 are characteristic of Carpo- daptes, as are the well-developed sixth apical cusp and the distinctly lower and well separated talonid cusp. These features are not found in Carpolestes. On the other hand, the large size, high crown, and the relative size of P4 to the molars are much more like Carpolestes. The species could therefore be regarded either as a less specialized representative of Carpolestes or as a large, relatively advanced representa- tive of Carpodaptes. The age of Divide Quany is not yet well established, but it appears to be later than 34 Bulletin Museum of Comparative Zoology, Vol. 147, No. 1 <+J Figure 21. Carpodaptes jepseni, new species, holotype, PU 20716, right dentary Crown view (above) and medial view (below). ■ 6. with P4-M Cedar Point Quarry, which yields Carpo- daptes hazelae, and earlier than Princeton Quarry, which yields Carpolestes dubius (D. C. Parris, in litt., 2/18/72). It appears, then, that C. jepseni is both structurally and stratigraphically intermediate between Carpodaptes and Carpolestes. It is probably in or very near the ancestry of Carpolestes. The posterior portion of a P4 (PU 20852) from Long Draw Quarry, Carbon County, Montana, is possibly referable to C. jepseni. Carpodaptes sp. Figure 10 A maxillary fragment (PU 14639) from Princeton Locality 11 in the Crazy Moun- tain Field, Montana, preserves left P4-M2. It is close in size to Carpodaptes hazelae, but P4 is more triangular (lingually com- pressed) and relatively more extended buccolingually than in C. hazelae. In the orientation of the three lingual cusps of P4, PU 14639 resembles the Cedar Point speci- Carpolestidae • Rose 35 2.0- - % u • o o§ »A • •• V XL V • t5 ]-5- (H . •. ^bV: A . CD ▲ • % Ji • d? *• . • \ • ^ - 1.0 1 ■ ■ i © 2.0 2.5 3.0 P4 Length Figure 22. Scatter diagram of P4 dimensions of Carpodaptes. Black circles = C. hazelae from Cedar Point Quarry. Open circles = C. hazelae from Scarritt Quarry. Black squares = C. hobackensis. Open triangles = C. cygneus from Swan Hills. Black triangles = C, cf. C. cygneus from Cypress Hills, Alberta. A = C. aulacodon, holotype. Circled J = C. jepseni, holotype. mens more closely than those of Scarritt Quarry. Locality 11 (Simpson, 1937b: 41) is approximately 1000 feet stratigraphically above the Scarritt Quarry, thus it is possible that this specimen represents a hitherto un- described species. Bell (1941), in an un- published thesis, based a new species on PU 14639. In view of the small number of maxillary dentitions known for this genus, and the high degree of intraspecific vari- ability present in large samples (of lower teeth, for example) it seems more appro- priate for the present to refer this to Carpo- daptes sp. than to propose a new taxon. 36 Bulletin Museum of Comparative Zoology, Vol. 147, No. 1 Carpolestes Simpson, 1928 Carpolestes Simpson, 1928: 7. Litotherium Simpson, 1929: 9. Type Species: Carpolestes nigridens Simpson, 1928 Type Species of Synonym: Litotherium compli- catum Simpson, 1929. Included Species: C. nigridens and C. dubius. Distribution: Latest Paleocene ( "Clarkf orkian" ) and earliest Eocene ( Wasatchian ) of Montana and Wyoming. Emended Diagnosis: Largest, most specialized carpolestids. Dental formula ?2.?1.?3.3 . . , _ , — . Anterior lower dentition much 2. 1. 2.3 as in Carpodaptes. Enlarged medial incisor anteriorly inclined, long, slender. Alveolus of I- indicating very reduced, peglike tooth. Canine and P3 tiny, single-rooted, crowns buttonlike; crown of C with slight anterior overhanging projection. P4 bladelike, very high crowned, relatively very large ( almost twice as long as Mi); apex considerably higher than any other tooth; enamel of buccal side extending downward much below« enamel of any other tooth, and relatively lower than in Carpodaptes. 8 to 9 apical cuspules closely spaced, varying in expression from tiny and sharp to nearly indistinguishable traces; talonid merging with main blade, height equaling that of trigonid of Mi; latter forming continuation of P4 blade. Trigonid of MT with paraconid situated directly anterior to protoconid, both cusps equally prominent. M2-3 slightly larger than in other two genera, sometimes with incipient mesoconid. At least 3 teeth present anterior to P2. P2 reduced, single-rooted, with one main cusp, as in Carpodaptes. P3-4 polycuspate, greatly enlarged, noticeably larger than molars; cuspule development as in Carpodaptes, but teeth larger, relatively longer; P3 larger than P4, with variable anteroexternal extension ("spur"). Upper molars larger but otherwise similar to those of Carpodaptes. Mandible deeper than in other two genera. Discussion: Carpolestes is well repre- sented by more than 80 specimens, many of them unusually complete. All of the lower teeth except I2 are known, and many more upper dentitions of this genus are preserved than of the other two combined. One specimen (PU 17978) includes the root or alveolus of each of three teeth anterior to P2, presumed to have been two incisors and a canine (see p. 37). Two nearly complete palates (PU Nos. 17709, 19422) of Carpolestes dubius reveal the presence of palatal vacuities. They may have been characteristic of all carpolestids, but no palatal evidence is available for the other two genera. There are no palatal vacuities in Plesiadapis. Szalay (1972b) proposed that Carpolestes be regarded as a junior synonym of Carpodaptes on the premise that the num- ber of apical cusps of P4 might not be a valid generic criterion. From the descrip- tions and discussions presented above, it is clear that this criterion (and others) are real distinctions separating the late Paleo- cene carpolestids into two groups which deserve generic rank. As noted above, Carpolestes is more specialized in several features than Carpodaptes. Carpolestes nigridens Simpson, 1928 Figures 23, 27A, 27B Carpolestes nigridens Simpson, 1928: 7. Litotherium complicatum Simpson, 1929: 10. Carpolestes aquilae Simpson, 1929: 10. Holotype: AMNH 22159, right mandible with Pi-Ms. Holotype of L. complicatum: AMNH 22196, left P3. Holotype of C. aquilae: AMNH 22233, right mandible with P4-M3. Hypodigm: Types and AMNH Nos. 22187- 22190, CM Nos. 11518, 11524, 11549, 11557, 11665, 11702, and PU No. 20718, from Bear Creek; PU Nos. 17978, 19543, from "Clark Fork" beds, Bighorn Basin; and YPM Nos. 24614 and 24615, from Paint Creek. Carpolestidae • Rose 37 Horizon and Locality: latest Paleocene ("Clark- forkian" ) of Montana and Wyoming and earliest Eocene (Wasatchian) of Wyoming: Eagle Coal Mine, Beer Creek, Montana (Fort Union For- mation); Paint Creek Locality and other sites in the Bighorn Basin, northwestern Wyoming (Polecat Bench Formation). Emended Diagnosis: P3 differing from that of C. dubius in having 5 prominent buccal cusps, with anterior one fully as developed as others, resulting in noticeably more elongate anteroexternal "spur" (pro- jection ) ; anterior 2 cusps slightly separated from posterior 3. Discussion: Simpson (1929: 10) diag- nosed C. aquilae as follows: "Length P4-M2, 6.2 mm. P4 about 10 % smaller than in C. nigridens and slightly smaller relative to the molars. Molars relatively narrower basally." It is now almost certain that the features cited by Simpson are typical of individual (intraspecific) variation, and that there are no characters (within the known, limited fossil evidence) that con- sistently separate these two species. They are therefore synonymized. Only one P3 (AMNH 22196) has been recovered from Bear Creek, the locality of the type mandible, but it is undoubtedly from the same species. Examination of all known specimens of CarpoJestes reveals no consistent differences in size or morphology between C. nigridens and C. dubius except in the form of P3 (Figs. 27, 28). AMNH 22196 has five distinct buccal cusps, the anterior two forming a pronounced antero- lateral projection. Consequently, the length of the buccal side of the tooth is nearlv twice that of the lingual side. Three other specimens (PU 17978, 19543, YPM 24614) reveal a P3 of similar form and are here referred to C. nigridens. Although some variation in the morphology of P3 does occur in C. dubius, no individual shows such a pronounced anteroexternal spur as in C. nigridens. It may be significant that all the speci- mens here allocated to C. nigridens come from sites generally regarded as later "Clarkforkian" or Wasatchian in age, while those that can be definitely referred to C. dubius are from the earliest "Clarkforkian" Princeton Quarry level (Silver Coulee beds ) . Since the one diagnostic feature of C. nigridens is a further specialization of P3 over that of C. dubius, it would be expected to occur stratigraphically higher. PU 17978 contains, in addition to P3, the root of P2, an alveolus for the canine (?), and roots of two more anterior teeth, in- terpreted as incisors. Unfortunately the maxillary-premaxillary suture is indiscern- ible, due to poor preservation. The root of the lateral incisor is round in cross section and smaller than that of the medial incisor; it is about the size of the root of P2. The root of the medial incisor is elliptical in cross section and oriented obliquely to the other anterior teeth, indicating that it is probably the most anterior tooth. The front of the specimen is crushed, but there is some indication that the left premaxilla has been superimposed onto the right incisor region, thus strengthening the supposition that the medial incisor is indeed the most anterior tooth. The crowns of the upper incisors are not preserved. Szalay (1972b: fig. 1-9) recently referred an upper (?) incisor from Bear Creek tentatively to C. nigridens. Although it is superficially similar to that of archaic primates such as Plesiadapis, the incisor is probably referable to the plagiomenid Planetetherium (Rose, 1973); consequently, the crowns of the upper incisors of carpolestids remain un- known. Carpolestes dubius Jepsen, 1930 Figures 5C, 24, 25, 27C, 27D, 29-31, 34E, 34F Carpolestes dubius Jepsen, 1930: 520. Holotype- PU 13275, right mandible with C, Pi-M3, roots of L and P3, alveolus of I2. Hypodigm: Type and PU Nos. 13276, 13305 (Jepsen's "allotype"), 14069, 14077, 14235, 38 Bulletin Museum of Comparative Zoology, Vol. 147, No. 1 cs * ,*x ■■> Figure 23. Carpolestes nigridens Simpson, holotype, AMNH 22159, right dentary with P4-M2. Crown view (above) and lateral view (below). X 6. Carpolestidae • Rose 39 14341_14347, 17646, 17647, 17707-17710, 17856-17861, 17926, 17927, 17963, 19013, 19018, 19023, 19030, 19031, 19064, 19080, 19081, 19095, 19109, 19349, 19385, 19406, 19409, 19422, 19436, 19849, 19886, 21548, 21549; MCZ Nos. 19435-19438; all from Princeton Quarry level. Horizon and Locality: latest Paleocene ( "Clark - forkian"), Polecat Bench Formation of Wyo- ming: Bighorn Basin, including Princeton Quarry, Schaff Quarry, Fritz Quarry, and other nearby sites of approximately equivalent strati- graphic level. Emended Diagnosis: P3 with four promi- nent buccal cusps, anterior one slightly separated from three succeeding cusps; incipient fifth cusp at anteroexternal edge; anteroexternal projection much less elong- ate than in C. nigridens. Discussion: As noted above, the only recognizable, consistent difference between this species and C. nigridens is in the morphology of P3 (Figs. 27, 28). The incipient anterior buccal cusp arises from a faint anteroexternal cingulum in some individuals. It is this cusp which enlarges in C. nigridens to produce the antero- external "spur." Jepsen (1930) presented a detailed de- scription of C. dubius and outlined its distinctions from C. nigridens. The features he listed do differentiate between the type specimens of the two, but when series are examined, it becomes evident that none of the features occur consistentlv. Particularlv J J variable are the prominence of vertical ridges and development of an internal cingulum on P4, the height of the trigonid in Mi, and the development of the hypo- conulid on Mi and M2. The tendency to- ward merging of the paraconid and meta- conid on M3 (as in the holotype and a few other individuals ) seems to occur as a result of anteroposterior compression of the trigo- nid. However, the trigonid is compressed but otherwise normal in other individuals, so this is not a specific feature either. The dental formula accepted here ( see diagnosis of Carpolestes) differs from Jepsen's inter- pretation (see discussion under Elphido- tarsius, cf. E. florencae). Jepsen (1930:521) presumed that P2 was two-rooted, but more complete material now available shows this to be a diminutive one-rooted tooth. He noted five cusps in the buccal row of P3. As mentioned above, the most anterior of these is usually a small, incipient cusp; it is never as prominent as the other four cusps. This is the most important distinction between C. dubius and C. nigridens; and, in fact, Jepsen did point out that in P3 of C. dubius, the anteroexternal projection is noticeably smaller than in the type of "Litotherium complicatum" ( = C. nigri- dens ) . Like the large sample of Carpodaptes hazelae, the sample now available for Carpolestes dubius offers an exceptional opportunity to observe the extent and kind of intraspecific variability which may occur in adult carpolestids. The population is essentially homogeneous in age; only adults are known (based on complete eruption of M33 and absence of specimens preserving identified deciduous or erupting teeth), and very few could be considered senes- cent. As in other species of carpolestids, there are only minor variations in the molars, which, except for Mi, differ little even interspecifically throughout the fam- ily. The specialized premolars, P4 and P3-4, display much more variability. Only one specimen (PU 14235) pre- serves the enlarged medial incisor (Figs. 5C, 24 ) . It is similar to that in Carpodaptes in being more slender (laterally com- pressed) and longer than in Elphidotarsius. In CarpoJestes the incisor is nearly twice as long from base to tip as in Elphidotarsius. Unlike Elphidotarsius, it does not broaden at the base of the crown, but is nearly as narrow (mesiodistally ) there as at the tip. The ridge which formed the lateral ( distal ) border of the dorsal face of the incisor in Elphidotarsius has migrated mesially in Carpolestes to occupy a more central posi- tion on the dorsal aspect of the tooth. As a result, what was the dorsal face in Elphidotarsius now inclines sharply toward 40 Bulletin Museum of Comparative Zoology, Vol. 147, No. 1 Figure 24. Above: Carpolestes dubius Jepsen, PU 14235, left dentary with medial incisor, M,_2, crown view. X 6. Below: Carpolestes dubius Jepsen, PU 19409, left dentary with C-M3, medial view. X 6. the midline in Carpolestes. At the base of There are no evident wear facets on this the crown, the ridge meets a weakly-de- incisor. veloped internal cingulum. The incisor is A few specimens preserve the reduced curved in two planes: dorsally at the tip anterior teeth (L», C, P3) which are all of the crown (more so than in Elphido- small, peglike, and single-rooted, as in tarsius), and toward the midline at the tip. Carpodaptes hobackensis. I2 is often dis- Carpolestidae • Rose 41 Figure 25. Carpolestes dubius Jepsen, PU 19409, left dentary with C-M3. (below). Schaff Quarry. X 6. Crown view (above), lateral view placed buccally (as deduced from the alveolus) in relation to the other anterior teeth (PU Nos. 17860, 19409). The enamel of the crown of the canine overhangs an- teriorly but in P3 this tendency is reduced (PU 17857) or lacking (PU 19409). P4 shows the greatest range of variation of any of the teeth. Size varies considerably 42 Bulletin Museum of Comparative Zoology, Vol. 147, No. 1 2.5- - © ■ ■ readth b ■ • ■ jP ■ ■■ ■ :■ DO 0? • ■«ft. ▲ • A A 1.6 i » i 1 -4. , . . — i 1 < 2.6 3.0 3.5 P4 Length Figure 26. Scatter diagram of P4 dimensions of Carpolestes. Black squares = C. dubius. Black circles C. nigridens. Black triangles = Carpolestes sp. Circled J = Carpodaptes jepseni, holotype. (see table). The apical cuspules may be sharp and distinct (PU Nos. 19409, 21549) or rather indistinct (holotype); their definition does not appear to be necessarily a function of the degree of wear. They are followed by a talonid cusp which smoothly continues the blade. There are usually eight apical cuspules, the eighth separated somewhat from the preceding seven and only slightly higher than the most anterior one. Some individuals show a faint indi- cation of a ninth cuspule just anterior to the talonid (MCZ 19437). On the other hand, only seven apical cusps are present in PU 21549. The homologies of the apical cusps are not at all obvious, as they were in Elphidotarsius, but it is likely that prolif- eration of apical cuspules has occurred primarily at the front of P4 (i.e., the original trigonid cusps are now situated at the end of the apical cusp sequence). A vertical ridge is present beneath each apical cusp on the lingual side. These ridges vary in distinction but are usually correlated with the degree of definition of the apical cusps. The ridge beneath the eighth cusp is al- ways faint since this part of the tooth is concave, forming a posterointernal de- pression. This excavation is always present but is variable in depth. As noted above, the sinuous shape of the crest of P4 ( viewed occlusally) is due in part to this depression. Carpolestidae • Rose 43 Vertical ridges may be present buccally as well, but are always much less distinct. Crown height of P4 is variable, and like cusp definition, does not appear to be necessarily correlated with age or wear. P4-M3 are exodaenodont (see above, p. 11), particularly the buccal side of P4. The enamel above the anterior root of P4 extends lower than that above the posterior root in some individuals (PU 19409) a condition more pronounced lingually than buccally in PU 19886. In others (e.g. PU 19030) the enamel over the posterior root descends lower. The development of an internal cingulum on P4 is variable. Some speci- mens, such as the holotype, have no internal cingulum (as noted by Jepsen, 1930). Jepsen apparently regarded the absence of an internal cingulum as a specific feature of C. dubius, but his paratype, PU 13284, has a weakly-expressed internal cingulum. It should be noted, however, that PU 13284 is recorded as coming from upper "Clark- forkian" beds, so it may represent C. nigri- dens. Other specimens that are undoubtedly associated with upper teeth referable to C. dubius have a basal cingulum extending across the entire lingual surface (PU Nos. 19031, 21549, MCZ 19438). The internal cingulum, when present, is usually better developed on the posterior half of the tooth (PU 17710, 19385). It may be developed anteriorly and posteriorly, but less so or not at all in the middle of the tooth (PU 17860). P4 varies also in shape at the base of the crown (viewed occlusally). The shape is roughly quadrate but slightly later- ally constricted at the midsection (between the roots ) . In some specimens, the posterior half of the tooth is more laterally com- pressed than the anterior half (PU 19409); in others the anterior half is more com- pressed (PU 17861). Some specimens of P4 are relatively quite narrow buccolingu- ally ( PU 17861 ) . These features have been cited to illustrate the exceptional variability of P4 in C. dubius. Although most other samples of carpolestid species are con- siderably smaller, it appears that similar Figure 27. Comparison of P3 in Carpolestes, to same scale. C. nigridens: A) PU 19543, left P3; B) YPM 24614, right P3. C. dubius: C) PU 19349, left P3; D) PU 14077, right P3. variation is typical of carpolestids in general. The lower molars of C. dubius are much less variable than P4. Variation in size is not so apparent as in P4. A faint external cingulum develops on the molars of some individuals, but it is never very distinct and is often absent. A weak internal cingulum may develop on the trigonid of Mi (holotype and PU 17710), but this, too, is often absent. The trigonid of Mi and the talonid of P4 are invariably the same height. (In PU 19409 [see Figs. 24, 25] the talonid of P4 appears to be lower than the trigonid of M4; this is due to breakage and subsequent distortion of this speci- men.) The posterior edge of P4 and the anterior edge of M4 are compacted, forming a continuous blade from the anterior edge of P4 to the protoconid of Mi. At least fifteen specimens of upper denti- tions (see Figs. 29, 30, 34F) are referable to C. dubius, more than are known in any other carpolestid. The most anterior tooth known is the diminutive P2. As afore- mentioned, Jepsen (1930) thought P2 was two-rooted, but specimens now known ( PU 44 Bulletin Museum of Comparative Zoology, Vol. 147, No. 1 Table 5. Metrical data for lower cheek teeth of Carpolestes dubws from Princeton Quarry LEVEL. X OR (mm) X s V length 26 3.0-3.6 3.32 ± .03 .172 ± .024 5.2 ± .7 P4 breadth 27 1.8-2.3 2.01 ± .02 .120 ± .016 6.0 ± .8 Mi length 23 1.6-2.0 1.87 ± .02 .105 ± .015 5.6 ± .8 breadth 23 1.5-2.1 1.67 ± .03 .150 ± .022 9.0 ± 1.3 Mo length 15 1.4-1.7 1.54 ± .02 .074 ± .014 4.8 ± .9 breadth 15 1.4-1.7 1.54 ± .02 .074 ± .014 4.8 ± .9 M3 length 6 2.0-2.3 2.13 ± .05 .121 ± .035 5.7 ± 1.6 breadth 6 1.3-1.4 1.38 ± .02 .041 ± .012 3.0 ± .9 Mandibulai depth 7 4.0-5.5 4.87 ± .18 .489 ± .131 10.0 ± 2.7 Nos. 19349, 19422) reveal that it is a one- rooted tooth with one main cusp. A cingu- him, faint buccally but well developed lingually, surrounds it; this rises at the anterior and posterior edges of the tooth, posteriorly giving the suggestion of a small second cusp. Little can be added to Jepsen's (1930: 521-522) accurate and detailed description of P3 and P4. Compared to P4, the morphol- ogy of P3 and P4 is relatively constant. Jcpsen described P3 as having five cusps in the external row, but as already noted, the first of these is incipient and never ap- proaches the magnitude achieved in C. nigridens. In PU 13305, on which Jepsen based his description, this cusp is larger than in most other specimens. The cusp is represented only by a slight rise of the anteroexternal cingulum in some others "D 03 CD 00 CO CL 2.5- 2.0 -1 1- -1 1 1 1 r- 2.5 3.0 P3 Length Figure 28. Scatter diagram of P3 dimensions of Carpolestes. Black circles = C. dubius. Open triangles = C. nigridens. Carpolestidae • Rose 45 Figure 29. Carpolestes dubius Jepsen. Above: PL) 19349, left maxilla with P2-M3, crown view. Below: PU 14077, right maxilla with P3-M3, crown view. X 6. 46 Bulletin Museum of Comparative Zoology, Vol. 147, No. Figure 30. Carpolestes dubius Jepsen, PL) 17709, palate. Above: right P3-M3, crown view. X 6. Below: left P3-M3, crown view. X 6. (PU 17709, 19439), suggesting that it cusps of the medial row in both P3 and P4 originated as a stylar cusp. P4 may show are difficult to distinguish; this row in both a small sixth cusp at the posterior end of teeth is more of a "rugose crest", as noted the external cusp row (PU 17709). The by Jepsen. In P3 the anterior part of this Carpolestidae • Rose 47 Table 6. Metrical data for upper cheek teeth of Carpolestes dubws from Princeton Quarry level. N OR (mm) X s V P3 length breadth 14 14 2.4-2.7 2.2-2.7 2.51 ± .03 2.41 ± .04 .107 ± .020 .133 ± .025 4.3 ± .8 5.5 ± 1.0 P1 length breadth 12 13 2.0-2.4 1.9-2.5 2.23 ± .03 2.17 ± .04 .087 ± .018 .144 ± .028 3.9 ± .8 6.6 ± 1.3 M1 length breadth 9 9 1.5-1.8 2.1-2.5 1.58 ± .03 2.27 ± .05 .097 ± .023 .158 ± .037 6.2 ± 1.5 7.0 ± 1.6 M2 length breadth 8 8 1.4-1.5 2.0-2.3 1.49 ± .01 2.19 ± .05 .035 ± .009 .136 ± .034 2.4 ± .6 6.2 ± 1.6 M3 length breadth 6 6 1.3-1.4 1.9-2.2 1.32 ± .02 2.02 ± .04 .041 ± .012 .098 ± .028 ' 3.1 ± .9 4.9 ±1.4 crest bears a distinct, well-developed cusp, and wear patterns suggest the presence of perhaps three additional cusps on this crest (PU 19349). In P4 the only recognizable cusp on the medial crest is approximately at the center of the crest. A faint external cingulum may develop posteriorly on P4. The enamel of both P3 and P4 is crenulated, particularly on the buccal side of each tooth and on the lingual faces of the external and medial cusp rows. P3 is always larger — both broader and longer — than P4. A strong external cingulum is present on the molars. Stylar cusps are extremely faint, if present at all. An incipient parastyle is present in several specimens, and a small mesostyle can be detected on M1 of PU 19349. Anterior and posterior cingula are usually present but less developed than the external cingulum. A faint internal cingulum may unite the anterior and posterior cingula (PU 17709). The hypo- cone, originating from the posterior cingu- lum, is usually well expressed, although decreasing in size posteriorly. The anterior cingulum in some individuals (PU 19349) gives rise to a vague, incipient pericone (referred to by Jepsen [1930: 522] as an ectocone); this is rarely detectable except on M1. One specimen (PU 19422, a snout with most of the palatal dentition, see Fig. 31) bears an anomalous small right M1 (left M1 is normal). The anomalous tooth is much narrower buccolingually than the other molars, its paracone and metacone aligning with the conules of M2 and M3. It lacks a hypocone. The paraconule is well developed, but the metaconule is indistinct. This morphology is strikingly similar to that of dP4 of Plesiadapis (Simpson, 1935a: 6). The presence in PU 19422 of the fully erupted, complete adult dentition, includ- ing right and left P3, P4, and M3, seems to preclude any possibility that this repre- sents a deciduous tooth, however. When Carpolestes dubius was described, PU 13305 was the only known maxillary fragment. Although several more specimens are now available, only a few add to our knowledge of palatal and skull anatomy. A partially crushed snout (PU 19422, see Figs. 31 and 32) reveals several details, but much of the structure remains vague as a result of post mortem distortion and the exceptionally hard matrix. The features discussed below are based primarily on this specimen, supplemented by evidence pro- vided from PU Nos. 17709 (a partial palate), 13305, 14077, and 17927. The maxillae contribute significantly to the face, apparently relatively more so than in Plesiadapis. A large infraorbital foramen opens above the posterior root of P3, or more often slightly posterior to that point. As noted by Jepsen (1930), the jugal process 48 Bulletin Museum of Comparative Zoology, Vol. 147, No. 1 V\ * c o ir > i? to a. 3 o c CM Z> Dl C 01 CO Q. CD CO ,3 ■Q ■Q CO Q) CO a> o Q. *-. ca O CO CD CD X 3 Carpolestidae • Rose 49 1 cm B Figure 32. Tentative reconstructions of A) snout, and B) palate of Carpolestes, based primarily on PL) Nos. 19422, 17709, 17978, and 13305. ef = emissary foramen. F = frontal, if = infraorbital foramen. J = jugal. M = maxilla. N = nasal. P = palatine, pf = palatal fenestrae. Pm = premaxilla. of the maxilla emerges approximately above M2. The jugal portion of the zygomatic arch is preserved in PU 19422 and is robust, as in Plesiadapis. The suture between the maxilla and the jugal is nearly horizontal (but slightly more dorsal anteriorly) and extends approximately from above M2 to above the anterior part of P4. Matrix ob- scures the extent of the lacrimal. The premaxillae, missing anteriorly in PU 19422, are relatively small, contrasting markedly with Plesiadapis (Russell, 1959, 1964), where they are very large and form a major part of the snout. A small (emissary?) foramen is present anterolaterally in each frontal, near but just outside the orbital margin. Jepsen (1930) has noted the peculiar palatal fenestrae in this species. The new material collected since his description unfortunately does not reveal more pre- cisely the position or extent of these fenes- trae. Their occurrence in carpolestids is unique among primates, and their function is unclear. One pair was situated approxi- mately internal to P3, bounded at least laterally by the maxillae. There is some indication of a more posterior pair which may or may not have been connected with the anterior pair. These are internal to M1 or M2 and appear to be within the palatine; they may possibly be enlarged posterior palatine foramina. The specimens here referred to C. dubius are all from Princeton Quarry and nearby sites of equivalent stratigraphic level, and only specimens from this earliest "Clark- forkian" horizon are included in the hypo- digm. All referred lower dentitions are from localities at which they are associated with maxillary dentitions containing P:{ diagnostic of C. dubius. P3 with C. nigri- dens morphology has not been found at this horizon. Separation of the two species 50 Bulletin Museum of Comparative Zoology, Vol. 147, No. 1 1 cm Figure 33. Tentative reconstruction of the skull of Carpolestes, based on PU Nos. 14235, 19409, 19422, 14077, 17978, and 19349. Angular process of mandible restored from PU 21399 (Carpodaptes hazelae); upper canine restored from AMNH 33980 {Carpodaptes hazelae). on stratigraphic criteria eventually may prove justifiable, but is precarious at present. C. dubius is surely the immediate ances- tor of C. nigridens. Future discoveries may reveal that P3 morphology is variable and not taxonomically significant, and that C. dubius should be placed in synonymy. As long as the two species are maintained, however, mandibular dentitions of Carpo- lestes will be assignable to species only if associated with P3. Consequently, about a dozen specimens, including those from Buckman Hollow (USNM 21280) and Togwotee Pass (AMNH 88198), and several from the Big Horn Basin, must, for the present, be referred to Carpolestes, species indeterminate. A tentative reconstruction of the skull of Carpolestes, based primarily on specimens of C. dubius, is presented in Figure 33. The Phyletic Position of Saxonella D. E. Russell (1964) described Saxonella crepaturae, from the Middle Paleocene of Walbeck, Germany, and referred it to the Carpolestidae. Differences between the new form and the North American carpo- lestids compelled Russell to propose two subfamilies, the Carpolestinae, for the North American genera, and the Saxonel- Carpolestidae • Rose 51 linae, for the new genus. Like the North "plagiaulacoid" (Abel, 1931; Simpson, American forms, Saxonella has a "plagiaula- 1933) in allusion to its occurrence in coid" dental complex in the mandible, but members of the multituberculate suborder the premolar specialization is not the same Plagiaulacoidea. The complex is present as in carpolestids. In Saxonella, P3 is en- in several other mammalian groups, includ- larged and trenchant and P4 relatively ing ptilodontoid multituberculates, caeno- small. Such a difference is fundamental lestoid marsupials (Simpson, 1933; Paula and precludes the possibility of very close Couto, 1952), phalangeroid marsupials, and relationship between Saxonella and carpo- the plesiadapoid Saxonella. Simpson (1933) lestids. Furthermore, the mandibular dental concluded that plagiaulacoidy was acquired formula, 1.0.2.3, is reduced beyond that in independently at least four times: probably even the most advanced carpolestids, al- once in multituberculates, twice or more in though Saxonella appears to have been marsupials, and once in primates ( Saxon- contemporary with the earliest known ella then being unknown). The differing North American species of the family. homologies of the sectorial tooth in Russell's assignment of Saxonella to the Saxonella and carpolestids make it virtually Carpolestidae was, nevertheless, accepted certain that two independent origins oc- by some authors (e.g., Romer, 1966; Mc- curred in primates. Kenna, 1967). The bladelike specialization in these Some features (e.g., molar and incisor mammals may involve more than one tooth, morphology, form of ?P4, and mandibular as in plagiaulacids such as Ctenacodon, dental formula) are equally or more sug- but is usually restricted to one. In the latter gestive of plesiadapid affinities, although case it is not always the same tooth: in the dental reduction in Saxonella exceeds Saxonella it is P3, not P4 as in carpolestids; even that in contemporary plesiadapids. in caenolestoids it is P3 or Mi, and in Moreover, Russell was surely correct in Australian marsupials it is P8 ( the last pre- suggesting that other early primates had molar). In ptilodontid multituberculates upper incisors of comparable structure. Van the blade has generally been assumed to be Valen (1969) formally transferred Saxonella P4, but there is some evidence to suggest from the Carpolestidae to the Plesiadapidae, that it is actually the first molar ( Bohlin, a view followed by Szalay (1970, 1972b), 1945; Sloan, Kielan-Jaworowska, personal Simons (1972), and Butler (1973). communication). The morphology of Saxonella points to The adaptive significance of the "pla- affinities with both carpolestids and plesia- giaulacoid" dentition may be generally dapids, but its specializations are compa- similar in the varied types possessing it, but rable in magnitude, although different dissimilar upper dentitions in most of these from, those of either family. Accordingly, mammals suggest that the function is vari- Saxonella should be separated from both able. Simpson (1933) concluded that it is at the family level, as the Saxonellidae, but an adaptation for herbivory, especially effi- grouped with them in the superfamily cient for dealing with coarse vegetation. Plesiadapoidea. The only extant mammals with the "plagiaulacoid" dentition are certain macro- MORPHOLOGY AND FUNCTION podid marsupials (e.g., Hypsiprymnodon, _, .,_ „ . . Dorcopsis, Aepypnimnus, Bettongia) and The "Plagiau aco d" Dent t on fl u , ' .Jn A . ,. \ a the phalangend Burramys, Australian forms The lower dentition in carpolestids is in which the sectorial lower tooth is op- characterized by a hypertrophied, trenchant posed by a similar trenchant tooth in the P4, an enlarged medial incisor, and reduced upper jaw. This complex is quite different dentition between these, a pattern termed from that in carpolestids, and inferences of Bulletin Museum of Comparative Zoology, Vol. 147, No. 1 carpolestid dental function or diet by analogy with these "plagiaulacoid" mar- supials are probably poorly founded. Furthermore, although most of these mar- supials feed on vegetation such as fruits, berries, and leaves, at least one form (Hypsiprymnodon) is known to ingest sub- stantial quantities of insects and worms (Walker et al., 1968). Simpson (1933) contrasted the unusual upper premolars of Carpolestes with the upper teeth of other "plagiaulacoid" mammals and speculated that carpolestids may have been more frugivorous than extant "plagiaulacoid" marsupials. The upper premolars of some ptilodon- toid multituberculates show superficial similarities to those of advanced carpoles- tids, but there are indications that multi- tuberculate dentitions functioned differently than those of carpolestids. Almost certainly the sectorial lower tooth of multitubercu- lates was more directly involved with powerful shearing than appears to have been the case in carpolestids.1 Carpolestids and multituberculates may have been to some extent mutually competitive, but dental resemblances between the two may indicate only that both were specialized herbivores. The upper teeth of Saxonella are less specialized than those of Carpodaptes and Carpolestes, but not unlike those of Elphidotarsius; and it may be postulated that dental function in Elphidotarsius and Saxonella was similar. Major Features and Trends in the Carpolestid Dentition and Skull The lower dental formula is 2.1.3.3 in Elphidotarsius and 2.1.2.3. in Carpodaptes and Carpolestes. Elphidotarsius has many ^Iahn (1971) illustrated teeth of the Upper Jurassic paulchoffatiid multituberculates that demonstrate that their premolars were not in- volved in shearing; they show considerable abra- sive wear and erosion at the apices of the cusps, somewhat as in carpolestid premolars (see below). similarities to the contemporary plesiadapid Pronothodectes, but its P4, enlarged and bladelike, is characteristically carpolestid. It retains three lower premolars, like Pronothodectes. P2, judging from its alveolus, was a small, single-rooted, peglike tooth. P3 is small relative to P4 but is comparatively unreduced and premolari- form. In Carpodaptes and Carpolestes, the mandibular dentition is similar to that of Elphidotarsius, but further specialization and hypertrophy of P4 has resulted in the loss of P2 and the simplification of P3 into a single-rooted peg. Upper dentitions of carpolestids (Fig. 34), as for many other fossil mammals, are much less well represented in collections than are mandibles. Upper teeth or partial dentitions are known for Elphidotarsius cf. florencae, Carpodaptes hazelae, C. cygneus, Carpolestes nigridens, and C. dubius, the latter species being the best known. The most anterior upper teeth are unknown in Elphidotarsius and Carpodaptes. Speci- mens of Carpolestes suggest an upper dental formula of 2.1.3.3 in this genus. The maxillary-premaxillary suture is poorly preserved and unclear in all specimens pre- serving this region, but its apparent position is consistent with the interpretation of the dental formula given here. A number of features of carpolestids ob- served to be more fully developed in the later, more specialized species may be regarded as trends. For example, size tends to increase through time. Carpodaptes is generally larger than Elphidotarsius, and Carpolestes is always larger than all but one of the species (Carpodaptes jepseni) of the other two genera. While this trend involves primarily the cheek teeth, P4 and pa-4 are particularly affected. Some ex- ceptions have already been noted: Carpo- daptes hobackensis and C. cygneus are smaller than the known species of Elphido- tarsius and may be representative of a side branch of the genus which was decreasing in size; Carpodaptes jepseni is fully as large as Carpolestes. Correlated with the gradual Carpolestidae • Rose 53 B 5mm Figure 34. Lower cheek teeth (P4-M3) and upper cheek teeth (P3-M3) of carpolestids, to same scale. A and B) Elphidotarsius (based on £., cf. E. florencae). C and D) Carpodaptes (based on C. hazelae). E and F) Carpo- lestes (based on C. dubius). increase in size of the teeth is increasing mandibular depth to accommodate the lengthening roots, particularly of P4. The mandible of Elphidotarsius is relatively shallow (2.8-3.5 mm); that of Carpodaptes is somewhat deeper (3.1-4.0 mm; 5.5 mm in C. jepseni); and the jaw of CarpoJestes is comparatively deep (4.0-5.5 mm). There is also a trend toward lengthening and slenderizing the enlarged medial incisor. Associated with size increase is the in- creasing specialization of P4 and P3-4 (see Fig. 34). The enlargement and special- ization of these teeth occurred, at least in part, at the expense of the anterior teeth (except the medial incisors), which underwent reduction through time. Special- ization was accomplished by the addition of accessory cusps, with the result that in the two more specialized genera cusp homologies are not entirely clear. As in most Paleocene primates, the antemolar modifications are more diagnostic than are the molars, which remain relatively conservative (paromomyids such as Phena- colemur and Micromomys [Szalay, 1973] have an enlarged P4, reduced anterior teeth except the incisor, and relatively conserv- ative molars). Specialization of P4 in carpolestids was accompanied by elongation of the trigonid of M4. The trigonid cusps of Mi in Elphidotarsius are arranged in a triangle, but in Carpodaptes and Carpolestes the paraconid and protoconid are drawn out, nearly or exactly in line, and colinear with the cusps of P4. The lengthening of the trigonid of Mi occurs at the expense of the length of the talonid and is relatively more marked in Carpolestes than in Carpodaptes. M3 tends to enlarge in the sequence El- phidotarsius-Carpodaptes-Carpolestes, par- 54 Bulletin Museum of Comparative Zoology, Vol. 147, No. 1 ticularly by elongation of the third lobe, a small number of individuals are in ad- The hypoconulid of M3 is typically vanced stages of wear. twinned, and there is often a small meso- The most conspicuous trend in mandibu- conid on the cristid obliqua in later forms, lar morphology is increasing depth, associ- Breadth of P4 and of the lower molars ated with the lengthening of the roots of shows a tendency to increase through time, P4- The mental foramen is variable in but this feature is among the most variable position beneath P2 or P3 in Elphidotarsius intraspecifically and hence of little taxo- and below C or P3 in Carpodaptes and nomic value. Carpolestes. As in most prosimians, the The clearest trend in the upper dentition, symphysis was unfused in carpolestids. The as noted above, is toward specialization and posterior part of the mandible is preserved hypertrophy of P3~4 (associated with the in near entirety only in a few specimens of enlargement of P4). In Elphidotarsius, these Carpodaptes hazelae (PU 20615, 21341, teeth are smaller than the molars (or P4 21399; see Fig. 35). It is closely compa- may be about the size of the molars), but rable to the back of the dentary in Plesi- in Carpodaptes and Carpolestes they are adapis (PU 21246). The coronoid process noticeably larger than the molars. P3 is high and broad. The angular process is changes the most. It is a small, relatively recurved and tapers posteriorly, forming a generalized, premolariform tooth in Elphi- hooklike process much like the angle in dotarsius. In Carpodaptes it is already other archaic primates such as Plesiadapis subequal to P4, while in Carpolestes it is and Phenacolemur, and in such Recent always larger than P4 and may display a insectivorans as Ptilocercus, Setifer, and conspicuous anteroexternal projection. P4 Neotetragus. It serves as a point of insertion is somewhat specialized in Elphidotarsius, for the internal pterygoid and part of the already at this stage longitudinally tri- masseter. The condyle in Carpodaptes, as partite. These three sections are much in Plesiadapis, is situated about midway more complex in Carpodaptes and Carpo- between the angular process and the top lestes. The latter two genera have a closely of the coronoid, about even with or just similar P4, the main distinction being the slightly above the tooth row. In Phena- greater anteroposterior length in Carpo- colemur (PU 21405) the condyle is markedly lestes. Both P3 and P4 in the two later higher than the tooth row. (This feature, genera are polycuspate, with numerous together with dental morphology, suggests accessory cusps. that Phenacolemur may have been more Intraspecific variability of the teeth has strictly herbivorous than either Plesiadapis been discussed in the systematics section. or the carpolestids.) The articular surface The much larger samples of carpolestids .c c l ^ ' ^, P . ^ . ., , , ., ,, , specimen (PU 21399), covers the entire now available provide a reasonably ade- jit.- r r . i i i r l -i t» i dorsal and posterior surfaces of the condvle; quate basis for concluding, as did Butler ^ transverse diameter is approximately' 1.6 (1963: 1), that "paleontologists have tended mm to underestimate the variability of fossil The few known details about the palate species. ' and front of the skull are based primarily It should be noted that all known speci- on specimens of Carpolestes dubius and mens of carpolestids, as far as can be have been discussed under that species in determined, represent adult individuals. In the section on systematics. Skulls of archaic no specimen is there evidence of deciduous primates are poorly known. Only that of or erupting teeth, and those preserving the Plesiadapis is reasonably well preserved posterior part of the jaw always reveal a (Russell, 1959, 1964). Phenacolemur (Simp- fully erupted M3 (or its alveolus). Only son, 1955; Szalay, 1972a) and Palaechthon Carpolestidae • Rose 55 Figure 35. Lateral view of posterior of mandible, not to same scale. A) Carpodaptes hazelae, PL) 21399. B) Plesiadapis rex, PL) 21246 (reversed). C) Phe- nacolemur sp., PL) 21405 (M3 restored). (Wilson and Szalay, 1972) are each known from one poorly preserved skull. Present evidence permits few comparisons between Carpolestes and these genera, but it is prob- ably valid to conclude that the palatal ( and preorbital) length was relatively much less in Carpolestes than in Plesiadapis, and was comparable to or slightly less than in Palaechthon. Occlusion and Dental Function Introduction. Recent cineradiography studies of living mammals have provided much information on the movement of the mandible and the function of the teeth during feeding (e.g. Ardran et al., 1958; Crompton and Hiiemae, 1970; Hiiemae and Ardran, 1968; Hiiemae and Crompton, 1971; Hiiemae and Kay, 1972, 1973; Kay and Hiiemae, 1974a, b; Ride, 1959). For example, it has been established that in the opossum, Didelphis, feeding involves four steps: ingestion, transfer of food to the molar region, mastication, and degluti- tion, sometimes including regurgitation for further preparation by the teeth (Hiiemae and Crompton, 1971 ) . Mastication, the step involving preparation of the food by the cheek teeth, can be separated into three components: the upstroke or preparatory stroke, the power stroke (occlusal or near occlusal stroke), and the downstroke or recovery stroke (Hiiemae, 1967; Hiiemae and Ardran, 1968; Crompton and Hiiemae, 1969a, 1970; Hiiemae and Crompton, 1971; Kay and Hiiemae, 1974a, b). Two stages of mastication have been observed. The first, a puncture-crushing stage involving no direct contact of the upper and lower teeth causes wear on the tips of the cusps, resulting eventually in exposure of the dentine through windows in the enamel. The term "abrasion" has been used to describe wear resulting from the puncture-crushing stage of mastication (Hiiemae and Kay, 1973; Kay and Hiiemae, 1974a). Gingerich observed striated wear facets on the posterior surface of the meta- cone and anterior surface of the protoconid in Adapis (Gingerich, 1972) and on the front of the main cusp of P4 in Phena- colemur (Gingerich, 1974a). He believes this wear to be a result of "orthal retrac- 56 Bulletin Museum of Comparative Zoology, Vol. 147, No. 1 tion", which took place during the puncture- crushing stage. The second stage of mastication is direct tooth-to-tooth occlusion during which shear- ing, crushing, and often grinding1 occurs (Crompton and Hiiemae, 1970; Kay and Hiiemae, 1974a). This results in striated wear facets on the crests which join the cusps of the teeth, and in the basins. The resulting wear has been called "attrition" (Kay and Hiiemae, 1974b). A number of recent investigations have made use of these facets in interpretations of the func- tion of molariform teeth in therian mam- mals (e.g., Butler, 1952, 1972, 1973; Crompton, 1971; Crompton and Hiiemae, 1969a, b, 1970; Gingerich, 1972, and in press; Hiiemae and Kay, 1973; Kay and Hiiemae, 1974b; Mills, 1955, 1966, 1967). The power stroke of the occlusal stage of mastication in primates has two com- ponents, each forming a set of matching occlusal wear facets on the upper and lower teeth. Mills (1955, 1967) has used the terms "buccal phase" and "lingual phase" to describe the movement of the mandible during the power stroke, and the resultant wear facets. According to him, the buccal phase on the active side coincides with the lingual phase on the opposite side, promot- ing balanced occlusion (and simultaneously formed buccal and lingual phase facets). Hiiemae and Kay (1972, 1973; and Kay and Hiiemae, 1974b), on the other hand, have shown convincingly that no such balanced occlusion occurs in the primates they studied cinefluorographically. Wear facets thus reflect successive phases of the power stroke on the same side of the jaw. Because buccal phase and lingual phase, as defined by Mills (1955), differ from ob- served mandibular movements and methods of formation of wear facets, Hiiemae and Kay (1972) have proposed the terms "Phase I" and "Phase II" to describe the successive phases of the power stroke. During Phase I 1 Precise definitions of these three terms are presented by Kay and Hiiemae ( 1974a ) . in primates, the active side of the mandible moves upward, medially, and slightly an- teriorly, until the teeth are in centric oc- clusion. The facets produced face buccally on the lower teeth, lingually on the uppers. In Phase II, the active side moves down- ward and further medially and anteriorly (in such forms as Ptilocercus and Erinaceas the jaws moves posteromedial^, rather than anteromedially, during Phase II [Mills, 1967]). Phase II facets face lingually on the lower teeth, buccally on the uppers. This phase involves a slight shift in direction from that of Phase I, which is reflected in a difference in the direction of Phase I and Phase II facets. However, wear facets with striations intermediate in direction between those of Phase I and Phase II have been observed in Plesiadapis, suggesting that the two phases in this form were parts of a single, transverse movement (Gingerich, in press). Carpolestids. Phase I and Phase II facets have been observed on upper and lower molars of carpolestids and are mapped in Figures 36-38. Facets have been numbered according to the scheme introduced by Crompton ( 1971 ) and followed by Ginge- rich ( in press ) and Kay and Hiiemae ( 1974a ) . Facets 1-7 are formed during Phase I, facets 9 and 10 during Phase II. Occluding wear surfaces on upper and lower teeth are given the same number. Molar function in all three genera of carpo- lestids was very similar to that in Plesia- dapis (Gingerich, in press). Discussion of the relative tooth movements involved in forming each facet have been detailed by Kay and Hiiemae ( 1974a ) and Gingerich (in press). Orthal retraction facets have not been detected in carpolestids. Lower Teeth (Figs. 36, 37). All seven Phase I facets occur on one or more of the molars of each genus. Facets 1 and 5, on the posterior surface of the trigonid, differ markedly in size on the different molars. An oblique ridge (formed by the contact of the cristid oblique with the protocristid ) Carpolestidae • Rose 57 Figure 36. Occlusal wear facets of lower teeth, P4-M3. Vertical lines denote Phase I facets; stippling denotes Phase II facets. Numbering of facets is after Crompton (1971) and Kay and Hiiemae (1974b). B) Carpodaptes; C) Carpolestes. A) Elphidotarsius; demarcates facet 1, on the posterior surface of the protoconid, from facet 5, on the posterior surface of the metaconid. On Mi, in which the protoconid is the highest cusp, facet 1 is as large as or larger than facet 5, while on M2 and M3, where the metaconid is as high or higher than the protoconid, facet 5 is larger than facet 1. The enlarge- ment of facet 5 at the expense of facet 1 appears to be typical of many primates. Facet 2 is lengthened on Mi by virtue of the elongation of the trigonid. In some specimens an additional shearing surface parallel to facet 2 is developed on the anteroexternal cingulum (Mi or Mo), as in Plesiadapis (Gingerich, in press). Facet 6 Bulletin Museum of Comparative Zoology, Vol. 147, No. 1 Figure 37. Occlusal wear facets 1 and 5 (Phase I) in Elphi- dotarsius (posterior oblique view of right M,_3). develops on the buccal surface of the entoconid and may extend to include the hypoconulid region. This facet is most extensive and best developed on M3. Facet 7, variably developed on the buccal surface of the paraconid and metaconid, is a con- tinuation of facet 6 of the next anterior tooth (Kay and Hiiemae, 1974a) and in some cases these two facets appear to be confluent with facet 2 ( on the paracristid ) . Shearing surfaces may be increased in num- ber by fenestration of any Phase I facets; this is, of course, prevalent in individuals at advanced stages of wear. Shearing facets (Phase I) on P4 are, surprisingly, poorly developed or absent from most specimens. A few specimens of Elphidotarsius reveal small striated facets, probably homologous with facets 1 and 5, on the posterobuccal surface of the two posterior apical cusps ( the presumed homo- logues of the protoconid and metaconid). Specimens of Carpodaptcs, as a rule, show no better-developed wear surfaces than in Elphidotarsius. In Carpolestes some in- dividuals have small striated facets on the buccal surface of apical cusps on the poste- rior two-thirds of the tooth. These small facets become confluent on a few heavily- worn individuals (e.g. PU 19030; see Fig. 39D ) . A few specimens of P4 in Carpolestes display heavy abrasive wear (Fig. 39). No P4's examined show such extensive Phase I facets as indicated by Butler (1973: fig. 11). Phase II facets are much less clear than those of Phase I. Facet 9, on the lingual surface of the hypoconid, has been detected on one or more molars of each genus, but facet 10, on the lingual surface of the protoconid, has been observed only on M8 on a few (e.g. Elphidotarsius, PU 14285). This is in accord with Butler's (1973) ob- servation that Phase II facets are best de- veloped on M3 and less so anteriorly. There are no Phase II facets on P4, i.e., there is no wear on the lingual side of P4. Upper Teeth (Fig. 38). Phase I facets 1-7, matching those of the lower molars, are developed on upper molars of all three genera. As in Palenochtha (Kay and Hiiemae, 1974a) and Plesiadapis (Ginge- rich, in press), facets 1-4 are developed on the conule crests as well as on the prepara- crista, centrocrista, and postmetacrista; parallel crests are labelled "a" and "b". For example, facet la forms on the prepara- crista, lb on the preparaconule crista, parallel to the preparacrista. Facets lb, 2b, 3b, and 4b are well developed on M1 and M2; hence the conules are large. On M3 facet 2b is not developed and the metaconule is small. Facet 5, on the lingual surface of the preprotocrista, is prominent on all three Carpolestidae • Rose 59 B Figure 38. Occlusal wear facets of upper teeth (P3-M3), depicted as in Figure 36. daptes; C) Carpolestes. A) Elphidotarsius; B) Carpo- molars and in Carpodaptes and Carpolestes crest joining the protocone and hypocone). extends onto the precingulum in many in- It is confluent with facet 7 on the lingual dividuals. Facet 6 forms on the postproto- surface of the hypocone. Facet 7 was not crista and on the "nannopithex fold" (the detected in Elphidotarsius. 60 Bulletin Museum of Comparative Zoology, Vol. 147, No. 1 As in the lower molars, Phase. II facets are less evident than Phase I facets, but facet 9, on the buccal surface of the proto- cone, is relatively distinct in many speci- mens. Facet 10, on the buccal side of the hypocone, was observed only in a few individuals of Carpolestes. P3-4 are somewhat difficult to interpret, for although numerous specimens are now available, many show little or no wear. Phase I facets are developed on the lingual surfaces of the posterior third of the buccal and central crests on P4 in all three genera. These match facet 2 on the trigonid of Mi and are homologous with facets 2a and 2b on the molars. The significant wear on P3 and P4 appears to be due to abrasion, caus- ing the tips of the cusps in the buccal and central crests to become blunt and fenes- trated (Figs. 39G, H). This type of wear in Carpodaptes and Carpolestes is particu- larly evident on the posterior half of P3 and anterior half of P4, i.e., the region which occludes with P4. Small striated wear facets form on the flattened periphery of these cavitations in some specimens, and in Carpolestes they may extend to the short crests between the central and lingual row of cusps. They are not so large as depicted by Butler (1973: fig. 11), indicating that Phase I wear in P3 4, although present, is of minor importance. No phase II facets have been detected on P3 or P4 of any carpolestid. Interpretation. These observations indi- cate that the molars of carpolestids func- tioned very much like those of Plesiaclapis, but Phase II was evidently less prominent than in the latter genus and seems to have been much less important than Phase I. The molars were the significant teeth involved in shearing (Phase I) and grinding (Phase II); the specialized premolars were less effective during this stage of mastication. Shearing capability in the molars was en- hanced by increasing the number of shear- ing surfaces in the same ways as has been observed in Palenochtha and Plesiadapis. One method, called "en echelon" shear (Hiiemae and Kay, 1972), involves the development of successive shearing surfaces (e.g., in upper molars, facets la and lb, or facet 5 on both preprotocrista and pre- cingulum; in lower molars, facet 2 on both paracristid and anteroexternal cingulum; see Figs. 36, 38). Individual shearing sur- faces were multiplied by fenestration as in Plesiadapis (Gingerich, in press). Grinding, the major function in Phase II, involved all the molars but was more prominent posteriorly, as noted also by Butler ( 1973 ) . The function of the peculiar, specialized P4 and P3 4 of carpolestids is of particular interest. My conclusions differ somewhat from those of Butler ( 1973 ) . Despite its bladelike appearance, P4 was apparently not used primarily in shearing. Phase I facets observed on a few specimens indi- cate that such function occurred to a limited extent, but the poor development or abJ sence of these facets on the majority of specimens indicates that P4 was not very important during Phase I. The absence of Phase II facets indicates that P4 was in- significant in Phase II function. The premolars of many mammals are involved chiefly in the preparation of food prior to the masticatory stage involving Phase I and II. This "puncture-crushing" stage results in abrasion. P4 and P3-4 of carpolestids display mainly abrasive wear, which has resulted in erosion of the tips of the cusps and exposure of dentine in heavily worn individuals (Fig. 39). The term "puncture-crushing" seems inappropriate to describe the probable use of these teeth in carpolestids, however. P4 probably func- tioned in a sawlike manner, cutting or tear- ing food which was held partly by the polycuspate P3~4. Phase I facets in P3~4, as in P4, are poorly developed and often ab- sent, suggesting that Phase I shearing was not the primary function of these teeth. Similarly, the absence of Phase II facets on P34 indicates that their function was in- significant during this stage of mastication. Carpolestidae • Rose 61 A AMNH 33887 B AMNH 33983 AMNH 22233 AMNH 33979 Figure 39. Abrasive wear in carpolestids, exposing dentine (black areas). Vertical lines in D and E are Phase I wear facets. A-C) progressive degrees of wear in Carpodaptes hazelae. A and B are right M,_2; C is right M,_3 (crown views). D-H) progressive stages of wear in Carpolestes. D and E are right P4-M, of C. dubius (crown views); F is right P4-M, of C. nigridens (lateral view); G and H are left P3"4 of C. dubius (crown views). Dietary Inferences Even in extant mammals it is often difficult, from dental and mandibular structure, to postulate specific feeding pref- erences. In extinct forms, which often have no living analogues, inferences regarding diet must be largely speculative. Concern- ing dietary habits of Paleocene mammals in general, Van Valen and Sloan (1966: 264) have remarked: "Upon considering the diversity of food habits of recent species with teeth generally similar among the species, we are impressed with the impossibility in the foreseeable future of reconstruct- ing in any detail the diets of Paleocene mammals. When there is a phyletic trend towards a more herbivorously adapted dentition, it is probably valid to say that most of the later members of this clade were herbivorous, or more Bulletin Museum of Comparative Zoology, Vol. 147, No. herbivorous than their ancestors. There are very few Paleocene mammals of any kind for which a more detailed statement on diet is possible." While specific food preferences of ex- tinct forms may be impossible to establish, some general comments may nevertheless be offered. As noted above, it is probably invalid to base inferences regarding the diet of carpolestids on extant "plagiaula- coid" marsupials. The functional inter- pretations discussed above show that carpolestids used their molars much as plesiadapids did. Unfortunately, this pro- vides little precise insight, of course, for the diet of plesiadapids can only be speculative. Carpolestid premolars were probably used primarily during the puncture-crushing stage of mastication, suggesting that the diet of carpolestids may have consisted of herbage, fruits, and seeds tougher than those eaten by Plesiadapis. Similarities in morphology and wear patterns between carpolestid molars and those of the Recent Tarsius, a predominantly insectivorous pri- mate, further suggest that carpolestids were capable of feeding on insects and may have included them in their diet. ORIGIN, AFFINITIES, AND INTERRELATIONSHIPS OF CARPOLESTIDS Simpson (1937a: 8) remarked: "In spite of the extraordinary specialization of the premolars, the facts now known about the carpolestid dentition seem to me to point to the early primates and to no other group." I agree totally with this statement. Some early authors (e.g., Matthew and Granger, 1921; Jepsen, 1930), however, did not recognize carpolestids as definite pri- mates, and some recent students have continued to question their primate status. Saban (1961) listed them as Mammalia, Incertae seclis. Martin (1968) and Charles- Dominique and Martin (1970) have sug- gested that allocation of Plesiadapis and its allies to the Primates should be reassessed. Cartmill (1972) recently advocated the more drastic step of transferring the "archaic prosimians" from the Primates to the Insectivora, an unsatisfactory solution which does not accurately reflect the phylo- genetic relationships of these forms. Al- though the evidence now at hand indicates that some archaic prosimians did not possess the three features that Cartmill selected as diagnostic of Eocene and later primates ( postorbital bar, petrosal bulla, and oppos- able hallux or pollex with nail ) , the dental evidence points overwhelmingly to primate, not insectivoran, affinities, and it is the dentition that constitutes most of the known evidence of the archaic prosimians. Known Paleocene primates are not generally con- sidered ancestral to later ones, and it is only to be expected that they should have remained primitive in some features that later, and perhaps independently, evolved in the primate stock. If one accepts Cart- mill's three features as definitive of the Primates, the only recourse would be to establish a new order for the Paleocene forms. In our present state of knowledge, such a move would be premature. I believe that the mainly Paleocene archaic prosim- ians are definitely primates and should be grouped in the infraorder Plesiadapi- formes (Simons, 1972 )\ It has long been recognized that carpo- lestids share numerous features with mem- bers of the Plesiadapidae, and comparisons of carpolestids with all other Paleocene primates indicate that the closest affinities are with that family. Simpson (1937b: 162) noted the striking likeness of Elphidotarsius to Pronothodectes. "M2-3 are almost exactly like those of Pronothodectes, so much so that were this form known from those teeth alone it would have to be defined as a species of Pronothodectes." He later wrote (Simpson, 1940: 205): "There is a remark- ably close resemblance in the lower molars xThe suborder Plesiadapoidea Romer, 1966, is rejected here because it was proposed condition- ally, and because it is a homonym of the super- family Plesiadapoidea Trouessart. Carpolestidae • Rose 63 between carpolestids and plesiadapids, but it does not extend to all parts of the dentition and may indicate nothing more than that all are early primates." This now seems to be an overlv cautious view. There J are, in fact, marked similarities of Elphido- tarsius to Pronothodectes extending beyond the lower molars (see Fig. 41). Upper teeth of Elphidotarsius, unknown when Simpson wrote, are even closer structurally to those of Pronothodectes than are the lowers. The upper molars are virtually indistinguishable (except in size) from those of Pronothodec- tes, and P3-4 are noticeably similar in both genera. I2, C, and Po are reduced in both forms, and the dental formulae are the same as far as can be determined. These resemblances, and the lack of such close similarity to other archaic primates, argue strongly for community of origin of the two families (as Simpson [1937a] suggested), probably in the Puercan. Van Valen (1969: 295) proposed that carpolestids evolved "from Pronothodectes or just possibly some unknown more primi- tive plesiadapid." The latter alternative seems more probable, inasmuch as Prono- thodectes is not known from strata below those which contain Elphidotarsius. A new, undescribed, plesiadapoid genus from Purgatory Hill (Sloan, 1969) may shed light on this problem. As demonstrated above, derivation of Elphidotarsius from a Pronotliodectcs-Mke form would not be difficult. Virtually no modification of the molars would be involved, the trigonid of Mi being only slightly extended in Elphido- tarsius. P4 of Elphidotarsius may be termed submolariform, the apical cusps homologous to the three trigonid cusps (with the addi- tion of an anterior accessory cuspule), and a talonid cusp behind. P4 is typically premolariform in Pronothodectes. In the Puercan Purgatorius unio, it has a dis- tinct paraconid and may show enamel thickenings in the metaconid region (Clemens, 1974). Of the upper premolars, P4 has three longitudinal crests in both Elphidotarsius and Pronothodectes. P;J| in the former, however, is not divided into three longitudinal parts, as in Pronothodec- tes (and later carpolestids). The simpler P3, like that of Elphidotarsius, was probably the ancestral condition. Carpolestids and plesiadapids should be placed in a super- family Plesiadapoidea (along with Saxon- ella), to reflect their close relationship. Intergeneric relationships of carpolestids are easily interpreted. The conclusion reached here is that the three genera are members of a single lineage. All three are found in the same general area and they constitute a morphologic sequence found in successive strata. It is highly probable that Elphidotarsius gave rise to Carpodaptes, and the latter gave rise to Carpolestes ( Fig. 40B). No known evidence contradicts this hypothesis. Some authors (e.g. Schaeffer et al., 1972) advocate a much more complex interpreta- tion of phylogeny, suggesting that taxa that seem to form a direct lineage with regard to morphology and stratigraphy may actu- ally represent only a few members of a much wider radiation. The application of this concept to carpolestid phylogeny might lead to the conclusion that the three genera do not lie in a single lineage ( Fig. 40A ) . For example, the sympatric occurrence of Carpodaptes and Elphidotarsius in the Late Torrejonian Shotgun Local Fauna may suggest that there were two carpolestid lineages, one involving only Elphidotarsius, the other comprised of Carpodaptes and Carpolestes. However, Elphidotarsius is known also from Torrejonian beds almost surely older than those at Shotgun, which is the earliest occurrence of Carpodaptes. The antecedents of Carpodaptes must have passed through a stage, structurally like Elphidotarsius, that would be considered in that genus as now defined. No features of Elphidotarsius now known exclude it from a position ancestral to Carpodaptes. Even closer resemblances between Car- podaptes and Carpolestes indicate that these genera must lie in an ancestor-descen- dant line. Simpson (1937b) was of the Bulletin Museum of Comparative Zoology, Vol. 147, No. 1 U CO "X avaidvavisgid X <0 —I o Q. (X. < u 01 (D (0 ■v to a> o Q. k. TO o CO CD 'c _o >. .c a o Q. o odaptes jepseni, new species, holotype PU 20716 3.2 2.2 1.9 Carpolestes nigridens Simpson, from Bear Creek AMNH 22159 3.3 1.9 AMNH 22187 3.1 1.8 AMNH 22188 3.1 2.0 AMNH 22190 2.9 1.7 AMNH 22233 3.0 1.9 PU 20718 3.0 2.1 CM 11518 CM 11524 3.3 1.9 1.9 1.8 1.8 1.8 1.8 1.7 1.6 1.4 1.5 1.6 1.6 1.5 1.4 1.4 1.4 1.8 1.8 1.1 1.1 1.7 1.7 c.5.5 1.6 1.6 5.1 1.5 1.3 5.0 1.6 1.5 2.3 1.4 5.0 5.0 Carpolestidae • Rose 73 Table I [continued] Spec. No. P,L P4B MjL M^B MX M„B M3L M3B MD CM 11549 3.4 1.9 CM 11557 1.8 1.4 CM 11665 3.2 1.8 CM 11702 3.3 1.9 Carpolestes chibuis Jepsen, from Princeton Quarry level PU 13275 3.3 2.0 1.9 1.8 1.6 1.6 2.2 1.4 PU 13276 3.6 2.0 c.1.9 c.1.7 c.1.5 c.1.6 PU 14235 1.9 1.8 1.5 1.6 PU 14341 2.0 1.7 1.6 1.5 • PU 14342 3.4 2.1 1.9 1.6 PU 14344 3.0 2.1 1.8 1.6 1.6 1.6 4.6 PU 14345 3.2 2.0 1.7 1.5 PU 14346 3.3 1.8 1.9 1.5 1.5 1.4 PU 17646 1.8 1.5 1.4 1.5 2.3 1.4 4.0 PU 17647 1.6 1.6 1.5 1.5 PU 17710 c.3.2 c.1.9 1.8 1.5 1.5 1.5 2.0 1.3 4.9 PU 17856 1.8 1.5 PU 17857 3.3 2.0 PU 17858 c.3.6 c.2.2 2.0 2.1 PU 17860 3.6 2.0 PU 17861 3.5 2.1 1.9 1.8 c.5.0 PU 17926 3.1 2.1 2.0 1.7 1.7 1.7 4.8 PU 19013 3.1 1.8 PU 19018 1.9 1.6 1.6 1.6 PU 19023 3.3 2.1 PU 19030 3.3 1.9 2.0 1.7 PU 19031 3.2 2.3 1.9 1.6 1.5 1.6 PU 19064 c.2.1 1.9 1.9 PU 19095 3.6 2.0 PU 19109 3.3 2.0 PU 19385 3.4 2.1 1.9 1.7 1.6 1.6 2.2 1.4 5.3 PU 19406 3.0 2.0 1.7 1.6 1.5 1.5 2.0 1.4 PU 19409 3.5 2.2 2.0 1.8 1.5 1.6 2.1 1.4 5.5 PU 19849 3.5 2.2 PU 19886 3.3 1.9 1.9 1.6 PU 21548 3.3 1.9 PU 21549 3.3 1.9 Carpolestes sp., from various localities PU 13284 3.0 1.7 PU 14853 3.5 2.2 1.7 1.5 PU 18102 3.1 c.1.7 PU 18315 3.8 2.5 2.0 1.8 PU 20720 1.8 1.5 PU 20721 3.2 1.8 1.9 1.5 PU 21227a 3.1 1.6 PU 21227b 3.2 2.0 1.9 1.6 PU 21227c 3.1 2.1 1.8 1.6 AMNH 88198 3.3 1.9 1.8 1.5 1.5 1.4 5.0 USNM 21280 2.7 1.7 74 Bulletin Museum of Comparative Zoology, Vol. 147, No. 1 Table II. Dimension's of maxillary teeth of carpolestids (mm). Spec. No. P3L P3B P*L P4B M'l M]B M-L M-'B 1.3 2.2 Carpodaptes hazelac Simpson, from Scarritt Quarry AMNH 33855 2.0 2.2 1.8 2.3 AMNH 33980 2.0 2.3 2.0 2.6 1.4 2.3 AMNH 33981 1.9 2.1 1.8 2.3 AMNH 33982 1.8 2.0 1.5 2.1 1.3 2.1 AMNH 33985 1.9 2.1 1.4 2.2 Carpodaptes hazclae Simpson, from Cedar Point Quarry PU 19939 1.9 2.0 1.7 2.2 1.3 1.8 1.2 1.8 PU 20065 1.8 2.0 1.7 2.2 Carpodaptes cygneus Russell, from Swan Hills ROM 05624 1.6 1.6 Carpodaptes sp., cf. C. cygneus Russell, from Cypress Hills, Alberta UA5857 1.7 1.8 UA5858 1.8 1.8 UA5861 1.7 2.1 UA5862[M2?] 1.3 2.0 UA 5872 Carpodaptes sp., from Princeton Locality 11 in the Crazy Mountain Field PU 14639 1.7 2.4 1.4 2.1 1.3 2.0 Carpoleptcs nigridens Simpson, from various localities AMNH 22196 3.2 2.4 PU 17978 3.1 2.3 PU 19543 2.8 2.3 YPM 24614 3.3 2.5 M3L 1.2 1.3 1.2 1.1 M3B FAphidotarsius sp., cf. E. florencae Gidley, from Rock Bench Quarry PU 17439 1.3 1.6 1.6 2.1 1.4 2.2 PU 17736 1.5 1.9 1.3 2.1 2.0 2.0 1.6 1.7 Carpolcstes dubius J epsen, from Princeton Quarry Level PU 13305 2.5 2.5 2.2 2.2 1.6 2.1 1.5 2.1 PU 14069 2.4 2.2 2.1 2.2 1.6 2.2 PU 14077 2.4 2.3 2.2 2.0 1.5 2.2 1.5 2.0 1.3 1.9 PU 14343 2.5 2.5 2.3 c.2.5 PU 14347 2.6 2.5 PU 17707 2.4 2.3 2.1 2.1 PU 17708 1.5 2.3 1.4 2.2 PU 17709 r 2.5 2.3 2.2 2.2 1.5 2.4 1.5 2.3 1.3 2.0 PU 17709 1 2.4 2.3 2.2 2.2 1.5 2.5 1.5 2.3 1.3 2.0 PU 17859 2.4 2.4 2.2 2.1 PU 17927 2.6 2.3 2.2 2.2 1.6 2.1 PU 17963 2.0 1.9 1.5 2.1 1.4 2.0 PU 19081 2.5 2.4 PU 19349 2.5 2.7 2.3 2.3 1.8 2.4 1.5 2.3 1.3 2.0 PU 19422 r 2.7 2.5 2.3 2.2 PU 19422 1 2.7 2.5 1.6 2.4 1.5 2.2 c.1.3 2.0 PU 19436 2.4 2.1 US ISSN 0027-4100 J3 ul let in of th seum Comparative Zoology Systematics and Distribution of Ceratioid Anglerfishes of the Genus Chaenophryne (Family Oneirodidae) THEODORE W. PIETSCH HARVARD UNIVERSITY CAMBRIDGE, MASSACHUSETTS, U.S.A. VOLUME 147, NUMBER 2 23 MAY 1975 PUBLICATIONS ISSUED OR DISTRIBUTED BY THE MUSEUM OF COMPARATIVE ZOOLOGY HARVARD UNIVERSITY Breviora 1952- Bulletin 1863- Memoirs 1864-1938 Johnsonia, Department of Mollusks, 1941- Occasional Papers on Mollusks, 1945- SPECIAL PUBLICATIONS. 1. Whittington, H. B., and E. D. I. Rolfe (eds.), 1963. Phylogeny and Evolution of Crustacea. 192 pp. 2. Turner, R. D., 1966. A Survey and Illustrated Catalogue of the Teredini- dae ( Mollusca: Bivalvia). 265 pp. 3. Sprinkle, J., 1973. Morphology and Evolution of Blastozoan Echinoderms. 284 pp. 4. Eaton, R. J. E, 1974. A Flora of Concord. 236 pp. Other Publications. Bigelow, H. B., and W. C. Schroeder, 1953. Fishes of the Gulf of Maine. Reprint. Brues, C. T., A. L. Melander, and F. M. Carpenter, 1954. Classification of Insects. Creighton, W. S., 1950. The Ants of North America. Reprint. Lyman, C. P., and A. R. Dawe (eds.), 1960. Symposium on Natural Mammalian Hibernation. Peters' Check-list of Birds of the World, vols. 2-7, 9, 10, 12-15. Proceedings of the New England Zoological Club 1899-1948. (Complete sets only.) Publications of the Boston Society of Natural History. Price list and catalog of MCZ publications may be obtained from Publications Office, Museum of Comparative Zoology, Harvard University, Cambridge, Massa- chusetts, 02138, U.S.A. © The President and Fellows of Harvard College 1975. SYSTEMATICS AND DISTRIBUTION OF CERATIOID ANGLERFISHES OF THE GENUS CHAENOPHRYNE (FAMILY ONEIRODIDAE) THEODORE W. PIETSCH1 Abstract. The ceratioid anglerfish genus Chaenophryne of the family Oneirodidae is revised on the hasis of all known material. Of the 18 available names, four are recognized as species. The five nominal forms, Chaenophryne longiceps, C. crossota, C. bicornis, C. quadrifilis, and C. crenata, previously forming the C. longiceps- group, and C. haplactis of the C. draco-group, are considered synonyms, taking the oldest available name, C. longiceps. The C. longiceps-grovp, now- being monotypic, is not recognized. The C. draco- group, however, is retained to include larvae and males easily distinguished from those of C. longiceps but not divisible into smaller taxonomic units. Included in the C. draco-group are three species based on metamorphosed females: C. draco, C. melanorhabdus, and C. ramifcra. Chaenophryne parviconus, C. atriconus, C. columni- fera, C. macractis, and C. melanodactylus are considered synonyms of C. draco. Chaenophryne melanorhabdus with its junior synonym, C. pterolophus, is resurrected from synonymy and given specific status. Chaenophryne pads is a synonym of C. ramifera. The tentative distri- bution of each species is plotted, evolutionary relationships are discussed, and a key to the species of the genus is provided. INTRODUCTION The genus Chaenophryne includes globose, bathypelagic anglerfishes, easily separated from members of allied genera by the absence of sphenotic spines, an operculum that is only slightly concave posteriorly, and peculiar, highly cancellous bones (Bertelsen, 1951:109; Pietsch, 1974a). 1 Museum Universitv. of Comparative Zoology, Harvard Bull. Regan (1925) established the genus Chaeno- phryne with the description of C. longi- ceps based on 14 specimens, only one of which actually represented the type spe- cies. The remaining 13 specimens were included in descriptions of 13 new species introduced by Regan and Trewavas (1932): C. bicornis, C. crenata, C. quadrifilis, C. haplactis, C. parviconus, C. atriconus, C. columnifera, C. melanodactylus, C. macrac- tis, C. melanorhabdus, C. pterolophus, C. fimbriata, and C. ramifera. Prior to Bertel- sen's ( 1951 ) monograph on the Ceratioidei, four additional forms were described: C. crossota Beebe, 1932; C. draco Beebe, 1932; C. intermedia Belloc, 1938; and C. pads Koefoed, 1944. The total number of nominal species is thus 18, 14 of which were based on one or two adolescent female specimens less than 20 mm standard length, none on more than nine specimens, and only one represented by a specimen larger than 24 mm standard length. From an examination of the extensive larval material in the DANA Collections, Bertelsen (1951) was able to divide the then known material of Chaenophryne into two species-groups based on the inner pig- ment layer and number of pectoral fin-rays. Within the C. longiceps-group he tenta- tively recognized five species: C. longiceps, C. quadrifilis, C. bicornis, C. crenata, and C. crossota; within the C. draco-group three species: C. draco, C. parviconus, and Mus. Comp. Zool., 147(2): 75-100. May, 1975 75 76 Bulletin Museum of Comparative Zoology, Vol. 147, No. 2 Tahle 1. Reallocation of nominal forms of CHAENOPHRYNE. longiceps-groug of Bertelsen, 1951 draco-group of Bertelsen, 1951 longiceps Regan, 1925 crossotus Beebe, 1932 bicornis Regan and Trewavas, 1932 crenata Regan and Trewavas, 1932 quadrifilis Regan and Trewavas, 1932 haplactis Regan and Trewavas, 1932 draco Beebe, 1932 parviconus Regan and Trewavas, 1932 atriconus Regan and Trewavas, 1932 columnifera Regan and Trewavas, 1932 melanodactylus Regan and Trewavas, 1932 macractis Regan and Trewavas, 1932 melanorhabdus Regan and Trewavas, 1932 pterolophus Regan and Trewavas, 1932 ramifera Regan and Trewavas, 1932 fimbriata Regan and Trewavas, 1932 intermedia Belloe, 1938 parts Koefoed, 1944 longiceps Regan, 1925 draco Beebe, 1932 melanorhabdus Regan and Trewavas, 1932 ramifera Regan and Trewavas, 1932 C. ramifera. Chaenophryne fimbriata and C. intermedia were considered synonyms of C. ramifera and all remaining available names were placed in the synonymy of C. parviconus. In the present paper it is shown that the available female material of Chaeno- phryne represents four well-defined species (Tabic 1): C. longiceps, represented by 33 specimens, collected from all three major oceans of the world; C. draco, represented by 46 specimens, also of world-wide dis- tribution; C. melanorhabdus, 23 .specimens, apparently restricted to the continental slope of western North America; and C. ramifera, 15 specimens of world-wide dis- tribution. The separation of these species is based almost entirely on the morphology of the esca. Significant differences exist, however, in the width of the escal bulb, the length of the illicium, jaw tooth counts, and fin-ray counts. Despite a greater than four-fold increase in material since Bertelsen's (1951) re- vision, males can be separated into only two taxonomic units corresponding to C. longiceps and the C. f/raeo-group. METHODS AND MATERIALS Standard lengths (SL) were used through- out. Measurements were taken on the left side whenever possible and rounded to the nearest 0.1 mm. To insure accurate fin-ray counts, skin was removed from the pectoral fins and incisions were made to reveal the rays of the dorsal and anal fins. Illicium length is the distance from the articulation of the pterygiophore of the illicium and the illicial bone to the dorsal surface of the escal bulb, excluding escal appendages. Terminology used in describing the various parts of the angling apparatus follows Bradbury (1967). Definitions of terms used for the different stages of develop- ment follow those of Bertelsen (1951:10- 11). Complete locality data for primary type material is entered in the synonymies. Drawings were made with the aid of a Wild M-5 Camera Lucida. Since nearly all the available collections of Chaenophryne were made with non- closing nets, the actual depth of capture is unknown. For those species represented by sufficient material, vertical distributions were analyzed by a procedure similar to that used by William H. Krueger (un- published manuscript) for determining depths of capture of Idiacanthus, and out- lined by Cibbs ( 1969; see also Pietsch, 1974a). Station data were taken from Schmidt (1929), Carlsberg Foundation (1934), and unpublished data for VELERO Anglerfishes of the Genus Chaenophryne • Pietsch 77 IV cruises of the University of Southern California. For each trawl, the number of hours at depth was multiplied by the area of the mouth of the net. The number of meter-hours for each depth interval, and the number of specimens caught at each depth interval were tabulated and expressed as a percentage of total for comparison. When the percentage of specimens caught at any depth is considerably greater than the percentage of meter-hours at that depth, it may be assumed that this represents a region of concentration. The reverse indi- cates that specimens recorded for that depth probably were caught while the net was being lowered or raised. Material used for the comparative os- teological investigation was cleared and stained with alizarin red S following the trypsin digestion technique (Taylor, 1967). In many cases, dissections were made of uncleared specimens to confirm observa- tions made on cleared specimens. Bone terminology follows Pietsch (1974a). The generic diagnosis for females is based largely on osteological evidence pre- sented elsewhere (Pietsch, 1974a). Generic and species descriptions are based on 116 metamorphosed females ranging from 11.0 to 170.0 mm ( morphometries and jaw tooth counts on specimens 20.0 mm and larger). Males and larvae were described by Bertelsen (1951:110-116). Study material is deposited in the following institutions: AMNH: American Museum of Natural Historv, New York. BMNH: British Museum (Natural History), London. BOC: Bingham Oceanographic Collec- tions, Peabody Museum of Natural History, Yale University. IOM: Institute of Oceanology, Academy of Sciences of the USSB, Moscow. IOS: Institute of Oceanographic Sci- ences, Surrey, England (formerly the National Institute of Ocean- ography ) . ISH: Institut fur Seefischerei, Ham- burg. LACM: Natural History Museum of Los Angeles County. MCZ: Museum of Comparative Zoology, Harvard University. MNLR: Musee d'Histoire Naturelle de La Rochelle. NYZS: New York Zoological Society. OSUO: Oregon State University, Depart- ment of Oceanography, Corvallis. ROM: Royal Ontario Museum, Toronto. SU: Stanford University (collections now housed at the California Academy of Sciences, San Fran- cisco ) . UMML: University of Miami Marine Lab- oratory. USNM: United States National Museum, Washington. UW: Fish Museum, University of Washington, Seattle. ZMB: Zoological Museum, University of Bergen. ZMUC: Zoological Museum, University of Copenhagen. OSTEOLOGY The osteology of Chaenophryne has been previously described in detail and com- pared with that of other oneirodid genera (Pietsch, 1974a; C. melanorhabdus mis- identified as C. parvieonus). Not mentioned before, however, is the conspicuous, honey- comblike network of ridges that partially form many of the bones of Chaenophryne. These highly cancellous bones include those that provide the dorsal surface of the cranium, excluding those that form the illicial trough: the dorsal surface of the frontals, the sphenotics, pterotics, parietals, and posttemporals; portions of the mandib- ular arch: the maxillaries, dentaries, and articulars; and parts of the opercular ap- paratus: the operculum, suboperculum, and preoperculum. Osteological preparations were made of all four species of Chaenophryne using the trypsin digestion technique of Taylor ( 1967 ) . Comparison of specimens of a similar size showed little interspecific vari- 78 Bulletin Museum of Comparative Zoology, Vol. 147, No. 2 PTEROSPHENOID SPHENOTIC FRONTAL ETHMOID CARTILAGE VOMER PARIETAL PTEROTIC PROOTIC SUPRAETHMOID PARASPHENOID A B D Figure 1. Lateral views of crania of Chaenophryne: A. C. longiceps. LACM 30196-26, 40.0 mm; B. C. draco, LACM 31436-2, 42.0 mm; C. C. melanorhabdus, LACM 34073-1, 35.0 mm; D. C. ramifera, MCZ 47556, 42.0 mm. Anglehfishes of the Genus Chaenophryne • Pietsch 79 Figure 2. Pelvic bones of Chaenophryne: A. C. draco, LACM 34072-1, 33.0 mm; B. C. melanorhabdus, LACM 30619-5, 58.0 mm; C. C. ramifera, ISH 765 68, 35.0 mm. ation. The cranium of C. ramifera, how- ever, is considerably more elongate and depressed than that of its congeners (Fig. 1 ) . This elongation and depression is mainly due to having longer frontal bones which are slightly concave on their dorsal surface compared to the dorsally convex frontals of all other species of Chaeno- phryne. As a result of this, the frontals of C. ramifera are in a more anterior position; the ventromedial extensions of these bones, which are by consequence longer than those of its congeners, help to form longer and considerably narrower orbital foram- ina. In an earlier paper (Pietsch, 1974a: 29, 30, 86, 89, fig. 56D, Table 23) I described Chaenophryne as having triradiate pelvic bones based on an examination of three cleared and stained specimens of C. melanorhabdus (59.0-96.0 mm; incorrectly referred to as C. parviconus) . Examination of additional material of all four species of Chaenophryne by clearing and staining and by dissection, has shown that the pelvic bones vary in shape from triradiate to broadly expanded distally. In some cases, the two extremes of this variation may occur in the same specimen (Fig. 2). SYSTEMATICS Genus Chaenophryne Regan, 1925 Females: Chaenophryne Regan, 1925:564 (type species Chaenophryne longiceps Regan, 1925, by original designation and monotypy). Himantolophus Regan, 1926:40 (in part; type species Himantolophus groenlandicus Reinhardt, 1837, by monotypy; separate in 1837, see Reinhardt, 1838). Males: Rhynchoceratias Regan, 1926:44 (in part; type species Rhynchoceratias hrevirostris Regan, 1925, by subsequent designation of Fowler, 1936). Trcmatorhynchus Regan and Trewavas, 1932:91 ( in part; type species Rhynchoceratias leuchor- hinus Regan, 1925, by subsequent designation of Burton, 1933). Diagnosis for females. The genus Chaeno- phryne is distinguished from all other genera of the family Oneirodidae by the absence of sphenotic spines, an operculum that is only slightly concave posteriorly, and highly cancellous bones (the latter not found in any other ceratioid). In addition, Chaenophryne is unique in having the following combination of characters: mouth 80 Bulletin Museum of Comparative Zoology, Vol. 147, No. 2 large, cleft extending past eye; dorsal pro- small anterior projection in some specimens file of frontal bones strongly convex; an- (Bertelsen, 1951:109, figs. 66, 67). terior end of pterygiophore of illieium Ulieinm length 20.1 to 47.4 percent of exposed (see generic description below), its SL, becoming longer proportionately with posterior end concealed under skin; illieium growth (Fig. 3); pterygiophore of illieium length greater than 20 percent of SL; 70 to 82 percent of SL ( less than 50 percent symphysial spine of lower jaw rudimentary; of SL in other oneirodids; Pietsch, 1974a: articular spines rudimentary; angular spine 18); anterior end of pterygiophore of absent; vomerine teeth present; pharyngo- preserved specimens usually concealed branchial I absent; pharyngobranchials II under skin within illieial trough, but ca- and III present and toothed; epibranchial pable of considerable forward extension (as teeth absent; hypobranchial II present; in Oneirodes and ceratiids; Bertelsen, 1943, pectoral lobe short and broad, shorter than 1951:18); posterior end of pterygiophore longest rays of pectoral fin; suboperculum concealed under skin; esca with a single elongate, upper end tapering to a point, (C. draco-group) or paired (C longiceps), lower part nearly circular with anterior conical to elongate, anterior appendage or projection in some specimens; anal fin with appendages, internally pigmented with one, 5 rays, rarely 4 or 6; skin naked, covering or three, round, translucent windows at tip; caudal fin to some distance from fin base, a filamentous medial appendage or append- Diagnosis for males. Nonparasitic; skin ages present (C. longiceps) or absent (C. between nostrils and between posterior draco-group ) ; a posterior appendage con- nostril and eye pigmented; 8 to 12 olfactory sisting of a swollen basal portion and a lamellae; upper denticular with 10 to 22, compressed distal crest bearing none to irregularly curved denticles, mutually fused numerous filaments anteriorly or distally, in a semicircular group; lower denticular with (C. draco-group ) or without (C. with 13 to 31, recurved denticles in 2 to 3 longiceps) a pair of fringed, anterior lobes; irregular series; skin black and naked; no one or two filamentous, anterolateral ap- teeth in jaws; shape of pectoral fin-lobe, pendages on each side (C. melanorhabdus opercular bones, and anal fin-ray counts as and C. ramifera); a paired series of fila- for females (taken largely from Bertelsen, ments arising laterally from anterior base 1951:109). of eseal bulb (C. ramifera); a sub- Gencrie description of females. Body cutaneous, internally pigmented appendage relatively short, globular; jaws equal an- emerging from base of esca and descending teriorly; symphysial spine of lower jaw along anterior margin of illieial bone (ex- small to absent; oral valves well-developed; tremely well-developed in some specimens eye subcutaneous, appearing through a of C. ramifera). circular, translucent area of integument; Teeth slender, recurved and all depress- gill opening oval in shape, situated postero- iblc, in overlapping sets as described for ventrad to pectoral lobe; skin naked (em- other oneirodids (Pietsch, 1972b:5, fig. 2; bedded dermal spines cannot be detected 1974a, b); teeth in lower jaw larger than microscopically in cleared and stained those in upper jaw; number of teeth in specimens); lateral line papillae as de- upper jaw 21 to 51, lower jaw with 26 to scribed for other oneirodids (Pietsch, 1969, 57 teeth; jaw teeth few and rudimentary in 1972b, 1974a, b); ovaries paired; pyloric specimens smaller than approximately 18 eaeea absent. mm SL; vomerine teeth 2 to 4 on each side, Operculum triangular in shape, only the longest outermost, slightly concave along posterior margin; Color in preservation dark brown to suboperculum elongate, upper end tapering black over entire external surface of body to a point, lower part nearly circular with except for bulb and usually appendages of Anglerfishes of the Genus Ciiaenophryne • Pietsch 81 X h- 3 28 24 20 16 • LONGICEPS * DRACO vMELANORHABDUS oRAMIFERA 20 30 40 80 90 100 50 60 70 STANDARD LENGTH IN MM Figure 3. Graph of relationship of illicial length and standard length for species of Chaenophryne. esca ( see species accounts ) ; oral cavity and guts except for outer surface of stomach wall unpigmented. D. 6-8; A. 5-6; P. 16-22 (Table 2); pelvics absent; C. 9 ( 2 + 4 + 3 ) ; branchios- tegal rays 6(2 + 4). Generic description of males. See Bertel- sen (1951:110, 112, 116, figs. 66-70, 72, 73). Key to Females of the Species of the Genus Chaenophryne The following key will differentiate adolescent and adult females only. For larvae and males see Bertelsen (1951:110). 1A. Esca with a pair of internally pigmented, anterior appendages; medial escal append- age or appendages present ( Fig. 7 ) ; width of escal bulb 5.3-11.4 percent of SL in specimens 20 mm and larger (Fig. 4); pec- toral fin-rays 17-22 rarely less than 18 (Table 2) ! .._ C. longiceps Regan, 1925. IB. Esca with an unpaired, internally pig- mented, anterior appendage; medial escal appendages absent (Figs. 8-10); width of escal bulb 2.1-6.6 in specimens 20 mm and larger (Fig. 4); pectoral fin-rays 16-19, rarely more than 18 (Table 2) C. draco-group _. __ 2. 2A. Esca without anterolateral appendages ( Fig. 8 ) ; ratio of number of teeth in upper jaw to number of teeth in lower jaw 1.08-1.45 in specimens 20 mm and larger (Fig. 5) _ C. draco Beebe, 1932. 2B. Esca with one to three anterolateral append- ages on each side (Figs. 9, 10); ratio of number of teeth in upper jaw to number of teeth in lower jaw .76-1.30 (Fig. 5) 3. 3A. Esca with a series of filaments arising laterally from anterior base of bulb (Fig. 10); ratio of number of teeth in upper jaw to number of teeth in lower jaw less than 1.00 in specimens 30 mm to approximately 60 mm (Fig. 5; specimens larger than 55.5 mm unknown); D. 8, rarely 7; A. 6, rarely 5; P. 17-19, rarely 16 (Table 2) C. ramifera Regan and Trewavas, 1932. 3B. Esca without a series of filaments arising from base of bulb ( Fig. 9 ) ; ratio of number of teeth in upper jaw to number of teeth in lower jaw greater than 1.05 in specimens 30 mm and larger (Fig. 5); D. 6-7, rarely 8; A. 5, rarely 6; P. 16-17, rarely 18 (Table 2) . C. melanorhabdus Regan and Trewavas, 1932. 82 Bulletin Museum of Comparative Zoology, Vol. 147, No. 2 6 - o * 4 pa _i => 03 < O w 3 •LONGICEPS a DRACO ^MELANORHABDUS ORAMIFERA 20 30 40 50 60 70 STANDARD LENGTH IN MM 80 90 100 Figure 4. Graph of relationship of escal bulb width and standard length for species of Chaenophryne. Chaenophryne longiceps Regan, 1925 Figures 6, 7 Chaenophryne longiceps Regan, 1925:564 (in part; original description; 14 specimens; lecto- type, ZMUC P92106, 20.0 mm; DANA Station 1203 (11), 7°30'N, 79°19'W; 3000 m wire; 1500 hr; 11 January 1922). Regan, 1926:31, pi. 6, fig. 2 (in part; description after Regan, 1925; 3 additional specimens). Parr, 1927:22, fig. 8 (4 additional specimens designated longi- ceps forma typica, var. quadrifilis n. var., and var. quadrifilis ?; description). Regan and Trewavas, 1932:85, 86, figs. 14, 135 (descrip- tion; lectotype designated as only representative of longiceps, 13 paralectotypes made types of Table 2. Fin-ray frequencies for metamorphosed females of species of Chaenophryne. Species Dorsal Anal Pectoral 6 7 8 5 6 4 14 16 2 3 16 2 20 1 4 14 1 16 2 2 10 2 10 16 17 18 19 20 > i 22 longiceps draco melanorhahdus r ami f era 1 4 5 16 5 6 11 2 1 7 3 Anglerfishes of the Genus Chaenophryne • Pietsch 83 < en LU X O o LU 3 2 < "3 o cc LU pa 3 — i r- 1 1 A 1 T 1 1 T T !. I ■ • LONGICEPS a DRACO " 1.2 A A A A A A ^MELANORHABDUS o RAMIFERA 77 l 1 A A A A V V 1.0 A V V V A O A A A V o V o ° • o .8 V A V • o • ° • .7 A A • V • V V • V v o V • • •• • " .6 • 1 ' ' ' 1 1 ' • 1 10 20 30 40 50 60 70 STANDARD LENGTH IN MM 80 90 100 Figure 5. Graph of relationship of ratio between number of upper and lower jaw teeth, and standard length for species of Chaenophryne. other species; in key). Bertelsen, 1951:111, 113, figs. 66A, 71, Table 23 (description; comparison with all known material; in key). Grey, 1956: 256 (synonymy; distribution). Pietsch, 1974a: 33 (listed). Himantolophus groenlandicus, Regan, 1926:40 (in part; larval female referred to longiceps-group by Bertelsen, 1951). Bhynchoceratias leuchorhinus, Regan, 1926:44 (in part; 2 males referred to longiceps-group by Bertelsen, 1951). Chaenophryne crossotus Beebe, 1932:83, fig. 21 (original description; single specimen; holotype USNM 170942 (originally NYZS 20809), 17.0 mm; Bermuda Oceanographic Expedition, Net 1015, 32°12'N, 64°36'W; 0-915 m; 15 June 1931). Regan and Trewavas, 1932:85, 86 {crossota; description after Beebe, 1932; in key). Bertelsen, 1951:111, 114, Table 23 (crossota; description; comparison with all known material; in key). Grey, 1956 -257 (crossota; synonymy; distribution). Pietsch, 1974a: 33 (listed). Chaenophryne bicornis Regan and Trewavas, 1932: 84, 85, fig. 133 (original description; 2 speci- mens; lectotype, ZMUC P92107, 14.0 mm; DANA Station 4005 (1), 4000 m wire; 1145 hr; Bertelsen, 1951:111, 113, fig. (description; comparison with terial; lectotype designated; 13°31'N, 18°03'W; 12 March 1930). 66C, Table 23 all known ma- key ) . Grey, in 1956:256 (synonymy; distribution). Pietsch, 1974a: 33 (listed).' Chaenophryne crenata Regan and Trewavas, 1932: 84, 86, fig. 134 (original description; single specimen; holotvpe, ZMUC P92108, 18.0 mm; DANA Station 3714 (10), 15°22'N, 115°20'E; 2000 m wire; 0245 hr; 20 April 1929). Bertel- sen, 1951:111, 114, fig. 66B, Table 23 (de- scription; comparison with all known material; in key). Grey, 1956:257 (synonymy; distri- bution). Pietsch, 1974a: 33 (listed). Chaenophryne quadrifilis Regan and Trewavas, 1932:85, 87, fig. 136 (original description based on longiceps quadrifilis n. var. of Parr, 1927; 4 specimens; lectotype, BOC 2910 (not 2007 as in 84 Bulletin Museum of Comparative Zoology, Vol. 147, No. 2 Parr, 1927), 20.5 mm; PAWNEE Station 58, 32°24'N, 64°29'\V; 3050 m wire; 20 April 1927). Parr, 1937:63 (listed; lectotype desig- nated). Koefoed, 1944:8, pi. I, figs. 2, 3 (de- scription of additional specimen). Bertelsen, 1951:111, 113, Table 23 (description; compari- son with all known material; in key). Grey, 1956:257 (synonymy; distribution). Pietsch, 1974a: 33 (listed).' Chaenophryne haplactis Regan and Trewavas, 1932:85, 87, fig. 137 original description; single specimen; holotvpe, ZMUC P92114, 11.0 mm; DANA Station 1152 (3), 30°17'N, 20°44'W; 3000 m wire; 1930 hr; 22 October 1921). Trematorhynchus leuchorhinus, Regan and Tre- wavas, 1932:91 (in part; 2 males referred to longiceps-groixp by Bertelsen, 1951). Chaenophryne longiceps-grovp Bertelsen, 1951:71, 110-114, 269, figs. 30, 66, 68-71, Table 23 ( osteological description of larval female; 5 nominal species grouped; common characters; all available material listed; description of larvae, males, metamorphosing females; in key). Grey, 1956:256 (after Bertelsen, 1951; synonymy; distribution). Pietsch, 1974a:33 (after Bertel- sen, 1951; included species listed). Material. Thirty-three metamorphosed females, 11.0-170.6 mm: BMNH 1 (14.0 mm); BOC 3 (20.0-21.0 mm); IOM 1 (23.0 mm); IOS 1 (22.5 mm); LSI I 4 (35.0-103.0 mm); LACM 2 (18.0^0.0 mm); MCZ 3 (17.0-39.0 mm); OSUO 1 (25.0 mm); ROM 1 (170.0 mm); SU 1 (36.0 mm); UMML 2 (19.0^49.0 mm); USNM 7 (13.5- 29.5 mm); ZMB 1 (25.0 mm); ZMUC 5 (11.0-19.5 mm). In addition to material listed by Bertelsen (1951:269) the following males have been examined: IOS 1 (11.0 mm); LACM 1 (12.0 mm); MCZ 2 (14.0-17.0 mm); USNM 1 (19.5 mm). Diagnosis. In addition to characters of the esca which separate C. longiceps from all other species of Chaeno))J}ryne (see Key and Generic description), this species can be further distinguished from C. draco and C. melanorhabdus by having a smaller ratio between the number of upper and lower jaw teeth (Fig. 5). The illicium of C. longiceps appears to be slightly shorter than that of C. ramifera and slightly longer than that of C. melanorhabdus (Fig. 3). Chaenophryne longiceps has fewer dorsal and anal fin-rays than C. ramifera, and a greater number of pectoral fin-rays than all other species of the genus (Table 2). Description. Esca with a pair of elongate, internally pigmented (except for tip) and bilaterally placed, anterior appendages (occasionally sharing a common base as in holotvpe of C. crenata; Regan and Trewa- vas, 1932:86, fig. 134), less than one-tenth to greater than length of escal bulb; one to three, transversely placed, medial append- ages, bifurcated at midlength to highly filamentous from base, darkly pigmented in some large specimens (102.0 and 103.0 mm ) ; medial appendages may arise at any point from distal surface of escal bulb to anterior margin of swollen basal portion of posterior appendage (Fig. 7); a posterior appendage consisting of a swollen basal portion and a membranous, distal crest, darkly pigmented in largest known speci- mens (102.0-170.0 mm), and bearing one to several filaments anteriorly or distally in some specimens; anterolateral appendages and basal series of filaments absent; in most specimens a subcutaneous, internally pigmented descending appendage (Fig. 7). Illicial length 22.4-40.3 percent of SL (Fig. 3); escal bulb width 5.3-11.4 percent of SL (Fig. 4); total number of teeth in upper jaw 28-^0, in lower jaw 34-57; ratio of number of teeth in upper jaw to number in lower jaw .70-94 (Fig. 5); vomerine teeth 4-8; D. 6-8; A. 5-6; P. 17-22 (Table 2). Rest of characters as for genus. Distribution. Chaenophryne longiceps has a wide horizontal distribution, oc- curring in all three major oceans of the world. It has been collected from both sides of the north Atlantic from the equator to the northern coast of Iceland. Three records are known from the Indo-Pacific region: in the1 Indian Ocean from the Arabian Sea (approximately 7°N, 65°E) and Bay of Bengal (9°N, 90°E), and from the South China Sea. In the eastern Pacific, records extend from as far west as the Hawaiian Islands, between approxi- Anglerfishes of the Genus Chaenophryne • Pietsch 85 x o X a> c c < CD .a ca LU c TO CM O CO cm I CO CO a. CD o B> c o Q) c »^ ■C Q. O C CD CO -c o CO CD en 86 Bulletin Museum of Comparative Zoology, Vol. 147, No. 2 Figure 7 Escae of Chaenophryne longiceps: A. MCZ 47561, 17.0 mm, left side; B. I0S uncatalogued, 22 5 mm, left side; C. MCZ 49854, 29.0 mm, left side; D. MCZ 49857, 39.0 mm, left side; E. ISH 237 73, 102.0 mm, anterior ' view; F. ISH 607/73, 103.0 mm, left side. mately 45°N off the coast of Oregon and fishing gear, metamorphosed female speci- 23°S off Chile. The leetotype of C. longi- mens of C. longiceps are vertically dis- ceps was collected from the Gulf of Panama tributed between approximately 500 m and (Fig. 11), an unknown lower limit. Eighty-eight Based on maximum depths reached by percent of the known material (28 in- Anglerfishes of the Genus Chaenophryne • Pietsch 87 dividuals) was captured by gear fished below S50 m. The largest known specimens (3 individuals, 102.0-170.0 mm) were captured by nets fished below 950 m. Material is not sufficient from any one geographic area for a more analytical treatment of distributional data. Comments. Bertelscn (1951) proposed the Chaenophryne longiceps-group to in- clude five nominal species described by Beebe ( 1932 ) and Regan and Trewavas (1932) on the basis of one to four adolescent female specimens less than 25 mm, and on relatively small differences in the morphol- ogy of the esca. The greater amount of material now available, providing some understanding of individual and onto- genetic variation, has shown that these differences in escal morphology must be regarded as variation exhibited by widely distributed conspecific populations. In the absence of significant differences, these forms are here synonymized with C. longi- ceps (Table 1). Chaenophryne haplaetis, known only from the holotype (11.0 mm), was placed in the C. r/raco-group by Bertelsen (1951) on the basis of absence of inner pigment on the caudal peduncle and a low pectoral fin- ray count ( 15 according to Regan and Trewavas, 1932). Re-examination of the holotype by Bertelsen (personal communi- cation) shows the somewhat bleached re- mains of large, stellate melanophores on the peduncle; I count IS pectoral rays. Further, the esca of this specimen clearly has the paired anterior appendages char- acteristic of C. longiceps (Regan and Trewavas, 1932:89, fig. 137). For these reasons, C. haplaetis is removed from the C. draco-group and placed within the synonymy of C. longiceps. Chaenophryne draco-group Bertelsen, 1951 Figures 8-10 Himontolophus groerilandicus, Regan, 1926:40 (in part; 2 larval females referred to f/raco-group by Bertelsen, 1951). Wujnchoceratias leuchorhinus, Regan, 1926:44 (in part; 3 males referred to draco-group by Bertelsen, 1951). Trematorhynchus leuchorhinus, Regan and Trewa- vas, 1932:91 (in part; 5 males referred to rfraco-group by Bertelsen, 1951). Chaenophryne f/raco-group Bertelsen, 1951:72, 110, 114-118, 270, figs. 30, 67, 72-75, table 23 ( osteological description of adolescent male; 3 nominal species grouped; common characters; all available material listed; description of larvae, males, metamorphosing females; in key). Grey, 1956:257 (after Bertelsen, 1951; syn- onymy; distribution). Maul, 1962:22, fig. 10 ( description of metamorphosing female ) . Pietsch, 1974a:33 (after Bertelsen, 1951; in- cluded species listed). Material. Metamorphosed female speci- mens are listed below by species. In addi- tion to material listed by Bertelsen (1951: 270), the following males of the C. draco- group have been examined: IOM 2 (12.5- 15.0 mm); IOS 16 (10.5-14.0 mm); LACM 7 (9.5-13.5 mm); MCZ 5 (7.0-14.5 mm). Comments. The Chaenophryne draco- group (Bertelsen, 1951) is retained to in- clude larvae and males readily separated from those of C. longiceps by larval pig- mentation and pectoral fin-ray counts, but not divisible into smaller taxonomic units. Within the C. f/raco-group are also included three species based on metamorphosed females that together differ from C. longi- ceps in the basic design of the esca (see Key, Generic description and Figs. 7-10), and in other characters listed in the key to species. These three species are associated with the larvae and males of the C. draco- group by retention of the characteristic larval pigmentation in small specimens, and by pectoral fin-ray counts (Bertelsen, 1951: 110; Table 3). Chaenophryne draco Figure 8 Beebe, 1932 Chaenophryne longiceps Regan, 1925:564 (in part; original description; 14 specimens; paralecto- types of longiceps subsequently made types of parviconus, columnifera, melanodactijlus by Regan and Trewavas, 1932). Regan, 1926:31 (in part; after Regan, 1925). SS Bulletin Museum of Comparative Zoology, Vol. 147, No. 2 Chaenophryne draco Beebe, 1932:84, fig. 22 (original description: single specimen; holotype, USNM 170943 (originally \YZS 22396), 16.5 inm; Bermuda Oceanographic Expedition Net 1181, 32°12'N, 64°36'W; 1100 m; 15 August 1931). Regan and Trewavas, 1932:85, 89 (de- scription after Beebe, 1932; in key). Bertelsen, 1951:115, 116, Table 24 (description; compari- son with all known material; in key). Grey, 1956:258 (synonymy; distribution). Pietsch 1974a: 33 (listed). Chaenophryne parviconus Regan and Trewavas, 1932:35, 85, 87, figs. 39, 41, 138 (original description, osteology of head, pectoral arch; 9 specimens; lectotype, ZMUC P92110, 12.5 mm; DANA Station 1209(2), 7°15'N, 78°54'W; 3000 m wire; 1845 hr; 17 January 1922). Beebe and Crane, 1947:158 (6 additional specimens; description; columnifera and melanorhahdus synonyms of parviconus). Bertelsen, 1951:115, 117, fig. 74B, C (in part; description; com- parison with all known material; in key; haplactis, atriconus, columnifera, melano- dactylus, macractis, melanorhabdus, pterolophus, pads synonyms of parviconus) . Grey, 1956:258 (synonymy; distribution). Pietsch, 1974a:33 (listed). Chaenophryne atriconus Regan and Trewavas, 1932:85, 87, fig. 139 (original description; single specimen; holotype, ZMUC P92111, 15.0 mm; DANA Station 3847(5), 12°02'S, 96°43'E; 1500 m wire; 2100 hr; 11 October 1929 ) . Chaenophryne columnifera Regan and Trewavas, 1932:85, 88, fig. 140 (original description; 3 specimens; lectotype hereby designated, ZMUC P92112, 14.5 mm; DANA Station 1208(15), 6°48'N, 80°33'W; 2600 m wire; 1715 hr; 16 January 1922). Chaenophryne melanodactylus Regan and Tre- wavas, 1932:85, 88, fig. 141 (original descrip- tion; single specimen; holotype, ZMUC P92116, 15.0 mm; DANA Station 1370(13), 36°36'N, 26°14'W; 3000 m wire; 1150 hr; 13 June 1922). Chaenophryne macractis Regan and Trewavas, 1932:85, 88, fig. 142 (original description; single specimen; holotype, ZMUC P92115, 14.0 mm; DANA Station 3561(4), 4°2()'S, 1 16°46'W; 2000 m wire; 0900 hr; 24 September 1928). Material. Forty-six metamorphosed fe- males. 11.0-42.0 mm: AMNII 2 (29.0-35.0 mm); BMNH 5 (12.0-14.0 mm); IOS 1 (30.0 mm); LACM 17 (13.0-42.0 mm); MCZ 4 (12.5-18.0 nun); NYZS 6 (13.5- 19.5); USNM 1 (16.5 mm); ZMUC 10 (11.0-15.0 mm). Diagnosis. In addition to the characters of the esca, which separate C. draco from all other species of Chaenophryne (see Key and Generic description), this species is distinguished by having the highest ratio between the number of upper and lower Jaw teeth of any species (Fig. 5). The illieium of C. draco appears to be slightly shorter than that of C. ramifera and slightly longer than that of C. melanorhalxJus (Fig. 3). Fewer pectoral fin-rays, and fewer dorsal and anal rays, help to distinguish C. draco from C. Jongiceps and C. ramifera, respectively (Table 2). Description. Esca with a single conical to elongate, internally pigmented (except for tip), anterior appendage, less than one- seventh to nearly one-third length of escal bulb; medial appendages absent; a poste- rior appendage consisting of a swollen basal portion, and a somewhat compressed distal crest with a posterior filament or filaments and a pair of anterior lobes each bearing none to numerous filaments; anterolateral appendages and basal series of filaments absent; in larger specimens, a subcutaneous, internally pigmented, descending append- age (Fig. 8). Illicial length 24.0-36.4 percent of SL (Fig. 3); escal bulb width 4.3-6.7 percent of SL (Fig. 4); total number of teeth in upper jaw 35-47, in lower jaw 31-38; ratio of number of teeth in upper jaw to number in lower jaw 1.08-1.45 (Fig. 5); vomerine teeth 4-8; D. 6-8; A. 5-6 (of 21 specimens only one had A. 6); P. 16-18 (Table 2). Rest of characters as for genus. Distribution. Chaenophryne draco has a wide horizontal distribution, occurring in all three major oceans of the world. It has been collected from both sides of the Atlantic: two specimens from off Bermuda (including the holotype), one from the Cape Verde Islands, and a fourth from 36°36'N, 26°14'W. Four records are known from the Indo-Pacific region: two on the equator at approximately 65°E, one from off the Cocos Islands (12°02'S, 96°43'E), and one from Sagami Bay, Japan (C. Anglerfishes of the Genus Chaenophryne • Pietsch 89 Figure 8. Escae of Chaenophryne draco: A. MCZ 49863, 12.5 mm, left side; B. holotype of C. macractis, ZMUC P92115, 15.0 mm, lateral and anterior views reversed from Regan and Trewavas, 1932; C. LACM 32788-2, 19.0 mm, left lateral and anterior views. macractis of Imai, 1942). In the western Pacific, records extend from the Hawaiian Islands (13 specimens) to the Gulf of Panama (18 records), between approxi- mately 21°N and 4°S (Fig. 11). Based on maximum depths reached by fishing gear, metamorphosed female speci- mens of C. draco are vertically distributed between approximately 350 m and an un- known lower limit. Eighty-nine percent of the known material (42 individuals) was captured by gear fished below 700 m. Sufficient material is known from the Gulf of Panama for analysis of vertical data by a procedure outlined by Gibbs ( 1969; see Methods and Materials). Results indicate a concentration between 700 and 1500 m (Table 3). Comments. Prior to his monograph on the Ceratioidei, Bertelsen ( 1951 ) had not seen the holotype of C. draco Beebe, 1932. Beebe's (1932:85, fig. 22) colorful descrip- tion and somewhat stylized figure of the esca did not fully agree with descriptions of other species introduced six months later by Regan and Trewavas ( 1932 ) . These forms were maintained as separate species pending examination of Beebe's type. Per- sonal examination of the holotype of C. draco ( 16.5 mm ) has confirmed Bertelsen's (1951) prediction that C. parviconus (and Table 3. Vertical distribution of Chaeno- phryne DRACO BASED ON SPECIMENS COLLECTED BY THE DANA IN THE GULF OF PANAMA. METER- HOURS AND SPECIMENS EXPRESSED AS PERCENT OF TOTAL. See text for METHODS of calculation. Depth (m) Meter-hours Specimens 0-100 16.5 0 101-200 5.9 0 201-300 0.1 0 301-400 15.4 0 401-500 0 - 501-600 2.4 0 601-700 0 - 701-800 1.6 5.6 801-900 0 - 901-1000 5.9 11.1 1001-1250 10.7 27.7 1251-1500 8.9 50.0 1501-1750 14.5 5.6 1751-2000 11.2 0 2001-3000 5.9 0 3000 0 - Number of specimens 18 Number of meter-hours 596.6 Number of hauls 83 90 Bulletin Museum of Comparative Zoology, Vol. 147, No. 2 most of its included synonyms; sec Table 1) is a junior synonym of C. draco. Chaenophryne melanorhabdus Regan and Trewavas, 1932 Figure 9 Chaenophryne melanorhabdus Regan and Tre- wavas, 1932:85, 89, fig. 143 (original descrip- tion; single specimen; holotype, ZMUC P92117, 40.0 mm; DANA Station 1203(14), 7°30'N, 79°19'W; 2500 m wire; 2030 hr; 11 January 1922). Chaenophryne pterolophus Regan and Trewavas, 1932:85, 89, fig. 144 (original description; single specimen; holotype, ZMUC P92118, 20.5 mm; DANA Station 1208 (16), 6°48'N, 80°33'W; 2100 m wire; 1715 hr; 16 January 1922). Chaenophryne parviconus, Grinols, 1966:161-165, fig. 1, Tables 1, 2 (misidentification; 2 speci- mens; description; escae figured; distribution). Pietsch, 1972a: 35, 36, 42, 45, fig. 24(6) (mis- identification; osteological comments; otolith described, figured). Pietsch, 1974a: 33, 109, figs. 26, 37, 39D, 46, 49, 51F, 52B, 56D (mis- identification; osteological description; compari- son with other oneriodid genera; phylogenetic relationships ) . Material. Twenty-three metamorphosed females, 11.0-97.0 mm: LACM 18 (11.0- 97.0 mm); SU 1 (20.0 mm); UW 2 (39.0- 68.0 mm); ZMUC 2 (20.5-40.0 mm). Diagnosis. The characters of the esca easily separate C. melanorhabdus from all other species of Chaenophryne (sec Key and Generic description). In addition, the ratio between the number of upper and lower jaw teeth is less than that of C. draco but significantly greater than that of C. longiceps and C. ramifera (Fig. 5). The illicium of C. melanorhabdus is shorter than that of its congeners (Fig. 3). Finally, fewer pectoral fin-rays, and fewer dorsal and anal rays, help to distinguish C. melanorhabdus from C. longiceps and C. ramifera, respectively (Table 2). Description. Esca with a single, elongate, internally pigmented (except for tip), an- terior appendage, less than one-fourth to nearly one-third length of escal bulb; medial appendages absent; a posterior appendage consisting of a swollen basal portion and a somewhat compressed, distal crest with a posterior filament or filaments and a pair of anterior lobes each bearing numerous filaments; a filamentous, antero- lateral appendage on each side; basal series of filaments absent; subcutaneous, inter- nally pigmented, descending appendage well-developed, often extending full length of illicium (Fig. 9). Illicial length 20.1^1.3 percent of SL (Fig. 3); escal bulb width 2.1-6.3 percent of SL (Fig. 4); total number of teeth in upper jaw 21^45, in lower jaw 26-42; ratio between number of teeth in upper jaw to number of teeth in lower jaw .78-1.30 (Fig. 5); vomerine teeth 4-7; D. 6-8 (of 19 specimens only one had D. 8); A. 5-6; P. 16-18 (Table 2). Rest of characters as for genus. Distribution. Chaenophryne melanorhab- dus appears to be restricted to the western continental slope of North and Central America, ranging from approximately 46°N, 125 °W in Pacific Subarctic Water, through the mixed Transition Zone of the California Current (18 specimens), into the Eastern Pacific Equatorial waters of the Gulf of Panama (type locality) (Fig. 11). Based on maximum depths reached by fishing gear, metamorphosed female speci- mens of C. melanorhabdus are vertically distributed between approximately 200 m and an unknown lower limit. Eighty-three percent of the known material (19 individu- als), including the largest specimens (35.0 mm and larger) was collected by gear fished below 450 m. Sufficient material is known from the Transition Zone of the California Current for analysis of vertical data by a procedure outlined by Gibbs (1969; see Methods and Materials). Results indicate a concentration between 300 and 1000 m (Table 4). Comments. Beebe and Crane (1947:158) synonymized C. melanorhabdus and C. columnifera with C. parviconus considering the material to represent stages of develop- ment. Bertelsen (1951:114, 117) agreed Anglerfishes of the Genus Chaenophryne • Pietsch 91 B Figure 9. Escae of Chaenophryne melanorhabdus: A. holotype of C. melanorhabdus, ZMUC P92117, 40.0 mm, lateral and anterior views reversed from Regan and Trewavas, 1932; B. LACM 30387-1, 96.0 mm, left side, showing extent of subcutaneous de- scending appendage. I s Bulletin Museum of Comparative Zoology, Vol. 147, No. 2 with this action and the reasons for it, adding a number of additional forms to the synonymy of C. parviconus, including C. pterolophus (Table 1). Recent collecting in the Eastern Pacific especially bv the VELERO IV of the Allan Hancock Foun- dation and the University of Southern California, has yielded a number of speci- mens of Chaenophryne, including a com- plete growth series (12.0-97.0) that have a pair of filamentous, anterolateral escal appendages like those present in Regan and Trewavas' (1932) C. melanorhabdus and C. pterolophus. In all other characters this new material compares very well with the type material. Consequently, C. melanor- habdus, with C. pterolophus as a junior synonym, is removed from the synonymy of C. pariconus and given specific status. Chaenophryne ramifera Regan and Trewavas, 1932 Figure 10 Chaenophryne ramifera Regan and Trewavas, 1932:85, 90, fig. 146 (original description; single specimen; holotype, ZMUC P92119, 17.0 mm; DANA Station 3550(6), 7°10'N, 78°15'W; 3000 m wire; 0145 hr; 5 September 1928). Belloc, 1938:305, fig. 29 (after Regan and Trewavas, 1932; comparison with C. intermedia sp. nov.). Bertelsen, 1951:115, figs. 67D, 75, Table 24 ( description; comparison with all known material; in key; fimhriata, intermedia synonyms of ramifera). Pietsch, 1974a:33 ( listed ) . Chaenophryne fimhriata Regan and Trewavas, 1932:85, 90, fig. 145 (original description; single specimen, holotype, ZMUC P92120, 16.5 mm; DANA Station 3917(3), 1°45'N, 71°05'E; 3200 m wire; 1800 hr; 5 December 1929). Belloc, 1938:305, fig. 27 (after Regan and Trewavas, 1932; comparison with C. inter- media sp. nov. ) . Chaenophryne intermedia Belloc, 1938:305, figs. 24, 28 (original description; single specimen; holotype, MNLR P449, 14.0 mm; PRESIDENT THEODORE TISSIER Station 708, 14°54'N, 23 15'W; 1000 m wire; 15 May 1936). Chaenophryne pads Koefoed, 1944:9, pi. II, figs. 6, 7 ( original description; single specimen; holo- type ZMB 4301, 15.0 mm; MICHAEL SARS Station 53, 34°59'N, 33°01'W; 2600 m wire; 8-9 June 1910). Table 4. Vertical distribution of Chaeno- phryne MELANORHABDUS BASED ON SPECIMEN COL- LECTED BY THE VELERO IV OFF SOUTHERN CALI- FORNIA. Meter-hours and specimens expressed as percent of total. See text for methods of calculation. Depth (m) Meter-hours Specimens 0-100 4.3 0 101-200 2.1 0 201-300 6.3 0 301-400 7.7 16.7 401-500 5.2 11.1 501-600 6.5 16.7 601-700 8.2 33.3 701-800 8.2 0 801-900 8.3 0 901-1000 6.0 11.1 1001-1250 14.4 0 1251-1500 17.7 11.1 1501-1750 3.4 0 1751-2000 1.1 0 2001-3000 0.7 0 3000 0 - Number of specimens 18 Number of meter-hours 10143.7 Number of hauls 547 Material. Fifteen metamorphosed fe- males, 13.5-55.5 mm: IOM 1 (16.0 mm); ISH 5 (26.0-55.5 mm); MCZ 3 (13.5-39.0 mm); MNLR 1 (14.0 mm); UMML 1 (36.5 mm); ZMB 1 (15.0 mm); ZMUC 3 (16.5- 33.5 mm). Diagnosis. The characters of the esca distinguish C. ramifera from all other spe- cies of Chaenophryne (see Key and Generic description). In addition, the low ratio be- tween the number of upper and lower jaw teeth further separates this species from C. draco and C. melanorhabdus (Fig. 5). The illicium of C. ramifera is longer than that of its congeners (Fig. 3). Chaenophryne ramifera has fewer pectoral fin-rays than C. longiceps, and a greater number of dorsal and anal rays than all other species (Tabic 2). Description. Esca with a single, elongate, internally pigmented, anterior appendage, approximately one-fourth to nearly one- Anglerfishes of the Genus Chaenophryne • Pietsch 93 B Figure 10. Escae of Chaenophryne ramifera: A. holotype of C. intermedia, MNLR P449, 14.0 mm, lateral view reversed from Belloc, 1938; B. holotype of C. fimbriata, ZMUC P92120, 16.5 mm, lateral view reversed from Regan and Trewavas, 1932; C. holotype of C. ramifera, ZMUC P92119, 17.0 mm, lateral view reversed from Regan and Trewavas, 1932; D. ISH 2247/71, 26.0 mm, anterior view; E. UMML 29702, 36.5 mm, left side. third length of escal bulb, trilobed distally with three round translucent windows; anterior appendage flanked laterally by an elongate, unpigmented swelling; medial appendages absent; a posterior appendage consisting of a swollen basal portion and a somewhat compressed distal crest with a terminal, anteriorly directed, crescent- shaped filament and a pair of anterior lobes each bearing none to several filaments; two or three, filamentous, anterolateral append- ages on each side; a basal series of seven to twelve filaments on each side; subcutane- ous, internally pigmented, descending ap- pendage extremely well-developed in some specimens (see Comments below), emerg- ing as a free, anteriorly and dorsally di- rected tentacle (Fig. 10). 94 Bulletin Museum of Comparative Zoology, Vol. 147, No. 2 Illicial length 32.8-47.4 percent of SL (Fig. 3); escal bulb width 4.5-6.5 percent of SL (Fig. 4); total number of teeth in upper jaw 25-51, in lower jaw 33-53; ratio between number of teeth in upper jaw to number of teeth in lower jaw .76-.9S (Fig. 5); vomerine teeth 4-8; D. 7-8; A. 5-6; P. 16-19 (of 15 specimens only one had P. 16 and one had P. 19) (Table 2). Rest of characters as for genus. Distribution. Chaenophryne ramifera has a wide horizontal distribution occurring in all three major oceans of the world. In the Atlantic it appears to be restricted to the eastern side; the northern and westernmost record is at approximately 35°N, 33°W (the holotype of C. pads); seven additional specimens were collected between approxi- mately 11°N and 8°S, between 26°W and 4°E in the Gulf of Guinea. In the Indo- Pacific region, five records are known ranging across the Indian Ocean between 9°N and 2°S. The holotype, collected from the Gulf of Panama, is the only known specimen from the Pacific Ocean (Fig. 11). Based on maximum depth reached by fishing gear, metamorphosed female speci- mens of C. ramifera are distributed verti- cally between approximately 200 m and an unknown lower limit. Large specimens may be captured at relatively shallow depths: a 35.0 mm specimen was collected by gear fished above 200 m, a 36.5 mm specimen by gear fished above 550 m. Eighty-seven percent of the known material (13 individu- als) was captured by gear fished below 550 m, 47 percent (7 individuals) by gear fished below 1000 m. Material is not sufficient from any one geographic area for a more analytical treatment of distributional data. Comments. I have not seen the holotype of C. pads. From Koefoed's (1944:9, fig. 6, 7 ) description and figures, there can be little doubt that this is a synonym of C. ramifera. The anterolateral appendages and basal series of filaments have apparently been lost; but the "trilobed stigma" unique to C. ramifera, with its three translucent windows at the tip of the unpaired an- terior escal appendage, is well-developed. Chaenophryne pads is removed from the synonymy of C. parviconus and placed within that of C. ramifera. The development of the subcutaneous, descending escal appendage of C. ramifera does not appear to be correlated with on- togeny as thought by Bertelsen (1951:118). Of the 15 known specimens of this spe- cies, only five have an esca with the descending appendage emerging as a free tentacle. These five specimens range from 13.5 mm to 36.5 mm. In all other known individuals the descending appendage is quite short, only 16.3 percent of the illicial length in the largest known specimen (55.5 mm ) . Species Incertae Sedis Chaenophryne galeatus Koefoed, 1944, nomen nudum. Chaenophryne galeatus Koefoed, 1944:8. Comments. This name was used by Koefoed in a manuscript dated 1918 (not seen by me), and later mentioned in pub- lished form (Koefoed, 1944:8) without ap- plication to a description or type. EVOLUTIONARY RELATIONSHIPS Chaenophryne is unique in several ways and does not appear to be phylogenetically closely related to any other oneirodid genius (see Pietsch, 1974a:89, fig. 104). Evolution- ary relationships among the species of Chaenophryne are deduced on the basis of eight morphological characters chosen for their intraspecific stability and interspecific variability. These characters are summa- rized below with a discussion of the evo- lutionary direction of each. Methods used in determination of directional change in character states follow Pietsch ( 1974a ) . For each character the primitive state is given a lower case letter, the derived state a capital letter, and a secondary derivation, a capital letter starred. Anglerfishes of the Genus Chaenophryne • Pietsch 95 Bulletin Museum of Comparative Zoology, Vol. 147, No. 2 1. Length and shape of frontal bones. Elongation and depression of the frontals is a general trend found within the family Oneirodidae and within several of its contained genera (Dolo- pichthys, Oneirodes; Pietseh, 1972b, 1974a). The relatively short, dorsally convex frontals of C. longiceps, C. draco, and C. melanorhabdus (Fig. 1A- C) are considered to represent a primi- tive state (a); the longer, dorsally con- cave frontals of C. ramifera (Fig ID) are a derived condition (A). 2. Anterolateral escal appendages. Within more speciose oneirodid genera (Dolo- pichthys, Oneirodes; Pietseh, 1972b, 1974a) there is a trend toward an increase in morphological complexity of the esca. The absence of antero- lateral appendages in C. longiceps and C. draco (Figs. 7, 8) is a primitive condition (b); whereas, the presence of these appendages in C. melanorhabdus and C. ramifera (Figs. 9, 10) repre- sents a derived condition (B). 3. Basal series of escal filaments. The absence of a series of filaments arising from the base of the escal bulb in C. longiceps, C. draco, and C. melanorhab- dus (Figs. 7-9) is a primitive condition (c); presence of this series in C. rami- fera (Fig. 10) is a derived state (C). 4. Escal bulb width. The width of the escal bulb of C. longiceps (Fig. 4) is like that of nearly all other ceratioids and is thought to represent a primitive condition (d). The considerably nar- rower bulb of C. draco, C. melanorJial)- dus, and C. ramifera (Fig. 4) is con- sidered a derived state (D). 5. Illicial length. Elongation of the illicium is a trend found in several oneirodid genera (Dolopichthys, Onei- rodes, Loplwdolos; Pietseh, 1972b, 1974a, b). The short illicium of C. melanorhabdus represents a primitive conditio] i (e), an illicium of inter- mediate length found in C. longiceps and C. draco represents a derived state (E), and the relatively long illicium of C. ramifera a further derived state (E*; Fig. 3). 6. Pectoral fin-rays. The loss of fin-rays is a general trend found within the Ceratioidei. Among species of Chaeno- pJiryne two character states are recog- nized: 18-22, rarefy 17 (C. longiceps) (f); 16-18, rarely 19 (C. draco, C. melanorhabdus, C. ramifera) (F; Table 2). 7. Dorsal and anal rays. The loss of fin- rays is a general trend found within the Ceratioidei. Among species of Chaeno- phryne two character states are recog- nized: D. 8, rarely 7, A. 6, rarely 5 (C. ramifera) (g); D. 6-7, rarely 8, A. 5, rarely 6 (C. longiceps, C. draco, C. melanorhabdus) (G; Table 2). 8. Batio between number of upper and lower jaw teeth. Nearly all ceratioids have a greater number of teeth in the lower jaw relative to the number of teeth in the upper jaw, resulting in a low ratio between these two counts. The low ratio found in C. longiceps and C. ramifera is considered a primitive state (h). The considerable higher ratio found in C. draco and C. melanorhal)- dus is the derived state (H; Fig. 5). A hypothetical phylogeny for the species of ChaenopJiryne was constructed on the basis of shared patterns of derived character states. Each level in the phylogeny (Fig. 12) represents the maximum number of shared derived states for the maximum number of taxa (Pietseh, 1974a: 87). For comparison of the four species with respect to evolutionary specialization, the following numbers were assigned to the character states: 0 = lower case letter, 1 = capital letter, 2 = capital letter starred. The maxi- mum index of specialization is 9. The actual totals for the taxa are: longiceps 2, draco 5, melanorhabdus 5, ramifera 7. The Anglerfishes of the Genus Chaenophryne • Pietsch 97 RAMIFERA 7 ABCE -* DRACO 5 MELANORHABDUS 5 Figure 12. Proposed phylogenetic relationships of species of Chenophryne. Each level in the phylogeny represents the maximum number of shared derived character states for the maximum number of taxa. See text for details. length of the line between any two levels in the phylogeny (Fig. 12) corresponds to the degree of evolutionary specialization found between those two levels. For example, the distance from the base of the phylogeny to ABCE* is seven units long, equalling the sum of the values of evo- lutionary specialization assigned to ad- vanced character states DFABCE*. The number of derived character states shared between all possible species pairs of Chaenophryne is summarized in Table 5. From this character analysis, it appears that C. longiceps is the least derived mem- ber of the genus having the least number of derived character states; it is most closely related to C. draco with which it shares a greater number of derived states than any other species. Chaenophryne ramifera is the most derived species, having the greatest number of derived character states; it is most closely related to C. melanorhab- dus with which it shares a greater number of derived states than any other species. Chaenophryne draco and C. melanorhabdus are more closely related to each other than to any other species, sharing the greatest number of derived character states of any species pair (Table 5). ACKNOWLEDGMENTS I thank Erik Bertelsen and Karel F. Liem for critically reading the manuscript and offering suggestions. Thanks are also due the following persons and their institu- tions for making material and collection data available: Donn E. Bosen and Gareth 98 Bulletin Museum of Comparative Zoology, Vol. 147, No. 2 I lble 5. Number of derived character states shared between all possible species pairs of Chaenophryne. Numbers in parentheses refer to characters summarized in the text. Species pair Number of derived states shared draco-melanorhabdus melanorhabdus-ramifera longiceps-draco draco-ramifera longiceps-melanorhabdus longiceps-ramifera 4(4,6-8) 3(2,4,6) 2(5,7) 2(4,6) 1(7) 0 ]. Nelson (AMNH); Alwyne Wheeler (BMNH); Keith S. Thomson (BOC); William N. Eschmeyer and Tomio Iwamoto (California Academy of Sciences, San Fran- cisco); Thomas A. Clarke (Hawaii Institute of Marine Science, University of Hawaii); T. S. Rass and N. V. Parin (IOM); Nigel Merrett and Julian Badcock (IOS); Gerhard Krefft (ISH); Robert J. Lavenberg and Jerry W. Neumann (LACM); Robert G. Schoknccht (MCZ); Michael Legand (Office de la Recherche Scientifique et Technique Outre-Mer, Noumea, New Caledonia); William G. Pearcy (OSUO); W. B. Scott (ROM); Catherine Rainwater (University of Southern California, Los Angeles); C. Richard Robins and Robert N. Lea (UMML); Robert H. Gibbs, Jr. and Stanley H. Weitzman (USNM); Richard H. Backus, James E. Craddick and Bruce H. Robinson (Woods Hole Oceanographic Institution); A. P. Andriashev (Zoological Institute, Academy of Sciences of the USSR, Lenin- grad); Hans Kauri (ZMB); and Erik Ber- telsen (ZMUC). In addition, I am grateful to Andrea West for typing the manuscript, and to Elizabeth Anne Hoxie for her excel- lent rendering of Figures 6, 7E and F. For gracious hospitality extended to me during visits to their institutions, I am particularly grateful to Erik Bertelsen (ZMUC); Gerhard Krefft (ISH); Nigel Merrett and Julian Badcock (IOS); and Alwyne Wheeler (BMNH). The work was supported by a grant from the National Science Foundation, No. GB- 40700, for studies of the Ceratioidei of the western North Atlantic Ocean. Partial as- sistance from the William F. Milton Fund of Harvard University, and the Johannes Schmidt Fund of the University of Copen- hagen is also gratefully acknowledged. LITERATURE CITED Beebe, W. 1932. Nineteen new species and four post-larval deep-sea fish. Zoologica, N. Y. 13: 47-107. Beebe, W., and J. Crane. 1947. Eastern Pacific Expedition of the New York Zoological So- ciety. XXXVII. Deep-sea ceratioid fishes. Zoologica, N. Y. 31 (11): 151-182. Belloc, G. 1938. Resultats des croisieres scien- tifiques du navire "President Theodore- Tissier." Liste des poissons pelagiques et bathypelagiques captures au cours de la cinquieme croisiere avec diagnoses prelimi- naries de deux especes nouvelles. Rec. Trav. Peches marit. Paris 11 (3): 281-313. Bertelsen, E. 1943. Notes on the deep-sea angler-fish Ceratias holboelli Kr. based on specimens in the Zoological Museum of Copenhagen. Vidensk. Medd. fra Dansk naturh. foren., Bd. 107: 185-206. . 1951. The ceratioid fishes. Ontogeny, taxonomy, distribution and biology. Dana Rep. 39, 276 pp. Bradbury, M. G. 1967. The genera of batfishes (family Ogcocephalidae). Copeia 1967: 399-422. Burton, M. 1933. Zoological Record 69 (13): 1-62. Carlsberg Foundation. 1934. Introduction to the reports from the Carlsberg Foundation's Oceanographical Expedition Round the World 1928-30 Dana Rep, 1, 130 pp. Fowler, H. W. 1936. The marine fishes of West Africa. Based on the collections of the American Museum Congo Expedition, 1909- 1915. Bull. Amer. Mus. Nat. Hist. 70 (2): 607-1493. Gibbs, R. H., Jr. 1969. Taxonomy, sexual di- morphism, vertical distribution, and evolution- ary zoogeography of the bathypelagic fish genus Stomias (Stomiatidae). Smithsonian Contrib. Zool. 31: 1-25. Grey, M. 1956. The distribution of fishes found below a depth of 2000 meters. Fieldiana, Zool. 36 (2): 75-337. Grinols, R. B. 1966. Addition of adult angler- fish, Chaenophryne parviconus Regan and Trewavas ( Pisces : Oneirodidae ) , to the east- ern subarctic Pacific Ocean. Calif. Fish and Game 52 (3): 161-165. AXGLERFISHES OF THE GENUS ChAENOPHRYNE • PietSch 99 Imai, S. 1942. On some deep-sea angler-fishes obtained in Sagami Bay and Suruga Bay. J. Dept. Agric, Kyusyu Imperial Univ. 7 (2): 37-48. Koefoed, E. 1944. Pedicnlati from the "Michael Sars" North Atlantic Deep-sea Expedition 1910. Rep. Sci. Res. "Michael Sars" Exped. IV. 2(1): 11-18. Maul, G. E. 1962. On a small collection of ceratioid fishes from off Dakar and two recently acquired specimens from stomachs of Aplianopus carbo taken in Madeira (Me- lanocetidae, Himantolophidae, Diceratiidae, Oneirodidae, and Ceratiidae) Bol. Mus. Mun. Funchal 16 (54): 5-27. Parr, A. E. 1927. Scientific results of the third Oceanographic Expedition of the "Pawnee" 1927. Bull. Bingh. Oceanogr. Coll. 3 (1): 1-34. . 1937. Concluding report on fishes. Bull. Bingh. Oceanog. Coll. 3 (7): 1-79. Pietsch, T. W. 1969. A remarkable new genus and species of deep-sea anglerfish (family Oneirodidae ) from off Guadalupe Island, Mexico. Copeia 1969: 365-369. . 1972a. A review of the monotypic deep-sea anglerfish family Centrophrynidae: taxonomy, distribution and osteologv. Copeia 1972: 17-47. — . 1972b. Ergebnisse der Forschungsrei- sen des FFS "Walther Herwig" nach Siida- merika. XIX. Systematics and distribution of ceratioid fishes of the genus Dolopichthys (family Oneirodidae) with the description of a new species. Arch. Fischereiwiss. 23 ( 1 ) : 1-28. — . 1974a. Osteology and relationships of ceratioid anglerfishes of the family Onei- rodidae, with a review of the genus Oneirodes Liitken. Xat. Hist. Mus. Los Angeles Co., Sci. Bull., 18: 1-113. . 1974b. Svstematics and distribution of ceratioid anglerfishes of the genus Lophodolos (family Oneirodidae). Breviora 425- 1-19. Regax, C. T. 1925. New ceratioid fishes from the N. Atlantic, the Caribbean Sea, and the Gulf of Panama, collected by the "Dana". Ann. Mag. Nat. Hist., Ser. 8, 8 (62): 561- 567. . 1926. The pediculate fishes of the sub- order Ceratioidea. Dana Oceanog. Rep. 2, 45 pp. Regax, C. T., and E. Trewavas. 1932. Deep- sea anglerfish (Ceratioidea). Dana Rep. 2, 113 pp. US ISSN 0027-4100 Bulletin of the Museum of Comparative Zoology The American orb-weaver genera Larinia, Cercidia and Mangora north of Mexico (Araneae, Araneidae) HERBERT W. LEVI HARVARD UNIVERSITY CAMBRIDGE, MASSACHUSETTS, U.S.A. VOLUME 147, NUMBER 3 23 MAY 1975 PUBLICATIONS ISSUED OR DISTRIBUTED BY THE MUSEUM OF COMPARATIVE ZOOLOGY HARVARD UNIVERSITY Breviora 1952- Bulletin 1863- Memoirs 1864-1938 Johnsonia, Department of Mollusks, 1941- Occasional Papers on Mollusks, 1945- SPECIAL PUBLICATIONS. 1. Whittington, H. B., and E. D. I. Rolfe (eds.), 1963. Phylogeny and Evolution of Crustacea. 192 pp. 2. Turner, R. D., 1966. A Survey and Illustrated Catalogue of the Teredini- dae (Mollusca: Bivalvia). 265 pp. 3. Sprinkle, J., 1973. Morphology and Evolution of Blastozoan Echinoderms. 284 pp. 4. Eaton, R. J. E., 1974. A Flora of Concord. 236 pp. Other Publications. Bigelow, H. B., and W. C. Schroeder, 1953. Fishes of the Gulf of Maine. Reprint. Brues, C. T., A. L. Melander, and F. M. Carpenter, 1954. Classification of Insects. Creighton, W. S., 1950. The Ants of North America. Reprint. Lyman, C. P., and A. R. Dawe (eds.), 1960. Symposium on Natural Mammalian Hibernation. Peters' Check-list of Birds of the World, vols. 2-7, 9, 10, 12-15. Proceedings of the New England Zoological Club 1899-1948. (Complete sets only.) Publications of the Boston Society of Natural History. Price list and catalog of MCZ publications may be obtained from Publications Office, Museum of Comparative Zoology, Harvard University, Cambridge, Massa- chusetts, 02138, U.S.A. © The President and Fellows of Harvard College 1975. CD E o o O CD CO D) CO E to O CD .c CD LU d 'c CD .C Q. CD Dl O O sz a. •d c CO o CO c o u CO o CD CO c CD CD CD O CD Q. en c O THE AMERICAN ORB-WEAVER GENERA LARINIA, CERCIDIA AND MANGORA NORTH OF MEXICO (ARANEAE, ARANEIDAE) HERBERT W. LEVI1 Abstract. There are three species of Larinia found north of Mexico, one of which, Larinia directa, occurs as far south as southern Brazil. The closest relatives are found in Africa. Cercidia prominens, possibly introduced from Europe, is found from the Great Lakes region to New England in relatively undisturbed habitats. Cercidia and Larinia are closely related offshoots of Araneus. While M angora species are found in all parts of the world, most are Neotropical. Of the seven species of M angora found north of Mexico, three are tropical. As a result of the discovery that Cercidia funebris is an older name for Hypsosinga singaeformis, this rarely used name must be changed to H. funebris. INTRODUCTION Revisionary studies of invertebrates try to accomplish several different tasks. The first task is to analyze a group of species and genera so that they can be arranged in phylogenetic order; the second is to devise a guide for the determination of the species; and the third, to discover generalizations of interest to biology. The first two aims are somewhat contradictory, since, in order to provide an ideal phylogenetic system, the group should be studied worldwide. Nevertheless, I have limited myself to the American fauna because a worldwide study would make the determination of the spiders in North America exceedingly cumbersome, as well as greatly delaying the publication of the results. An additional 1 Museum University. of Comparative Zoology, Harvard problem, which does not plague students of vertebrates, is the difficulty of obtaining sufficient museum specimens of the genera under study, because most collections are not and cannot be sorted to genera. But for nonspecialists to be able to determine spiders of North America is an urgent necessity, as arachnologists all have teach- ing, research, and curatorial obligations, and do not have time to determine all the specimens submitted by entomologists in- terested in insect ecology and insect preda- tors. Ironically, a taxonomic revision some- times is the stimulus for swamping its author with requests for determinations of the studied species. To facilitate determinations the taxo- nomic studies of the American orb-weavers will be published in two series: first, the species of araneid genera north of Mexico; later, the very many more species from the rest of the Americas. For the purposes of determination, it is obviously much easier to keep the fauna of America north of Mexico distinct from the rich tropical fauna. However, it is difficult to revise North American species without reference to tropical species and the better known temperate species of Europe. Even though I am limiting this work to the Araneidae of northern parts of North America, I am examining and illustrating all the types of species names from other parts of America to minimize later changes and to get a better idea of the diversity of species and Bull. Mus. Comp. Zool. 147(3): 101-135, May, 1975 101 102 Bulletin Museum of Comparative Zoology, Vol. 147, No. 3 genera. Unfortunately, many of these types are known from only one sex or from im- matures. Of course, I am also examining Specimens from other parts of the world, as they can be found in American col- lections. Perhaps it is the broader aspect of the study, the revision of the genera within the family, that makes me uneasy about using the small genera in Grasshoffs excellent studies (1970a, b, c), even though each is a natural group of closely related species. Pending better knowledge of spiders, it may be better and easier to keep genera large. Also, judging from my study, it appears that Larinia and Cercidia are veiy closely re- lated to Araneus, but Mangora is a special- ized group far removed from Araneus. Nevertheless, for this paper, and until I know more about the groups, I will keep Larinia and Mangora together, following Simon (1895) and Grasshoff. Larinia does not rest in the center of the web, but on vegetation to the side, as do most other relatives of Araneus. Mangora rests in the hub, and, like many other Araneidae that rest in the center (e.g., Argiope, Gasteracantha, Acanthepeira, Cyclosa, Leucauge and Mierathena), is more specialized in structure. Why there should be a relationship between the rest- ing place and the appearance of the spider I do not know, except perhaps for the obvious explanation that an exposed spider would be more subjected by predators to selection pressure favoring inconspicuous- ness or the ability to disappear rapidly. Specimens for this study were made available by J. A. Beatty of the Southern Illinois University, D. Bixler, J. E. Carico, R. E. Crabill of the U. S. National Museum, C. D. Dondale of Canadian National Col- lections, A. Timotheo da Costa of the Museu Nacional, Rio de Janeiro, H. Dybas of the Field Museum of Natural History, W. J. Gertsch of the American Museum of Natural History, M. Grasshoff of Sencken- berg Naturmuseum, M. Hubert of Museum National d'Histoire Naturelle, Paris, D. Lowrie, W. Peck of the Exline-Peck col- lection, S. Rieehert, V. D. Roth, R. X. Schick of the California Academv of Sciences, W. Sedgwick, K. Stone, W. Star^ga of the Polish Academy of Science, F. Wanless and D. Norman of the British Museum (Natural History), H. V. Weems of the Florida State Collection of Arthro- pods. T. Kronestedt of the Natural History Museum, Stockholm, provided information, and F. Enders transmitted specimens from South American museums. Lorna R. Levi corrected the syntax. The study and its publication were supported in part by National Science Foundation research grant GB-36161. Larinia Simon Larinia Simon 1874, Arachnides de France, 1: 115. Type species by monotypy Epcira lineata Lucas, 1846. The gender of the generic name is feminine. Drcxdia McCook, 1892, Proc. Acad. Natur. Sci. Philadelphia 1892: 127. Type species by mono- typy Epeira dirccta Hentz. In several excellent studies, Grasshoff (1970a, b, c, 1971) split Larinia into nu- merous genera (Kilima, Larinopa, Siwa, Paralarinia, Faradja, Mahembea, Larinia- ria), each including related species. But if I were to follow Grasshoff, I would have to make additional new genera for the tropical American species. Perhaps the Grasshoff genera should be considered subgenera (Grasshoff, in letter). Even this would be unsatisfactory, however, as it is not possible to express all phylogenetic relationships through nomenclature. For the time being I prefer to use Larinia in its broad sense. The type species of Larinia Simon is L. lineata, occurring in the Mediterranean region. It has recently been illustrated by Grasshoff ( 1970b ) , as have most species other than American ones. In this discus- sion I refer to the various papers of Grass- hoff. Description. Larinia species arc elongate, the abdomen longer than wide, often pointed anteriorly above the carapace and projecting behind and above spinnerets Larinia, Cercidia, and Mangora • Levi 103 (Figs. 5, 6). Behind the median eyes the yellowish carapace may have a double line that fuses to a .single median longitudinal line. The legs are not banded, but may have ventral and sometimes dorsal black spots. The sternum may be all black, or may have the sides but not the anterior black. The abdomen has a pattern, usually variable within a species, of longitudinal lines, sometimes with two rows of black spots; the sides are white. The venter of the abdomen is gray to black, the dark area surrounds the spinnerets. Anteriorly, the dark area encloses a longitudinal white patch that is often divided by a median black line (Figs. 6, 16, 25). The height of the clypeus equals about the diameter of the anterior median eyes. The carapace has a median longitudinal thoracic line more pronounced in males. There are four teeth on the anterior margin of each chelicera. Genitalia. The female epigynum has a scape, which I think is torn off in speci- mens I have examined of L. lineata of the Old World. The internal genitalia have a duct connecting the large opening with the seminal receptacles ( Figs. 9, 12, 19, 28, and illustrations of Grasshoff's papers). Along the duct are numerous, irregularly placed, kernel-like structures, each connected with the lumen of the connecting duct. Pre- sumably these structures are glands that have the outer layer sclerotized. The embolus of the palpus is ribbon- shaped with the edge facing mesally, the broad side apically. The embolus tip breaks off during mating and stays behind in the female duct [Figs. 31-33 and Grass- hoff, 1970a, fig. If for Larinia (Kilima) conspersa; 1970b, fig. 7d for Larinia chloris; 1971, fig. 33d for Larinia ( Drexelia ) trifida; fig. 38b for Larinia (Drexelia) bifida: and fig. 44a for Larinia (Drexelia) ishaniio.} The tip contains a duct ( Figs. 31-33 ) . The palpus has a median apophysis projecting to the side of the palpus, a subterminal apophysis and a terminal apophysis ( Fig. 3). I am not certain that I have been consistent in naming the subterminal apophysis in various genera. There is a hematodocha between embolus and radix, and a large distal hematodocha between the sclerotized portion of the terminal and subterminal apophysis and the embolus. In Kilima, according to Grasshoff, tegulum and stipes are fused. The structure of the male palpus is so similar to that of Araneus that I believe Larinia is close to Araneus. If this is the case, the tip or cap of the embolus, which breaks off during mating, probably has a lumen in Araneus too. Grasshoff found a lumen in the embolus tip of Araneus, but I have been unable to confirm its presence in this more sclerotized structure. Diagnosis. Besides having diagnostic ventral coloration very different from that of Araneus and Metepeira (which has a white median ventral longitudinal band framed by black), Larinia differs from Araneus by the eye arrangement. The anterior median eyes are the largest, sepa- rated by more than one diameter; the posterior median eyes are subequal to an- terior medians or slightly smaller (to 0.8 diameter), and are almost touching or are separated from each other by their radius at most. The anterior median eyes are 1.2 to 1.5 diameters apart. The median ocular quadrangle is more than twice as wide in front as behind (Figs. 5, 15, 24). Males are smaller than females and have longer legs. In males of American species the palpal femur has a basal ventral tooth (as in Araneus); the palpal patella has two long macrosetae (one in Kilima, Grasshoff, 1970a); the first coxa lacks a hook (but has one in Kilima ) ; the first or second legs have stronger macrosetae than in the female. Note. Several South American Eustala species resemble Larinia, including the commonest one, E. fuscovittata. However, the genitalia differ; the Eustala epigynum has an anteriorly directed scape and the palpus has a large, nonsclerotized median apophysis parallel to the long axis of the appendage. The closeness of the posterior 104 Bulletin Museum of Comparative Zoology, Vol. 147, No. 3 mm. • .-* i f • ■i - -sfc,'gfc* Plate 1. Larinia directa, penultimate female, from North Carolina in the laboratory. median eyes to each other may be an adaptation to the leg position of the resting animal (Plate 1). Relationship. I believe this genus is close to Araneus but not to Mangora. Larinia lacks the high carapace and the feathered trichobothria characteristic of Mangora, and the structure of the genitalia differs considerably. Traditionally Larinia has been placed close to Mangora. Archer (1951b: 26) also considered Larinia close to Araneus. Grasshoff (1971) shows that the species closest to the North American ones are found in Africa. I assume that of the three species found north of Mexico, Larinia directa, with a range extending from south- ern Brazil to southern United States, is the ancestral one. Larinia boreal is and L, fanuilatoria appear to be two temperate species, one eastern, one western, derived from L. directa. Now the ranges of all three overlap in the southwestern United States. Natural History. Specimens of Larinia are commonly collected by sweeping. Larinia directa is found in open areas; it sits in the hub of the web at night and in vegetation to the side of the web by day. ( Frontispiece and W. Eberhard, personal communication). Distribution. Worldwide but absent from central and northern Europe. Misplaced species. Larinia forata Key- serling 1893 ( = Epeira forata McCook 1893) is Araneus niveus (Hentz). NEW SYNONYMY. The specimen has a Marx label "Epeira muraria Mx, Sta. Rosa, Cala. no. 442" and a pencil label by McCook Epeira forata. The locality appearing in the publication and on the label is in error; Araneus niveus and close relatives are not found in California. The specimen is in the United States National Museum. Larinia nigrofoliata Keyserling, 1884 = Araneus miniatus (Walckenaer, 1841). Separating species. The three species found in the Nearctic area differ in propor- tions, size, and in the genitalia (the base and scape of the epigynum, and the shape of the embolus ) . The palpus is similar and lightly sclerotized in all three species, and the median apophysis has greater variation within each species than among the three species (Figs. 37-46). Key to American Species of Larinia Nohth of Mexico 1 First patella and tibia of female 1.6 to 2.2 times carapace length, distal rim of epigynal scape wide and smooth ( Figs. 7, 10, 11); first patella and tibia of male 2.0 to 2.4 times carapace length, New Jersey, Arkansas, southern Arizona, south- ern California to southern Brazil ( Map 1 ). directa First patella and tibia of female 1.2 to 1.6 times carapace length, distal rim of scape not wide and smooth (Figs. 17, 20, 21, 26); first patella and tibia of male 1.4 to 1.7 times carapace length. 2 2(1) Sternum black, venter of abdomen with black marks (Fig. 25); females with transverse striations on base of scape, and lateral lip of epigynal base diagonal ( Figs. Larinia, Cercidia, and Mancora • Levi 105 • •/ \ • »-• •— ■'— * a* Map 1. Distribution of Larinia directa (Hentz). 26, 29, 30 ) ; total length of males less than 3.3 mm; Wyoming, Idaho, southern Cali- fornia to Chiapas (Map 2). famulatoria Sternum with sides black at darkest, venter of abdomen with gray marks (Fig. 16); females with striations of scape curved posteriorly at middle, and lateral lip of epigynal base framing opening laterally and posteriorly forming a right angle (Figs. 17, 20, 21); total length of males more than 3.9 mm long; New England, southern Canada to eastern Washington, south to Virginia, New Mexico and Chihuahua ( Map 2 ) . borealis Larinia directa (Hentz) Frontispiece, Plate 1; Figures 1-12, 31, 34, 37-41; Map 1 Epeira directa Hentz, 1847, J. Boston Soc. Natur. Hist., 5: 478, pi. 31, fig. 21, 9. Type speci- mens from South Carolina and Alabama de- stroyed. Epeira rubella Hentz, 1847, J. Boston Soc. Natur. Hist., 5: 478, pi. 31, fig. 22, 9. Type speci- mens from Alabama destroyed. Epeira tetragnathoides O. P. -Cambridge, 1889, Biologia Centrali-Americana, Araneidea, 1: 16, pi. 7, figs. 9-10, 9 c5 • Female, male syntypes from Guatemala and Panama in the British Museum, Natural History. Name preoccupied by E. tetragnathoides Walckenaer. Epeira intercisa O. P. -Cambridge, 1889, Biologia Centrali-Americana, Araneidea, 1: 18, pi. 5, fig. 11, $. Male holotype from Bugaba, Panama in the British Museum, Natural History, lost. Drexelia directa, - McCook, 1892, Proc. Acad. Natur. Sci. Philadelphia, 127. McCook, 1893, American Spiders, 3: 249, pi. 6, figs. 10, 11, pi. 22, fig. 3, 9. F. P.-Cambridge, 1903, Bio- logia Centrali-Americana, Araneidea, 2: 461, pi. 106 Bulletin Museum of Comparative Zoology, Vol. 147, No. 3 -43. figs. 12, 13, 9,6- Roewer, 1942, Katalog der Araneae, 1: 764. Bonnet, 1956, Bibliogra- phia Araneoruin, 2: 1609. Grasshoff, 1971, Senckenbergiana Biol., 52: 93, fig. 45, 9. Epeira deludens Keyserling, 1893, Spinnen Ameri- kas, 4: 261, pi. 13, fig! 195, $, $. Male and female syntypes from Florida and Guatemala (Marx collection) in the U.S. National Museum, examined. Larinia bellona Banks, 1898, Proc. California Acad. Sci., ser. 3, 1(7): 257, pi. 15, fig. 6, 9, $. Female, male syntypes from Tepic, Mexico in the Museum of Comparative Zoology, examined. NEW SYNONYMY. Larinia directa, - Emerton, 1902, Common Spiders, p. 182, figs. 424-427, 9 , $ . Comstock, 1912, Spider Book, p. 508, fig. 545, 9 • Comstock, 1940, Spider Book, rev. ed., p. 521, fig. 545, 9 . Larinia cymotypa Chamberlin, 1924, Proc. Cali- fornia Acad. Sci., 12: 649, fig. 93. Female holotype from Gulf of California in the Cali- fornia Academy of Sciences, examined. NEW SYNONYMY. Larinia albonigra Franganillo, 1934, Mem. Soc. Cubana Hist. Natur., 8: 158. Male holotype from Cuba in Cuban Academy of Science, Havana, but with code number only, and no key to code. Franganillo, 1936, Los Aracnidos de Cuba, 81, fig. 38,