PI *£L ^ m — ^iinms^ m in \ ^ c/> — in _ If) RIES SMITHSONIAN INSTITUTION NOliniliSNI NVINOSHimS S3 1 H Vd 8 11 LIBRARIES SMITHSONIAN INSTITUTIO in 2 w z w z .... in MioiiiuMw b-> X in 2: t z t z M;:? > > '^0^ "S. > '00 5 ^ 5 Z W ^ z in 2 CO ^ 2 — liSNi NviNOSHiiws S3iavaan libraries Smithsonian institution NoiiniiiSNi nvinoshiiws siiavnai in ^ CO _ CO — CO O ^ 2 _i \ R I ES^SMITHS0NIAN“'|NSTITUTI0N NOIiniliSNl“'NVINOSHillMS S3 I dVd 8 II LI B RAR I ES^SMITHS0NIAN“|NSTITUTI0 2 [I 2: 2 r- 2 CD •; > CO ± CO X CO liSNI NVINOSHilWS SBIdVdail LIBRARIES SMITHSONIAN INSTITUTION NOliniliSNI NVINOSHimS S3IHVaa Z W ^ Z , W ^ ^ 2 » CO 2 CO i ^ ^ " S .-ife < X in O _ > 5 ■ 2 - 2 xijosHjS/ > R I ES'^SMITHSONIAN INSTITUTION N0linillSNI_NVIN0SHllWs‘^S3 I d Vd a 11 LI B RAR I ES^SMITHSONIAN INSTITUTIO CO — (O 2 \ ^ 2 ^ ^ H C H VvJ . Q '^-v Q “ ' q n^osv<^ -I 2 _j 2 _l 2 2 IlSNI NVINOSHIIWS S3ldVdail LIBRARIES SMITHSONIAN INSTITUTION NOliniliSNI NVINOSHlItMS S3IdVdai - 2 2: I- ^ 2 - - ro 00 /SiieTO^A ,- / A A'=:kw^'^\ .- 30 > 33 m >»; CO m m m C/5 A z CO ± CO X CO RIES SMITHSONIAN INSTITUTION NOliniliSNI NVINOSHIUNS S3ldVdail LIBRARIES SMITHSONIAN INSTITUTIC 1 iw I . 1 u I iv.yni^ 2 CO 2 .... CO 2 CO i\ o CO X >y t: ^ >-2 CO 2 CO ' y CO IlSNI NVINOSHIIWS S3ldVdail LIBRARIES SMITHSONIAN INSTITUTION NOliniliSNI NVINOSHIIIMS S3ldVda — "' — in — CO — CO I ■ t 1 1 I v'^«cv/4Z'**x. cn LU in R I ES^SMITHS0NIAN“'lNSTITUTI0N^N0linillSNl“NVlN0SHlllAIS~S3 I dVd 8 11 LI B RAR I ES~SMITHSON!AN_INSTITUTIC 2 <2 ^ _ _ O _ 2'?o>IoX O xSviTnTx _ jn. ^ O s m NLivAstiix m ^ NLjvAsv^ix rn ^ * rn ^ CO X CO — CO — CO IlSNI NVINOSHIIWS S3ldVdan LIBRARIES SMITHSONIAN INSTITUTION NOliniliSNI NVIN0SH1I^^JS S3ldVda 2 CO 2 , CO ^ ^ 2 » CO 2 g2 *>' ■" #23^ ? I < * s £ X CO o > 2 > to 2 CO ' z to ■ ■■’■ 2 CO RIES SMITHSONIAN INSTITUTION NOliniliSNI NVINOSHIIINS S3ldVdail LIBRARIES SMITHSONIAN INSTITUTIO c/) rr VA^>^ m rn CO ' — CO t: CO VARIES SMITHSONIAN INSTITUTION NOliniliSNI NVINOSHIIWS S3iavaan LI (f) 7' , CO Z CO BRARIES SMITHSONIAN INSTITUl CO ^ o \iVA2:;^ Q 'CiNoi^ — o z ■ -1 Z _J z RARIES SMITHSONIAN INSTITUTION NOliniliSNI MVINOSHIIWS S3iavaan NSTITU' miiisNi NviNOSHiiiAis S3iavaa z CO m xjA m CO — CO n LIBRARIES SMITHSONIAN INSTITUTION NOliniliSNI NVINOSHIIIAIS S 3 1 a V d Z CO Z V Z ^ ^ CO < . 2 _ CO z RARIES SMITHSONIAN . z UJ 2 > (O ‘ Z ‘ Z CO NSTITUTION NOliniliSNI NVINOSHIIWS S3iavaai1 libraries SMITHSONIAN CO ~ . CO 2 UJ ZtvIivArx “ .>sS!^ -> UJ ZCAi^vTrx^ tn NSTITU O DC^ _ x^oisgx o '‘X ” X^Aix^SZ O _j z _i z"^_i ^ -J ^ LnillSNI NVINOSHIIINS S3iavaan libraries SMITHSONIAN^INSTlTUTION^NOlinillSNI^NVINOSHlUMS^SS i a va ro ' ^ TO W r CO C/) — c/> ± cn RARIES SMITHSONIAN INSTITUTION NOliniliSNI NVINOSHlItMS S3iavaai1 LI "W BRARIES SMITHSONIAN INSTITU cc ^ ^ z ■rJIsTIT(J>x <; > z CO ■■" LnillSNI NviNOSHiiws S3iavaa > _ ^ CO T^LI B RAR I Es'^SMITHSONIAN INSTITUTION NOlinillSN CO X CO UJ CO ^ CO -I CC > z CO NVINOSHIIINS S3iaVi X N CO 1. .S» Uv h\ V:if ^■' .'•■(' ., ,'w „• ■>».’ iii», ftSSt'fe'VV' • , m-. :. ;»]?' ■ I :'V«'./, .' ,;. >!( -.V- I,.-|» ' VA^:;?A ■X . r-'-*' ‘ ’■':"■ ... ■„■*■■ C3 :■■ ^ ' Y ••iWllii * A> ' ■'■'. * ; ;■’ .' ".'M'l. . .llv.I'l^*jl^''^lil^■If'! iiiC;:fe'4 . .jv;i'^'%:;j^: -V. .j'l •J'i'f m 'V:iiiV''^:j:ii'’' T fi.-iiii'' Ai y »;V.. li ' *i:A..i'iii ,J / / " i- ' APPENDICULAR MYOLOGY AND RELATIONSHIPS OF THE NEW WORLD NINE-PRIMARIED OSGINES (AVES; PASSERIFORMES) NUMBER 7 ROBERT J, RAIKOW .y . PITTSBURGH, 1978 BULLETIN of CARNEGIE MUSEUM OF NATURAL HISTORY APPENDICULAR MYOLOGY AND RELATIONSHIPS OF THE NEW WORLD NINE-PRIMARIED OSCINES (AYES: PASSERIFORMES) ROBERT J. RAIKOW Research Associate, Section of Birds, Carnegie Museum of Natural History, and Department of Biological Sciences, University of Pittsburgh, Pittsburgh, Pennsylvania 15260 NUMBER 7 PITTSBURGH, 1978 BULLETIN OF CARNEGIE MUSEUM OF NATURAL HISTORY Number 7, pages 1-43, figures I-IO, tables 1-5 Issued 17 November 1978 Price: $3.50 a copy Craig C. Black, Director Editorial Staff: Hugh H. Genoways, Editor; Duane A. Schlitter, Editor; Stephen L. Williams, Associate Editor; Teresa M. Bona, Technical Assistant. © 1978 by the Trustees of Carnegie Institute, all rights reserved. CARNEGIE MUSEUM OF NATURAL HISTORY, 4400 FORBES AVENUE PITTSBURGH, PENNSYLVANIA 15213 CONTENTS Abstract 5 Introduction 5 Problems of Phylogeny Construction 5 Monophyly of the New World Nine-primaried Oscines 7 Materials and Methods 10 Dissection 10 Nomenclature 10 Determination of Primitive and Derived States 10 Muscles of the Forelimb H Muscles of the Hindlimb 17 Characterization of Taxa 25 Vireonidae 25 Parulidae 26 Thraupidae 28 Coerebidae 3I Emberizinae 31 Cardinalinae 3I Icteridae 32 Carduelinae 32 Drepanididae 33 A Phylogeny of the New World Nine-primaried Oscines 34 Cluster 1 35 Cluster 2 35 Cluster 3 3g Cluster 4 39 Cluster 5 39 Cluster 6 39 Proposed Classification 40 Conclusions 4I Acknowledgments 42 Literature Cited 42 ABSTRACT The gross morphology of the forelimb and hindlimb muscles was studied in approximately 100 species of songbirds, and ana- lyzed cladistically to construct a phylogeny of the New World nine-primaried oscines. Methods and problems of phylogenetic analysis are discussed, and the rationale for the proposed phy- logeny is presented. It is suggested that the Parulidae are the most primitive family of the group, the Thraupidae somewhat more advanced, and the Fringillidae and their descendents the most highly derived. The Icteridae may be the sister group of the Emberizinae, with Spizu as a link. The Drepanididae arose from the Carduelinae. The position of various problematic gen- era is discussed. INTRODUCTION This is a study of the evolutionary relationships in a large assemblage of songbirds, the New World nine-primaried oscines. The relationships among passerine birds continue to be unclear despite many studies attempting to unravel the pattern of their affinities. This is because of the large number of species, genera, and families involved, the general lack of distinctive characters defining specific groups, the tendency of groups to intergrade, and the high frequency of parallel and convergent sim- ilarities, especially in the suborder Oscines (Pas- seres). These problems are apparent in the confused taxonomic situation. On the one hand there has been a tendency to combine great numbers of spe- cies into large and unwieldy families, as in the treat- ment of the broadly defined Muscicapidae and Em- berizidae of the Check-list of Birds of the World. On the other hand many small (often monotypic) families have been created for forms whose rela- tionships to larger groups are undetermined, such as the Tersinidae, Catamblyrhynchidae, and Zele- doniidae. The family names used herein are those of Wetmore (1960) unless otherwise stated. If any sense is to be made of the oscine problem, the first task will have to be to cluster the large families into groups of apparently related forms, and then to ana- lyze these groups individually. Affinity in this case should be based on evolutionary rather than purely phenetic relationships, that is, superfamilial assem- blages should be hypothesized to be monophyletic. If the hypothesis of monophyly withstands scruti- ny, the next step should be to analyze evolutionary relationships within each assemblage, that is, to de- velop a phylogeny of the families involved. In the course of this process an attempt should be made to determine to which large family each small family has its closest relationships, so that they may be combined where possible, and the number of fam- ilies reduced. Once the phylogeny of several such assemblages has been determined, then the phylo- genetic relationships between them may be ana- lyzed, and gradually an overall phylogeny of the suborder Oscines can be constructed. Any attempt to work out relationships within the whole suborder by individual family comparisons is probably doomed to failure because of the complexity of the situation and the number of families involved. This paper reports the results of an attempt to analyze the relationships within one large suprafa- milial assemblage of the suborder Oscines. The New World nine-primaried oscines have tradition- ally been regarded as some sort of “natural” group, although various authors have differed on which families should be included, as well as on the rank to be given various groups. For the purpose of anal- ysis I have included all of the groups that various authors have considered part of the assemblage. These are the Vireonidae (including the Vireolani- idae and Cyclarhidae), Parulidae, Zeledoniidae, Thraupidae, Coerebidae, Tersinidae, Catambly- rhychidae, Fringillidae (including the Emberizinae, Cardinalinae, and Carduelinae), Icteridae, and Drepanididae. Together these comprise a signifi- cant number of species, approximately 955, which amounts to roughly 1 1% of living species of birds, 19% of living species of passeriformes, and 24% of living species of oscines. Problems of Phylogeny Construction In any analysis of the evolutionary history of a group of organisms, several problems must be over- come, and the first is to choose a method of anal- ysis. Assuming that evolution really has occurred, there must be a true phylogeny of the taxa under study, but we have no sure way to know what this is, or whether we have found it. In practice, then, a phylogeny is a hypothesis that may be presented for subsequent corroboration or refutation. Any method of analysis may be used to hypothesize a phylogeny. Many workers have constructed 5 6 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 7 branching diagrams (dendrograms) that they hy- pothesize to be phytogenies on the basis of shared similarities. In current terminology this would be termed a phenetic method, whether done by tradi- tional methods of clustering taxa by resemblance, or by modern computer techniques. A more rigor- ous approach is advocated by followers of the cla- distic school, who construct dendrograms by clus- tering “sister groups” on the basis of shared derived character states (synapomorphies) only. The differences between the phenetic and cladistic philosophies have engendered much vigorous and often rancorous discussion, though one has the impression that the philosophical differences are greater than the results that the two methods pro- duce. It appears to me that both approaches rep- resent extremes to which practical students need not limit themselves, and I expect that some inter- mediate position, containing contributions from both camps, will probably become the standard ap- proach in the coming years. In the present study I will use mainly the cladistic method of analysis, but where this falters I will not hesitate to employ more traditional approaches. I see no justification for ig- noring information that cannot be forced into a cla- distic mold. While one good synapomorphy may be better than many “similarities,” several solid sim- ilarities are still probably better in suggesting rela- tionships than a trivial or contrived synapomorphy. Once cladistics has been chosen as the principal method of phytogeny construction, several other problems arise: (1) It must be established that the taxon under analysis is monophyletic; (2) A suffi- cient number of characters must be found whose primitive/derived polarity can be determined with confidence; (3) Character conflicts must be re- solved; (4) A consistent philosophy of evolutionary probabilities must be maintained. These points will be discussed individually. (1) One could probably do a cladistic analysis of any collection of taxa, but unless the group can be shown to be probably monophyletic, there is little reason to suppose that the result represents a true phylogeny. To demonstrate monophyly, the most convincing arguments are demonstrations that the taxa comprising the group share one or more syn- apomorphies. If this is done, the existence of phe- netic similarities will strengthen the hypothesis of monophyly. A hypothesis based on such similarities alone, however, is less convincing. My analysis is based mainly on new information from the limb muscles, but first I will review other kinds of information to establish the likelihood that the New World nine-primaried assemblage is in- deed monophyletic. Nevertheless, at the outset the hypothesis of monophyly should be presented as such, and after the analysis of new data is com- pleted, this hypothesis should be reexamined to see whether the new information supports or refutes it. If it is supported, then the hypothesis of monophyly is strengthened, the proposed phylogeny may be taken as reasonable, and the whole exercise is jus- tified. If it is refuted, then the major result of the study may be to show that a group formerly con- sidered monophyletic or “natural” is not so, lead- ing to the necessity for a reevaluation of its rela- tionships. (2) One of the greatest problems in cladistic anal- ysis is finding a sufficient number of characters that can be analyzed to determine primitive to derived polarities. The more characters available, the great- er the number of branching points that can be placed in a proposed phylogeny. Also, the validity of branching points is increased if more than one character shift can be indicated at each, especially if they are not parts of a single adaptive complex. The lower the categorical level of the group being analyzed, the more difficult it is to find many char- acters useful in analysis. The New World nine-pri- maried oscines are a particularly difficult group in this respect because they differ little in major fea- tures, while at the same time having undergone an enormous amount of speciation. (3) Character conflicts arise when different char- acters indicate different branching patterns. Some characters must, therefore, have arisen indepen- dently in different lineages by convergence, parallel- ism, or evolutionary reversals, giving rise to false synapomorphies. This is especially troublesome in the lower taxonomic categories with groups that share a similar genetic background, and when the character states themselves are of a relatively sim- ple nature, such as losses of structures or minor structural modifications. Although we may realize that these conflicts can result from recognized bi- ological processes, in many specific cases it will be difficult or impossible to determine which alterna- tive branching pattern is most likely to reflect the actual evolutionary history of the group. In such cases the expedient solution is to adopt the simplest pattern, or to attempt to correlate different char- acters so as to arrive at a reasonable solution. I emphasize strongly that it is impractical to expect that any phylogeny will be without some character 1978 RAIKOW— OSCINE APPENDICULAR MYOLOGY 7 conflicts, but if it is recognized that these situations arise through ordinary biological processes, one need not refrain from developing a phylogenetic hypothesis that reflects the best fit available with the data at hand. (4) In phylogenetic analysis, as in any branch of science, there may be a tendency to lose sight of the overall problem when confronted with a mass of individual data. In cladistics the problem is to maintain an overview of the whole evolutionary pic- ture while dealing with many individual character phylogenies. In analyzing the New World nine-pri- maried oscines I will adhere to a model of evolution that postulates (a) that new major adaptations arise by the gradual modification of preexisting states, and (b) that the development of a new major adap- tive feature may be followed by extensive radiation into specialized subdivisions of the new adaptive zone. Thus, in developing a phytogeny of these birds, I will hypothesize that the major family and subfamily groups represent secondary radiations in new adaptive zones associated primarily with feed- ing specializations. This hypothesis will be tested by comparing the correlation between feeding ad- aptations and modifications of the locomotor ap- paratus. Monophyly of the New World Nine-Primaried Oscines This group is generally but uncritically regarded as monophyletic, an idea based mainly on several lines of evidence that I will review below. Some of this evidence can be interpreted in terms of primi- tive and derived character states, even though it may previously have been presented in more tra- ditional phenetic terms, and such evidences give the most convincing arguments in favor of monophyly. Other evidences cannot be interpreted in this way, and must be considered only in terms of general similarity. I do not agree with those who believe that such data are of no use in phylogenetic analy- sis. Such information may not be helpful in deter- mining the exact pattern of branching points in a phylogeny, but it does indicate genetic similarity between taxa and can at least be used to support hypotheses about clusters of related forms within a larger group. The purpose of this section is not to analyze relationships within the New World nine- primaried complex, but to examine the likelihood that the group as a whole is monophyletic, which is a necessary precondition for a cladistic analysis of its subgroups. This hypothesis of monophyly will be reexamined after the limb muscle data have been analyzed to see whether the new information will corroborate or refute the idea of monophyly. It will be seen that neither earlier studies nor the present investigation support the inclusion of the Vireoni- dae in the New World nine-primaried oscine assem- blage. Nevertheless, they are included here because they have traditionally been grouped with this as- semblage. umber of primaries . — The number of primary feathers varies among birds, but is usually constant within a family. The number may be reduced (or occasionally increased) among flightless forms, but in flying birds it varies from 10 to 12. In passeri- formes the number is 10, but the tenth (outermost) primary is sometimes reduced to a vestige, and such birds are referred to as nine-primaried (Van Tyne and Berger, 1976:127-129). In all, this suggests that the nine-primaried condition among passerines is a derived state, and that a "functional” tenth primary is primitive. Most of the New World nine-primaried oscines have only a vestigial tenth primary, except for some of the vireonids (Cyclarliis, Vireolanius, and some vireos, Mayr and Amadon, 1951:27). There are sev- eral other nine-primaried oscine families as well, such as the Zosteropidae, Hirundinidae, Alaudidae, and Motacillidae. These families do not appear to be closely related to the New World nine-primaried oscines, however, and most probably acquired a reduced tenth primary independently of them. Thus, except for the vireonids, the New World nine-primaried oscine assemblage is linked by this condition, which may be reasonably interpreted as a synapomorphy supporting the hypothesis that the group is monophyletic. Pneumatic fossa. — A well-developed second pneumatic fossa of the humerus is a derived con- dition, whereas the primitive state is the presence of only one fossa (Bock, 1962:437). The derived state occurs in several oscine families, some of which are not closely related, suggesting parallel evolution of this state. Bock (1962:432) pointed out that the New World nine-primaried oscines ”. . . appear to be the only larger subgroup within the oscines, that is rather uniform in having a fully de- veloped condition.” Again, the Vireonidae are an exception, having ”... only the beginnings of the second fossa” (Bock, 1962:432). A few genera of Icteridae also have only a small second fossa, but because their familial position is unquestioned, this is best interpreted as a case of secondary reduction. 8 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 7 Tongue apparatus . — In a study of the skeleton and certain muscles of the tongue of songbirds, George (1962, 1968) found two correlated differ- ences between the New World nine-primaried os- cines and other families. The basihyale, an unpaired bone forming part of the central axis of the tongue skeleton, is laterally flattened in the Coerebidae, Drepanididae, Parulidae (except Peucedranuis), Ic- teridae, Catamblyrhynchidae, Thraupidae, and in the cardueline, cardinaline, and emberizine finches. In 20 other oscine families, including the vireonids, cyclarhids, vireolaniids, and Tersinidae, it is cylin- drical rather than flattened in cross section. There is a related variation in the hyoglossus obliquus muscle (which George termed the hypoglossus pos- terior). This muscle arises from the posterior pro- cesses of the paraglossalia, passing transversely be- neath the basihyale without inserting on it (condition A) in most families examined. However, it is partly or completely attached to the basihyale (condition B) in the New World nine-primaried os- cines except for the Cyclarhidae, Vireonidae, and Peucedramus (not determined for Vireolanius or Tersina). The cylindrical basihyale always occurs together with the unattached muscle (condition A) except in swallows (cylindrical and B), whereas in the Certhiidae a compressed basihyale also occurs with condition B. Because neither swallows nor creepers are closely related to the New World nine- primaried oscines, these exceptions need not con- cern us. The correlation of the compressed basihyale with the attached hyoglossus obliquus suggests a functional relationship, so these two characters will be regarded as a single complex character. The fact that the New World nine-primaried assemblage as a whole stands apart from many other oscine families in this character argues, at least on pho- netic grounds, that they form a related group. This suggestion would be greatly strengthened if it could be shown that the tongue condition in the New World nine-primaried oscines is a derived state. George (1962, Table 1) recognized two char- acter states for the hyoglossus obliquus as already noted — A (not inserted on the basihyale), and B (inserted on the basihyale). This is a simplification, however, as there are variations in condition B. George recorded three variations, as shown in his figs. 5B-D and 6B-D. One is a condition peculiar to the Hirundinidae, which need not concern us. The condition shown in Eigs. 5C and 6C has some fibers inserting on the basihyale, while a deeper lay- er passes beneath the bone. The condition shown in Eigs. 5D and 6D has all fibers inserting on the basihyale, with none passing beneath it. There is some confusion in interpreting George's explana- tion because in his Table 1 he lists only the condi- tions A and B noted above, whereas in the text he notes conditions A (as in Table 1), B (as in the Hir- undinidae, conesponding to Eigs. 5B and 6B), and C (corresponding to Figs. 5C, D and 6C, D). Thus several variants are combined under condition B of Table 1, and it is not apparent which variant occurs in any given species having this general condition B. These variations may be interpreted as stages in a morphocline. Either the hyoglossus obliquus is primitively attached to the basihyale, and is pro- gressively losing this connection in some forms, or alternatively it is primitively not attached and is progressing to an inserted state. Which is the most likely direction of evolutionary change? I believe that the condition in the New World nine-primaried oscines, with the hyoglossus obli- quus attached to a flattened basihyale, is the de- rived state for the following reasons: (1) The cylin- drical basihyale with no muscle insertion occurs in Alaudidae, Corvidae, Paridae, Sittidae, Chamaei- dae, Cinclidae, Troglodytidae, Mimidae, Turdidae, Sylviidae, Motacillidae, Dulidae, Bombycillidae, Ptilogonatidae, Laniidae, Cyclarhidae, and Vireo- laniidae (George, 1962, Table 1). These groups rep- resent a great variety of different adaptive types with many different feeding specializations. It would seem most likely that the occurrence of the same structural conditions in all these different groups would be due to inheritance from a common ancestor, with the different condition in the New World nine-primaried oscines being related to some specialization in early members of this group. George (1962) argues that it is a specialization for more firmly attaching the movable tongue to the hyoid skeleton. (2) George (1962:27) notes that in the development of "'Setophaga picta" (^Myio- borus pictus) (Parulidae) from the nestling to the fledgling stage, the developing basihyale becomes progressively more flattened. Possibly this ontoge- netic change parallels a phylogenetic change in the shape of the bone from a primitive to a derived state. As with any criteria for determining morpho- cline polarity, these are not without uncertainty, but on the whole it appears more likely than not that the condition of the tongue apparatus in the New World nine-primaried oscines is derived rela- tive to the condition in oscines generally. Egg-white proteins. — Sibley (1970) studied rela- 1978 RAIKOW— OSCINE APPENDICULAR MYOLOGY 9 tionships among passerine birds by electrophoresis of egg-white proteins. The New World nine-pri- maried oscines were found to have similar electro- phoretic patterns, even to the point where some groups generally recognized as families were vir- tually indistinguishable. Data of this sort cannot be analyzed cladistically because the direction of evo- lutionary change of the protein molecules is not known; indeed differences in molecular structure are only indirectly estimated by the procedure. This is because the technique does not examine molec- ular structure directly, but only electrophoretic mo- bility, which has a close but imperfect correlation with molecular structure as coded in the genes. Therefore, as Sibley (1970:21) makes clear, what is measured is “genetic relatedness,’’ not genealogy. Nevertheless, this information is useful in the pres- ent context since I am concerned not only with cla- distic genealogy, but also with the nature and amount of evolutionary change in evolving lineages between branching points. The close similarity in proteins indicates a close similarity in the genes coding for their production, and by any philosophy this is a strong indication that the various members of the New World nine-primaried oscine assem- blage share a common genetic background. This strongly supports the idea of monophyly of the as- semblage, even though it does not contribute to an analysis of its phylogenetic branching pattern. Distribution and adaptive diversity. — The New World nine-primaried oscines occur mainly in the New World as their name implies; the exceptions are probably offshoots from this center of distri- bution and presumed origin. The coherence of the group suggests that the various families are inter- related as parts of a single adaptive radiation in the New World, and that the family groups correspond roughly to major adaptive niches based mainly on feeding specializations. Thus the Parulidae are mainly insectivorous; the Thraupidae, rather omniv- orous but with a strong reliance on fruits; the Em- berizinae may be characterized as mainly terrestrial foragers on relatively small seeds; the Carduelinae are mainly arboreal feeders on larger seeds; the Ic- teridae are marked by a specialized biomechanical system involving forceful jaw gaping into crevices, for opening fruits, and so forth. The Drepanididae are a special case; founded by a cardueline ancestor (Raikow, 1911b), the family radiated into an unusu- ally wide variety of feeding niches in the absence of ecological competitors. Sibley (1970:107) warns against overreliance on feeding specializations; the possibility of conver- gence provides a danger of classifying feeding niches rather than organisms. Nevertheless, there is also a possibility that useful taxonomic characters may be needlessly shunned by the fear that if adap- tive, they are subject to convergence. Some organ- isms must be similar because of ancestry rather than convergence, and 1 think it probable that the major families of New World nine-primaried os- cines represent the products of secondary radia- tions, arising when an ancestor entered a new adap- tive zone. In the present case this idea is generally supported by the anatomy of the limb muscles as discussed below; this is important because these muscles are not part of the adaptive complex of the feeding mechanism. These New World families often appear to be ecological equivalents of unrelated Old World fam- ilies that have independently radiated into similar broadly-defined adaptive niches. Thus, for exam- ple, the Parulidae correspond to the Sylviidae, and the “New World finches” (Emberizinae and Cardi- nalinae) to the Ploceidae and Estrildidae. The Car- duelinae, however, are widely distributed and eco- logically diverse in both areas, and especially in the Old World. The Icteridae correspond to the Stur- nidae. In general, then, the cohesive assemblage of New World nine-primaried oscines presents a picture of distribution and specialization that is most reason- ably interpreted as resulting from an evolutionary radiation into various broad feeding niches, with secondary radiations within each adaptive zone, and paralleling a separate pattern of similar radia- tions among other taxa in the Old World. This pic- ture supports the idea that the group is monophy- letic. Pterylosis. — The unity of the nine-primaried New World oscine group is also attested to by the pter- ylosis. According to Mary H. Clench (personal communication) “aside from the distinctive pattern found in vireos (including Cyclarhis and Vireolan- ius) the body pterylosis of the New World nine-pri- maried oscines is relatively uniform. Minor differ- ences in tract geometry and numbers of feathers may serve to group some genera, or to indicate cer- tain family divisions, but the overall similarity of the pterylosis of this assemblage is striking.” Indistinct boundaries . — Although the New World nine-primaried oscines are usually classified in sev- eral separate families, the boundaries between them are often indistinct and arbitrary, many genera 10 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 7 being intermediate between the typical members of different families. This indicates that the group is cohesive and closely interrelated. Indeed, some au- thors have suggested merging them into a small number of families in the light of these difficulties. For example Sibley ( 1970) reduced the assemblage (excluding the vireonids) to a single family Fringil- lidae, with three subfamilies and nine tribes. In the preceding discussion the hypothesis that the New World nine-primaried oscine assemblage (possibly excluding the Vireonidae) is monophyletic has been supported by various lines of evidence. Some of these, including the number of primaries. the pneumatic fossa of the humerus, and the tongue apparatus, appear to be derived character states grouping these taxa apart from other oscines. Other evidences, including the indistinct boundaries be- tween families, the egg-white proteins, the ptery- losis, and the geographic distribution and pattern of adaptive diversity, are not readily interpretable in cladistic terms, but nevertheless offer supporting evidence in favor of the theory of monophyly. In- dividually, each type of evidence generates a cer- tain degree of confidence in the hypothesis; taken together they reinforce each other to further strengthen the hypothesis. MATERIALS AND METHODS Dissection The forelimb and hindlimb muscles were dissected in nearly 100 species of songbirds (listed in Tables 2 and 3). Dissection was carried out under a stereomicroscope at magnifications of 6x to 25x, and the muscles were stained with iodine (Bock and Shear. 1972) to render fiber arrangement more visible. Detailed descriptions of the muscles were prepared for a reference spe- cies, Loxops virens, and are presented elsewhere (Raikow, 1976, I977u). In the present study the muscles in each species are compared to those of L. virens and only the differences are not- ed. Drawings were made with a camera lucida attachment to the microscope. Berger (1969) reviewed the variations in passerine appendic- ular muscles. He listed various muscles for which knowledge of structure and variation is uncertain or ambiguous. In the follow- ing account further information will be provided to clarify these problems with regard to the groups studied herein. Nomenclature The nomenclature of avian myology has a long and confusing history. Many workers have applied different names to the same muscle and the same name to different muscles. Some have oc- casionally attempted to stabilize the nomenclature for groups of muscles, but with only partial success. Most recently Berger (George and Berger, 1966) presented a standardized nomencla- ture for the entire avian muscular system. In the present study I have made some changes from that system to conform with a new system of names being developed by the International Com- mittee of Avian Anatomical Nomenclature for publication as the Nomina Anatomica Avium (N.A.A.), which is intended to sta- bilize avian anatomical nomenclature in all branches of avian biology. The names for muscles used herein are those tentatively adopted for the N.A.A. at the time this is written. Synonyms with the nomenclature of Berger are given in my previous papers (Raikow, 1976, 1977a). Determination of Primitive and Derived States There are no certain methods for determining the direction or polarity of evolutionary change in an evolving character, but several criteria have become generally accepted as being rea- sonably reliable. The most common is out-group comparison. If a character varies within a group, and one of the variants also occurs in a related outside group, then the character state that occurs in both groups is considered primitive within the group being studied. Another method is in-group comparison. Within the group under study, the primitive state is considered likely to be that which is distributed among a variety of different subgroups. Ross (1974) and Kluge (1977) discuss the rationales underlying these and other methods of determination. Out-group comparisons are the principal method used in this study to de- termine character states, while in-group comparisons are used occasionally. I have tried to explain in the text the basis for each individual determination so that its validity may be examined. This practice is not always followed by students of cladistics. For doing out-group comparisons it is important to choose taxa that are appropriate for comparison to the in-group. The general opinion is that proper out-groups should be of about the same categorical level as the in-group, and closely related to it so as to minimize the chances of convergence. Ideally an out- group should be the sister group of the in-group, but in practice this may be impossible to carry out because the sister group is not known. The practical approach is to choose groups that ap- pear to be closely related to the in-group even though their pre- cise cladistic relationships are unknown. In this study I have chosen certain likely out-groups for examination and have con- sidered in addition the literature on more distant relatives. Thus, the muscle variations in birds generally are first examined, with particular emphasis on passerines (information summarized by George and Berger, 1966). To this I have added new observa- tions on limb muscles from unpublished studies now underway in my laboratory. Clench and Austin (1974) summarized the ideas of various workers on the relationships of passerines as expressed in linear arrangements, and this has aided me in choosing groups for comparison. In their own arrangement the Nectariniidae, Estrildidae, and Ploceidae are grouped with the New World nine-primaried oscines, along with a few other fam- ilies, so I have included a nectariniid in my dissections, and also have used the extensive observations of Bentz (1976) on the Ploceidae and Estrildidae. No principally New World oscine family appears close to the New World nine-primaried oscines, and in the Old World, many families are either highly specialized 1978 RAIKOW— OSCINE APPENDICULAR MYOLOGY (for example, Paradisaeidae, Hirundinidae) or are small or of limited distribution (for example, Artamidae, Irenidae) and, therefore, seem to be unlikely candidates for a sister-group re- lationship with the New World nine-primaried oscines. I have, therefore, concentrated my dissections on a small group of rel- atively unspecialized, mainly insectivorous, and evolutionarily successful Old World forms in the Sylviidae. These appear, rath- er intuitively, to lie near the origin of the New World nine-pri- maried oscines. My dissections of the limb muscles of these groups, and the work of Bentz (1976) on the ploceid-estrildid complex confirm that these groups are close to the New World types because their limb muscles in general are very similar. I also included some thrushes and mimids as examples of some- what more distant but not highly specialized forms. Altogether I believe that this collection of out-groups gives a good idea of the ancestral muscle forms in the general group from which the New World nine-primaried assemblage arose. It is in any event the best solution to the problem of out-group selection that avail- able information and materials permitted. MUSCLES OF THE FORELIMB Detailed descriptions and illustrations of the fore- limb muscles in a reference species, Loxops virens , were given previously (Raikow, 1977u). In the fol- lowing section only variations from these condi- tions are given; where no variations are noted, the condition in ail forms studied was as described for Loxops . M. latissimus dorsi. — This muscle has two sep- arate parts in most birds, and sometimes also in- cludes cutaneous slips that are not considered in this study. Of the two main parts, pars cranialis occurs in all forms studied here. Pars caudalis is a parallel-fibered, strap-shaped muscle that arises from the neural spines of the dorsal vertebrae cau- dal to the origin of pars cranialis. It passes crani- olaterally to insert on the humerus deep to the in- sertion of pars cranialis. Pars caudalis is present in many nonpasserine orders as well as some families of passerines (George and Berger, 1966:293; Ber- ger, 1969:220; Table 2) and, therefore, its absence in Passeriformes is considered due to loss and is a derived state. Pars caudalis occurs consistently in the Vireoni- dae, where it is present in all genera (Table 2). This supports the theory that the several genera are closely related, and set apart from the remainder of the assemblage. Among the Parulidae pars caudalis occurs only in Peiicedramus , which supports the idea that the genus is misplaced in this family. Pars caudalis also occurs in a few species of Thraupidae and Icteridae, in some cases on one side of the body only (Table 1). This peculiar distribution could rep- resent either the retention of a primitive state or its secondary reappearance (Raikow and Borecky, manuscript), but in view of its pattern of correlation with other myological characteristics, the latter ex- planation is more probable. M. tensor propatagialis . — A scapular tendon (Raikow, 1977a) occurs in all members of the New World nine-primaried oscines (Fig. 1) except the Vireonidae. It was also found in some Turdidae and Mimidae, but not other out-groups (Table 2). Its presence in the New World nine-primaried oscines may be primitive, or the structure could have arisen independently in the other forms. There is no com- pelling evidence allowing a choice of these alterna- tives. At any rate its absence does distinguish the Vireonidae from the other families. M. deltoideiis major. — In most forms the caudal head inserts fleshy. InPsarocolins (Icteridae) it nar- rows distally and inserts by a stout tendon. This ap- pears to be an autapomorphic state in this genus only. M. deltoideiis minor. — This muscle arises from the pectoral girdle and inserts on the head of the hu- merus. In most cases the origin is from the scapula only, but sometimes there is also an origin from the adjacent coracoid. In the latter case the coracoidal origin may form a somewhat separate head, or a sin- gle continuous belly may be present. Because I sus- pect that the separation of the two heads may be an Table 1. — Occurrence of M. latis.'iiinus dorsi caudalis in some species of Thraupidae and Icteridae. ‘ Species No. of specimens Left side Right side Thraupidae Thraupis virens 7 + -H Thraupis palmarum 2 -1- -1- Thraupis palmarum 1 - -1- Thraupis cyanocephala 3 - - Icteridae Cacicus cela 1 -r -1- Cacicus cela 1 -1- - Cacicus haemorrhous 1 - - Psarocolius angustifrons 1 -1- -r Psarocolius decumanus 1 - - ' + = muscle present; - = muscle absent. 12 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 7 Table 2. — Major variations in pectoral musculature. For each muscle, the character states and character phylogenies are as follows: M. latissimus dorsi pars caudalis: + = present (primitive). - = absent (derived): M. tensor propatagialis, scapular tendon: + = present, — = absent (polarity undetermined): M. deltoideus minor, coracoidal head: + = present (derived). — = absent (primi- tive): M. coracobrachiaiis cranialis, + = well developed (primitive), ± = vestigial (derived). — = absent (further derived): M. pronator profundus: I = single belly (primitive), 2=2 bellies (derived): M. flexor digitorum profundus: I = belly wide (primitive), 2 = belly narrow (derived). Species M. latissimus dorsi pars caudalis M. tensor propatagialis scapular tendon M. deltoideus minor, M. coraco- coracoidal brachialis head cranialis M. pronator profundus M. flexor digitorum profundus Sylvia atricapilla Sylviidae + 1 1 Apalis fiavida -1- - - 1 1 Orthotomus atrogularis -1- - + 1 1 Regains satrapa - 7 - ■± 1 1 Grouped data on 41 genera from Bentz +(29) -(41) Ploceidae/Estrildidae + (4) +(2) 1(32) 7 ( 1976). No. of genera indicated in paren- theses Tardus migratorius -(12) + 9 -(37) Turdidae -(39) 2(8) 1 2 Sialia sialis - + - - 1 2 Catharus ustulatus + + - - 2 2 Dumetella carolinensis + + Mimidae ± 2 1 Nectarinia sperata + 9 Nectariniidae — 1 7 Vireo olivaceus + Vireonidae + 1 1 Vireo pallans + - - + 1 1 Hylophilus poicilotis + - - ±_ 1 1 Cyclarhis gujanensis + - - I 1 Vireolanius pulchellus + - - + 1 1 Dendroica coronuta + Parulidae 2 1 Geothlypis semiflava - + - - 1 1 Icteria virens - + + - 2 I Mniotilta varia - + - - 2 1 Myioborus miniatus - 7 - - 2 1 Oporornis tolmei - + - - 2 1 Peucedramus taeniutus + + - - 1 1 Seiurus aurocapillus - + + - 2 1 Wllsonia pusilla - + - - 2 1 Basileuterus rujifrons - + - - I 1 Zeledonia coronata _ + Zeledoniidae + 1 1 Sericossypha albocristata 7 Thraupidae 1 1 Rhodinocichla rosea - + + + 1 I Euphonia laniirostris - + - ± 1 1 Tangara cyanicollis - + - ± 1 1 Piranga ludoviciana - + - 2 1 Rhamphocelus passerini - + - - 2 1 1978 RAIKOW— OSCINE APPENDICULAR MYOLOGY 13 Table 2. — (Continued) Species M. latissimus dorsi pars caudalis M. tensor propatagialis scapular tendon M. deltoideus minor, coracoidal head M. coraco- brachialis cranialis M. pronator profundus M. flexor digitorum profundus Thraupis virens + + 2 I Tachyphonus rufus - + - ± 2 1 Urotliraupis slolzmanni - + - 1-2 1 Nephelornis oneilli + Tersinidae 1-2 1 Tersina viridis + Coerehidae I 2 Conirostrum speciosum - + - ± 2 1 Coereba fluveola - + - 7 2 1 Dacnis cayana - + - 2 1 Diglossa barbitula - + - ±_ 2 1 Cycinerpes cyaneus - + - 7 2 1 Chlorophanes spiza - + - 7 2 1 Euneornis cumpestris + Catamblyrhynchidae 2 1 Catamblyrhynchus diadema + + Fringillidae: Emberizinae 2 I Aiinophilu rujicauda - + + - 1 1 Arremonops conirostns - + + 2 1 Calcarius lapponicus - + + - 2 1 Chlorura chlorura - + + - 2 1 Emberiza flaviventris - + - 2 1 Geospiza fuliginosa - + - 2 1 Junco hyenialis - + + ± 2 1 Loxigilla portoricensis - ■) - 2 1 Passerella iliaca - 7 + ± 2 1 Pleclroplienax nivalis - + + ± 2 1 Zonotrichia capensis + + - Fringi!lidae:Cardinalinae 2 1 Passerina cyunea - + - 2 1 Pheucticus ludovicianus - + + 2 1 Cardinalis cardinalis - + + 2 1 Saltator maxinuts - + - + 2 1 Guiraca caenilea + FringillidaetCardueiinae 2 1 Leucosticte australis - + + 7 7 1 Leucosticte tephrocotis - + + 2 1 Carpodacus cassini - + - ■> 2 1 Pinicola enucleator - + + ± 2 1 Serinus mozambicus ? + ? 7 7 1 Serinus serinus - + + + 2 1 Carduelis carduelis - + + 7 2 1 Chloris chloris - + + ± 2 1 H esperiphona vespertina - + + 2 1 Loxia curvirostra - + + 2 1 Pyrrhula pyrrhula - + + 2 1 Carduelis pinus - + + ■± 2 1 Eringilla coelebs - + - *> 1 14 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 7 Table 2. — (Continued) Species M. latissimus dorsi pars caudalis M. tensor propatagialis scapular tendon M. deltoideus minor, coracoidal head M. coraco- brachialis cranialis M. pronator profundus M. flexor digitorum profundus Heniignathus procerus + Drepanididae + - 2 I Hemignathus wilsoni - 7 + 7 2 1 Puroreomyza maculala bairdi - 7 + ± 2 1 Loxops virens n ilsoni - + + 2 1 Psittirostra cantans cantans - + + + 2 1 Psittirostru cantans ultima - -H + 7 2 1 Psittirostra psittacea - -1- + ± 2 1 Hiniatione sanguinea - + + 7 2 1 Vestiariu coccinea - + + + 2 1 Palmeria dolei - + + ± 2 1 Cacicus cela + ■ + Icteridae 1 2 Cassiculus melanicterus - + - -H 2 2 Dolichonyx oryzivorus - + + ± 2 2 Psarocolius decuinanus - + - ■±_ 1 2 Quiscalus quiscula - + + 2 2 Spiza americana - + + 2 2 Sturnella magna - + + 2 2 Sturnella neglecta - + •? 7 7 2 Agelaius phoeniceus - + + ±_ 2 2 Icterus parisoruin - + - 2 2 Molothrus ater - + + 2 2 artifact resulting from manipulation, I have not dis- tinguished these variants. The significant variation in this muscle, therefore, is the presence or absence of a coracoidal head. I have described and illustrated this variation previously (Raikow, 1977/?). A cora- coidal head is absent in most out-groups and in most New World nine-primaried oscines, including the generally primitive Vireonidae and most Parulidae. For these reasons its absence is considered primitive and its presence derived in the assemblage under in- vestigation. The only groups in which a coracoidal head is vir- tually universal are the Carduelinae and Drepanidi- dae, which argues for their close relationship (Rai- kow, 1977b). In the Parulidae it occurs only in Icteria, whose position in the family is uncertain, and in Seiurus, which is also aberrant in some hind- limb muscles. Among the Emberizinae the coracoid- al head is absent only in a few genera. In the Thrau- pidae the coracoidal head is absent, but the scapular head is relatively slender, which is apparently also a derived state. In Saltator and Catamblyrhynchus the muscle is also slender, supporting their connec- tion to this family. On the other hand, the peculiar Rliodinocichla , which is currently included in the Thraupidae with uncertainty, possesses a coracoidal head. See Table 2. M. coracobrachialis cranialis . — This is a small (ca. 3-mm long) parallel-fibered muscle of the shoul- der. It arises from the cranial surface of the head of the coracoid and inserts on the humerus between the articular surface and the pectoralis insertion. The muscle is buried in the tissue of the coracohumeral joint capsule, which must be dissected away to re- veal the tiny belly (Fig. 2). It is well developed in Vireo and Vireolanius , but in Hylophilus and Cy- clarhis it is pale in color and takes up little or no stain as compared to normal muscles. It arises by a well- defined tendon, but the belly is soft and difficult to separate from surrounding connective tissue, though slightly more dense than the latter. This appears to be a degenerate condition, with perhaps an absence or poor development of muscle fibers, which may be partly or mostly replaced by connective tissue. This would have to be determined by histological exam- ination. In any event I term this condition vestigial, as it appears to represent a stage leading to complete loss. The muscle is well developed in Zeledonia, 1978 RAIKOW— OSCINE APPENDICULAR MYOLOGY 15 TENSOR PROPATAGIALIS PARS LONGA Fig. I. — Dorsal view of the superficial muscles of the shoulder in Hesperiphonci vespertina showing the scapular tendon passing from the origin of the two parts of M. tensor propatagialis to the dorsal surface of the scapula. Staining deeply. It is apparently absent in Parulidae, where I could find no trace. It is present in Saltator and Catamhlyrhynchus . Otherwise it is demonstra- ble in most other forms as a vestige, or not at all, especially in smaller species. This muscle is present in many nonpasserine or- ders (George and Berger, 1966:313-315); hence its reduction and loss are a derived trend in the group studied. Its relative retention in Zeledonia and Ca- tamblyrhyuchus set these forms apart as primitive offshoots of the Parulidae and Thraupidae, respec- tively, with which they are currently associated. M. s err at us superficialis . — In Catamhlyrhynchus the caudal head of pars cranialis is wider than usual, arising not only from the first true rib (and its uncin- ate process) but also from the second true rib. Pars costohumeralis generally arises from the third true rib, but from the fourth in Tangara and Tersina. M. rhomhoideus superficialis . — In most cases this muscle inserts only on the scapula, but in a few cases the area of insertion is extended cranially so that a few fascicles also insert on the clavicle. This inser- tion is on the medial surface of the head of the clav- icle and the shaft just distal to the head. It occurs in Sialia, Nectarinia, and to a small extent in Quis- calus. In Zonotrichia, Chlorura, and Emheriza , in contrast, the muscle is shorter, its anterior extreme ending about 2 mm short of the clavicle. M. serratus profundus. — In most cases the cranial head arises by two slips, one each from the penul- timate and antepenultimate cervical vertebrae. An origin from only the penultimate vertebra was found in Wilsonia, Loxops, and Himatione . M. expansor secondariorum . — In Rhodinocichla this inserts only on the proximal two secondaries, rather than three as in most species. 16 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 7 HUMERUS Fig. 2. — Deep shoulder muscles in Vireolanius pulchellus showing the well-developed coracobrachialis cranialis muscle. M. pronator profundus. — As noted by Berger (1969:221), this muscle arises in part by a tendon from the humerus and in part fleshy from the adjacent humeroulnar pulley. These two parts tend to form separate bellies, but there is a distinct variation in the degree of separation in the bellies (Fig. 3). in Type I the two parts are adjacent and more or less fused together, so that the belly is essentially single, flat, and fan shaped. It spreads out to an essentially con- tinuous line of insertion along the radius. In Type 2, however, the two bellies become separated. The proximal belly inserts proximally on the radius by a flat aponeurosis, while the distal belly inserts more distally on the radius by a narrow tendon that arises along its cranial surface. Between these two bellies there stretches a very thin connective tissue mem- brane that may represent a vestige of the formerly continuous tendon of insertion, but has no mechan- ical strength in this form, and indeed may merely be derived from fascia. The effect of this change is to make two distinct muscular bellies rather than one, presumably with a capability of acting more inde- pendently of each other in contraction, and thus in- creasing the functional versatility of the pronator profundus. In their review of this muscle, George and Berger (1966:346-347) did not mention a separation of this muscle into two bellies in nonpasserine birds. Most other passerines studied have Type 1 , but a few have Type 2. This suggests that the separation into two bellies (Type 2) is a derived state in passerines, prob- ably arising several times (Table 2). M. flexor digitorum profundus. — In all Icteridae examined (including 5p/za) this muscle narrows dis- tally so that its caudal margin does not overlie the distal part of the belly of the underlying M. ulno- metacarpalis ventralis (Fig. 4). A similar condition was seen only in Tersina, among the New World nine-primaried oscines (Table 2). M. extensor digitorum communis . — In Tersina the belly is longer than usual, extending nearly the entire length of the ulna. M. interosseus dorsalis. — In Zeledonia the belly was lacking, the fine tendon being anchored in con- nective tissue. In Catamblyrhynchus the belly is re- duced to a tiny vestige, the tendon still being present. M. ulnimetacarpalis ventralis . — This muscle aris- es from the ulna, and may have one or two moder- ately separate heads. The two-headed condition may simply represent a v-shaped origin of the single bel- 1978 RAIKOW— OSCINE APPENDICULAR MYOLOGY 17 Fig. 3, — Variation in M. pronator profundus. Above, the primitive Type I with a continuous belly, shown in Vireo olivaceous . Below, the derived Type 2 with separate bellies, shown in Dolichonyx oryzivorus . ly, however, distinctly separate heads being unusu- al. Most forms studied have the v-shaped origin, but some members of most families have the single con- dition, which occurs most consistently in the Drep- anididae. Although Berger ( 1968:221) suggested that this variation might prove taxonomically useful, the extent of intergradation and apparently random dis- tribution make it of no value in the present study. MUSCLES OF THE HINDLIMB Detailed descriptions and illustrations of the hind- limb muscles have been given previously (Raikow, 1976) for the reference species Lo.vopx vire/is . In the following section I report variations in some of these muscles; where no variations are noted, the muscle in all forms examined was like that in L. virens. There are also some comments related to muscle variations in passerines generally, which will add to the review given by Berger (1969). M. iliotibialis lateralis. — In some passerines the postacetabular portion of the muscle is absent (Ber- ger, 1969:221) but both pre- and postacetabular por- tions were present in the New World nine-primaried oscines. Stallcup (1954:165) reported that in Vireo the muscle lacks a central aponeurotic portion found in other forms, but 1 found it present and identical to the condition in the other species, including some that Stallcup also dissected. In Sericossypha a small part of the dorsal fleshy part is absent between the pre- and postacetabular parts, and reaching as far distally as the distal central aponeurosis of the mus- cle. M.femorotibiaiis extenuis . — No differences were noted. Berger (1969:221) reported a deep distal head in some passerines, which I also described in L. vi- rens. Because Berger did not name these divisions. 18 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 7 Fig. 4. — Variation in M. flexor digitorum profundus. Above, the derived state with narrowed distal portion of the belly, shown in Spiza americana . Below, the primitive state in Sericossypha albocristata . I propose the terms pars proximalis (the superficial part) and pars distalis (the deep, distal head). Stall- cup (1954) did not describe pars distalis, but appar- ently overlooked it as I found it in many species that he studied. I also found it in Hirundo (it was not mentioned by Gaunt, 1969), and in some Sturnidae, Mimidae, Corvidae, Turdidae, Alaudidae, Paridae, and Nectariniidae. The division of M. femortibialis externus into proximal and distal heads is probably common in passerines but has been overlooked by most workers. M. puho-iscliiofemoralis . — Gaunt (1969) found that pars cranialis and pars caudalis are fused to- gether in the Hirundinidae, but they are separate in the forms studied here. Stallcup (1954:168) claimed that the origin is modified in Vireo, but I could not find any difference from the other forms examined. M. obturatorius lateralis. — This muscle has dis- tinct dorsal and ventral bellies (Eig. 5). Pars ventralis is present in all forms studied, but pars dorsalis is absent in some. Because it occurs in many groups of birds, its absence in some genera of the families stud- ied herein is undoubtedly due to loss and is derived. When present, pars dorsalis varies in size. I term it small when the area of origin is not caudal to the ob- turator foramen; medium when the origin lies be- tween the obturator foramen and the midpoint of the ilioischiatic fenestra; and large when the origin lies caudal to this point. Pars dorsalis is large in most families studied herein, and in most of the out-groups as well. Thus its reduction in size in the Carduelinae and Drepanididae must be a derived state, and one that supports the theory (Raikow, \911b) that these families are sister groups. Reduction continues to a point of complete loss in some cardueline genera. The pars dorsalis is also independently lost in a few genera of families where it is otherwise well devel- oped (Table 3). Bentz (1976) found a similar tenden- cy toward loss in the Old World finches. M. gastrocnemius . — This muscle has three sepa- rate bellies with a common tendon of insertion. Pars externa and pars intermedia do not vary, but pars interna is highly variable and of some taxonomic use (Pig. 6). Pars interna is the most superficial muscle on the medial surface of the crus, arising by two dis- tinct heads. The superficial head arises in part from the inner cnemial crest of the tibiotarsus, while a band of fibers (the patellar band) may extend around Fig. 5. — Deep muscles of the hip in four species of New-World, nine-primaried oscines, showing variation in the size and occurrence of M. obturatorius lateralis pars dorsalis. A) Cyanerpes cyaneus, showing muscle of large size; B) Loxops virens wilsoni, showing muscle of medium to large size; C) Pinicola enudeaior . showing muscle of small size; D) Hesperiphona vespertina , showing muscle absent. Abbreviations; F, femur; IlF, ilio-ischiatic fenestra; ITC, M. iliotrochanterichus caudalis; OLD, M. obturatorius lateralis pars dorsalis; OLV, M. obturatorius lateralis pars ventralis; OM, tendon of M, obturatorius medialis; PIFC, M. pubo-ischiofemoralis pars cranialis. the knee, arising from the patellar tendon. The deep head of pars interna arises from the medial surface of the head of the tibiotarsus. The two heads fuse and the common belly extends caudally. Stallcup ( 1954) noted some variation in pars interna, but a more detailed classification of this variation is given here. In Type 1 the superficial head is present, in- cluding a patellar band. In Type 2 the superficial head is present but lacks a patellar band. In Type 3 the su- perficial head is absent. In forms having Type 1 the relative size of the patellar band varies. This is ex- pressed in Table 3 as the percentage of the distance between the patellar crest and the patella which the origin of the patellar band covers (Fig. 7). This vari- ation is only of occasional usefulness, but variation in the Types 1 through 3 is more valuable taxonomi- cally. The three types appear to be stages in a mor- phocline. I believe that Type 1 is the ancestral state in the New World nine-primaried oscines because of out-group comparisons (Table 3), because it is the most common condition within the group, and be- cause it occurs in families (for example, Parulidae) generally considered on other grounds to be primi- tive within the assemblage. The derived state, loss of the patellar band, oc- curs in some cardinalines and thraupids, most car- duelines, and all icterids. If this were considered a synapomorphy, however, it would conflict with oth- 20 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 7 Table 3. — Major variations in pelvic musculature. For each muscle, the character states and character phylogenies are as follows: M. gastrocnemius pars interna: I = superficial head present, including patellar hand (primitive), 2 = superficial head present but patellar band absent (derived), 3 = superficial head absent (further derived): Patellar band: decimal value = width of hand as % of distance between patella and patellar crest (polarity undetermined): M. ohturatorius lateralis pars dorsalis: — = absent (derived), size classes defined in text (polarity undetermined): M. plantaris: + = present (primitive), — = absent (derived); M. peroneus brevis tibial head: - = absent (primitive), ± = partially developed (derived), + = fully developed (further derived): M. flexor digitorum longus: ABB (primitive), other patterns (derived). Species M. gastro- cnemius pars interna Patellar band M. obturatorius lateralis pars dorsalis M. plantaris M. peroneus brevis tibial head M. flexor digitorum longus Sylvia atricapilla 1 0.95 Sylviidae Ige. -1- ABB Apalis flavida 1 0.80 Ige. -I- - ABB Orthotomus atrogularis 1 1.00 Ige. -t- - ABB Regulus satrapa 1 0.60 Ige. - -1- ABB Grouped data on 41 genera from Bentz 1(7) 0-1.00 Plocidae/Estrildidae -(26) -r(37) -(41) ABB(37) ( 1976). No. of genera indicated in paren- theses. Some genera have more than one condition in different species. Turdus migratorius 2(30) 3(8) 1 0.50 sml.( 10) med.(5) Ige. (3) Turdidae Ige. -(4) + ABC{3) AAA(I) ABB Sialia sialis 1 0.50 Ige. + - ABB Catharus ustulatus 1 0.80 Ige. -1- - ABB Dumetella carolinensis 1 1.00 Mimidae Ige. -1- _ ABB Nectarinia sperata 1 0.50 Nectariniidae Ige. -r — AAB Vireo olivaceus 1 0.70 Vireonidae med. -1- AAB Vireo pallens 1 0.90 Ige. -1- - AAB Hylophilus poicilotis 1 0.80 med. -1- - AAB Cyclarhis gujanensis 1 0.30 med. + - AAB Vireolanius pulchellus I 1.00 sml. -1- - AAB Dendroica coronata I 0.70 Parulidae med. -1- ABB Geothylpis semiflava 1 0,50 Ige. -t- - ABB Icteria virens I 0.90 med. - ABB Mniotilta varia 1 0.80 med. -1- - ABB Myiohorus miniatus 1 0.80 med. -t- - ABB Oporornis tolmei 1 0.90 med. + - ABB Peucedramus taeniatus 1 0.90 Ige. -1- - ABB Seiurus aurocapillus 1 0.90 - -1- + ABB Wilsonia pusilla 1 1.00 med. -1- - ABB Basileuteurus rufifrons I 1.00 med. -f- - ABB Zeledonia coronata 1 1,00 Zeledoniidae med. -1- — ABB Sericossypha albocristata 2 Thraupidae med.-lge. -1- ACB Rhodinocichia rosea 1 0.60 Ige. + - CBB Euphonia laniirostris 1 0.50 med. + - ABB Tangara cyanicollis I 0.60 med. -1- ABB Piranga ludoviciana 2 - med. -1- - ABB 1978 RAIKOW— OSCINE APPENDICULAR MYOLOGY 21 Table 3. — {Continued) Species M. gastro- cnemius pars interna Patellar band M. obturatorius lateralis pars dorsalis M. plantaris M. peroneus brevis tibial head M. flexor digitorum longus Rhaniphocelus passerini 2 _ med. -1- _ ABB Thraupis virens 2 - Ige. -1- - ABB Tachyphonus ntfus 3 - Ige. + - ABB Urothraupis stolzmanni 1 0.80 Ige. + ABA Nephelornis oneilli 1 0.80 Ige. Tersinidae + ABB Tersina viridis 3 med. Coerebidae -r ACB Conirostriun speciosiini 1 0.40 med. -1- - ABB Caere ba flaveola I 1.00 Ige. + ABB Dacnis cayana 1 0.40 med. + - ABB Diglossa harbitida 1 1.00 Ige. - - ABB Cyanerpes cyaneus 3 - Ige. - - ABB Chlorophanes spiza 1 0.30 Ige. -1- - ACB Euneornis campestris 1 0.80 Ige. -1- Catamblyrhynchidae ABB Catamblyrhynchus diadema 1 1 .00 Ige. Fringillidae: Emberizinae ABA Ai mop hi la ruficaiida 1 0.60 Ige. + - ABB Arremonops coniroslris 1 1.00 med. + - ABB Calcarius lapponicus 1 0.70 med. + - BBB Chlorura chlorura 1 1.00 ■> + - ABB Emberiza flaviveniris 1 0.40 Ige. + - ? Geospiza fuliginosa 1 1.00 Ige. -f- - ABB Jiinco hvemalis 1 1.00 Ige. -1- - BBB Loxigilla portoricensis I 0.60 Ige. + - ABB Passerella iliaca 1 0.80 Ige. -1- - BBB Plectrophenax nivalis 1 0.30 Ige. -1- - 7 Zonotrichia capensis ' 0.70 med. -1- Fringillidae;Cardinalinae ABB Passerina cyanea 2 - Ige. + - ABB Pheucticiis ludovicianus 3 - Ige. -1- - ABB Cardinalis cardinalis 1 0.25 Ige. -1- - ABB Saltator maximus 1 0.25 sml. -t- - ABB Guiraca caerulea 2 Ige. + FringillidaeiCarduelinae AAB Lencosticte australis 1 0.70 sml. + - BBB Leiicosticte tephrocotis I 0.60 - -1- - ABB Carpodacus cassini Y - sml. - - ABB Pinicola enucleator 2 - sml. -1- - ABB Serinus mozambicus 2 - med. -h ABB Serinus serinus 2 - sml. - 7 Carduelis carduelis 3 - med. - -t- ABB Chloris chloris 3 - - -1- -1- ABB Hesperiphona vespertina 3 - - -1- - ABC Loxia curvirostris 3 - sml. - 4- ABB Pyrrhula pyrrhula 3 - - + ABB Carduelis pinus 3 - sml. - 4- ABB Fringilla coelebs 3 - sml. 4- - ABB BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 7 Table 3. — (Continued) Species M. gastro- cnemius pars interna Patellar band M. obturatorius laterialis pars dorsalis M. plantaris M. peroneus brevis tibial head M. flexor digitorum longus Hemignathus procerus 1 0.75 Drepanididae med. -1- ABB Hemignathus wilsoni 1 0.60 sml. -1- + ABB Paroreomyza nuicuUila bciirdi 1 0.60 sml. -1- + ABB Loxops virens wilsoni 1 0.75 med. - -f- ABB Psittirostra cantuns cantans 1 0.50 sml. -1- -1- ABA Psitlirostrci cantans ultima 1 0.50 sml. + -1- ABA Psittirostra psittacea 1 0.40 sml. -1- -t- ABB Himatione sanguinea 1 0.60 med. - -1- ABB Vestiaria coccinea 1 0.70 sml. - -1- ABB Palmeria dolei 1 0.75 med. - -r ABA Ciridops anna 1 >0.50 med. 7 ■7 ■? Cacicus cela 2 Icteridae Ige. -r ABA Cassiculus meianicterus 2 - med. -1- - ABB Doli clionyx oryzi vorus 2 - med. + - ABB PsarocoUus decunianus 2 - Ige. -1- - ABB Quiscalus quiscuia 0 - Ige. + - ABB Spiza americana 2 - med. + - ABB Sturnella magna T - - + - ABA Sturnella neglecta 2 - 7 + 7 7 Agelaius phoeniceus 3 - Ige. + - ABA Icterus parisorum 3 - med. -1- - ABB Molothrus ater 3 - med. + - ABB * Reduced. er data as discussed below, so most probably the loss of the patellar band has occurred independently in these three groups. M. peroneiis brevis. — This muscle shows an im- portant variation not previously described in birds. In most forms studied it arises from the fibula alone, or also the adjacent tibial shaft. In some, however, there is also a fleshy origin from the caudal margin of the lateral cnemial crest, just caudal to the origin of the tibial head of M. tibialis cranialis. This tibial head passes distad superficial to the femoral head of the tibialis cranialis, and joins the usual fibular head to form a single belly. I have illustrated this muscle and discussed its mode of origin elsewhere (Raikow, \911b). Its occurrence is listed in Table 3. Because it does not occur in other birds so far as known, and because it is found here in the cardue- lines and drepanidids, two groups of advanced po- sition in the New World nine-primaried assemblage, the presence of a tibial head is clearly derived and argues for a sister-group relationship of these two families (Raikow, \911b). A partial or complete tib- ial head was also found in a few other genera (Table 3) where it presumably arose independently. M. flexor perforatus digiti II. — In Rhodinocichla rosea the muscle was distinct, having a small sec- ondary head originating from the surface of M. flex- or hallucis longus, and joining the main head along its medial border. This condition was not observed in any other form. Berger (1969:222) reported that the tendon of insertion in this muscle is not perfo- rated by the tendons of M. flexor perforans et per- foratus digiti II or M. flexor digitorum longus in various passerines. Stallcup ( 1954: 171) claimed that it is so perforated in the species he studied. I believe that in all the species I studied the tendon is not perforated, and that Stallcup made an error in in- terpretation. It appears from his description of the insertion, when compared to observation of speci- mens, that what he considered a sheathlike termi- nation of the tendon of M. flexor perforatus digiti II is actually the tendon sheath of the first phalanx. Although this is in contact with the tendon of M. flexor perforatus digiti II at its insertion, the sheath 1978 RAIKOW— OSCINE APPENDICULAR MYOLOGY 23 Fig. 6. — Medial view of the knee area showing variation in M. gastrocnemius pars interna. A) Sericossypha ulbocristuta , showing the superficial head present as in Types I and 2. B) Icterus parisorum, showing Type 3 with the superficial head lacking. itself should not be considered part of that tendon. Such a sheath covers all the flexor tendons on the plantar surface of each phalanx of every digit. Stall- cup ( 1954: 171) stated that this muscle is not in con- tact with M. flexor hallucis longus in Vireo, but I found that it is in such contact as in the other spe- cies studied. M. plantaris. — This muscle is designated by the letter E in muscle formulae, and until recently was considered present in all passerines. However, Gaunt (1969) found it absent in most Hirundinidae, and Bentz (1976) in some estrildids. 1 also found it absent in some New World nine-primaried oscines (Table 3). Because M. plantaris is present in many passerine and nonpasserine families, its absence here is considered a derived state due to loss. M. flexor hallucis longus. — Berger ( 1969:22) not- ed that the muscle may have one, two, or three heads of origin in passerines. Stallcup (1954:174) reported the lateral head absent in Vireo, but I found it present. According to Stallcup (1954) the usual origin is by two heads, but Berger found three heads in Agelaius (George and Berger, 1966:443) and Dendroica (Berger, 1968:613). My findings agree with Berger’s descriptions, and it appears that Stallcup failed to distinguish between the interme- diate and medial heads described by Berger. M. flexor digitorum longus. — There is a third, femoral head of origin in some birds. Among the forms included in this study, a femoral head was found only in the tanager Sericossypha , arising from the lateral condyle of the femur and passing caudad to join the fibular head. The tendon of this muscle divides into three branches inserting on the three anterior digits. Variations in this pattern of insertion (Fig. 8; Table 3) are sufficiently constant in recognized groups to be useful in this study. The usual pattern in the New World nine-primaried os- cines is ABB (letter designations are shown in Fig. 8), the main exception being the Vireonidae with AAB. In other families there are occasional varia- tions, usually involving the addition of delicate ac- cessory vincula, but only in the Vireonidae is the variation consistent. It is difficult to determine the polarity of this character as the structural changes involved are minor, but out-group comparisons (Table 3) suggest that ABB is probably primitive. M. flexor hallucis brevis. — In the Vireonidae this muscle is greatly enlarged and somewhat bipennate in structure, while in most groups it is quite small (or even apparently absent) and appears to be nar- rowly fan-shaped (Fig. 9). This sets the Vireonidae apart from the New World nine-primaried oscines, and also clusters the four genera of the Vireonidae, attesting to the unity of the group. The large size is probably a derived specialization of the Vireonidae as it did not occur in other groups studied. M. extensor hallucis longus . — This muscle con- 24 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 7 Fig. 7. — Cranial view of the knee in four species showing variation in the size and occurrence of the patellar band of M. gastrocnemius pars interna. A) Coereba flaveola. Type I, patellar band size 1.00; B) Leucosticte tephrocotis , Type 1, patellar band size 0.60; C) Saltator ma.ximiis. Type 1, patellar band size 0.25; D) Carpodacus cassini. Type 2, patellar band absent. See text for explanation. Abbreviations: P, patella; PBG, patellar band of M. gastrocnemius. sists of proximal and distal heads. The proximal head is as described by Berger ( 1968). A distal head has been described in a few nonpasserine birds (George and Berger, 1966:454-455), but apparently has not previously been found in passerines. It can be demonstrated only under high magnification (25 X) and with the iodine stain. It is a fan-shaped muscle about 3 mm long, and arises from the medial surface of the distal end of the shaft of the tarso- metatarsus just proximal to the first metatarsal. It passes distad alongside the tendon of the proximal head and inserts into the joint capsule between metatarsal I and the proximal phalanx of the hallux (Fig. 9). 1978 RAIKOW— OSCINE APPENDICULAR MYOLOGY 25 DIGIT II A. DIGIT III DIGIT IV B. C=3C3CT C. (R^ac=a^q Fig. 8. — Diagram showing the patterns of insertion of the three branches of M. flexor digitorum longus on digits II, III, and IV. The formula for insertions can be determined from this diagram. For example, ABB, the most common pattern, means that digit II has the form shown in line A, whereas digits III and IV have the forms shown in line B. The proximal head shows little variation except that in the vireonid group it is relatively larger com- pared to the other forms studied. This enlargement is not as striking as the enlargement of M. flexor hallucis brevis in the same group, but is perhaps related to it as part of an adaptation for increased strength in movement of the hallux. M. lumhricalis . — Stallcup (1954) described this muscle in the forms he studied, but Berger (1966, 1968) did not find it in Agelains or Dendroica . 1 found the muscle to be universally present in the forms studied here. It is small and strap-shaped, and lies deep in the plantar surface of the foot. It arises from the tendon of M. flexor digitorum longus just proximal to the point at which that tendon tri- furcates, and inserts on the joint pullies of digits III and IV. No variation was observed. The muscle is very easily stretched, suggesting that in passerines it may be partly or completely converted to an elas- tic ligament, but this would have to be determined histologically. CHARACTERIZATION OF TAXA ViREONIDAE The genera Vireo, Hylophilus , Cyclarhis, and Vireolanius are often included in one family (Mayr and Amadon, 1951; Blake, 1968), but the latter two are sometimes placed in separate families near the Laniidae (Wetmore, 1960). In terms of limb mus- cles, the four genera are generally primitive in most characters that vary significantly in the New World nine-primaried oscines, but also show some unique derived character states (Tables 3 and 4). The su- perficial head of M. gastrocnemius pars interna is more clearly separate from the deep head than in the other families. The insertion pattern of M. flex- or digitorum longus is AAB compared to the usual ABB pattern of other families. M. flexor hallucis brevis is greatly enlarged in all four genera; in no other family does it approach this size, and indeed is often so small as to be difficult to identify. These are interpreted as derived states. The tensor pro- patagialis scapular anchor is absent in the vireonids, but present in the other families. I cannot determine whether this is a derived state, but at least it does separate the vireonids from the other groups. M. coracobrachialis cranialis is well developed in Vireo and Vireolanius , but is reduced in Hylophilus and Cyclarhis, the latter being a derived state shared with most of the assemblage. This could mean that reduction of the muscle occurred independently in the vireonid radiation and the other families, or that the vireonid genera with reduction are closer to the 26 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 7 EXTENSOR HALLUCIS LONGUS Fig- 9. — Muscles of the tarsometatarsus in Vireolcinius ptilchelliis , showing the well-developed flexor hallucis brevis characteristic of the Vireonidae. origin of the remainder of the assemblage. The la- tissimus dorsi pars caudalis is present in all vireo- nids, but otherwise almost always absent in the nine-primaried groups. Are Cyclarhis and Vireolaniiis closely related to Vireo and Hylophilus , and should one family or three be recognized? The derived conditions of Mm. gastrocnemius pars interna, flexor digitorum longus, flexor hallucis brevis, and perhaps tensor propatagialis tend to cluster these genera apart from the other families, and indicate that the group as a whole is monophyletic. Retention of M. latissimus dorsi pars caudalis supports this, though less strongly as it is a shared primitive character. Fur- thermore, these genera lack various derived states of several muscles that occur in many nine-primar- ied families, such as those of Mm. gastrocnemius, obturatorius lateralis, plantaris, and peroneus brev- is. They are, therefore, clearly related, and a single family is desirable to emphasize both their close- ness to each other, and their distinctiveness from the remainder of the assemblage. A second question concerns the possible rela- tionship of the Vireonidae to the shrikes (Laniidae). Early workers suggested a close relationship, though this is now usually regarded as unlikely, and based mainly on the convergent development of a “toothed" bill. I dissected a typical shrike {Lanins vittatiis: Laniinae) and a bush shrike (Telophoriis dohertyi'. Malaconotinae) for comparison. Table 4 shows that the two shrikes are unlike one another, and indeed they represent subfamilies that some workers consider unrelated. This is a separate prob- lem, however. With respect to the vireonids, the main emphasis should be placed on Lanins as rep- resentative of the "true” shrikes, to which a vir- eonid relationship is presumably suggested. Table 4 shows little support for this idea. The derived con- ditions of Mm. gastrocnemius, flexor digitorum lon- gus, and flexor hallucis longus possessed by the vir- eonids are lacking in the shrikes. The shrikes have a reduced flexor perforatus digiti III that is not seen in the vireonids. Lanins has a large femoral head of origin on M. flexor digitorum longus, which is lack- ing in the vireonids. The only character shared by the shrikes and vireonids is the retention of M. la- tissimus dorsi pars caudalis, a primitive character state in the passeriformes. In short, the myological evidence fails to support the idea of a close vireo- nid-laniid relationship. Parulidae Except for certain genera that appear to be either aberrant or misplaced in this family, and which are discussed separately below, most of the wood war- blers examined are myologically similar to each oth- er and generally primitive in their appendicular musculature. They are similar to the Emberizinae, 1978 RAIKOW— OSCINE APPENDICULAR MYOLOGY 27 Table 4. — Myological comparisons of the Vireonidae with other taxa. Taxa M. gastro- cnemius pars interna, superficial head M. flexor digitorum longus femoral head M. flexor digitorum longus insertion M. flexor perforatus digit! 3 M. flexor hallucis brevis M. latissimus dorsi caudalis Vireo Separate - AAB Normal Large + Hylophilus Separate - AAB Normal Large -1- Cyclarhis Separate - AAB Normal Large -(- Vireolanitis Separate - AAB Normal Large -t- Lanins Partly fused •f AAA Reduced Small -1- Telophorus Partly fused - ABB Reduced Medium + New World, nine-primaried oscines Partly fused or lost ABB' Normal Small _2 ‘ The few exceptions (Table 3) do not affect the distinction between this group and the Vireonidae. ^ Occasionally present (Tables 1 and 2). although their muscles are often less robust than those of the latter group, which may reflect their more actively arboreal habits and frequently small- er body size. The latissimus dorsi caudalis has been lost, a derived state, whereas the deltoideus minor coracoidal origin is absent, which is a primitive con- dition in the New World nine-primaried oscine as- semblage. M. coracobrachialis cranialis appears to have been entirely lost, as I could not find even a vestige in these forms, though one generally re- mains in other groups (Table 2). It is possible, how- ever, that some trace of this muscle does remain but was not detected because of the small size of most species. The pronator profundus has the de- rived condition of two bellies (Type 2) except in Geothlypis and Basileutems . This important char- acter state difference suggests that Lowery and Monroe (1968) erred in combining Oporornis and Geothlypis. M. flexor digitorum profundus, how- ever, is invariably primitive. In the hindlimb the typical parulids are primitive in ail the variable mus- cles described herein, altogether this places the family low on an evolutionary scale within the nine- primaried assemblage. Seiunis is distinct in several features. In the fore- limb of S. aurocapillus M. deltoideus minor pos- sesses an origin from the coracoid, which is lacking in typical parulids, as well as in S. motacilla and S. noveboracensis . In the hindlimb the obturatorius lateralis pars dorsalis is absent (a derived state), except in S. motacilla . Most unusual is the presence of a tibial head of M. peroneus brevis. Such a head occurs in other families, but its form here is differ- ent. Instead of arising independently from the lat- eral cnemial crest, it arises from that area in com- mon with the tibial head of M. tibialis cranialis. This indicates that its occurrence here is an independ- ently derived state, and is not homologous with the condition in the cardueline/drepanidid group. It was found in 5. aurocapillus (two specimens) and S. motacilla (one specimen), but not in 5. novehora- censis (one specimen). The significance of these peculiarities is unknown. Possibly they link 5ez7//7/5 with some unanalyzed oscine family. Alternatively Seiunis may indeed be a parulid, whose peculiari- ties are in some way functionally related to its lo- comotor habits, as it is much more terrestrial than other parulids. This appears reasonable because Seiunis has hybridized with Dendroica (Short and Robbins, 1967). Icteria is larger than other parulids and has been thought not to belong to this family (Ficken and Ficken, 1962; Eisenmann, 1962rz; Clark, 1974), but its pelvic muscles are like those of typical parulids. In the forelimb, however, M. deltoideus minor arises in part from the coracoid, a derived condition otherwise found only in some Seiunis among the parulids dissected. Peucedramiis is another genus whose placement in the Parulidae has been questioned. George ( 1962, 1968) suggested that it may be more closely related to the Sylviidae. Its pelvic musculature is like that of typical parulids, but so is that of the Sylviidae examined for the most part (Table 3). In the fore- limb M. latissimus dorsi caudalis is retained, which links it more with the Sylviidae than the Parulidae. Likewise, Peucedramiis has a Type 1 (primitive) pronator profundus as in the Sylviidae, rather than 28 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 7 the Type 2 found in most (but not all) typical paru- lids. These data are consistent with George's hy- pothesis of a sylviid relationship. Zeledonia coronata, the Wrenthrush, was for- merly thought to be related to the Turdidae, but Sibley (1968, 1970) studied the egg-white proteins and concluded that it properly belongs among the New World nine-primaried oscines, and probably closest to the Parulidae. Ames (1975) reported that thrushes have a derived condition of the syringeal muscles that is lacking in Zeledonia. Hunt (1971) presented life history evidence supporting place- ment of Zeledonia in the New World nine-primaried assemblage. In the thrushes that I examined M. ob- turatorius lateralis pars dorsalis is very large, aris- ing from the entire ventral margin of the ilioischiad- ic fenestra as well as its caudal margin. \n Zeledonia and the nine-primaried types it arises only from the ventral margin of the fenestra, not the caudal mar- gin. In the general structure of hindlimb muscles Zeledonia is typical of the more primitive New World nine-primaried oscines, such as the Parulidae (Table 3). Its large patellar band resembles that of Basileuterns , a parulid genus to which relationship of Zeledonia has been suggested (Hunt, 1971). In the forelimb Zeledonia lacks the latissimus dorsi caudalis as do the parulids. This muscle was present in two of the three turdids examined (Table 2). The flexor digitorum profundus in Zeledonia is Type 1 as in the Parulidae, whereas in the Turdidae examined it is Type 2. The most peculiar condition in Zele- donia is that the coracobrachialis cranialis is large and stains like a muscle, which is a primitive con- dition. In both the Parulidae and Turdidae exam- ined this muscle is either reduced to a vestige or lost. Thus, in most respects, the limb myology of Zeledonia is similar to that of the Parulidae, but differs from that of the Turdidae. Compared to the genera of Parulidae dissected, Zeledonia is most similar to Basilenterns . Both have a maximally developed (1.00) patellar band, which is smaller in the other genera examined ex- cept Wilsonia. Both also have a Type 1 (primitive) pronator profundus, whereas most other genera (except Geothlypis) have a Type 2. {Peucedramus also has Type 1, but as discussed above, is probably not a parulid.) These similarities may be significant because Hunt (1971) showed life history similarities of Zeledonia and Basileuterus , and the two genera also have a similar dorsal coloration and nasal oper- culum, though these could be convergent (Sibley, 1968). Thus, the results of the present study support the hypothesis that Zeledonia is not a thrush, but is a member of the New World nine-primaried os- cines, close to the Parulidae and perhaps especially to Basilenterns . Thraupidae This family is a classic example of taxonomic confusion resulting from the variety of plumages, sizes, and feeding adaptations that it exhibits (Sto- rer, 1969). The problem is not eased by reducing the group to subfamily status as some have advo- cated. Not only are relationships within the family difficult, but the affinity of the thraupids to the car- dinal finches, parulids, and other groups are also problematical. A number of genera seem to connect the more typical tanagers to one or another of these other groups, and the allocation of these genera forms an additional difficulty. I have dissected only a few of the many tanagers, but even this small sample shows considerable diversity in appendicu- lar muscles. My analysis will not begin to solve the problems surrounding this group, but should con- tribute to their eventual resolution. In the forelimb M. latissimus dorsi caudalis is generally absent, but was found either bilaterally or unilaterally in some species of Thranpis (Tables 1 and 2). The presence of this muscle is inconsistent with the apparent phylogenetic position of the fam- ily (see below), and as in the case of its presence in some Icteridae (see discussion under that family) I suspect that its appearance here may be a case of the reestablishment of a previously lost muscle. The coracoidal head of the deltoideus minor (derived) was found only in the aberrant genus Rhodinoci- chla \ it was absent in all the more typical tanagers examined. Indeed, in these other forms the scapular portion of the muscle was invariably rather slender compared to its form in other families. The pronator profundus occurs in both primitive and derived states (Table 2). In the hindlimb (Table 3) both primitive and de- rived states of the gastrocnemius were found, but otherwise the musculature is generally primitive. Enphonia and Tangara are myologically the most primitive forms studied, with their Type 1 pronator profundus and Type 1 gastrocnemius pars interna. These muscles in the other typical thraupids ex- amined are more highly derived. Nephelornis oneilli , a new genus and species of nine-primaried oscine, was recently described from Peru (Lowery and Tallman, 1976). N. oneilli is a rather nondescript, moderately thin-billed species 1978 RAIKOW— OSCINE APPENDICULAR MYOLOGY 29 Table 5. — Myological comparisons o/ Nephelornis with oilier taxa. Muscle Nephelornis Urothraiipis Parulidae Thraupidae Emberizinae Gastrocnemius Type 1 Type 1 Type 1* Types 1*. 2, 3 Type 1* Gastrocnemius. 0.80 0.80 0.50-1.00* 0.50-0.60 0.40-1.00* patellar band Obturatorius Lge. Lge. Med. ( 1 Lge.) Med., Lge.* Lge.* (most) lateralis p. dorsalis Flexor digitorum longus ABB ABA ABB* ABB* (most) ABB*. BBB Peroneus brevis ± + -d-M — (few±)* - Pronator profundus 1-2 1-2 2* (few 1 ) 1, 2* 2* (one 1) Deltoideus minor - - -*(2 + ) * + (3 — ) coracoidal head Deltoideus minor narrow narrow wider narrow* wider * Indicates correspondence with Nephelornis . living in highland cloud forests. Lowery and Tail- man (1976) were unable to determine in which family it should be placed. Walter J. Bock examined the jaw and tongue muscles and confirmed that Nephe- lornis is a primitive member of the nine-primaried group (cited in Lowery and Tallman, 1976) but left the question of family position open. The hindlimb myology suggests that Nephelornis is closely relat- ed to Vrothraupis , which has been placed in the Thraupidae as recently as 1969 (Storer, 1969), but has been transferred to the Emberizinae by Paynter (1970) without explanation. Relevant comparisons are given in Tables 2, 3, and 5. The gastrocnemius pars interna is of the primitive type (Type 1) with a relatively wide (0.80) patellar band, in both Nephelornis and Urothraiipis . This falls within the range of variation in Parulidae and Emberizinae. Most Thraupidae have a more de- rived condition (Type 2 or 3) but Tangara and Eii- phonia have the primitive state. M. obturatorius lat- eralis pars dorsalis is large in Nephelornis and Urothraiipis as in many thraupids and most ember- izines. The insertion pattern of the flexor digitorum longus is ABB in Nephelornis , and ABA in Uro- thraiipis, which is the only notable difference be- tween the limb muscles of the two genera. ABB is the primitive and usual condition in the New World nine-primaried oscines, and ABA and other vari- ants are rare and show no pattern, so the condition in Urothraiipis is probably significant only at the generic level. One possibly significant feature is found in M. peroneus brevis. In both Nephelornis and Uro- thraiipis there is a small band of fibers arising from the ligament to the head of the tibiotarsus. As noted (Raikow \911h) this is probably an early stage in the development of a tibial head to the muscle. A rudiment such as this was also seen in Tangara, Coereha , and Sericossypha . In the forelimb the pronator profundus is inter- mediate in form between the primitive Type 1 and the derived Type 2, but closer to the latter. The deltoideus minor in both genera lacks a coracoidal head, which is a primitive state, whereas most em- berizines possess one, as do a few parulids but no thraupids. Eurthermore it is rather slender in these two genera, thus resembling the condition in thrau- pids. The latter is a rather subjective character, however. It is apparent that the limb muscles do not pro- vide sufficient information to definitely ally Nephe- lornis with any given family. However, they do show the following points; (1) Nephelornis is nearly identical in limb musculature with Urothraiipis', it is not as similar to any other genus examined. (2) Nephelornis (and Urothraiipis) are clearly very primitive members of the New World nine-prima- ried assemblage, close to the base of the Thraupidae and Emberizinae. (3) In Table 5 1 have marked with an asterisk the families to which each muscle indi- cates a probable close relationship. There are five similarities to the Parulidae, five to the Emberizi- nae, and seven to the Thraupidae. The latter include the probably significant similarities in the peroneus brevis and deltoideus minor. (4) The other families in the assemblage are more highly derived in var- ious characters and are, therefore, of no concern to this problem. The bill of Nephelornis is rather slender, and the jaws show little deflection, which suggests little if 30 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 7 any adaptation for seed-eating. Lowery and Tail- man (1976) reported that it fed mainly on insects, and found no seeds in stomachs examined. This suggests that it is more primitive in feeding habits than typical emberizines, though it is clear that more investigation of both its feeding habits and feeding mechanism are needed. There is thus a conflict of evidence concerning familial position. The limb myology favors inclusion of Nephelornis and Urothraupis in the Thraupidae. This family is traditionally regarded as being char- acterized by fruit-eating in addition to insectivory. I have no information on the food of Urothraupis and the few data on Nephelornis suggest a concen- tration on insects, along with a small amount of plant material. On the basis of the foregoing discussion I con- clude that Nephelornis is a very primitive member of the New World nine-primaried oscine assem- blage, and that it is little modified from the primitive basal stock from which the early Thraupidae and Emberizinae arose. Nephelornis is thus of interest because it more closely approaches this hypotheti- cal ancestor than any other genus examined, except Urothraupis . On the basis of limb musculature alone, I would hesitate to separate the genera. I suggest on the basis of present knowledge that for the purpose of classification Nephelornis be consid- ered a very primitive member of the Thraupidae, and that Urothraupis be returned to this family, ad- jacent to Nephelornis . Rhodinocichla has been placed in various fami- lies but no solution has been entirely satisfactory. Skutch (1962) argued on the basis of behavior that its affinities lie with the Mimidae, whereas Eisen- mann (1962^) concluded mainly from morphology that it is a tanager; this was supported by Clark (1974). A study of the limb myology only confuses the picture more. The latissimus dorsi caudalis is absent as in most tanagers, whereas in Duinetella it is present. The deltoideus minor has a coracoidal head, unlike either Duinetella or thraupids. The pronator profundus in Rhodinocichla is the Type 1, whereas in Duinetella it is Type 2. Both types occur among Thraupidae, but Type 1 was found only in a few forms. These data tend to support the thrau- pid theory rather than the mimid theory, but the limited evidence is hardly convincing. The problem is complicated because Rhodinocichla has several unique myological conditions that are not found in either the Mimidae or Thraupidae. The flexor digi- torum longus has a CBB insertion pattern (Fig. 8), which I found in no other form examined. The ex- pansor secondariorum inserts only on two second- aries, rather than the usual three. The flexor per- foratus digit! II has a secondary head as described above, which is not found in any other form dis- sected. The limb myology, then, shows Rhodino- cichla to be even more distinctive than previously supposed, but does not help in determining in which family it should be placed. For the present it is most practical to retain Rhodinocichla in the Thraupidae until other evidence suggests an alternative. Sericossypha is the most myologically aberrant tanager dissected, with its centrally reduced iliotib- ialis lateralis, and well-developed femoral head of M. flexor digitorum longus, both unique conditions in the New World nine-primaried oscines studied. It also has a partial tibial head to M. peroneus brev- is, and unusually large Mm. flexor hallucis brevis and extensor hallucis longus. This pattern of pe- culiar features does not resemble any other form studied, and so its relationships remain obscure. Catamblyrhynchus is a distinctive genus of un- certain affinities, sometimes being placed in the Thraupidae and sometimes in a family Catambly- rhynchidae. The interosseus dorsalis is vestigial, an autapomorphic character that does not assist in de- termining the relationships of the genus. The cora- cobrachialis cranialis is well developed, a primitive condition. The deltoideus minor, which lacks a cor- acoidal head, is slender as in the Thraupidae. The pelvic musculature is primitive with respect to Mm. gastrocnemius, obturator lateralis, and per- oneus brevis. The ABA insertion pattern of the flex- or digitorum longus does not occur in thraupids, but was found in several other families as a variant, and probably has no significance. The absence of the plantaris is a derived state. Altogether, the limb myology places Catamblyrhynchus within the nine- primaried assemblage as a relatively primitive, but in some ways distinct form, lying close to the paru- lid/emberizine/thraupid group. The robustness and length of the shank muscles, however, set it apart from the Parulidae. The bill, although superficially finchlike, differs in detail from that of true finches. The tip of the bill is abruptly squared off when seen from above, and there are peculiar grooves running forward from the nostril on either side. Further- more there is hardly any indication of an angled commissure, as is typical of finches. All this sug- gests that the resemblance of the bill is a superficial 1978 RAIKOW— OSCINE APPENDICULAR MYOLOGY 31 case of convergence, and does not indicate that Catamblyrhynchus is a modified emberizine, as sug- gested by Paynter (1970). On the basis of these observations my inclination is to regard Catamblyrhynchus as most probably an early offshoot of the Thraupidae, and to regard it taxonomically as rather distinct within that family. 1 do not advocate giving it full family status as that obscures its probable affinity with the Thraupidae. The Swallow Tanager, Tersina viridis, is unusual because of its flycatching behavior. It is sometimes included in the Thraupidae, and sometimes placed in a monotypic family Tersinidae. Tersina was found to have a few unusual myological character- istics. M. serratus superficialis pars costohumeralis arises from the fourth true rib, rather than the third as in most species studied. M. flexor digitorum pro- fundus narrows distally rather as in Icteridae (Type 2), though the similarity must certainly be conver- gent. M. extensor digitorum communis has an elon- gated belly. The flexor digitorum longus has an ACB insertion pattern, rather than the usual ABB. Otherwise the myological conditions fall within the usual range of variation in the Thraupidae (Tables 2 and 3). The several distinctive conditions in the forelimb could possibly represent modifications somehow related to flycatching. COEREBIDAE This family was long maintained for a group of nectar-feeding genera though it was recognized that some were close to the Parulidae and some to the Thraupidae. Following Beecher (195k/) Coereba and Conirostrum (including Ateleodacnis) are now generally placed in the Parulidae, and Cyanerpes, Diglossa, Dacnis, Chlorophanes, Euneornis, Hemi- dacnis, Iridophanes, Xenodacnis, and Oreomanes in the Thraupidae. It is generally believed that nec- tar feeding arose at least twice and perhaps several times independently in this assemblage. I have dis- sected the first seven of the above 1 1 genera. There are some variations, but no derived specializations that would cluster the genera into a separate family. Coereba and Conirostrum fit easily into the range of variation in the Parulidae, being more like typical wood warblers than are such aberrant forms as Seiurus and even Icteria . The only peculiarity is a small partial tibial head of the peroneus brevis in Coereba , a feature that occurs in some members of several other groups. The Thraupidae have more limb muscle variation than do the Parulidae, but again the related “coer- ebid” genera fit easily among them. All have the derived state of the pronator profundus, while both primitive and derived forms occur among the Thraupidae. Dacnis, Diglossa, Chlorophanes, and Euneornis retain the patellar band of the gastrocne- mius, whereas in Cyanerpes and several typical tan- agers it is lost. The essential coherence of this group is further attested by the successful hybridization of such diverse genera as Cyanerpes and Tangara (Delacour, 1972). Emberizinae The emberizine finches show great uniformity in their appendicular musculature. They are similar to the Parulidae except that their muscles tend to be heavier and the shank muscles tend to extend far- ther along the length of the tibiotarsus than in the wood warblers. In the forelimb the latissimus dorsi caudalis is absent, M. coracobrachialis cranialis is vestigial or lost, and with one exception the pron- ator profundus is of the derived type. The emberi- zines agree with the parulids in these derived states, and also share the primitive condition of M. flexor digitorum profundus. However, unlike typical paru- lids the deltoideus minor has in most cases an ex- panded area of origin from the coracoid, a derived state. In the hindlimb all forms studied have the prim- itive state in Mm. gastrocnemius, obturatorius lat- eralis, plantaris, and peroneus brevis. The patellar band and the dorsal head of M. obturatorius later- alis are both large, showing no trends toward re- duction. However, the emberizine finches are gen- erally primitive and show little variation in their appendicular muscles. Cardinalinae The cardinal finches show a small amount of myological diversity in both limbs. The deltoideus minor coracoidal head may be present or absent. Saltator has a well-developed coracobrachialis cranialis, a muscle that is reduced to a vestige in most forms studied. In the hindlimb the gastrocne- mius pars interna varies from the most primitive to the most derived conditions. As a whole, this group cannot be distinctly separated from either the em- berizine finches or the tanagers on the basis of limb myology. 32 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 7 ICTERIDAE The Icteridae are of particular interest because the family has undergone a considerable adaptive radiation in feeding specializations and related be- havior (Beecher, 1951/?; Lowther, 1975). It includes both terrestrial and arboreal types ranging from smaller, short-billed genera like Spizci, Dolichony.x, and Molothrns that closely resemble emberizine finches, to large long-billed arboreal foragers like Cacicus and Psarocolius . In the forelimb the Icteridae show a modification of the flexor digitorum profundus in which the cau- dal border of the muscle is narrowed (Fig. 4). Though this is not a profound modification, it is found without exception in all forms examined, and this consistency suggests that it is a reliable char- acter useful in defining the family. The pronator profundus and deltoideus minor show the derived state in most forms, but are primitive in a few. The occurrence of M. latissimus dorsi pars cau- dalis in a few icterids is especially intriguing (Table 1). This muscle occurs consistently only in the Vir- eonidae (Table 2). Its presence in a few members of an otherwise advanced group is difficult to ex- plain. Possibly the muscle was retained in the his- tory of the group, and was lost independently in many lineages, including most of the icterids, but I believe it possible that its presence here may be an example of a reestablished condition. In other words, the muscle (or rather its expression in the phenotype) may have been lost early in the evolu- tion of the nine-primaried assemblage, but the ge- netic information for its production could have been retained and later reactivated in a few forms. This is suggested by the distribution of the muscle in the Icteridae (Table 2). It occurs in some species of a genus, and not in others; it may even occur on one side of the body only in some specimens. This sug- gests an easily perturbed genetic mechanism con- trolling its appearance. A similar phenomenon was also found in some Thraupidae. The reappearance of “lost" ancestral muscles in birds is discussed elsewhere (Raikow, 1975; Raikow and Borecky, in preparation). The muscles of the hindlimb are similar to those of the emberizines except that the patellar band of the gastrocnemius pars interna is always absent (Type 2 or 3), a derived condition whereas in the emberizines it is Type 1. Spiza requires special mention because of its con- troversial taxonomic status. Sibley (1970) and Tor- doff (1954) placed it among the cardinaline (rich- mondenine) finches, whereas Sushkin (1925) considered it intermediate between the Emberizi- nae and Icteridae and arbitrarily placed it in the former group for convenience. Beecher (1951/?) considered it an icterid on the basis of jaw muscles and the horny palate. The forelimb myology tends to support the icterid theory. Spiza has a Type 2 flexor digitorum profundus as in the Icteridae, whereas the Cardinalinae have a Type 1 muscle. In Sturnella the forelimb is typically icterid, but the hindlimb is distinctive in several ways. The dor- sal head of the obturatorius lateralis is absent, un- like all other genera studied, where it is relatively large. The aponeurosis of origin of M. adductor fe- moris pars caudalis is very short. The superficial head of M. gastrocnemius pars interna is very well developed and separate for most of its length from the deep head. The tendon of M. flexor perforans et perforatus digiti III is ossified in the shank and tarsus, and the tendon of M. peroneus longus is ossified from the belly to its bifurcation near the tibial cartilage. The lateral head of M. flexor hal- lucis longus arises about 3 mm proximal to the ilio- fibularis insertion, whereas in other forms it arises just distal to this insertion. The intermediate and medial heads of M. flexor hallucis longus are not clearly separable. The significance of these varia- tions is unknown, but perhaps they are functionally related to the terrestrial habits of this form. In this regard it may be significant that in Dolichonyx , an- other terrestrial genus, the gastrocnemius pars in- terna has a form similar to that in Sturnella. Carduelinae The cardueline finches show little variation in the forelimb musculature but a great deal in the hind- limb. The only variation found in the forelimb is in M. deltoideus minor — most forms possess a cora- coidal head (derived state) but it is lacking in Frin- gilla and in Carpodacus cassini and C. piirpureus (but was present in C. mexicanus). This suggests that its absence in the latter genus, which on the basis of its pelvic myology is a fairly advanced member of the family, is probably due to secondary loss in some species. Its absence in Fringilla, how- ever, is probably primitive, as is discussed below. There is considerable diversity in the hindlimb muscles (Table 3). M. gastrocnemius pars interna shows a complete range of structure from the most primitive through the most derived conditions. The dorsal head of M. obturatorius lateralis is present in some forms, and absent in others. When present. 1978 RAIKOW— OSCINE APPENDICULAR MYOLOGY 33 however, it is of medium or (usually) small size, showing a general trend throughout the family for reduction and loss of the muscle. The tibial head of M. peroneus brevis may be absent, partially devel- oped, or completely developed. M. plantaris may be present or absent. Furthermore, the diversity is increased by the several combinations of variations found in different species. This diversity character- izes the Carduelinae as a progressive and rapidly evolving group. There are several taxonomic problems involving the Carduelinae to which the present study can con- tribute useful insights. First, are the carduelines more closely allied with the Old World finches (Plo- ceidae and Estrildidae) or the New World finches (Emberizinae and Cardinal inae)? Tordoff (1954) argued on the basis of palatal osteology that the group is unrelated to the latter and placed it as a subfamily of the Ploceidae, but few workers have supported this. Bentz (1976) described the limb myology of the Ploceidae and Estrilididae and some of his findings are briefly summarized in Tables 2 and 3. Many of the Old World forms retain the la- tissimus dorsi pars caudaiis, which is totally absent in the carduelines. Bentz did not find the tensor propatagialis scapular tendon in his study, whereas it is always present in the Carduelinae. The deltoi- deus minor lacks a coracoidal head in most Plocei- dae/Estrilididae, but is almost universally present in the Carduelinae. The pronator profundus is Type i in all ploceids and most estrildids, but is Type 2 in all carduelines. The peroneus brevis has a tibial head in many carduelines, but never in the Plocei- dae or Estrildidae. Altogether, the limb myology supports the theory that the Carduelinae are part of the New World nine-primaried oscine assemblage and are not closely related to the ploceid/estrildid complex. A second problem is the relationship of the Car- duelinae to the genus Fringilla. Sibley (1970:100- 103) reviewed this problem in detail. Bock (1960:476) suggested that Fringilla is intermediate between the Emberizinae and the Carduelinae, and the limb muscle data are consistent with this view. The deltoideus minor lacks a coracoidal head in Fringilla, a primitive state found in a few emberi- zines (Table 2), but only in some species of one of the cardueline genera studied. The gastrocnemius pars interna shows a derived state in Fringilla, like most carduelines rather than emberizines, but in other respects its limb myology is generally primi- tive. In terms of pelvic muscle structure Leucostic- te is even more primitive than Fringilla , as it is the only cardueline retaining the Type 1 gastrocnemius pars interna. Leucosticie also has a relatively prim- itive tongue structure, more like that of some em- berizines than that of the more advanced cardue- lines (Raikow, 19776). Tordoff ( 1954) showed that in the carduelines the head of the humerus is rela- tively broader than in the emberizines, but that this trait was least developed in Leiicosticte . These con- siderations suggest that Leucosticie rather than Fringilla might be better regarded as the most prim- itive of the cardueline genera investigated. Finally, there has been much discussion of the relationship of the Carduelinae and the Drepanidi- dae. The drepanidid genus Psit tiros tra is anatomi- cally essentially a cardueline, and the two families appear to be sister groups. This is discussed further below. Drepanididae The history of drepanidid classification was re- viewed by Sibley (1970: 104). Early workers placed the different genera in several families, allying the finchlike Psittirostra with the Fringillidae, and the nectar-feeding forms with Old World nectar-feeders (Dicaeidae and Meliphagidae). Later they were rec- ognized as constituting a single assemblage despite their diversity. Some workers were impressed by the nectar-feeding adaptations of many Drepanidids and supposed their ancestors to be either the Coe- rebidae or Thraupidae. Others have been more im- pressed by similarities between the finch-billed Drepanididae and the cardueline finches. Beecher ( 1953) found similarities in the jaw muscles between the groups. Sushkin ( 1929) allied the Drepanididae and Carduelinae on the basis of the bill, skull, and horny palate. Bock ( I960) suggested that the Car- duelinae lack specialized features that would pre- clude their ancestry of the Drepanididae, and that they have a tendency to wander erratically in flocks, a habit that might be expected in a coloniz- ing group. The pelvic musculature of the Drepanididae is relatively uniform (Table 3), including both subfam- ilies. This reinforces the idea that the family is de- rived from a single founding species. The M. gas- trocnemius pars interna in all forms studied is Type 1, the most primitive sort. Among the Carduelinae only Leucosticie shows this condition; all other genera examined have derived conditions 2 or 3 for this character. The dorsal head of M. obturatorius lateralis is present in all forms; again this is the 34 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 7 ancestral state, whereas both conditions occur among carduelines. The plantaris muscle is present (primitive state) in most Drepanididae, but absent (derived state) in three species; the Carduelinae also include both conditions. Of greatest signifi- cance is the M. peroneus brevis. In all Drepanididae examined this muscle has a fully developed tibial head. As noted above, this is a derived character state of infrequent occurrence. The only other New World, nine-primaried group that regularly shows this character state is the Carduelinae, where it occurs in a majority of genera. These observa- tions suggest strongly that the Carduelinae are the closest relatives of the Drepanididae. The carduelines have evolved considerably since the time that they split off from the ancestral Drep- anididae. The derived state of M. gastrocnemius pars interna must have arisen after this separation. The ancestral state was retained in the Drepanidi- dae, but only the genus Leucosticte among the forms studied still shows it in the Carduelinae; other genera have progressed to more derived states. Partly on the basis of the evidence cited above I feel that the Drepanididae are best regarded as an offshoot of a fairly primitive cardueline species, though one in which the tibial head of M. peroneus brevis had already developed. They have remained relatively conservative in their pelvic myology while radiating spectacularly in their feeding appa- ratus. The Carduelinae in contrast have radiated more in their pelvic musculature since the separa- tion of the two lineages, but less so in their feeding apparatus. A more detailed analysis of the drepa- nidid-cardueline relationship is presented elsewhere (Raikow 1977/?). A PHYLOGENY OF THE NEW WORLD NINE-PRIMARIED OSCINES 1 will now present a phylogeny of the New World nine-primaried oscines as a hypothesis of the pat- tern of ancestral relationships in the group to the degree that present information makes possible. This model phylogeny is based on the idea that the major groups are the products of adaptive radia- tions into discrete adaptive zones defined mainly in terms of feeding specializations, and involving the structure of the feeding apparatus, the methods of foraging, and the types of foods taken. Many writ- ers have suggested that feeding specializations are “adaptive" and hence poor indicators of relation- ships because of the supposed ease with which di- vergent groups may come to resemble each other through convergence. Thus "bill shape," formerly an important taxonomic character, is now in dis- repute. These criticisms are valid only insofar as comparison is superficial and limited to simple structures, but, if a whole adaptive complex is ex- amined in some detail, it should be possible to rec- ognize convergence in most cases. This can be tested by comparing the feeding system with infor- mation from other sources, such as the limb mus- cles studied herein. The present classification of the New World nine-primaried oscines is to a great ex- tent based on feeding adaptations, so the present analysis is an attempt to see whether this concept is a valid model for a theory of phylogenetic rela- tionships within the group. This type of evolutionary pattern has been dis- cussed by various workers including Schaefer (1976), Mayr (1976), and Bock (1965). Basically, it is suggested that an evolving lineage may enter a new adaptive zone by developing a new structural and behavioral specialization that allows it to ex- ploit the environment in a manner not highly com- petitive with other organisms. This could involve the use of a new kind of food, or the ability to feed in places formerly unreachable. When this happens, the way is open for an extensive radiation of the pioneering group into a variety of specialized sub- zones of the general adaptive zone. Eor example, once a mechanism was developed for cracking hard-shelled seeds, specializations could occur in the relative size of the bill, in bill shape, in the me- chanics of jaw action, in the structure of the tongue, and in the locomotor apparatus, making possible a radiation into specific feeding niches defined by the size and hardness of the seeds taken, the substrate on which foraging occurs, and so forth. Fig. 10 is a cladogram representing the phylogeny of the New World nine-primaried oscines analyzed at the generic level for most of the forms dissected in the present study. Especially uncertain relation- ships are indicated as dotted lines; some of these show alternative possibilities. This phylogeny is not intended as a final solution to the problem of rela- tionships among the birds studied; rather it is a hy- 1978 RAIKOW— OSCINE APPENDICULAR MYOLOGY 35 pothesis representing what I consider to be the most reasonable interpretation of the information avail- able at the present time. It is presented as a frame- work for discussion of the possible history of the New World nine-primaried oscines and to point out specific problem areas where further research is particularly needed. We must first deal with the problem of the Vir- eonidae. Sibley (1970:170) reviewed the taxonomic history of this family; the general opinion today is that the family Vireonidae is related to the New World nine-primaried oscines, but is somewhat set apart from the other families. The remainder of the group is clustered by a collection of synapomor- phies and corroborating noncladistic data that ex- clude the Vireonidae. Both in the characteristics reviewed earlier, and in the limb muscle structure reported herein, the Vireonidae do not appear to be part of this otherwise monophyletic group. They share none of the evolutionary trends in limb mus- cle structure seen in the other families. In addition, they have several derived states not found in the other families, including the separation of the su- perficial head of M. gastrocnemius pars interna, the enlargement of flexor hallucis brevis, and the flexor digitorum longus insertion. Eurthermore they lack the tensor propatagialis scapular anchor and retain the latissimus dorsi caudalis. Except for Beecher’s ( 1953) belief that the jaw muscles show an ancestral relationship to the Parulidae, there is no compelling evidence, and most importantly, no shared derived character states that group the Vireonidae with the other New World nine-primaried oscines. There- fore, it must be concluded that the Vireonidae have at most only a distant phylogenetic link to this group, and should not be included within it. Cluster I With the exclusion of the Vireonidae, the re- mainder of the New World nine-primaried assem- blage clearly appears monophyletic on the basis of apparent synapomorphies (reduced tenth primary, pneumatic fossa of humerus, compressed basihyale with attached hyoglossus obliquus, and loss of la- tissimus dorsi caudalis) supported by noncladistic evidences (egg-white proteins, pterylosis, and pat- tern of adaptive diversity) as discussed in detail above. The Parulidae are in general the most primitive group of this assemblage on the basis of their limb muscles, although a few genera depart from the typ- ical condition of the family in certain muscles. This general condition is correlated with the widely held idea that purely insectivorous habits are more prim- itive than feeding on plants. Dendroica, Mniotilta, Oporornis, Myiohoriis, and Wilsonia are “typical” wood warblers showing the derived state of M. pronator profundus (la), but not separable from each other by any characters analyzed herein. Geothlypis and Basileutenis are similar except for possession (presumably retention) of the primitive state of M. pronator profundus. Zeledonia is very primitive myologically, even to the retention of a well-developed M. coracobrachialis cranialis. It is placed beside Basdeutenis , with which it possibly has a relatively recent common ancestor as dis- cussed above. The reduction of its flight mechanism is autapomorphous (lb). The “coerebid-parulids” Coereha and Coniros- truin are essentially typical parulids in their limb muscles. They are clustered here (Ic) because they share nectar-feeding adaptations. It is uncertain whether this is really a synapomorphy; quite pos- sibly they evolved this specialization independently and do not share a nectar-feeding common ances- tor. Limb myology cannot resolve this problem, but a study of the feeding apparatus might possibly do so. Seiiinis and Icteria are clustered by the common possession of a coracoidal origin of M. deltoideus minor (Id). However, each is so distinctive (le. If, and discussion above) that I consider it doubtful that they are closely related; more likely they evolved this derived state independently, especially because it only occurs in one of the three genera of Seiuriis examined. Their placement here is very speculative, and mainly serves to point out the need for additional studies of their relationships. Nephelornis and Urothraupis are morphological- ly similar to each other, but their exact position remains uncertain pending further studies. How- ever, they are myologically primitive and must lie somewhere close to the position shown, near the border of the Parulidae and Thraupidae. Cluster 2 The Parulidae are almost entirely insectivorous, whereas the emberizine, cardinaline, and cardue- line finches eat seeds as well, as do the more prim- itive members of the highly diversified Icteridae and Drepanididae. This provides a spectrum of feeding specializations in the New World nine-primaried assemblage; the problem has been to determine which is the primitive end of the spectrum (for ex- 36 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 7 Fig. lOA. — A cladogram suggesting phylogenetic relationships in the New World nine-primaried oscines. Numbers in circles designate presumed shared derived character states. See text for discussion. 1. Reduced tenth primary; fully developed pneumatic fossa; com- pressed basihyale with attached hyoglossus obliquus; latissimus dorsi caudalis lost. Supportive noncladistic data; egg-white protein similarities; indistinct family boundaries; pattern of adaptive diversity; pterylography. la. Type 2 pronator profundus, lb. Reduction of wings and near loss of flight. Ic. Nectar feeding. Id. Deltoideus minor coracoidal head added, le. Peroneus brevis tibial head added; obturatorius lateralis dorsalis lost. If. Large size; deltoideus minor coracoidal origin. 2. Feeding on fruits added to insectivory; bill heavier but lacking pronounced deflection. 2a. Type 2 pronator profundus. 2b. Patellar band of gastrocnemius lost. 2c. Plantaris lost; nectar feeding. 2d. Nectar feeding. 2e. Plantaris lost. 2f. Flexor digitorum longus type CBB; deltoideus minor coracoidal head added; flexor perforatus digiti 2 extra head added. 2g. Serratus supeificialis from 4th rib; flexor digitorum profundus type 2; flexor digitorum longus ACB; wide bill and flycatching. 2h. Interosseus dorsalis vestigial; coracobrachialis cranialis well developed; flexor digitorum longus ABA; plantaris lost. 3. Bill shorter and deeper, with pronounced deflection; seeds eaten, nutcracker method; deltoideus minor coracoidal head added. 3a. Bill still heavier; feed on larger percentage of seeds and often larger seeds, fewer insects. 3b. Patellar band of gastrocnemius lost. ample, Mayr, 1955:34). The limb muscle data ana- lyzed herein provide a solution to this problem; as most workers have suspected at least intuitively, the seed-eating habit is a derived specialization as shown by its correlation with limb muscle evolu- tion. However, seed-eating is a highly specialized condition requiring extreme structural modifica- tions of the feeding apparatus, and it is unlikely that it evolved suddenly in insectivorous forms. More likely birds first began eating the softer tissues of fruits, and gradually moved up to the harder seeds. Eruit-eating is a likely intermediate stage in the evo- lution of herbivorous habits, and thus the Thrau- pidae appear to represent a probably intermediate stage between purely insectivorous types (Paruli- dae) and the advanced granivores. The Thraupidae are a diverse group, some members being quite thin-billed and difficult to separate from the Paru- lidae, others being more intermediate, and still oth- ers closely approaching the cardinaline and ember- 1978 RAIKOW— OSCINE APPENDICULAR MYOLOGY 37 Fig. lOB. — 3c. Ground-foraging habits, sometimes with double-scratch method. 3d. Patellar band of gastrocnemius lost; flexor digitorum profundus Type 2. 3e. Rictal bristles reduced or lost. 3f. Obturatorius lateralis dorsalis lost. 3g. Primarily arboreal habits. 3h. Pronator profundus Type 1. 3i. Deltoideus minor coracoidal origin lost. 4. Bill heavier, with vise method of seed cracking; obturatorius lateralis dorsalis reduced. 4a. Patellar band of gastrocnemius lost. 4b. Plantaris lost. 5. Peroneus brevis tibial head added. 5a. Patellar band lost. 5b. Obturatorius lateralis dorsalis lost. 5c. Plantaris lost. 6. Colonization of Hawaiian Islands. Supportive noncladistic data: near uniformity of limb muscles throughout family. 6a. Bill enlarged with increased overlap of upper mandible. 6b. Bill, tongue, and nares elongated. 6c. Tubular tongue; enlargement of nasal operculum with convex margin. 6d. Plantaris lost. 6e. Nasal operculum larger and more flattened; plumage harder and less fluffy. 6f. Nasal operculum still larger, anterior notch added; primaries truncate. 6g. Bill elongated and decurved. izine finches. Thus the family seems to contain genera representing various stages in the evolution of the seed-eating types. In their limb myology the thraupids are generally rather primitive, but various genera differ in the possession of several derived conditions. More than any other family, this one is difficult to characterize as monophyletic, because the Thraupidae are not clustered by any important synapomorphies. The best solution, which 1 rec- ognize as very imperfect, is to regard the Thraupi- dae as probably representing more a structural/be- havioral grade than a monophyletic taxon, and to place it in this position as being more highly derived than the Parulidae but more primitive than those families with clear specializations for seed cracking. The following discussion and conclusions are rather tentative because 1 have studied only a small extent of the range of diversity that occurs in the Thrau- pidae. Many more genera must be examined in or- der to clarify the problems in this group. Point (2) indicates the development of herbivo- rous habits and in correlation, a heavier bill. The forms radiating from point (2) constitute the heter- ogeneous thraupid assemblage, plus the ancestral lineage of the seedeaters. Tangara and Euphonia are the most primitive genera in limb myology; their Type 1 pronator profundus is more primitive than even the typical parulid condition. If we assume that the evolution of herbivory in addition to insec- tivory occurred only once, so that the thraupid as- semblage is monophyletic in the strictest sense, then these genera will arise as shown from point (2) in Fig. 10. This assumption is certainly parsi- monious, but is also very simplistic and probably false. On the basis of their primitive limb myology these two genera could also have arisen earlier within the primitive insectivorous ancestry of the assemblage, as shown by the alternative pathway leading from point (1). In this case they would have developed herbivorous adaptations independently of other Thraupidae. Here is a case where a study of the feeding mechanism could prove helpful. Tachyplwnus, Thraupis, Rhamphocoelus , and Piranga are “typical" tanagers and essentially alike in limb myology. They share synapomorphous states of M. pronator profundus (2a) and M. gas- 38 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 7 trocnemius (2b). Cyanerpes has these features plus the loss of M. plantaris and the addition of nectar feeding (2c). Again, this group is part of the radia- tion of a monophyletic Thraupidae if we hold to the assumption of a single origin of herbivory. How- ever, if we do not insist on this, then the genera arising from (2b) could, on the basis of limb myol- ogy, also have arisen from point (la) instead, as shown by the dotted line in Eig. 10. The “coerebid-thraupids" Dacnis, Cliloro- phaiies, and Eunerornis are similar in limb myolo- gy; Diglossa has an additional derived state in the loss of M. plantaris (2e). These forms are clustered on the basis of a presumed common origin of nec- tar-feeding adaptations (2d). As with Coereba and Conirostruin in the Parulidae, this is an expedient, parsimonious, and tentative placement pending fu- ture analysis of the feeding mechanism. Again, these forms could arise via (la) rather than (2a) if the assumption of monophyletic herbivory is not held to, as shown in Eig. 10. If Cyanerpes is also hypothesized as part of a single radiation of nectar- feeding forms, then its loss of the patellar band (2b) must have occurred independently. As discussed earlier, Rliodinocichia, Tersina, and Catamblyrhynchus lie within the thraupid as- semblage, but on the basis of limb myology these traditionally problematic forms remain troublesome (2f, 2g, 2h). The depiction of the thraupid radiation arising from a single point (2) implying a single common ancestor for the group is undoubtedly a gross over- simplification. It appears intuitively likely that sev- eral lineages independently developed herbivorous feeding adaptations, as suggested by the alternative phytogenies shown in Eig. 10. Only a few thraupid genera were dissected in this study, and the true phytogeny of the group must be far more complex than that shown here. The limited analysis of limb myology has not solved this problem, but it has clearly shown that the analysis of the limb muscles can support a hypothesis of a polyphyletic family Thraupidae. Cluster 3 These groups are clustered by a feeding appara- tus adapted for cracking seeds. Exceptions are some icterids and drepanidids with feeding mecha- nisms otherwise specialized, but which are clearly derived within their families, the more primitive members of which possess the seed-eating adapta- tions. This adaptation is presumably derived by fur- ther specialization from the principally fruit-eating Thraupidae. Bock (1960) analyzed the functional anatomy of the feeding mechanism. The emberizines show a relatively modest development of this sys- tem, and feed on relatively smaller seeds, using a biomechanical system that Bock termed the “nut- cracker” method. Cardinalines use the same meth- od, but have larger bills and eat larger seeds. This group is also clustered by possession of a derived character state in a wing muscle, the pres- ence of a coracoidal head to M. deltoideus minor. There are a few exceptions, however. A few genera of most of these families have the primitive state, and a few parulids and one (presumed) thraupid have the derived state (Table 2). Nevertheless, nearly all parulids and thraupids are primitive, and nearly all the remainder are derived in this char- acter. This is an example of the type of character conflicts that were disucssed earlier, and because they are few, I think it probable that cluster 3 is correctly grouped by this synapomorphy. Thus cluster 3 is grouped by two independent characters (seed-eating adaptations and coracoidal head of the deltoideus minor). Three lineages are shown arising from this branching point, though strict cladistic methodolo- gy demands only dichotomous branches. In order to resolve this it would be necessary to demonstrate some synapomorphy clustering two of the three branches, but I do not know of one. The Cardinalinae resemble the Thraupidae in their colorful plumages and arboreal habits, but in their enlarged bills and habit of eating larger seeds (3a) appear to be more highly derived in feeding specializations than the emberizines. There is some variation in limb muscles (3b) but as only a few genera were dissected, the range of variation in the group is not well known. The Emberizinae and Icteridae share a special- ized ground foraging technique with both structural and behavioral aspects (3c). The other taxa in this assemblage are highly arboreal and seldom forage to any extent in open terrestrial habitats. They are also characterized in general by brightly colored, conspicuous plumages, at least in males. The em- berizines in contrast are more terrestrial in foraging habits, and less brightly colored. Their plumages emphasize browns, grays, blacks, and sometimes yellows, and their backs are commonly streaked with brown and gray. Presumably this is associated with their ground foraging habits by making them less conspicuous to predators. The Icteridae appear 1978 RAIKOW— OSCINE APPENDICULAR MYOLOGY 39 to have been derived from the emberizines through such forms as Spiza, Dolichonyx , and Molothrus , which resemble them in bill form and habits. Eur- thermore, many emberizines forage with a special- ized type of movement, the bilateral scratch, in which they jump first forward and then backward with both legs simultaneously, scratching the sub- strate on the second jump to scatter surface litter and reveal food. At least two icterid genera, Mol- othrus and Agelaius, are known to use a similar foraging technique (Greenlaw, 1976). The emberizines are very uniform in limb myol- ogy and most genera are inseparable on this basis. Geospiza, Einberiza, and Loxigilla lack the cora- coidal origin of M. deltoideus minor. This could be either a primitive state or a derived (secondarily primitive) condition. I have chosen the second al- ternative as shown in Eig. 10 (3i) because they so closely resemble the other emberizines in general, but the matter is uncertain. In view of the wide geographic separation between these genera, it is likely that this derived state arose independently in the three groups, so that cluster 3i may well be a false synapomorphy. The Icteridae are clustered by two derived states, the loss of the patellar band and the Type 2 flexor digitorum profundus (3d). In most icterids the rictal bristles are either vestigial or completely lost (3e) as noted by Ridgway (1902:169). However, I have observed that the rictal bristles are well developed in Spiza, which myologically is allied with the Ic- teridae (3d). This supports the idea that Spiza is a primitive icterid, and a link to the emberizines. The assumption of (3c), that ground-foraging is a synapomorphy of the Icteridae and Emberizinae, is contradicted by the arboreal habits of some ic- terids (3g) and the primitive Type 1 pronator pro- fundus of a few (3h). If this suggestion is correct, then the apparently primitive states of cluster (3g) are presumably due to evolutionary reversal. The conflict is clearly unsettled, but I have chosen as most probable the scheme shown in Eig. 10 because (1) Spiza is so clearly an intermediate form between the two groups, and (2) the grossly enlarged bill of most genera in cluster (3g) is highly specialized and most certainly derived within this assemblage. Cluster 4 This group includes the cardueline finches and the Drepanididae. The carduelines and the drepa- nidid genus Psittirostra are seedeaters that possess a more specialized biomechanical seed-cracking system than that found in the emberizines and car- dinalines. This was studied by Bock (1960), who termed it the “vise” method. It includes a stronger skull ossification, a restriction of the mobility of the upper jaw, and other features, and is thought to have evolved from the less specialized “nutcrack- er” system of the other members of cluster (3). It must be noted, however, that Bock's concept of “nutcracker” and “vise” systems of jaw biome- chanics has been questioned (Zusi, 1961). The car- duelines and drepanidids are also clustered by the reduction in size (leading in some cases to complete loss) of a hip muscle, M. obturatorius lateralis dor- salis. Despite their diversity, the Carduelinae appear to be a coherent group on the basis of various struc- tural and behavioral features (see Tordoff, 1954) as well as the presence of the vise method of seed cracking, and so the parsimonious suggestion that the group is monophyletic is a reasonable one. For this reason they are shown as having arisen from a single lineage (Fig. 10). Several cardueline genera appear relatively prim- itive by the absence of a derived state, the tibial head of M. peroneus brevis. In terms of limb myol- ogy Leucosticte is the most primitive cardueline studied (4); the structure of the tongue also supports this view (Raikow, \911b). Several other genera are more advanced in limb myology (4a. 4b). Cluster 5 The development of the tibial head of M. pero- neus brevis, a derived state, occurs in several other genera (cluster 5), which may be further clustered on other myological synapomorphies (5a, 5b, 5c). There is a character conflict in that the plantaris was apparently lost twice (at least) in the cardue- lines (4b, 5c), but the arrangement shown here is the most reasonable one given the distribution of the tibial head of the peroneus brevis. Cluster 6 The derivation of the Drepanididae is treated in considerable detail elsewhere (Raikow, \911b) and therefore will be considered here only briefly. It is difficult to demonstrate by strictly cladistic methods that the family Drepanididae is monophyletic; be- cause of their incredible diversity, resulting from their adaptive radiation, they do not all share any clear-cut derived character states not also found among the Carduelinae. The argument for mono- phyly is based instead on the geographical restric- 40 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 7 tion to the Hawaiian Islands, the morphological in- tergradation between distinct adaptive types, and the remarkable uniformity of their limb muscles (Raikow, 1976, \911a, \911b). Psittirostra is a finch-billed, seed-eating genus, whereas the other genera have bills adapted to a variety of feeding specializations. Psittirostra is clearly the most primitive drepanidid genus, little modified from the cardueline ancestor that founded the family. A phylogeny of the genera of the Drep- anididae is shown in Fig. 10 for the sake of com- pleteness, but as I have analyzed it in detail else- where (Raikow, 1977/?), I will not do so here. PROPOSED CLASSIFICATION Because of the diverse philosophies of classifi- cation now current, it is desirable to explain briefly the basis for a proposed classification. The school of phylogenetic systematics based on the ideas of Hennig includes both a methodology for recon- structing phylogenies, and a technique for con- structing classifications based on those phyloge- nies. I accept the first but reject the second. Thus, although my phylogeny was derived by mostly Hen- nigian methods, my proposed classification is not. In the Hennigian or cladistic method of classifi- cation the hierarchy of taxonomic categories is based directly on the branching pattern of the dado- gram without regard to the nature or degree of evo- lutionary change (which may be quite variable) oc- curring between branching points. In a group of any size this will result in a large number of categories and of named taxa. The resulting classification is so complex and unwieldy as to be wholly impractical. Consider, for example, the classification of mam- mals proposed by McKenna (1975). This ineludes the categories (in alphabetical order) class, cohort, grandorder, infraclass, infraorder, legion, magnor- der, mirorder, order, parvorder, subclass, suble- gion, supercohort, superlegion, and superorder. These are just the categories; the named taxa are much more numerous. The method does have the advantage that one can specify any monophyletic group in the cladogram by name. However, this can also be done more simply by referring to the group by the number of the branching point from which it arises. For instance, in referring to the group con- taining the Drepanididae plus those carduelines pos- sessing the peroneus brevis tibial head (Fig. 10), one could specify, say, the "infrasupercohort Car- dueloidida," but it is simpler (dare I say more par- simonious?) merely to say "cluster 5." I also reject Hennigian classification because it is redundant, being merely verbal restatement of the cladogram. I prefer a classification in which evo- lutionary changes are considered, so that each tax- on is characterized by some adaptive or other pe- culiarity that sets it apart from other taxa. If we are to have both phylogenies and classifications, why not let them serve different purposes? Critics claim that this results in subjectivity because different workers, given the same data, might produce dif- ferent classifications. This may be true, but I see no fault in it. Different workers can emphasize the events that they consider to have been of the great- est biological significance in the history of the group. Mayr (1974) has effectively criticized cla- distic classification, so I will not belabor the point further. For most families I have not suggested subfami- lies because I consider the basis for them to be un- certain or ambiguous. Future studies of intergeneric relationships may make subdivision of more fami-' lies possible. My elassification is not greatly differ- ent from those generally in use at the present time, as the limb myology has, in general, confirmed pre- vious ideas of relationships. Several small families, such as the Cyclarhidae, Vireolaniidae, Tersinidae, and Catamblyrhynchidae, listed by Wetmore (I960), are not recognized because I have included their genera in other families of which they appear to be aberrant members. Inclusion of Fringilla with the cardueline finches unfortunately necessitates the use of the name Fringillinae for the combined group, although I would have preferred to retain the name Carduelinae. Each family recognized appears to be a coherent group. The Vireonidae includes Cyclarhis and Vir- eolanius because of their anatomical similarity, as pointed out by several other authors. The Parulidae are essentially arboreal insectivores, with occasion- al lines specializing in different directions, as with Seitiriis and the nectar-feeders. The Thraupidae are essentially arboreal fruit and insect eaters. This family is still the least understood, and may by poly- phyletic, but recognition of a single family is the most practical course on the basis of our present 1978 RAIKOW— OSCINE APPENDICULAR MYOLOGY 41 understanding. The Eringillidae are seed-eaters. The Cardinalinae are basically heavy-billed types using the nutcracker method of seed breaking, and with colorful plumages. The Emberizinae are es- sentially smaller-billed nutcracker types, with em- phasis on ground foraging. The Eringillinae (Car- duelinae) are mainly short-legged, arboreal seedeaters with a heavy bill, using the vise method of seed cracking. The Icteridae, despite their con- siderable diversity, are marked by a specialized bill- gaping feeding mechanism. The line between Erin- gillinae and Drepanididae is vague and admittedly arbitrary. However, I feel that the Hawaiian Hon- eycreepers merit family rank because they are un- questionably the product of a single adaptive radia- tion in a geographic region where no other members of the assemblage occur naturally. The sequence of families given here is, in my opinion, the most reasonable solution to the prob- lem of expressing a multidimensional branching pat- tern in a linear order. In any linear sequence of the families of oscines, the sequence from the Parulidae onward should be kept intact, with no other families interposed. This follows from the monophyly of the group. The position of the Vireonidae, however, remains problematical because their affinities are still obscure. Proposed Classification: Vireonidae Parulidae Thraupidae Eringillidae Cardinalinae Emberizinae Eringillinae Drepanididae Icteridae CONCLUSIONS 1. The hypothesis that the New World nine-pri- maried oscines form a monophyletic group is supported by the comparative anatomy of the limb muscles for all groups except the Vireoni- dae. Eor all other groups the muscles show a common general pattern with a series of gradual evolutionary changes consistent with a scheme of adaptive radiation presented in a phylogenetic hypothesis. 2. The Vireonidae do not share derived character states with the other groups and cannot be in- cluded with them in a single phylogeny. The shrike-vireos and peppershrikes share unique derived states with the vireos, and it is recom- mended that all be included in a single family whose monophyletic nature is indicated by their synapomorphies. 3. The limb myology indicates that the Parulidae are a primitive but fairly cohesive group which is probably monophyletic. This conclusion is based more on the general similarity of most gen- era than on the existence of any clear-cut syn- apomorphies. The “coerebid” genera Caere ha and Conirostriim are included. Peucedramus does not fit into this family but probably belongs with the Sylviidae. Icteria and Seiurus are aber- rant in limb muscles but the significance of this is uncertain. Zeledonia is myologically primitive. and may reasonably be included in the Parulidae near Basileiitenis . 4. The Thraupidae are myologically heterogeneous and may be polyphyletic. Nephelonds and Uro- thraupis are very similar and relatively primitive. Their exact position is uncertain but is at the parulid/thraupid border. It is somewhat arbitrar- ily suggested that they be placed in the Thrau- pidae. The “coerebid” tanagers fit easily within the Thraupidae, but the question of their mon- ophyly as nectar feeders is unsettled. Rhodino- ciclda, Tersina, and Catamhlyrhynchus are best treated as aberrant thraupids whose intergeneric affinities are obscure. 5. The Emberizinae and Icteridae may be sister- groups on the basis of terrestrial foraging habits and techniques. The limb myology is consistent with the hypotheses that both are monophyletic groups. Spiza is a primitive icterid close to the Emberizinae. The orioles, caciques, and oropen- dolas are secondarily arboreal. I. The Cardinalinae are somewhat diverse in limb myology and their phylogenetic position is prob- ably close to the emberizine/icterid group. 7. The Carduelinae are diverse in limb myology, with Leiicosticte and Fringdla being relatively primitive within the group. The hypothesis of monophyly is supported by a derived jaw mech- 42 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 7 anism and a pattern of derived limb muscle fea- tures. 8. The Drepanididae are almost certainly mono- phyletic as indicated by geographical distribution and morphological intergradation. They arose from a single cardueline founder species. 9. Based on the foregoing phylogenetic study, the general taxonomic recommendation arising from this investigation follows from a belief that the family category should be used for grouping rather than separating forms, and that families with one or a very few genera should be avoided as much as possible. Eor the most part I will not suggest subfamily groups because I consider the basis for most to be uncertain or ambiguous. Proposed Classification; Vireonidae Parulidae Thraupidae Eringillidae Cardinalinae Emberizinae Eringillinae Drepanididae Icteridae ACKNOWLEDGMENTS Extensive anatomical studies would not be possible without the generous cooperation of individuals and institutions in pro- viding loans of specimens for dissection. Materials used in the present study were provided by A. J. Berger (University of Ha- waii); P. J. K. Burton (British Museum, Natural History, Tring); Mary H. Clench (Carnegie Museum of Natural History, Pitts- burgh); Ned K. Johnson (Museum of Vertebrate Zoology, Uni- versity of California, Berkeley), George H. Lowery, Jr., and John P. O’Neill (Louisiana State University Museum of Zoolo- gy, Baton Rouge); Charles G. Sibley (Peabody Museum of Nat- ural History, Yale University, New Haven); and Richard L. Zusi (National Museum of Natural History, Washington, D.C.). The services of the Museum of Vertebrate Zoology are supported by NSF Grant BMS 7200102. I am grateful to Kenneth C. Parkes, Harrison B. Tordoff, and Richard L. Zusi for reading and criti- cizing the manuscript, and to Mary H. Clench for many helpful suggestions. Supported by NSF grants DEB76 20337 and BMS74 18079. LITERATURE CITED Ames, P. L. 1975. The application of syringeal morphology to the classification of the Old World insect eaters (Muscicap- idae). Bonn. Zool. Beitr., 26:107-134. Beecher, W. J. 1951«. Convergence in the Coerebidae. Wilson Bull., 63:274-287. . 19516. Adaptations for food-getting in the American blackbirds. Auk, 68:411^140. . 1953. A phylogeny of the oscines. Auk, 70:270-333. Bentz.G. D. 1976. The appendicular myology and phylogenetic relationships of the Ploceidae and Estrildidae ( Aves: Passer- iformes). Unpublished Ph.D. dissert. Univ. Pittsburgh, Pitts- burgh, Pennsylvania, 157 pp. Berger, A, J. 1966. The musculature. Pp. 224-473, Avian myol- ogy (J. C. George and A. J. Berger, eds.) Academic Press, New York and London, 500 pp. . 1968. Appendicular myology of Kirtland’s Warbler. Auk, 85:594-616. . 1969. Appendicular myology of passerine birds. Wilson Bull., 81:220-223. Blake, E. R. 1968. Family Vireonidae, peppershrikes, shrike- vireos, and vireos. Pp. 103-138, Check-list of birds of the world (R. A. Paynter, Jr., ed.), Mus. Comp. Zool., Cam- bridge, Massachusetts, 14:1-433. Bock, W. J. 1960. The palatine process of the premaxilla in the Passeres. Bull. Mus. Comp. Zool., 122:361-488. . 1962. The pneumatic fossa of the humerus in the Pas- seres. Auk, 79:425^443. . 1965. The role of adaptive mechanisms in the origin of higher levels of organization. Syst. Zool., 14:272-287. Bock, W. J., and R. Shear. 1972. A staining method for gross dissection of vertebrate muscle. Anat. Anz., 130:222-227. Clark, G. A., Jr. 1974. Foot-scute differences among certain North American oscines. Wilson Bull., 86:104-109. Clench, M. H., and O. L. Austin, Jr. 1974. Passeriformes. En- cyclopedia Brittanica, Inc., 5th ed., pp. 1052-1066. Delacour, J. 1972. Hybrids Sugar-bird x tanager (Cvanerpes cy emeus x Tangara nigrocinctci fnmciscae) . Avicult. Mag., 78:187-188. Eisenmann, E. 1962a. On the genus “C/iawact/i/vpiY" and its supposed relationship io Icteria. Auk, 79:265-267. . 19626. On the systematic position of Rhodinocichla ro- sea. Auk, 79:640-648. Ficken, M. S., and R. W. Ficken. 1962. Some aberrant char- acters of the Yellow-breasted Chat, Icteria virens. Auk, 79:718-719. Gaunt, A. S. 1969. Myology of the leg in swallows. Auk, 86:41- 53. George, J. C., and A. J. Berger. 1966. Avian myology. Ac- ademic Press, New York and London, 500 pp. George, W. G. 1962. The classification of the Olive Warbler, Peucedramus taeniatus . Amer. Mus. Novit., 2103:1-41. . 1968. A second report on the basihyale in American songbirds, with remarks on the status oi Peucedramus . Con- dor, 70:392-393. 1978 RAIKOW— OSCINE APPENDICULAR MYOLOGY 43 Greenlaw, J. S. 1976. Use of bilateral scratching behavior by emberizines and icterids. Condor, 78:94-97. Hunt, J. H. 1971. A field study of Wrenthrush, Zeledonia co- ronata. Auk, 88: 1-20. Kluge, A. G. 1977. Concepts and principles of morphologic and functional studies. Pp. 1-27, in Chordate structure and func- tion (A. G. Kluge, ed.), Macmillan Publ. Co., New York, 2nd ed., 628 pp. Lowery, G. H., Jr., and B. L. Monroe, Jr. 1968. Family Paru- lidae. Wood Warblers. Pp. 3-93, in Check-list of birds of the World (R. A. Paynter, Jr., ed.), Mus. Comp. Zool., Cam- bridge, Massachusetts, 14: 1-433. Lowery, G. H., Jr., and D. A. Tallman. 1976. A new genus and species of nine-primaried oscine of uncertain affinities from Peru. Auk, 93:415-428. Lowther, P. a. 1975. Geographic and ecological variation in the family Icteridae. Wilson Bull., 87:481-495. Mayr, E. 1955. Comments on some recent studies of song bird phylogeny. Wilson Bull., 67:33^14. . 1974. Cladistic analysis or cladistic classification? Zool. Syst. Evol.-forsch., 12:94-128. . 1976. The emergence of evolutionary novelties. Pp. 88- 1 13, in Evolution and diversity of life (E. Mayr, ed.). Har- vard Univ. Press, Cambridge, Massachusetts, 721 pp. Mayr, E.,andD. Amadon. 1951. A classification of recent birds. Amer. Mus. Novit., 1496: 1-42. McKenna, M. C. 1975. Toward a phylogenetic classification of the Mammalia. Pp. 21^6, in Phylogeny of the Primates (W. P. Luckett and F. S. Szaly, eds.). Plenum Publ. Corp., New York, 483 pp. Paynter, R. A., Jr. 1970. Subfamily Catamblyrhynchinae. P. 215, in Check-list of birds of the World (R. A. Paynter, ed.), Mus. Comp. Zool., Cambridge, Massachusetts, 13:1^43. . 1970. Subfamily Emberizinae, Buntings and American Sparrows. Pp. 3-214, in Check-list of birds of the World (R. A. Paynter, ed.), Mus. Comp. Zool., Cambridge, Massachu- setts, 13:1-443. Raikow, R. j. 1975. The evolutionary reappearance of ancestral muscles as developmental anomalies in two species of birds. Condor, 77:514-517. . 1976. Pelvic appendage myology of the Hawaiian Hon- eycreepers (Drepanididae). Auk, 93:774-792. . 1977fl. Pectoral appendage myology of the Hawaiian Honeycreepers (Drepanididae). Auk, 94:331-342. . 1977b. The origin and evolution of the Hawaiian Hon- eycreepers (Drepanididae). The Living Bird, 15:95-117. Ross, H. H. 1974. Biological systematics. Addison-Wesley, Reading, Massachusetts, 345 pp. Ridgway, R. 1902. The birds of North and Middle America, Part II. Bull. U.S. Nat. Mus. 50:1-834. Schaefer, C. W. 1976. The reality of the higher taxonomic cat- egories. Z. Zool. Syst. Evolut.-forsch., 14: 1-10. Short, L. L., Jr., and C. S. Robbins. 1967. An intergeneric hy- brid Wood Warbler (Seiurus x Dendroica) . Auk, 84:534- 543. SiBLEY, C. G. 1968. The relationships of the “wren-thrush," Zeledonia coronata Ridgway. Postilla, 125: 1-12. . 1970. A comparative study of the egg-white proteins of passerine birds. Bull. Peabody Mus. Nat. Hist., Yale Univ., 32:1-131. Skutch, a. F. 1962. On the habits of the Queo, Rhodinocichla rosea . Auk, 79:633-639. Stallcup, W. B. 1954. Myology and serology of the avian fam- ily Fringillidae, a taxonomic study. Univ. Kansas Publ., Mus. Nat. Hist., 8: 157-211. Storer, R. W. 1969. What is a tanager? The Living Bird, 8: 127- 136. SuSHKiN, P. P. 1925. The Evening Grosbeak (Hesperiphona), the only American genus of a Palearctic group. Auk, 42:256- 261. . 1929. On the systematic position of the Drepanididae. Proc. Internat. Omith. Congr., 6:379-381. Tordoff, H. B. 1954. A systematic study of the avian family Fringillidae based on the structure of the skull. Mus. Zool. Misc. Publ., Univ. Michigan, 81:1-42. Van Tyne, J., and A. J. Berger. 1976. Fundamentals of orni- thology. John Wiley & Sons, New York, 2nd ed., 808 pp. Wetmore, a. 1960. A classification for the birds of the world. Smithsonian Misc. Coll., 139(ll):l-37. Zusi, R. 1961. [Review ot] Bock, W. J., 1960, The palatine pro- cess of the premaxilla in the passeres. Bull. Mus. Comp. Zool., 122:361-488. Auk, 78:101-102. ' > . ^ ►; I* ‘ ■ r-?; ?t V. -( '-'■A ■ I Copies of the following Bulletins of Carnegie Museum of Natural History may be obtained at the prices listed from the Publications Secretary, Carnegie Museum of Natural History. 4400 Forbes Avenue, Pitts- burgh, Pennsylvania 15213. 1. Krishtalka, L. 1976. Early Tertiary Adapisoricidae and Erinaceidae (Mammalia, Insectivora) of North America. 40 pp., 13 figs $2.50 2. Guilday, J. E., P. W. Parmalee, and H. W. Hamilton. 1977. The Clark’s Cave bone deposit and the late Pleistocene paleoecology of the central Appalachian Mountains of Virginia. 88 pp., 21 figs. $12.00 3. Wetzel, R. M. 1977. The Chacoan peccary, Cu?agou//i' vvugut'/v (Rusconi). 36 pp., 10 figs. .. $6.00 4. Coombs, M. C. 1978. Reevaluation of early Miocene North American Moropus (Perissodactyla, Chalicotheriidae, Schizotheriinae). 62 pp., 28 figs $5.00 5. Clench, M. H., and R. C. Leberman. 1978. Weights of 151 species of Pennsylvania birds analyzed by month, age, and sex. 87 pp $5.00 6. Schlitter, D. A. (ed.). 1978. Ecology and taxonomy of African small mammals. 214 pp., 48 figs. $15.00 SKULL AND RELATIONSHIPS OF THE UPPER JURASSIC SAUROPOP AMro^t^ (REPTILIA, SAURISCHIA) DAVID S BERMAN AND JOHN S. MCINTOSH NUMBER 8 PITTSBURGH, 1978 \ k a BULLETIN of CARNEGIE MUSEUM OF NATURAL HISTORY SKULL AND RELATIONSHIPS OF THE UPPER JURASSIC SAUROPOD APATOSAURUS (REPTILIA, SAURISCHIA) DAVID S BERMAN Assistant Curator, Section of Vertebrate Fossils JOHN s. McIntosh Wesleyan University , Middletown, Connecticut 06457 NUMBER 8 PITTSBURGH, 1978 BULLETIN OF CARNEGIE MUSEUM OF NATURAL HISTORY Number 8, pages 1-35, figures 1-11, table 1 Issued 17 November 1978 Price: $3.00 a copy Craig C. Black, Director Editorial Staff: Hugh H. Gcwov^ays. Editor; Duane A. Schlitter, Editor: Stephen L. Williams, Aiioc/a/e Editor; Teresa M. Bona, Technical Assistant. © 1978 by the Trustees of Carnegie Institute, all rights reserved. CARNEGIE MUSEUM OF NATURAL HISTORY, 4400 FORBES AVENUE PITTSBURGH, PENNSYLVANIA 15213 CONTENTS Abstract 5 Introduction 5 Abbreviations 6 Historical Review 6 Previous Collections and Descriptions 6 Discussion 1 1 Skull of Diplodocus 12 External Features 13 Braincase 19 Probable Skull of Apatosaurus 21 Description of Skull CM 11162 21 Comparison of CM 11162 W\\h Diplodocus 24 Additional Evidence on the Skull of Apatosaurus 27 Comparison of Postcranial Skeletons 30 Vertebral Column 30 Appendicular Skeleton 31 Relationships of Apatosaurus 32 Acknowledgments 34 Literature Cited 34 M' ■ •>£ '■'S ABSTRACT Evidence is presented to show that \) Apatosaurus probably possessed a Diplodocus -like, rather than Camarasaurus-Wke, skull, and 2) Apatosaurus and Diplodocus are closely related and well separated from Camarasaurus . A Diplodocus-\\kc skull attributed Xo Apatosaurus by W. J. Holland over a half century ago is described for the first time. A cranium and a pair of quad- rates that are very similar to those of Diplodocus are also de- scribed and shown probably to belong io Apatosaurus . Inaccur- acies and omissions in previous descriptions of the skull of Diplodocus have necessitated a redescription of much of its ex- ternal features and braincase. Differences between the skull at- tributed here to Apatosaurus and that of Diplodocus are of a subtle proportional and structural nature. Comparisons of the postcranial skeletons of the Jurassic Apatosaurus, Diplodocus, and Camarasaurus demonstrate that the former two genera share a large number of significant features and are quite distinct from the latter. Apatosaurus and Diplodocus . along with the Jurassic Baro- saurus, Cetiosauriscus, Mumenchisuurus, and Dicraeosaurus and the Cretaceous Nemegtosaurus , should be grouped under Diplodocidae Marsh, 1884. INTRODUCTION Of the six well-established sauropod genera from the Upper Jurassic Morrison Formation of North America, Brachiosauriis, Haplocanthosaurus , and Barosaurus are rare and the latter two incompletely known. The other three, Camarasaurus, {=Moro- saurus, —Uiutasaurus), Apatosaurus (=Brontosau- rus), and Diplodocus are common and well known. The last three genera were described a century ago on small but diagnostic portions of skeletons (Cope, 1877u; Marsh, 1877/?, 1878u). Although large col- lections of numerous isolated elements and some partial skeletons of Camarasaurus, Apatosaurus, and Diplodocus were made during the next quarter century, adding greatly to the knowledge of these genera, significant misinterpretations concerning their morphology and relationships arose. Despite detailed descriptions (Gilmore, 1925, 1936; Hatch- er, 1901u, 19036; Holland, 1906) of excellent spec- imens of all three genera even after this period, some of the misinterpretations were so entrenched in the literature that they persist today. The mis- understandings discussed here regard two impor- tant aspects of the morphology and relationships of Apatosaurus . First, since the original restorations of "^Brontosaurus" published by Marsh (1883, 1891), the skull Apatosaurus has been incorrectly depicted as being CamarasaurusAWse and the alter- native suggestion presented by Holland (1915c/) and supported here, that it is Diplodocus-Wkt, has been almost totally ignored. Second, Apatosaurus has been falsely viewed as more closely related to Cam- arasaurus than to Diplodocus, despite the fact that descriptions of their postcranial skeletons demon- strate just the opposite. In an earlier review of the nature of the skull of Apatosaurus, we (McIntosh and Berman, 1975) dis- cussed the controversy raised by Holland (1915u, 1924), who refuted Marsh's (1883, 1891) use of probable Camarasaurus skulls in his restorations of Apatosaurus , pointing out that the skulls were not found directly, or even closely, associated with postcranial skeletons of this genus. Holland argued that the skull of Apatosaurus is probably like that ok Diplodocus . His opinion was based almost solely on a very large Diplodocus-Wkc skull that was very closely associated with two nearly perfectly pre- served postcranial skeletons of Apatosaurus in the quarry at what is now Dinosaur National Monu- ment near Jensen, Utah. This skull was never de- scribed by Holland and his assertion was almost totally disregarded, because it ran counter to the then well-established view that Apatosaurus is structurally more similar and more closely related to Camarasaurus than to Diplodocus', this skull is described here for the first time. A pair of quadrates and the greater portion of a cranium that are nearly indistinguishable from those of Diplodocus are also described as probably belonging to Apatosaurus . These elements were found near Morrison, Colo- rado, by Marsh's collectors in 1877 and evidence is presented that suggests that they belong to the ho- lotype of A. ajax. Previous descriptions of the skull of Diplodocus are inaccurate and incomplete, ne- cessitating a redescription of most of its external features and braincase. A comparison of the post- cranial skeletons of the three common, Morrison Formation sauropods confirms that Apatosaurus and Diplodocus are very similar and are distinct from Camarasaurus . The close resemblance of the skull and postcra- nial skeleton ok Apatosaurus to those ok Diplodocus clearly indicates that Apatosaurus is more closely 5 6 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 8 related to Diplodociis thantoCamarasaurus. Apato- saurus and Diplodociis, along with five lesser known genera, Barosaunis, Cetiosaiiriscus, Ma- menchisaiirus, Dicraeosaiirus, and Nemegtosau- riis, are grouped under Diplodocidae Marsh, 1884, for the first time. ABBREVIATIONS AMNH, CM, USNM, and YPM refer to collections at the American Museum of Natural History, the Carnegie Museum of Natural History, the National Museum of Natural History, and the Yale Peabody Museum, respectively. Abbreviations used in figures are as follows; aca canal for anterior cerebral artery Bo basioccipital bpp basipterygoid process Bs basisphenoid ca crista antotica cp crista probtica Eo exoccipital F frontal gr pn groove for palatine branch of facial nerve ica foramen for internal carotid artery J jugal jv foramen for jugular vein L lacrimal Ls laterosphenoid M maxilla mpa canal for median palatine artery N nasal Op opisthotic Os orbitosphenoid P parietal pa foramen for palatine artery pao preantorbital opening Pf prefrontal pf posttemporal fenestra Pm premaxilla pn foramen for palatine branch of facial nerve Po postorbital Pr prootic Ps parasphenoid Pt pterygoid Q quadrate Qj quadratojugal So supraoccipital Sq squamosal s-Sq sutural surface for squamosal 1-Xll foramina for cranial nerves HISTORICAL REVIEW Previous Collections and Descriptions In order to understand fully and, therefore, hope- fully, to resolve remaining areas of confusion con- cerning the morphology of Apatosaurus and its re- lationships to other sauropods, particularly to Camarasaurus and Diplodociis, it is necessary to give a detailed chronicle of the circumstances and events surrounding the collecting and description of those specimens pertinent to this topic. In July, 1877, Marsh (I877u) described a large incomplete sauropod sacrum (YPM 1835) as Ti- tanosauriis montanus, which was found by Arthur Lakes and H. C. Beckwith at what was later des- ignated as YPM quarry 1 north of Morrison, Colo- rado (Ostrum and McIntosh, 1966). As Lydekker (1877) had used the same generic name several months earlier in describing two caudals and a chevron of a different species of sauropod, Titano- sauriis indiciis. Marsh (1877^) altered the name of his Morrison specimen to Atlantosaurus montanus in December of the same year. Earther on in the same publication Marsh also described a second sauropod sacrum and vertebrae as representing a new genus. Apatosaurus ajax. This specimen, also discovered by Lakes in another quarry near Mor- rison, later designated YPM quarry 10, had origi- nally been sent to E. D. Cope for identification. However, when Marsh purchased the specimen from Lakes, Cope sent it to the Yale Peabody Mu- seum at Lake's request. Marsh kept accurate rec- ords of his collections and an accession number was placed on all fossils arriving at the museum as the boxes containing them were unpacked. Later, when the bones were identified and studied, they were given a catalogue number. Tho Apatosaurus sacrum from YPM quarry 10 which originally had been sent to Cope, was included as part of accession no. 993 and was later catalogued as YPM 1860. Lakes was to make a number of additional shipments from a total of 11 separate quarries at Morrison. In the ship- ments that followed from quarry 10 there was a very large femur that was attributed by Marsh (1878a) to a new species of Atlantosaurus, A. invnanis, and 1978 BERMAN AND McINTOSH— AEAJ05AC/i? [75 RELATIONSHIPS 7 was later catalogued as YPM 1840. The numerous shipments from Morrison included many more ele- ments from quarry 10. By 1883 the Morrison col- lections had been prepared and it was evident that the material from quarry 10 belonged to two very large skeletons. The smaller of the two, which in- cluded the Apatosaurus ajax sacrum YPM 1860, was from a dark clay layer that imparted a black color to it. The larger skeleton, which included the Atlantosaurus immauis femur YPM 1840, came from a light colored sandstone immediately over- lying the clay layer and its elements are light col- ored. S. W. Williston, who, as Marsh’s assistant, sorted out the bones of these two specimens in 1883, noted in a memorandum to Marsh their close similarity and believed them to belong to the same species. Included in the shipments from Morrison were cranial materials that have a direct bearing on the controversy about the nature of the skull of Apato- saurus. The second Morrison shipment sent by Lakes and B. F. Mudge in 1877 and assigned acces- sion no. 1002 contained material from quarries 1, 8, and 10. Among this material was the greater part of a cranium on which was originally marked only the accession number. The importance of this specimen was apparently not realized at the time of its re- ceipt, because the box number, which would have indicated from which of the three quarries it came, was not recorded on the cranium. In sorting out the collections from Morrison, Williston assigned the cranium to the ^"Atlantosaurus immanis" speci- men, indicating that he believed it was found in quarry 10. Marsh (1896) later figured the cranium as Atlantosaurus montanus , which would appear to indicate that he believed it to be from quarry 1 . The cranium, as will be shown in a later section, is Dip- lodocus-like in structure and its quarry origin is, therefore, of great importance. Adding to the con- fusion, when the quarry 10 material was catalogued the number YPM 1860 was placed on not only the bones of the holotype of Apatosaurus ajax, but also on those of ""Atlantosaurus immanis." It is not known why, when or by whom this was done. Also from quarry 10 at Morrison was a pair of very large Diplodocus -Wko quadrates. Although the catalogue number YPM 1860 is marked on both, only the left one bears the accession no. 1052 and the box no. 53, which definitely identifies its origin as quarry 10. The quadrates, therefore, provide important evidence on the nature of the skull of Apatosaurus . In the summer of 1879 two of Marsh’s foremost collectors, W. H. Reed and E. G. Ashley, discov- ered the major portions of two very large sauropod skeletons in the same stratum of two adjacent quar- ries at Como Bluff, Wyoming. These were de- scribed by Marsh as two species of a new genus. Brontosaurus . The more perfect skeleton, YPM 1980 from Como Bluff quarry 10 (to date, one of the most complete sauropod skeletons ever found), he described (1879u) as the holotype of the type species Brontosaurus excelsus, whereas the other, YPM 1981, from Como Bluff quarry II, he de- scribed (1881) as the type ofB. ampins. YPM 1980 lacked the skull, first few cervicals, posterior half of the tail, ulna, and all the bones of both the manus and pes except the astragalus; YPM 1981 possessed only one bone not represented in YPM 1980, the second metacarpal. In 1883 Marsh published a res- toration of B. excelsus, the first for any sauropod dinosaur. Though his restoration was quite good overall, it contained numerous errors, most of which depicted ""Brontosaurus" as having Camara- saurus-Uke features. The feet were incorrectly re- stored with a full complement of phalanges and, in Camarasaurus fashion, two proximal carpals and tarsals were attributed to ""Brontosaurus" \ Apato- saurus has only one each of these elements, the astragalus and "scapholunar.” The crushed ulna and manus used in Marsh’s restoration belonged to a partial skeleton of a large adult Camarasaurus, YPM 4633, from YPM Como Bluff quarry lA. De- tailed drawings of these elements, which were pre- pared for Marsh for a proposed sauropod mono- graph, have been reproduced by Ostrom and McIntosh (1966). The narrow, elongated metacar- pals and the slender ulna of YPM 4633 are in sharp contrast to the short, stout metacarpals and the ex- tremely robust ulna of Apatosaurus . When YPM 1980 was mounted at the Yale Peabody Museum the ulna and manus of YPM 4633 were used to com- plete the skeleton. Marsh was unaware of the cor- rect number of cervical and caudal vertebrae of ""Brontosaurus" and restored the neck and tail after ""Morosaurus " that is, Camarasaurus , with too few vertebrae. The neck was shown as having only 12 vertebrae as in Camarasaurus , rather than the cor- rect number of 15. Marsh also did not know that ""Brontosaurus" had a long, “whip-lash” tail, con- taining as many as 82 vertebrae, almost twice the number found in Camarasaurus . Most importantly, for the missing skull of his restoration of YPM 1980 Marsh used a large, incomplete Camarasuarus-fike skull, YPM 1911, from YPM Como Bluff quarry 13, 8 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 8 located about 4 mi from and in a stratum distinctly lower than that of YPM quarry 10 from which came "fi.” excelsus (Ostrom and McIntosh, 1966). Quar- ry 13 had yielded four partial skeletons of Camara- saiirus, including the type of '’‘’Morosaitriis'" len- tus (YPM 1910), and a quadratojugal and caudal centrum of Diplodocus, but no identifiable remains of Apatosaurus . The skull YPM 1911 consists of premaxillae, maxillae, lacrimals, vomers, dentaries, loose teeth, and some fragments; its massive jaws and spatulate teeth are prominent Cainarasaurus features. In 1891 Marsh published a revised recon- struction of "'Brontosaurus, " which was in some ways less accurate than his first (Riggs, 1903u). Al- though he added a thirteenth vertebra to the cer- vical series, he also increased the number of dorsals from the correct 10 to 14, which is closer to the 12 possessed by Cainarasaurus . In his second resto- ration Marsh used a different skull, USNM 5730, from YPM-USNM Canyon City quarry 1 at Garden Park, Colorado. It is about the same size as the skull YPM 191 1 but somewhat more complete, con- sisting of maxillae, premaxillae, squamosal, denta- ries, cranium, and perhaps a quadrate. USNM 5730 also has the distinctive massive jaws and spatulate teeth of Cainarasaurus . The skull was found iso- lated from other skeletal materials in quarry 1, rep- resenting as many as five or six sauropod genera. Although Apatosaurus was one of those genera present, quarry maps indicate no reason to believe that the skull was associated with any remains of this genus. Most, if not all, of the vast collections of sauro- pods made by the American Museum of Natural History from 1897 to 1905 at Bone Cabin quarry, Como Bluff, and nearby localities in southeastern Wyoming, belonged to the three common genera of the Morrison Formation. A partial skeleton of Apato- saurus from Como Bluff was described by Osborn (1898) as Cainarasaurus and, although not explic- itly stated, he strongly implied that Cainarasaurus and ""Brontosaurus" were very closely related, if not synonymous. The greater part of the collections from Bone Cabin quarry consisted of limbs, feet, and tail segments. The feet were sometimes artic- ulated with the limbs, but more often not. For un- known reasons the numerous undersized limbs were separated out as ""Morosaurus," the large ro- bust limbs as ""Brontosaurus" and the large slender limbs as Diplodocus; these assignments caused problems. The robust hindlimb bones of Cainara- saurus are very similar to those of Apatosaurus and some of the large hindlimb elements of the former were assigned to the latter. Also resulting in mis- identifications, the forelimb bones of Cainarasau- rus are slender and the radius and ulna, in partic- ular, resemble those of Diplodocus . Further, it was not known that the metacarpals of Cainarasaurus were much longer and more slender than those of Diplodocus . Important to the discussion here was the misidentification of a right radius, ulna and ma- nus of Cainarasaurus AMNH 965 from Bone Cabin quarry. Osborn (1904) originally described the ma- nus correctly as ""Morosaurus" but apparently re- considered his identification about a year later when he sent a cast of it to the Carnegie Museum in re- sponse to their request for a manus to complete the Diplodocus being mounted there. A reduced model of the manus was not only used in the Carnegie Museum exhibit, but also in 10 casts of the entire skeleton sent to museums throughout the world. Osborn also sent photographs of AMNH 965 to Abel, who not only published them (1910) as the manus of Diplodocus, but also used them in his res- toration of this genus. Forelimbs of Diplodocus and forefeet of Cainarasaurus from Bone Cabin quarry were also mistakenly associated as composite spec- imens of Diplodocus and sent by the American Mu- seum of Natural History to a number of museums throughout the world. In the early 1900s numerous important discov- eries were made that revealed errors in Marsh’s (1883, 1891) restorations of Apatosaurus . Most sig- nificantly, these discoveries not only removed some of the erroneous resemblances between Cainara- saurus and Apatosaurus that were suggested by Marsh's restorations of the latter, but also disclosed some important features shared by Apatosaurus and Diplodocus . Hatcher (19016, 1902) described the forelimb and, more importantly, the forefoot of Apatosaurus correctly, using associated material, CM 563, now mounted at the University of Wyo- ming, Laramie. Hatcher, however, failed to notice that the forefeet of Apatosaurus and Diplodocus are much closer in structure than either is to that of Cainarasaurus . Riggs (19036) not only showed that Brontosaurus is a junior synonym of Apatosaurus, but also demonstrated that Apatosaurus possesses 10 dorsal vertebrae, that the number of sacral ver- tebrae of sauropods is not a valid generic character as Marsh believed, and that the chevrons of the midcaudals of Apatosaurus are Diplodociis-\xV.t in having fore and aft distal processes. Undoubtedly, the most important event with regard to this dis- 1978 BERMAN AND MclNTOSH— APATOSAURUS RELATIONSHIPS 9 cussion here was the discovery in 1909 by Earl Douglass of the Carnegie Museum of the well- known, richly fossiliferous, dinosaur quarry at what is now Dinosaur National Monument, near Jensen, Utah. The first specimen discovered and excavated from this quarry, known then as Carnegie quarry, was important not only in being the most complete Apatosaurus skeleton ever found, but in having a large skull closely associated with it. In 1915 Hol- land (1915^) not only described the postcranial skel- eton, CM 3018, as a new species, A. louisae , but, on the basis of the skull associated with it, he (1915«) also challenged Marsh's (1883, 1891) origi- nal identifications of the skull of ""Brontosaurus The type of A. louisae, which was designated field no. 1, was found (Fig. 1) largely articulated, but with the trunk, neck and forelimb somewhat dis- placed. A second, almost as complete and articu- lated skeleton of this species, field no. 40, lay beside CM 3018 and, although an adult specimen, was 15 to 20% smaller than the type; this specimen is now at the Los Angeles County Museum. Lying beside cervicals 12 and 13 of field no. 40 and about 4 m from the atlas of CM 3018 was a large Diplodocus- like skull without mandibles, CM 11162. Though the posterior portion of a medium-sized Diplodocus skeleton (field no. 60) lay only about 3 m from the skull CM 11162 (Fig. 1), their size difference pre- cludes any possibility that they were associated. Noting the close proximity of the skull CM 11162 to the skeleton CM 3018, their position in the same layer and the exact fit of the occipital condyle of the skull into the articular cup of the atlas of CM 3018, Holland (1915u:274) concluded that the Di- plodocus-like skull represented the true skull of Apatosaurus, stating that “Had nothing in the past been written in reference to the structure of the skull of Brontosaurus the conclusion would natu- rally and almost inevitably have been reached that this skull belongs to the skeleton the remainder of which has been recovered." However, when the skeleton of A. louisae CM 3018 was mounted at the Carnegie Museum, Holland considered using this skull but refrained from doing so apparently at the insistence of Osborn (Holland, 1915«) and, instead, the skeleton stood headless for more than 20 years (Gilmore, 1936). After Holland’s death in 1932 a cast of the Camarasaurus skull CM 12020 was used to complete the mount. This large, incomplete skull was collected at Carnegie quarry as part of field no. 240, which included the greater part of an adult Camarasaurus skeleton, and originally both the skull and postcranial skeleton received the same catalogue number, CM 11393. There is no reason to believe that the skull and postcranial skeleton did not belong to the same individual and, further, no Apatosaurus material was found nearby to suggest that the skull might pertain to this genus. It is also important to mention here that Holland (1915u) de- scribed a second feature of Apatosaurus that fur- ther helps to substantiate its closeness in structure to Diplodocus , the presence of a "whip-lash” type of tail. This structure was clearly documented not only in A. louisae CM 3018, but in a medium-sized specimen, CM 3378, found isolated at the far west- ern end of Carnegie quarry and consisting of a ver- tebral column complete and articulated from the mid-cervical region to the eighty-second caudal. Three other specimens found closely associated at Carnegie quarry (Fig. 1) are pertinent to this dis- cussion. A partial skeleton, field no. 24, of a small, juvenile Apatosaurus, CM 3390, was found lying near the cervicals of A. louisae CM 3018. CM 3390 consists of the complete dorsal vertebral series, sacrum, caudals 1-12, left pelvic bones, right is- chium, and a few ribs. Earl Douglass, who directed the Carnegie quarry excavation, estimated its total length to be about 5 m. Most interesting, however, in the records of the collection from the quarry Douglass states that "About 20 feet east of here [field no. 24], ten or more connected cervicals of a small dinosaur (field no. 37) were found, also the anterior portion of a small jaw with pencil-like teeth (field no. 35). I worked out nos. 24 and 37 later when in the museum in Pittsburg and this confirmed the surmise that these belonged to the same indi- vidual.” If these three specimens belonged to the same individual, then the Diplodocus -like teeth of the jaw provides additional evidence on the nature of the skull of Apatosaurus . Carnegie quarry has also been important in yield- ing excellent skulls of Diplodocus; two of these are relied on heavily in redescribing the skull of Di- plodocus in a later section. The complete and un- crushed skull and mandible CM 11161 was discov- ered beside the anterior caudals of the nearly complete vertebra! column of the medium-sized Apatosaurus CM 3378 found isolated at the far western end of the quarry. Earl Douglass viewed this association as evidence that Apatosaurus pos- sessed a Diplodocus-Wkt skull (McIntosh and Ber- man, 1975). The skull, however, was the basis of Holland’s (1924) description of the skull of Diplod- ocus. The palate and lower jaw of CM 11161 were 10 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 8 Fig. 1. Relative quarry positions of various sauropod specimens discussed in text from Dinosaur National Monument. Redrawn from quarry map on file at Carnegie Museum of Natural History showing specimens removed from quarry by that institution. Field nos. 1 and 40, postcranial skeletons of type of Apatosaurus louisae, CM 3018, and specimen now at the Los Angeles County Museum; CM 11162, skull very probably belonging to field no. 1 or 40; field no. 60, postcranial skeleton of Diplodociis: field no. 24, portion of postcranial skeleton of a }u\eni\e Apatosaurus (CM 3390), field no. 37, series of cervical vertebrae, and field no. 35, anterior portion of small jaw with Diplodocus-like teeth, probably belonging to one individual of Apato- 1978 BERMAN AND McINTOSH— APAr05At/7?D5 RELATIONSHIPS 11 recently redescribed (McIntosh and Berman, 1975). CM 3452, consisting of skull, mandible, and the first six cervicals in articulation, is a very important specimen in that it is the only instance in which a Diplodocus skull has been found articulated with postcranial elements. This specimen was sketchily illustrated and briefly referred to a few times by Holland ( 1924) in his description of the skull of Di- plodocus. Some of its disarticulated palatal bones were recently described (McIntosh and Berman, 1975). Discussion Numerous factors can be attributed to the origin of the false notions that Apatosaurus possessed a Camarasaurus -Wkt skull and was more closely re- lated to Camarasaurus than to Diplodocus . Not least among these is that the first descriptions of Apatosaurus Marsh (\%llb) and Camarasaurus Cope (1877a) were based on only very small por- tions of the type skeletons, were very brief, and were without illustrations. The reason for this was the Cope-Marsh feud at that time (Romer, 1964). In their zeal to be first in describing the large sauro- pods of North America, Marsh and Cope rushed out descriptions on the first few elements of the type skeletons they received from their collectors, even though the greater portions of the skeletons were still being excavated. During the next year and a half, as more material was collected and prepared, these descriptions were only slightly amplified (Cope, 1877c, 1878a; Marsh, 1879^), including fig- ures of a few bones of both genera, but neither ge- nus received the description it merited. Thereafter, both genera were largely ignored. Marsh’s (1878^, 1879a) descriptions of two new sauropods, Moro- saurus and Brontosaurus, further complicated the picture. Both genera were based on good material and were described in detail with many excellent illustrations. However, of the half dozen or more partial skeletons Marsh had identified as Morosau- rus, all were juveniles or subadults and were con- siderably smaller than the two large skeletons of Brontosaurus he had. As a result, the few adults Camarasaurus specimens he had were apparently misidentified as Brontosaurus because of their large size. Had he realized that '"Monosaurus" attained the same size as ""Brontosaurus," he might not have used the large Camarasaurus-\\k.Q skulls YPM 191 1 and USNM 5730, and the ulna and manus of the partial skeleton of the adult Camarasaurus YPM 4633 in his (1883, 1891) restorations of Bron- tosaurus . Although YPM 191 1 and USNM 5730 rep- resent individuals much larger than any of the spec- imens in Marsh’s collection, which he recognized as Morosaurus, they are not too large when com- pared with the type skeleton of C. supremus de- scribed by Cope ( 1877a), for which he did not report any skull parts (portions of the skull, including up- per and lower jaws with teeth, were later described by Osborn and Mook, 1921). It can also be pointed out that Marsh never indicated his use of referred specimens, seemingly selected on purely conjectur- al grounds, to complete his restorations of Bronto- saurus . This practice undoubtedly helped to per- petuate many of the misconceptions about the structure of Apatosaurus . Adding to the confusion, in 1898 Osborn described a ""Brontosaurus" skel- eton as Camarasaurus, apparently believing the two genera to be synonymous. Even after a thor- ough study of the type of Camarasaurus by Osborn and Mook (1921) showed it to be a senior synonym of Morosaurus , the erroneous concept of a close relationship between Camarasaurus and Apatosau- rus persisted. This was due largely to 1) the rec- ognition that the skeletons of Apatosaurus and Camarasaurus are very robustly constructed and their hindlimbs are nearly indistinguishable, where- as the skeleton of Diplodocus is very slender in structure and its hindlimbs are easily identified and 2) the continued acceptance of the false notion that Apatosaurus possessed diCamarasaurus-\\kQ skull. It is remarkable that Holland’s claim that Apato- saurus possessed a Diplodocus-Wke. skull continued to fail to receive serious consideration even after accurate restorations of the postcranial skeletons of Apatosaurus (Gilmore, 1936), Camarasaurus (Gil- more, 1925) and Diplodocus (Hatcher, 1901a , 1903/?; Holland, 1906) became available. Despite the fact that even a cursory comparison of their postcranial skeletons, excepting their hindlimbs, shows that Apatosaurus is not only quite distinct from Camarasaurus, but shares a great number of significant features with Diplodocus, such obser- vations have, to date, not been made. Examination of the type skeletons of Apatosau- rus ajax YPM 1860 and Atlantosaurus immanis YPM 1840, the only postcranial specimens from YPM quarry 10 at Morrison, Colorado, substanti- ates Williston’s observation that they belong to the same species and the latter is considered to be a junior synonym of A. ajax. The pair of Diplodocus- like quadrates from YPM quarry 10 are identical in size, color and morphological detail, leaving almost 12 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 8 no doubt they belong to the same individual. Fur- ther, their black color suggests that they belong to the identically colored skeleton YPM 1860, rather than to the light colored skeleton YPM 1840. The greater portion of cranium contained in the ship- ment of specimens Marsh received from Lakes and Mudge in 1877 from the Morrison quarries 1, 8, and 10 is also Diplodocus-Wko. and is also thought to belong to YPM 1860. If the cranium is from quarry 10, as Williston must have thought in assigning it to Atlantosaurus immanis YPM 1840, then it probably belongs to YPM 1860, because both have the same black coloring. It is also important to point out that the quadrates and the partial cranium are of the appropriate sizes to have belonged not only to the same animal, but to a skeleton the size of YPM 1860 or YPM 1840. The other possible, but far less likely, origin of the cranium is YPM quarry 1, as Marsh (1896) apparently thought when describing it as At- lantosaurus montanus . In addition to the type sac- rum YPM 1835 of A. montanus, YPM quarry 1 has yielded Camarasaurus-\\kt vertebrae. Examination of the sacrum YPM 1835 reveals that it is too frag- mentary to permit generic identification and it could conceivably belong to either Apatosaurus, Diplod- ocus, or Camarasaurus, which have been found at the 11 quarries at Morrison. It is not unlikely, how- ever, that either Williston or Marsh may have ob- tained more precise locality information for the cra- nium from Lakes or Mudge well after it arrived at the Yale Peabody Museum and that this was never recorded in the catalogues. The cranium now bears the catalogue number YPM 1860, but we do not know when, by whom, or on what basis it was given this number. It is possible that this was done at the same time that this number was placed on the bones of both the type of Apatosaurus ajax and ""Atlan- tosaurus immanis" (=A. ajax) YPM 1840. White (1958), on the basis of the catalogue number YPM 1860 on the cranium, quite reasonably assumed that it was part of the type of A. ajax and that Marsh (1896) had a lapsus calami in describing it as Altan- tosaurus montanus . Further, White considered the cranium to closely resemble that of Camarasaurus and, therefore, to provide evidence of a close re- lationship between Apatosaurus and Camarasau- rus. It is surprising that White appears not to have been aware of the pair of quadrates from YPM quar- ry 10; had he examined them, he surely would have immediately recognized their D/p/oc/ocw5-like struc- ture and so might have noticed the Diplodocus -like nature of the cranium. White also mentioned that parts of both pterygoids, which embrace the re- cesses basipterygoideus, are present with the cra- nium. We found one of these elements but were unable to either confirm or give an alternative to his identification. The probable Apatosaurus skull CM 11162 has never been described and only a small portion of it has been illustrated. In a discussion of tooth re- placement in Diplodocus, Holland (1924:Fig. 5) il- lustrated a small part of the anterior end of the right maxilla, where, due to the loss of surface bone, the replacement pattern is clearly seen. Additional preparation of CM 11162 has revealed considerable plaster restoration. Further, the Carnegie Museum of Natural History collections include a plaster cast of the skull in its restored state and it is suspected that the restoration and cast were done at the re- quest of Holland, who (1915u:277) stated that at times he was inclined to mount it on the postcranial skeleton of A. louisae CM 3018 on exhibit at Car- negie Museum. After a thorough review of all the evidence we (McIntosh and Berman, 1975) con- curred with Holland that the skull CM 11162 prob- ably YQpxtsenis Apatosaurus , if not the type of A. louisae. SKULL OF DIPLODOCUS The skull here described as probably belonging to Apatosaurus , CM 11162, is so close to that of Diplodocus in structure that the problem of distin- guishing between them is difficult. In light of this problem, comparisons using previous descriptions of skulls of Diplodocus are made somewhat ten- uous, because these skulls were found isolated, leaving some doubt as to their identification. Most importantly, previous descriptions of the skull of Diplodocus are incomplete and contain numerous inaccuracies. The Diplodocus skull CM 3452 from Dinosaur National Monument is, therefore, empha- sized here, because it represents the only known direct association of skull and postcranial skeleton of this genus. In Holland’s (1924:P1. XL, fig. 2) il- lustration of this specimen only the damaged and partially disarticulated right side of the skull is shown. Further preparation reveals that the left 1978 BERMAN AND MclNTOSH— APATOSAURUS RELATIONSHIPS 13 side, occiput, and roof of the skull are nearly per- fectly preserved. The essentially complete and un- crushed skull CM 11161 is here considered to rep- resent Diplodocus even though it was found closely associated with the nearly complete vertebral col- umn of Apatosaurus CM 3378 isolated at the far western end of Dinosaur National Monument quar- ry. This assignment is justified by the common pos- session of characters of CM 1 1161 and the unques- tionable Diplodocus skull CM 3452 that are not seen in the prohdih\t Apatosaurus skull CM 11162. An excellently preserved cranium, CM 26552 from Di- nosaur National Monument, on which our descrip- tion of the braincase of Diplodocus mainly rests, is assigned to this genus on the same grounds; CM 26552 has not been previously described. Similarly, the posterior portions of the skulls CM 662 used by Holland (1906) and AMNH 694 by Osborn (1912) to describe the braincase of Diplodocus exhibit a clos- er resemblance to CM 3452 than to CM 1 1 162. The skulls USNM 2672 and USNM 2673 used by Marsh (1884, 1896) and AMNH 969 used in part by Hol- land (1906, 1924) in reconstructions of the skull of Diplodocus are also tentatively accepted as belong- ing to this genus. External Features The shape and proportions of the skull of Diplod- ocus have been accurately reconstructed by Marsh (1884, 1896; see also Ostrom and McIntosh, 1966) and Holland (1906, 1924); in some details these as- pects of Marsh’s reconstructions are more exact. One apparent error in Marsh’s restorations of the shape of the skull, however, should be noted. In his dorsal views of the skull the steeply pitched, lateral surface below the larger, posterior, antorbital open- ing is incorrectly shown as being broadly bowed laterally, rather than flat. Whereas Marsh’s resto- rations omit some of the sutures and inaccurately show the courses of a few, Holland’s not only in- correctly show the extent of many of the bones, but erroneously depict the presence of others, such as a supraorbital and postfrontal. Most of the errors in the literature that pertain to the external features of the skull of Diplodocus concern the sutural pat- tern of the posterior half of the skull; the description that follows is mainly intended to resolve this con- fusion. Most of this information is readily visible in Figs. 2 and 3. In dorsal view the cranial roof is dominated by the broad, flat frontals. They contact the fused pa- rietals posteriorly in a nearly straight, transverse suture that extends laterally to nearly the upper end of the supratemporal fossa; at this point the frontal- parietal contact continues a short distance as it turns abruptly anteriorly to skirt the upper end of the fossa. Lateral to the frontal-parietal suture the frontal is drawn outward into a transversely orient- ed, nearly vertical wing that extends ventrally to contact on its posterior surface a dorsal, medially expanded process of the postorbital; the plane of their contact is oriented obliquely anteroventrally in sagittal section. The anterior surface of the lateral wing of the frontal forms the posterodorsal portion of the orbital border and wall. The posterior surface of the frontal wing is extensively overlapped by the postorbital and their surface line of contact extends outward and downward along the anterodorsal mar- gin of the supratemporal fossa so that the frontal makes little or no contribution to the fossa wall. Seen from above the frontal portion of the orbital rim is deeply concave. The nasal-frontal suture is sinuous and extends laterally, meeting the prefron- tal a short distance posterior to its medialmost level of projection. The anterolateral corner of the frontal is deeply incised by the narrowly triangular, pos- teromedially directed, posterior half of the prefron- tal. The fused parietals are narrowly exposed on the skull roof, where they taper somewhat as they ex- tend toward the supratemporal fossa. A vertically oriented, lateral wing of the parietal, forming the posterior wall of the supratemporal fossa, has an extensive occipital exposure. The lateral wing of the parietal is triangular in cross-section and thins toward its outer edge, which in occipital view is greatly expanded dorsolaterally into a smooth, broadly convex border. In CM 3452 the intersection of the median union of the frontals with the fused parietals is well pre- served and, as Holland ( 1924) pointed out, a parietal opening is absent. Holland also noted the absence of such an opening in the Diplodocus skull CM 662, but claimed that a medial opening was present in the parietal region of the skull roof of the probable Diplodocus skulls CM 11161, AMNH 969, USNM 2672 and USNM 2673, and the probable Apu/osc/a- rus skull CM 11162. We can find no evidence that this opening existed in these specimens. Though in none of the specimens illustrated here can the pres- ence of a midline suture of the nasals be verified, it is assumed that Marsh (1884, 1896) and Holland (1906, 1924) were correct in describing the nasals as paired. The paired nasals form the posterior mar- gin of the narial opening; each is slightly concave 14 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 8 10 cm 10 cm Fig. 2. — Skulls of Diplodocus . A, CM 3452, and B, CM 1 1 161. along this border, so that together they project somewhat forward into the narial opening along the midline. Laterally the nasal continues anteroven- trally as a narrow, strip-like process that contacts the anterior half of the medial border of the pre- frontal while bounding the posterolateral border of the narial opening. The anterior end of this process in CM 11161 is strongly beveled anteroventrally to form a short, sharply pointed lappet of bone that overlaps the upper end of the ascending process of the maxilla and contacts the anterior margin of the upper end of the lacrimal. The anteroventral pro- 1978 BERMAN AND MclNTOSH— APATOSAURUS RELATIONSHIPS 15 Fig. 3. — A, cranial roof, and C, occipital views of Diplodocus skull CM 3452. B, occipital view of Diplodonis braincase CM 26552. D cranial roof, and E, occipital views of probdib\e Apatosaurus skull CM 11162. cess of the left nasal of CM 3452 appears to differ from that of CM 11161 by terminating in a border that is normal to its long axis. The prefrontal enters the anterodorsal margin of the orbit and in lateral view its sharply pointed an- terior end wedges between the nasal and lacrimal. In CM 11161 the prefrontal-lacrimal suture extends directly posteromedially across the dorsal wall of 16 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 8 Fig. 4, A, ventrolateral, and B, anterolateral stereo views of Diplodocits braincase CM 26552. Structural features indicated in Fig. 6. 1978 BERMAN AND McINTOSH— APArOSAf/7?t/5 RELATIONSHIPS 17 Fig. 5. — Anterior stereo view of Diplodocus braincase CM 26552. Structural features indicated in Fig. 6. the orbit, whereas in CM 3452 there is a sharply angular jog in this suture. The jugal makes a small contribution to the anterior end of the ventral mar- gin of the orbit. From the anteroventral corner of the orbit the jugal-lacrimal suture, best preserved in CM 11161, runs a short distance anteriorly, then swings abruptly dorsally, forming the posterior bor- der of a narrow, dorsal process of the jugal, as it extends for a somewhat longer distance to reach the antorbital opening. The bluntly rounded, distal end of the dorsal process of the jugal, incomplete in the skulls CM 1 1 161 and CM 3452 but well illustrated in the restorations by Marsh (1884, 1896; see also Ostrom and McIntosh, 1966) and Holland (1906), projects a short distance dorsally and slightly an- teriorly into the posterior corner of the antorbital opening. The dorsal process of the jugal and a broadly convex expansion of the ventral margin of the ascending process of the maxilla opposite this process greatly constrict the antorbital opening. Su- tural contacts of the jugal with the maxilla and quadratojugal are accurately depicted in the resto- rations by Marsh and Holland and are well pre- served in CM 3452. The postorbital is basically triradiate in shape. A thick, medially expanded, dorsal blade forms the anterior wall of the supratemporal fossa and con- tacts the parietal in a nearly vertical suture on the medial wall of the fossa. In lateral view the post- orbital-frontal suture extends obliquely across the posterolateral edge of the orbit at about its mid- height, then turns sharply ventromedially across the posterior wall of the orbit (Figs. 4-6). Thus the postorbital forms the ventral half of the posterior orbital wall. A short, broadly triangular, posterior process of the postorbital overlaps the squamosal. A third, greatly attenuated, anterior process of the postorbital bounds almost the entire ventral margin of the orbit; its suture with the jugal is clearly pre- served in CM 3452 and its essentially vertical ori- entation is in marked contrast with the nearly hor- izontal contact depicted in previous accounts. In CM 11161 this suture is not well defined and the anterior process of the postorbital appears to have a wedge-shaped, overlapping contact with the jugal. The squamosal can also be described as consisting of basically three processes, all emanating from the posteroventral corner of the lateral side of the skull. A narrow, dorsomedially directed, occipital process of the squamosal is described in detail below. The narrow, ventral border of the supratemporal fossa is bounded by an anterodorsal process of the squa- mosal that is interposed between the distal end of the lateral wing of the parietal and the superior bor- der of the posterior process of the postorbital. A third, tongue-like process is directed anteroventral- ly along the lateral margin of the proximal end of the quadrate. The proximal head of the quadrate, which fits into a shallow concavity on the ventral surface of the squamosal, is narrowly exposed by 18 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 8 1978 BERMAN AND McINTOSH— APAr05Af//?t/5 RELATIONSHIPS 19 a concave notch at the posterior end of the ventral margin of the quadrate process of the squamosal. The upper half of the anterior surface of the flared, distal end of the paroccipita! process abuts against a narrow, slightly concave recess on the postero- medial edge of the squamosal below its occipital process and also buttresses the head of the quad- rate; the lower half of the anterior margin of the flared end of the paroccipita! process is free of con- tact. The posterior end of the quadrat ojugal has an extensive, overlapping contact with the lateral sur- face of the distal end of the quadrate. The ventral margin of the quadratojugal curves very slightly ventrally as it crosses the quadrate only a short dis- tance above its distal end; the quadratojugal then turns dorsally for a considerable length, where it terminates by smoothly tapering toward the poste- rior margin of the quadrate. Although the quadra- tojugal and the squamosal closely approach each other along the lateral margin of the quadrate, they do not meet. Braincase The first detailed description of the braincase of Diplodocus was given by Holland (1906) and was based on a well-preserved posterior portion of skull, CM 662 (now at Houston Museum of Natural Sci- ence). A critique of this paper was published by Hay (1908), who challenged many of Holland’s de- terminations of the bones and foramina. In partic- ular, he pointed out that the supraoccipital occupies a much larger area than that assigned to it by Hol- land, and that Holland’s failure to recognize the presence of the prootic was the basis of many of his errors. Both interpretations, however, contain a great number of significant errors. Osborn (1912) presented an illustration of the braincase of AMNH 694 in sagittal section, which accurately depicts the internal positions of most of the cranial foramina. However, in this figure and in one other, showing the same specimen in frontal view, he not only omitted the sutures of almost all of the bones, but also incorrectly indicated the extent of others. Von Huene ( 1914) gave a rough sketch of the lateral as- pect of the braincase of the same specimen used in Holland’s (1906) study. Though most of the foram- ina are indicated and correctly identified, his essen- tially diagrammatic illustration does not show some of the cranial sutures. In Marsh’s restorations of the skull of Diplodocus (1884, 1896; see also Ostrom and McIntosh, 1966) no attempt was made to de- note the bones of the occiput. Finally, in a recon- struction by Marsh of the skull in midsagittal sec- tion, published for the first time by us (McIntosh and Berman, 1975), the braincase is depicted in only a general way. Figs. 3-6 may largely take the place of a detailed description of the braincase. The occiput (Fig. 3), whose shape and general features are best preserved in CM 3452, is of typical sauropod form. The occiput is subrectangular in outline and is formed by the basioccipital, exoccip- itals, supraoccipital, fused parietals, and a small process of the squamosal. In none of the specimens (CM 26552, CM 11161, and CM 3452) has it been possible to trace the suture between the exoccipital and the opisthotic; thus these two elements became completely fused early. It is assumed, however, that a large but undetermined portion of this com- pound element exposed on the occiput and the part which encloses the two openings for cranial nerve XII at the extreme posteroventral corner of the lat- eral wall of the braincase represent the greater part of the surface extent of the original exoccipital. The original opisthotic, bounded by the exoccipital pos- teriorly and the prootic anteriorly, undoubtedly in- cluded the narrow span between the foramen for cranial nerves IX-XI and the fenestra ovalis and then presumably extended dorsolaterally as a flat- tened process that covered the anterior surface of the exoccipital in the paroccipital process. The ex- occipital bounds the lateral wall of the foramen magnum, is separated from its mate in the roof of the foramen magnum by the supraoccipital and in the floor by the basioccipital, and makes a small contribution to the dorsolateral surface of the oc- cipital condyle and its articular surface. Only in CM 26552 is the supraoccipital-exoccipital suture clear- ly preserved. The exoccipital extends directly out- ward from the foramen magnum, making contact with the supraoccipital above in a broadly undulat- ing suture that laterally reaches the dorsal margin of the small posttemporal fossa. At the level of the upper end of the posttemporal fossa the exoccipital- oposthotic complex bends downward to form a broad paroccipital process, constricting in breadth somewhat as it bounds the medial side of the post- temporal fossa, then expanding again distally where it contacts the squamosal and quadrate. A hook-like process at the ventrolateral corner of the supraoc- cipita! forms the dorsolateral margin of the posttem- poral fossa. The remainder of the lateral border of the fossa is completed by a narrow, occipital pro- cess of the squamosal. It enters the occiput by ex- tending medially between the ventrolateral corner 20 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 8 of the lateral, occipital wing of the parietal and the dorsal margin of the expanded end of the paroccip- ital process, then turns upward along the ventro- medial side of the parietal to meet the hook-like process of the supraoccipital. From this point the occipital process of the squamosal continues a very short distance, still in contact with the parietal but in front of the hook-like process of the supraoccip- ital; the end of the occipital process of the squa- mosal contacts a small area on the superior edge of the anterolateral face of the prootic (Fig. 3). In ventrolateral view the exoccipital pedicle is triangular, widening toward its ventral contact with the basioccipital where it encloses three foramina and bounds the posterior wall of a fourth. The most posterior of these openings is oval and carried the posterior branch of cranial nerve XII. Immediately anterior to this is a very small, round opening pre- sumably for an anterior branch of the same nerve. The largest and most anterior foramen enclosed by the opisthotic is elongate (about 17 mm high and about 4 mm wide), is inclined posterodorsally, and faces somewhat posteroventrally; often referred to as the jugular foramen, it presumably transmitted nerves IX-XI and probably the jugular vein. An- terior to the jugular foramen the narrow fenestra ovalis opens on the boundary between the opis- thotic and prootic. The prootic, an extensive ele- ment, is surrounded by the opisthotic and supraoc- cipital behind, the basioccipital and basisphenoid below, the laterosphenoid in front, and the parietal above. The prootic is tightly sutured to all of these elements except the laterosphenoid and parietal, with which it has abutment contacts. Exposed mainly as a broad, flat plate that faces anterolater- ally, the prootic extends only slightly onto the an- terior surface of the proximal end of the paroccipital process. Below the level of its contribution to the paroccipital process the prootic is deeply exposed posteriorly, forming a narrow, laterally projecting lamina of bone that extends downward to the ven- tralmost point of contact of this bone with the ba- sisphenoid. The lateral edge of the lamina is deeply emarginated into a smooth, broad, concave arc; the lamina forms the body of the crista prootica. The lower portion of the prootic encloses one foramen and impinges on two others. Posteriorly it forms the anterior border of the fenestra ovalis. From the dor- sal border of this opening two shallow, parallel grooves of nearly equal dimensions, one on either side of the probtic-opisthotic suture, extend pos- terolaterally to about where the suture between them turns abruptly upward across the anterior face of the proximal end of the paroccipital process. Directly anterior to the fenestra ovalis the posterior face of the crista prootica is perforated by a small round foramen for the VII nerve. A large, subcir- cular exit for nerve V, measuring about 8 mm in diameter, is positioned on the boundary between the probtic and laterosphenoid and at the same level as the facial nerve opening. The laterosphenoid is a narrow, wing-like struc- ture that is principally exposed as a flat, anterola- terally facing surface. It is strongly sutured to the basisphenoid below, but has an abutment contact with the probtic behind, the orbitosphenoid in front and the postorbital above. A short distance above its narrow contact with the basisphenoid the latero- sphenoid forms the anterior margin of the trigeminal foramen and the posterior margin of the oculomotor foramen; the latter foramen is about 1 1 mm high and about 4 mm wide. A smooth, concave notch in the laterosphenoid margin of the trigeminal foramen probably allowed the forward passage of the ophthalmic branch of nerve V. Extending down- ward from the ventral border of the trigeminal fo- ramen along the lateropehnsoid-orbitosphenoid contact is a deep channel that probably carried the maxillary and mandibular branches of this nerve. An ovate foramen for the trochlear nerve, measur- ing approximately 7 mm high and 4 mm wide, opens on the laterosphenoid-orbitosphenoid suture dorsal to the oculomotor foramen. As the laterosphenoid extends above the level of the trigeminal foramen it expands outward to form a thick, laterally arching lamina of bone, the crista antotica. The convex dor- sal edge of the crista antotica fits into a shallow, concave channel on the ventromedial edge of that part of the postorbital forming the anterior wall of the supratemporal fossa. In anterior view of the braincase the lateral wing of the frontal, which forms the posterodorsal wall of the orbit, nearly hides from view the laterosphenoid-postorbital con- tact. The lateral wing of the frontal tapers to a thin edge toward the ventromedial margin of the postero- dorsal wall of the orbit and does not make sub- stantial contact with the laterosphenoid. The orbi- tosphenoid, which forms the anteriormost eomponent of the lateral wall of the braincase, may possibly include some portion of the presphenoid. Its posterior portion, which forms the anterior mar- gins of the openings for nerves III and IV, is in the form of a stout vertical pillar whose expanded ends have a digitating suture with the frontal above and 1978 BERMAN AND MclNTOSU— APATOSAURUS RELATIONSHIPS 21 the fused basisphenoid-parasphenoid below. A short, thick process projects posteriorly from about mid-length along its posterior margin to separate the oculomotor and trochlear foramina. Immediately anterior to the oculomotor foramen and close to the midsagittal plane is a large, anterolaterally directed opening for the optic nerve, measuring about 8 mm high and 5 mm wide. At the level of the optic fo- ramen the orbitosphenoid extends forward a short distance as it converges on the midline to unite with its mate. Dorsally the united orbitosphenoids form the ventral borders of the very large, nearly co- alesced canals for the olfactory tracts, whereas ven- trally they extend in front of and a short distance below the optic foramina to contact the basipara- sphenoid complex; the portion of the paired orbi- tosphenoids forming the anteromedial borders of the optic foramina is nearly missing in CM 26552, but is well preserved in CM 11161 (not shown here). Small, blunt processes on the orbitosphenoid mar- gins of the olfactory canals, one on either side of the midline suture, form a small cleft through which probably passed the anterior cerebral artery, a branch of the internal carotid artery. A short dis- tance anterior to the orbitosphenoid in CM 11161 are fragments of a thin vertical plate of bone (not shown here) oriented on the midsagittal plane of the skull, which may represent remnants of the pre- sphenoid portion of the interorbital septum. The basioccipital appears to form the greater part of the articular surface of the condyle. The condyle is convex posteriorly and ventrally, and flattened dorsally. The long axis of the condyle is oriented at about a right angle to a plane passing through the jaw margins, indicating that the head was tilted at about a right angle to the neck. Between the con- dyle and the basal tubercles the inferior face of the basioccipital arches anterodorsally, then curves smoothly downward and slightly backward to form the caudal halves of the basal tubercles. The tuber- cles diverge slightly ventrolaterally and their pos- terior surfaces are separated by a deep furrow. Forming the cranial floor anterior to the basioccip- ital is the basisphenoid; it is completely fused with the parasphenoid, which is principally represented by the parasphenoidal rostrum. The parasphenoidal rostrum is broken off at its base in CM 26552 but is well exhibited in CM 11161 (Fig. 8). The basi- sphenoid forms the anterior halves of the basal tu- bercles, the long, slender, anterolaterally directed basipterygoid processes (broken off at their bases in CM 26552) and the inferior margin of the crista prodtica. There is a deep, smooth spheroidal depression between the bases of the basipterygoid processes; this depression is bounded anterolater- ally by a narrow, ridge-like projection along the an- teroventral surface of the proximal third of the ba- sipterygoid process which merges with the ventral edge of the narrow, blade-like parasphenoidal ros- trum. Above the basipterygoid processes the lateral walls of the basisphenoid converge anteromedially to become smoothly continuous with the rostrum. The entrance for the internal carotid artery and the palatine branch of the facial nerve into the basicra- nium, via the vidian canal, is in its normal location on the ventrolateral surface of the basisphenoid be- tween the inferior border of the crista prootica and the base of the pterygoid process. The vidian canal begins at the upper end of an approximately 5-mm- wide groove that extends a short distance antero- ventrally onto the base of the basipterygoid pro- cess. A faint groove, extending ventrally from the lower rim of the facial foramen to a smooth notch in the dorsal edge of the vidian canal (Fig. 4), pre- sumably traces the course of the palatine branch of the facial nerve. Directly below the foramina for cranial nerves 111 and V is a small, round opening for the exit of the palatine branch of the carotid artery, the palatine branch of the facial nerve and probably the abducens nerve. On the midline of the basiparasphenoid complex and immediately above the dorsal edge of the adjoining rostrum is a narrow, vertical opening that probably transmitted the paired, median palatine branches of the carotid ar- teries. PROBABLE SKULL OF APATOSAURUS Description of Skull CM 11162 The large skull CM 11162 that was closely asso- ciated with the postcranial skeletons of Apatosau- rus louisae field no. 1 (type, CM 3018) dind Apato- saurus field no. 40 at Carnegie quarry (Fig. 1) and presumed to belong to one of these specimens, con- forms closely to the skull of Diplodocus , despite some postmortem distortion. The skull is missing the lower jaw and has been variably crushed dor- soventrally (Fig. 7); though the right side of the 22 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 8 Fig. 7. — A, right lateral, B, dorsal, C, left lateral, and D, ventral views of probabXt Apatosaurus skull CM 11162. 1978 BERMAN AND MclNTOSH— APATOSAURUS RELATIONSHIPS 23 skull has retained, for the most part, its proper pitch, the left side has undergone flattening, giving it a broader appearance, especially in the snout re- gion. Distortion and incomplete preservation make impossible determination of the exact outlines of all the major skull openings except for the left supra- temporal fossa. However, there is no structural evi- dence that they differed in any important way from those in Diplodocus; this is certainly true of the supratemporal fossa. Skull dimensions of CM 1 1 162 are given in Table 1 . In lateral view the angles sub- tended between the meeting of the projections of the occipital plane with a plane passing through the ventral margins of the maxillae, the occipital plane with the cranial roof, and the cranial roof with the dorsal margin of the snout are about 75, 120, and 140 degrees, respectively. A restoration of CM 1 1 162 in lateral view is given in Fig. 8. The incompletely preserved external dermal bones of CM 1 1 162 do not differ greatly from those of Diplodocus . The premaxillae are well preserved and show that each possessed four or five functional teeth, represented by their bases. The maxillae are fairly well represented except for two important structures. The upper ends of their ascending pro- cesses have been lost, so their sutural relationships with the nasals and lacrimals are indeterminate. Also, it is not certain if the smaller, more anterior of the two antorbital openings of the maxilla that is characteristic of Diplodocus is present. The poste- rior rim of the preantorbital opening of the left max- illa is hesitantly identified in Fig. 7C; its position would approximate that in Diplodocus. A small, isolated fragment of bone lies in what would be the position of this opening. A tooth count is possible for only the right maxilla, where it is based on re- placement teeth, or in most instances their impres- sions, which have been exposed by the loss of sur- face bone of the maxilla; the right maxilla may have held as many as 12 or 13 functional teeth. The teeth of CM 11162 are identical to the very slender, cy- lindrical teeth of Diplodocus. The jugal and quad- ratojugal, best preserved on the right side of the skull, show their boundaries with each other and the maxilla. The distal end of the dorsal process of the right jugal projects into the antorbital opening as in Diplodocus . Remnants of the lacrimals re- main. The contacts of the left postorbital are clear except for that with the parietal, which in Diplod- ocus extends vertically down the innermost level of the supratemporal fossa wall. The right squamosal is fragmentary; the left is nearly complete except Table I. — Measurements (in mm) of skulls assigned here to Di- plodocus Apatosaurus. I , skull length, measured from snout tip to posterior margin of occipital condyle: 2, skull width, mea- sured at ventralmost level of quadrates; 3. greatest length of quadrate, measured through shaft; 4, skull length to quadrate length ratio; 5, length to distal width ratio of quadrate. Taxa and catalog numbers Measurements 1 2 3 4 5 Diplodocus CM 11161 515 178 185 .36 .20 CM 3452 440 — 167 .38 .19 USNM 2672 550 190 183 .33 .22 USNM 2673 600 — 215 .36 .22 Apatosaurus CM 11162 650 280 185 .28 .30 for the loss of the distal end of its anteroventrally directed quadrate process and has suffered little distortion, exhibiting the same basic relationships with is bordering elements as in Diplodocus . All that remains of the cranial roof (Fig. 3D) is the greater part of the left side. The left parietal is nearly complete, missing only a small portion of its medial boundary; the other sutural boundaries of the left parietal are distinct except for the dorso- medial end of its contact with the supraoccipital on the occiput and its contact with the postorbital. Approximately a fourth of the left frontal is lost along its medial border. The preserved portion of its posterior suture with the parietal and postorbital is distinct and rather dentate for about the medial half of its length; the orbital margin of the frontal exhibits the same concave emargination seen in Diplodocus. Also as m Diplodocus, the triangular, posterior half of the prefrontal penetrates deeply posteromedially into the anterolateral corner of the frontal. The posterior two thirds of the prefrontal projection into the frontal is well defined, but its anteromedial contact with the frontal and nasal is hesitantly traced. The anterior portion of the pre- frontal is absent. Only a short, narrow strip of the nasal is preserved along the medial border of the prefrontal; its contact with the frontal is not clear. The somewhat abraded occiput (Fig. 3E) is nearly intact on the left side, whereas the greater part of the right side is absent. Only the supraoccipital-ex- occipital suture of the occiput cannot be found. The median, nuchal crest on the supraoccipital above the foramen magnum is strongly developed. The small, left posttemporal fossa is clearly visible and is identical to that of Diplodocus in outline and in the way its borders are formed. The articular sur- face of the occipital condyle is hemispherical except 24 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 8 Fig. 8. — Restoration of probable skull CM 11162. for its flat, even possibly slightly concave, dorsal margin. The axis of the condyle is directed poster- oventrally at about 120 degrees from the long axis of the skull; this angle may have been exaggerated by dorsoventral crushing. A widely open fracture that extends across the floor of the foramen mag- num, giving this otherwise round opening a verti- cally elongate appearance, continues outward and slightly upward through the posterior process of the postorbital on the lateral side of the skull. The ba- sicranium (Fig. 7D) contains a number of major and minor fractures along some of which there has been displacement. Except for proportional differences discussed below, the basicranium of CM 11162 ex- hibits no marked structural differences from that of Diplodocus . With the exception of proportional differences, the partially preserved palate of CM 1 1 162 (Fig. 7D) exhibits no noticeable departures from that de- scribed (McIntosh and Berman, 1975) for Diplodo- cus. The pterygoids are essentially complete, un- distorted and, as in Diplodocus, form a midventral, dihedral angle between them of about 60 degrees. The palatines, which presumably would have oc- cupied the acute angle formed between the narrow, transverse process and the anterior process of the pterygoids, are absent. Both vomers are lost, but they undoubtedly occupied the same position as those in Diplodocus (McIntosh and Berman, 1975); there the flat, narrowly triangular vomers articulate with the lateral surface of the broadly concave, an- teroventral portion of the pterygoid borders for most of their length and their anterior ends are clasped together at the midline by broad, medially directed processes of the maxillae. The maxillary processes are well preserved in CM 11162 and the medial gap between them, which held the anterior ends of the vomers, is somewhat wider than normal due to the dorsoventral flattening of the snout. Both ectopterygoids appear to be absent, although they may be buried in the remaining matrix. The right quadrate is essentially complete, undistorted and in its proper orientation, whereas the left is badly crushed and missing a large central section. Comparison of CM 1 1 162 with Diplodocus Though the incomplete preservation of the skull CM 11162 eliminates many opportunities for de- tailed comparisons, this skull is obviously very close to that of Diplodocus . Comparisons between CM 1 1 162 and Diplodocus skull CM 3452 and those skulls very likely belonging to Diplodocus have re- vealed a number of subtle proportional and struc- tural differences. Some of these differences, how- ever, have to be evaluated with caution because they may be the result of postmortem distortion of CM 1 1 162. It will also be noticed that the obviously 1978 BERMAN AND McINTOSH— APAJ05ADRD5 RELATIONSHIPS 25 greater general robustness of CM 11162 is a fun- damental aspect of many of the features used below to contrast it with the skull of Diplodociis . In lateral view the occiput of the probable Apato- saurus skull CM 11162 slopes anterodorsally at an angle of about 75 degrees to the horizontal pass- ing through the ventral margins of the maxillae as compared to its right angle orientation in Diplodo- cus (Fig. 2); differences they exhibit in the angles subtended between the occipital plane and the cra- nial roof, and the cranial roof and the snout are too small to be safely considered as diagnostic. In CM 11162 the axis of the occipital condyle is inclined posteroventrally at an angle of about 120 degrees to the long axis of the skull in eontrast to its approx- imately right angle orientation in Diplodocus; the larger angle of the former, however, may be partly due to crushing. The triangular, posterior process of the postorbital of CM 1 1 162 differs from that of Diplodocus in being more broadly developed and in extending to a level posterior to the supratemporal fossa. In the probable Apatosaurus skull there is also a greater vertical development of the squa- mosal below the posterior proeess of the postorbit- al, which has resulted in a corresponding length- ening of its posterior contact with the distal end of the paroccipital process; the quadrate proeess of the squamosal is also proportionally broader in CM 1 1 162 than in Diplodocus . Differences are also seen in the occipital views of the skulls (Fig. 3). In this aspect CM 11162 ap- pears rather dome-shaped in outline, whereas the skull of Diplodocus CM 3452 is subrectangular in outline; this difference can be attributed mainly to development of th§ lateral, occipital wing of the parietal. In Diplodocus its free, superior border arches smoothly and rather strongly dorsolaterally, completely hiding the supratemporal fossa from oc- cipital view. The lateral wing of the parietal in CM 1 1 162 is much narrower and its nearly straight, ven- trolaterally sloping, free border allows the supra- temporal fossa to be partially seen in occipital view. Further, in CM 11162 the distal end of the parietal wing does not encroach as greatly upon the squa- mosal as in Diplodocus and, as a result, in the for- mer the dorsolateral process of the squamosal, which forms the ventral border of the supratem- poral fossa, is wider and the contact between the occipital process of the squamosal and the occipital wing of the parietal is considerably shorter. In Di- plodocus the lateral surface of the skull below the supratemporal fossa meets the occiput in a sharp. right angle corner, whereas in CM 11162 this inter- section is somewhat rounded. As a consequence, the probable Apatosaurus skull can also be distin- guished from that of Diplodocus by its greater ex- posure of the squamosal and its partial exposure of the posterior process of the postorbital in occipital view. In Diplodocus all that can be seen of the squa- mosal in this view is its narrow, occipital process. In addition to this process in CM 1 1 162, the antero- dorsal process of the squamosal and a wide margin along its posterior contact with the flared, distal end of the paroccipital process are also clearly visible in occipital view. The most marked proportional differences be- tween CM 11162 and the skull of Diplodocus CM 11161 are in the palate, quadrate, and braincase (Fig. 9). Though proportionally the lengths of their braincases, measured from the back of the condyle to either the base or the tip of the paraphenoid ros- trum, are very similar, the basipterygoid process of CM 11162 is shorter and stouter, and the condyle is much more massive. In CM 11162 there is a marked flaring of the distal end of the basipterygoid proeess, whereas in CM 1 1 161 there is only a slight swelling. In these features of the braincase the Di- plodocus skulls CM 3452 and CM 11161 are iden- tical. The quadrate of CM 11162 is proportionally shorter and more massive at its distal end than in CM 11161, CM 3452, USNM 2672, and USNM 2673 (Table 1). A proportionally shorter quadrate in CM 1 1 162 is reflected in a more posterior position of its contact with the quadratojugal than in CM 11161. Although the lengths of these contacts along the posterior borders of the quadrates of both speci- mens are proportionally very similar, if not equal, the posterior margin of the quadratojugal in CM 11162 is at a level slightly anterior to the median union of the basipterygoid processes, whereas in CM 11161 it is considerably posterior to this level. In this feature the Diplodocus skull CM 3452 is identical to CM 11161. In palatal view the angle formed between the basipterygoid processes in CM 11162 is about 60 degrees, whereas it is about 40 degrees in CM 11161. As a result, the end of the basipterygoid process in CM 1 1 162 is brought closer through a horizontal plane to the distal end of the quadrate. The basipterygoid process and the quad- rate are also brought closer together in CM 11162 because the medial surface at the distal end of its quadrate does not curve slightly laterally as in CM 11161. In CM 3452 the angle between the basipter- ygoid processes is about 35 degrees and the medial 26 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 8 Fig. 9. — Palatal views of skulls of A, Diplodocus CM 11161, and B, prohahXQ Apatosaurus CM 11162 reduced to same size in order to demonstrate proportional differences. Horizontal lines connect identical topographical points. 1978 BERMAN AND McINTOSH— APATOVA 177? t/5 RELATIONSHIPS 27 surface of the quadrate also curves slightly laterally at its distal end. As a result of the proportionally shorter postpterygoid structures in CM 11162, its pterygoid occupies the more posterior position in the skull than that in CM 11161. At the level oc- cupied by the vomers there is a disproportionate, longitudinal lengthening of the palate in CM 11162 over that of CM 11161 to the extent that, anterior to this level none of their palatal structures shows any appreciable differences in anteroposterior po- sition. Holland (1915«) stated that maxillary teeth of CM 11162 did not insert vertical to the jaw line as in Diplodocus , but were more or less procumbent; we cannot find any evidence to support this observa- tion. Further, our earlier observation (McIntosh and Berman, 1975) that the probable Apatosaurus skull CM 11162 may differ from Diplodocus in the reduced size and more anterior position of the fore- most antorbital opening also cannot be verified with further preparation. Additional Evidence on the Skull of Apatosaurus A pair of quadrates and the greater part of a cra- nium were found by Marsh’s collectors at Morri- son, Colorado, which provide additional evidence, though circumstantial, that the skull of Apatosaurus was Diplodocus-V\ke in structure. The catalogue number YPM 1860, which both quadrates bear, is very likely their correct assignment. In reviewing the ambiguities surrounding the locality data of the cranium, it was concluded that there is not only strong reason to believe that it came from YPM quarry 10, but that it also belongs to the type of A. ajax, YPM 1860. Both the quadrates and the cra- nium represent a skull larger than any skull here- tofore identified as Diplodocus and are also slightly larger than the presumed Apatosaurus skull CM 11162. The left quadrate from Morrison (Fig. 10) is com- plete except for the thin, anteriorly directed plate of bone on whose medial surface the pterygoid ar- ticulated, the right quadrate is missing not only the pterygoid process, but also a little over 20% of the upper, proximal end of its main shaft. The quad- rates are not only near duplicates of those of Di- plodocus and the probable Apatosaurus skull CM 1 1 162, but are readily distinguishable from those of Camarasaurus . White’s (1958) detailed description and illustration of the quadrate of Camarasaurus, as well as the excellent illustrations of this bone given by Ostrom and McIntosh (1966, PI. 4), make a close comparison here between the quadrate of this genus and the quadrates YPM 1860 unneces- sary. Viewed laterally the posterior margin of the quadrate shaft curves smoothly and gently antero- ventrally and has a greatest length of about 21 cm. In anteroposterior length the distal end of the shaft, measuring 3.8 cm, does not greatly exceed the prox- imal end, which measures 2.8 cm. The lower half of the shaft is expanded into a strongly ridged, ar- ticular surface for the posterior end of the quadra- tojugal; the upper, narrowly tapering end of this sutural scar extends slightly onto the posterior sur- face of the shaft. That part of the quadratojugal, which articulated with the quadrate, was undoubt- edly like that in Diplodocus and the probable Apa- tosaurus skull CM 11162 in its shape and sutural relationship with the quadrate. Beginning at the anterodorsal margin of the shaft, a deep channel extends a considerable distance ventrally as it curves gradually onto and across the lateral surface of the shaft; the channel certainly held the same narrow, quadrate process of the squamosal seen in skulls of Diplodocus and in CM 11162. Further, the course of the channel indicates that a small portion of the lateral surface of the proximal end of the quadrate was exposed at the posterior end of the ventral margin of the quadrate process of the squa- mosal as in Diplodocus . In posterior view, the quadrate appears club-shaped. From the greatest width of about 4.8 cm near its distal, lower end, it gradually narrows dorsally to about 1.5 cm at a point approximately three fourths its height, then widens slightly to about 2.0 cm. The lower, wider portion of the posterior surface is very slightly con- cave except along the distal margin of the shaft, where it is moderately convex; above this region the posterior surface becomes flat and remains so until near the proximal end of the shaft, becoming here a pronounced ridge. For most of its length the medial surface of the shaft consists of a strongly developed ridge; dorsally the ridge merges with the nearly flat, narrow, proximal end of the shaft. Only the base of the anteriorly projecting pterygoid pro- cess is preserved, but the general outline and ori- entation of the process can be largely deduced from its remaining margins. Anterior view of the quad- rate reveals the base of the pterygoid process as a rather thin plate that is broadly bowed laterally. The base of the process is thinnest at mid-height, thick- ening only somewhat dorsally, but greatly thick- ening ventrally. Below the pterygoid process the 28 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 8 ABC D Fig. !0. — A, lateral, B, anterior, C, medial, and D, posterior views of left quadrate probably belonging to the holotype of Apatosaurus ajax YPM 1860. Missing anterior pterygoid process indicated in A. Distal end toward bottom. anterior surface of the shaft is excavated into a shal- low depression, giving the articular surface a kid- ney-shaped outline. In anterior view the condylar surface slopes ventromedially. In Marsh’s ( 1896) description of the cranium from Morrison, Colorado, as ^"Atlantosaums montanus" the only feature he commented on was a so-called pituitary canal leading from the brain cavity down through the base of the skull. The cranium was il- lustrated by Marsh (1896:274, PI. XV) in posterior and ventral views only and with just a few of the sutures and foramina indicated; further preparation has revealed almost all of these features clearly (Eig. 11). The only major portions of the occiput not represented are the distal end and superior mar- gin of the lateral, occipital wing of the parietal, and the occipital process of the squamosal that forms the lateral border of the posttemporal fossa. Only the left side of the cranial roof is present but this includes most of the parietal and frontal, and the posterior third of the prefrontal. The parietal-frontal juncture has been destroyed, separating the cranial roof from the principal portion of the cranium; these have been reunited in what is thought to be their correct relative position. The more complete left cranial wall includes the exoccipital-opisthotic com- plex, prootic, laterosphenoid, and the base of the orbitosphenoid. Of the cranial floor elements, the basioccipital is complete, the basisphenoid lacks mainly the basipterygoid processes and the para- sphenoidal rostrum is broken off at its base. The occipital condyle is as in Diplodocus in its shape and orientation and the skull must have been di- rected at nearly a right angle to the neck. The bas- ioccipital-basisphenoid suture is not detectable. Two features of the cranial roof clearly distinguish the cranium from that of Camarasaunis and give it a distinctly Diplodocus-\\kt character: 1) the pos- terior end of the prefrontal is triangular and projects posteromedially into the anterolateral corner of the frontal, and 2) a moderately deep, concave emar- gination of the orbital margin of the frontal occurs 1978 . posterior. B, ventral. C. dorsal, and D. left lateral (stereo pair) views of hraincase probably belonging to holotype of Apatosaurus aja.x 30 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 8 at a level just behind the posterior end of the pre- frontal. Both these features are present in the prob- able skull CM 1 1 162. Only a few minor differences between the Morrison cranium and that of Diplodocus can be noted. Small proportional dif- ferences, such as the greater size of the condyle and basal tubercles, are to be expected in the Mor- rison cranium because of the greater robustness of Apatosaurus over Diplodocus . The prootic-latero- sphenoid suture is tightly closed and is not an abut- ment contact as in Diplodocus; this probably indi- cates a fully grown individual. The transverse opening of the “pituitary canal” between the basal tubercles that was noted by Marsh (1896) is most likely the result of incomplete preservation. The thin cranial floor in this region and the fact that the cranium was glued along a number of breaks that intersect this opening could account for the absence of bone here. One other piece of evidence suggests that the COMPARISON OF POS If Apatosaurus possessed a DiplodocusAWso., rather than a Camarasaurus-WkQ , skull, it may be asked how the postcranial skeletons of these three sauropods compare. Comparisons are made possi- ble by detailed accounts of the postcranial skeletons of Diplodocus (Osborn, 1899; Hatcher, 1901u, 1902; Holland, 1906; Gilmore, 1932), Apatosaurus (Gil- more, 1936), and Camarasaurus (Gilmore, 1925). The postcranial skeletons of Apatosaurus and Cani- arasaurus have been generally considered more similar to each other than to Diplodocus because of their much greater robustness. It is in fact difficult to distinguish between isolated hindlimb bones of Apatosaurus and Camarasaurus , especially if these elements are imperfectly preserved. Excepting this superficial resemblance between Camarasaurus and Apatosaurus, the postcranial skeletons of Apa- tosaurus and Diplodocus share a large number of characters that set them widely apart from Cama- rasaurus. Diplodocus and Apatosaurus , in contrast to Camarasaurus, have relatively very long necks, short trunks, and very long tails, unusual anterior caudal vertebrae and midcaudal chevrons, shorter forelimbs and metacarpals, and reduced number of carpal and tarsal elements. Vertebral Column The cervical vertebrae, particularly the posterior ones, are among the most diagnostic bones in the skull of Apatosaurus was like that of Diplodocus . Examination of the partial skeletons field nos. 24 (CM 3390) and 37, found about 6 m apart at Dino- saur National Monument (Fig. 1), convinces us that Douglass (see Historical Review) was probably cor- rect in his conclusion that these specimens belong to the same juvenile individual of Apatosaurus. Unfortunately, although the anterior portion of a small jaw possessing Diplodocus-Wko teeth (field no. 35) that was noted by Douglass in the collection records as having been found with field no. 37 was prepared at the Carnegie Museum, it cannot be lo- cated. If Douglass’ observation on the nature of the teeth of the jaw is correct — there is no reason to doubt it — and if the jaw and field specimens nos. 24 and 37 were part of one individual, then it follows that Apatosaurus had Diplodocus-Wko teeth, rein- forcing our conclusion that the skull of Apatosaurus is Diplodocus-Wkt. RANIAL SKELETONS sauropod skeleton. Apatosaurus and Diplodocus possess 15 cervicals, Camarasaurus, 12. In all three genera the neural spines of the posterior cervicals and the anterior dorsals are deeply cleft; in Apa- tosaurus and Diplodocus the clefts are V-shaped, whereas those of Camarasaurus are more U- shaped. The cervicals of Apatosaurus are propor- tionally shorter and more solidly constructed than those of either Diplodocus or Camarasaurus . In Apatosaurus and Diplodocus the cervical ribs are much shorter than in Camarasaurus and do not ex- tend beyond the posterior end of the centrum from which they originate, whereas in Camarasaurus some cervical ribs, such as the ninth, may reach a length of about two and a half times the length of the centrum. The cervical ribs of Apatosaurus are considerably stouter than those of either Diplodo- cus or Camarasaurus . Apatosaurus and Diplodocus have 10 dorsal vertebrae that exhibit similar region- al variations. Their anterior dorsal centra are op- isthocoelous; posteriorly they are amphiplatyan or amphicoelous. In the shoulder region the neural spines in both are low but rise posteriorly to be- come very high and slender in the sacral region. Camarasaurus has 12 dorsals, all of which are opis- thocoelous. The dorsal neural spines exhibit little change in height posteriorly and at the posterior end of the series they are much lower, stouter, and lat- erally expanded above than in the other two genera. 1978 BERMAN AND MclNTOSH— APATOSAURUS RELATIONSHIPS 31 There are five sacral vertebrae in all three genera. In Apatosaurus and Diplodocus , the sacral centra and ribs are hollow, the second and third spines are united and there is a tendency for the fourth to unite with the third. In Cainarasaurus the centra are solid or have much smaller cavities and four or even all five spines may fuse. The tails in Apatosaurus and Diplodocus reach enormous lengths, up to 82 caudals in the former and 73 or more in the latter; the caudals of approx- imately the posterior third of their tails consist es- sentially of elongated rods that form a "whip-lash” structure. The centra in both are generally amphi- platyan throughout, although there is a tendency, particularly in Diplodocus, for the anterior centra to be somewhat procoelous. In both genera the transverse processes in the anterior part of the tail form thin, vertically expanded, wing-like plates that more closely resemble the sacral ribs than the trans- verse processes of the remainder of the tail. In Apatosaurus the first three or four caudals show this development; in Diplodocus it occurs in the first 12 or more. In both the caudal spines are very high anteriorly and their distal ends are not ex- panded transversely. Their anterior caudal chev- rons are of normal structure, the laminae of the chevrons being united by a bridge of bone above the haemal canal but joining below the canal to form a simple, laterally flattened spine. The chevrons of the mid-caudal region, however, are unusual in that they lack the bridge of bone above the haemal canal and the laminae do not unite immediately below the canal but at the ends of well-developed anteriorly and posteriorly directed processes that originate at their distal ends. This character is more pronounced in Diplodocus . The presence of these unusual mid- caudal chevrons in Apatosaurus was not discussed by Gilmore (1936) because only the anterior three chevrons are present in the specimen (CM 3018) studied by him, but he did indicate them in his post- cranial restoration. The "double-arch” type of mid- caudal chevron was, however, briefly noted by Riggs (19036) and shown in his restoration of Apa- tosaurus . In some features the caudals of Apato- saurus and Diplodocus are distinct. The anterior caudal centra of Diplodocus have deep lateral pleu- rocoels and ventral excavations; the ventral exca- vations occur well into the midcaudal region. The centra are even more elongated in Diplodocus than 'm Apatosaurus, a feature which becomes more pro- nounced posteriorly, especially in the whip-lash portion of the tail. In Apatosaurus the caudal centra are not excavated laterally or ventrally except for possibly the first few, which may have small irreg- ularly placed cavities, and the anterior centra have a blunt, ventral keel. The tail of Cainarasaurus dif- fers from those of Apatosaurus and Diplodocus in having only 53 much shorter vertebrae. The trans- verse processes of even the anteriormost caudals are simple. The anterior neural spines are not unusually high and distally are expanded transversely into ball-like structures. The centra are amphicoelous, unexcavated and without a ventral keel, and the chevrons are unspecialized throughout and almost never enclose the haemal canal above. Appendicular Skeleton The marked difference in massiveness of the ap- pendicular skeletons of Diplodocus and Apatosau- rus is more apparent than their similarities. The scapulae are distinct in all three genera. The broad plate, which extends anterodorsally from the base of the scapular blade, is more vertically expanded and the prominent ridge, which divides its external surface into two broad muscle fossae, is less de- veloped and makes a larger angle with the shaft in Apatosaurus and Cainarasaurus than in Diplodo- cus. The upper end of the scapula is greatly ex- panded in Cainarasaurus, much less so in Diplod- ocus, and only slightly expanded in Apatosaurus . The coracoids are dissimilar in outline in all three, being quadrangular in Apatosaurus, roughly trian- gular in Cainarasaurus, and intermediate between these two in Diplodocus . The ilia in all three genera are similar. The pubes of Apatosaurus and Diplod- ocus are relatively slender compared to those of Cainarasaurus . Diplodocus has a pronounced hook-like process for the ambiens muscle on the upper, anterior margin of the pubis; this process is much less prominent in Apatosaurus and is nearly absent in Cainarasaurus . Confusion has occurred through the legend of figure 37 in Gilmore’s ( 1936) description of Apatosaurus in which the figured pubis of Diplodocus carnegiei CM 94, exhibiting a very prominent ambiens process, is mistakenly identified as Apatosaurus e.xcelsus CM 563. The fig- ure given by Hatcher (1903a, PI. IV, Fig. 1) of the pelvis of CM 563 also tends to exaggerate this fea- ture \n Apatosaurus . The ischium is one of the most diagnostic bones in the sauropod skeleton. \n Apa- tosaurus and Diplodocus the blades of the ischia are tilted ventromedially and their expanded distal ends contact each other along a wide margin of the ventral borders of their medial surfaces. The more 32 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 8 slender blades of the ischia in Camarasauriis are not expanded distally and are twisted along their long axes so that the ends of the blades come to lie in a horizontal plane with their inferior margins con- tacting each other medially. Perhaps the most striking and significant feature separating the limbs of Apatosaurus and Diplodo- cus from those of Camarasaunis is their forelimb to hindlimb length ratio; the humerofemoral length ratio is 2/3 in both Apatosaurus and Diplodocus and 4/5 in Camarasaurus . The forelimb of Apatosaurus is robust, rivaled in this feature only by that of the South American Titanosaurus australis (von Huene, 1929). Eorelimbs of Camaraurus resemble those of Diplodocus more nearly in their overall slender- ness, although the humerus of Camarasaurus is somewhat more robust and the medially projecting process at the upper end of its ulna is also more pronounced. The manus of Apatosaurus and Di- plodocus are similar so far as known. A single car- pal bone remains in Apatosaurus', the condition in Diplodocus is unknown. The metacarpals of Apato- saurus are short and robust and in Diplodocus they are short but more slender. Camarasaurus has two carpals and its metacarpals are very long and slender. The hindlimb bones of Apatosaurus and Camarasaurus are about equal in their much great- er robustness than those of Diplodocus. Despite RELATIONSHIPS It has become common practice (Janensch, 1935; Nopcsa, 1930; von Huene, 1948) to divide the Sau- ropoda into two families primarily on the basis of dentition. Though a variety of family names has been employed, these classifications are in essential agreement in their separation of the broad, spatu- late-toothed forms such as Bracliiosaurus and Cam- arasaurus from the slender-toothed forms such as Diplodocus. The long-standing conclusion that Apatosaurus had a Camarasaurus -\ikt skull and dentition was the major reason for its alliance with the former group. Romer (1956) divided the sauro- pods into the Brachiosauridae and Titanosauridae; Apatosaurus was assigned to the latter family even though Nopcsa ( 1930) and von Huene ( 1948) placed it in the former. White ( 1958), believing he had sub- stantiating evidence that the skull of Apatosaurus was Camarasaurus-\\k&, recommended the removal of Apatosaurus from the Titanosauridae and place- ment in the subfamily Camarasaurinae of the Bra- chiosauridae. In a later classification Romer (1966) this feature, the hindlimbs of Apatosaurus and Di- plodocus can be distinguished from those of Cam- arasaurus. The femur of Camarasaurus has a straight shaft, whereas femora of Apatosaurus and particularly of Diplodocus exhibit a slight sigmoid curve. In Camarasaurus the cnemial crest of the tibia is relatively less pronounced and the muscle scar on the lateral surface of the fibula is much more strongly developed than in the other two genera. The pes of Apatosaurus and Diplodocus exhibit dif- ferences from that of Camarasaurus . No calcaneum has yet been found associated with any Diplodocus ox Apatosaurus pes, and, although the question of its existence has not been settled, the evidence strongly suggests that the only tarsal element they possess is an astragalus. In Camarasaurus the tar- sus consists of an astragalus and a small, spherical calcaneum. The metatarsals of Apatosaurus and Diplodocus are very similar except that the third and fourth are more slender in the latter. In both, metatarsals III and IV are the longest, the fourth often being slightly longer, and metatarsal I is un- usual in having a process on the posteroventral mar- gin of its lateral surface. In Camarasaurus metatar- sals II and III are equal in length and the longest, and metatarsal I does not possess the above-men- tioned process. 0¥ APATOSAURUS referred Apatosaurus to the Titanosauridae and Brontosaurus to the Brachiosauridae even though Riggs (19306) had clearly demonstrated that the lat- ter genus is a junior synonym of the former. It is beyond the scope of this paper to present a revised classification of the sauropods, but we reject the commonly used, two-family division, which artifi- cially associates widely divergent forms. Apatosaurus and Diplodocus are morphologically very similar, and the former is quite different from Camarasaurus, to which it has been closely allied by many authors. Equally important. Apatosaurus and Diplodocus share a suite of characters that can be seen in various combinations in five other less well known sauropod genera — Barosaurus, Cetio- sauriscus, Marne nchisaurus , Dicraeosaurus, and Nemegtosaurus . These genera are judged by us to be very closely related and quite distinct from all other adequately known sauropods and deserving of familial separation. The oldest valid name avail- able for this group is Diplodocidae Marsh (1884). 1978 BERMAN AND MclNTOSH— APATOSAURUS RELATIONSHIPS 33 The only other family names that could be consid- ered, Atlantosauridae (Marsh, 1877/?) and Amphi- coeliidae Cope (18771?) are rejected because their type genera are indeterminate. In the case of Atlan- tosauridae the type genus Atlantosaurus (first de- scribed as Titanosaums Marsh, 1877a) cannot be adequately defined and has to be considered a no- men dubium. The type species, A. montanns, is based on only an incomplete sacrum (YPM 1835) that cannot be clearly distinguished from those of a number of sauropods, including genera outside the new family grouping proposed here. The adop- tion of Amphicoeliidae has the same drawback as Atlantosauridae; the type genus cannot be ade- quately defined. The family was established by Cope (1877/?) to include two species of a new genus, Amphicoelias altus, the type species, and A. latns; a third species, A. fragillimns, was later added by Cope (1878/?). In a restudy of these species, all of which are represented by single specimens, Osborn and Mook (1921) concluded that A. altiis represents a young individual of Camarasaurns and suggested ihoi A. fragillimns should be provisionally referred to A. altns; we are in agreement with these conclu- sions. Though the type of A. altus is referable to the Diplodocidae, as defined below, its incomplete- ness does not allow it to be distinguished from either Diplodocus or Barosaurus . Diplodociis was first described by Marsh (1878/?) on the basis of a small, but adequately diagnostic portion of the post- cranial skeleton, yet it was not until his later de- scription of the skull (1884) that he considered the genus unique and the type of a new monotypic fam- ily, Diplodocidae. Though Marsh (1898) later trans- ferred Barosaurus from the Atlantosauridae to the Diplodocidae, subsequent classifications of the sau- ropods by von Huene (1927a, 1927/?) continued to recognize the latter family as monotypic. In a recent catalogue of the dinosaur genera White (1973), with- out giving a revised or expanded definition of the family Diplodocidae, included within it a great va- riety of genera, many of which are too divergent to be grouped together at this taxonomic level. The availability of the family name Diplodocidae is fortunate because Diplodocus is well known and very representative of the new family grouping pro- posed here, for which we offer the following revised definition. Diplodocidae, Marsh, 1884 Definition. — Skull: nares superior in position; quadrate directed anteroventrally; basipterygoid processes elongated; definition of weak, pencil-like teeth. Vertebral Column: midpresacrals exhibit tendency toward “cervicalization” to produce long neck; midpresacral spines cleft; sacral spines very high; anterior caudals with broad, wing-like trans- verse processes; midcaudal chevrons having distal, fore and aft directed processes; tail consisting of large number of vertebrae, forming a “whip-lash” structure. Appendicular Skeleton: Forelimbs short with a humerofemoral length ratio of 2/3; tar- sus and at least in some cases the carpus reduced to single element; distal ends of ischia expanded in vertical plane and contacting each other along a wide, ventral margin of their medial surfaces; pro- cess on posteroventral edge of lateral face of meta- tarsal I; metatarsals III and particularly IV longest. Remarks. — Although our inclusion of Apatosau- rus in the Diplodocidae is obvious, assignment of the other genera to this family must be justified. The brief comments that follow are intended to serve this purpose. The Upper Jurassic Barosaurus Marsh, 1890, is structurally very close to Diplodocus and is distin- guished mainly by its enormously elongated cervi- cal vertebrae and slightly less developed caudal neural arches and spines; its limb elements are scarcely distinguishable from those of Diplodocus . In Barosaurus cervicalization of the midpresacrals is evident, the anterior caudals have wing-like transverse processes, the midcaudal chevrons pos- sess the Diplodocus-Wkt fore and aft processes, the distal ends of the ischia are expanded and contact each other on their ventromedial surfaces, and metatarsal I has a distinct process on the postero- ventral edge of its lateral surface. Cetiosauriscus von Huene ( 1927/?) has not pre- viously been associated with the members of the family group proposed here, but a number of char- acters indicate that this Upper Jurassic genus should be considered a primitive member of the Diplodocidae. Except for several posterior dorsal centra, its presacral vertebrae, which are very di- agnostic among the sauropods, are otherwise un- known. The anterior caudals, although incomplete- ly known, appear to possess wing-like transverse processes, the midcaudal chevrons are Diplodocus- like in that their distal ends possess fore and aft directed processes and there is a whip-lash devel- opment of the tail. The humerofemoral length ratio is 2/3. The calcaneum appears to be absent in the tarsus and the astragalus is the only tarsal element. Metatarsals III and IV are the longest and metatar- 34 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 8 sal I clearly exhibits a process on the posteroventral margin of its lateral surface. The Upper Jurassic Mainenchisaurus Young, 1954, tentatively referred to Diplodocidae, has a long neck with 19 cervicals and there are 11 dorsal vertebrae, which possess cleft neural spines. The length of the tail is unknown, but the midcaudal chevrons possess the distal fore and aft directed processes as in Diplodociis . Though the humerus and femur are not known for any one specimen, the humerofemoral length ratio is considered to be a little greater than 2/3 in the type genus. This is based on the fact that the height of the sacral neural spines is relatively somewhat less than in other members of the family and that there exists a direct correlation between the height of the sacral spines and the humerofemoral length ratio. This ratio may vary among specimens referred to this genus and only articulated material will reveal its true value. Dicraeosaunis Janensch, 1914, is a somewhat puzzling. Upper Jurassic genus and is tentatively referred to this family. The neck is short; the num- ber of both the cervical and dorsal vertebrae is 12. Surprisingly, the dorsal vertebrae do not possess pleurocentral cavities; the skull and teeth, however, are distinctly diplodocid. The neural spines of the presacrals are more deeply cleft than in any other sauropod and the sacral spines are high. The ante- rior caudals have wing-like, transverse processes and the midcaudal chevrons are Diplodocus -Wke.. The distal ends of the ischia are greatly expanded. The forelimb is short, which, along with the high sacral spines, suggests that the humerofemoral length ratio may be close to 2/3. Finally, the Upper Cretaceous Nemegtosaurus Nowinski, 1971, known only by the skull, which is distinctly diplodocid in structure, including the teeth, is referred to Diplodocidae. ACKNOWLEDGMENTS We are indebted to Dr. John H. Ostrom of Yale Peabody Museum for allowing us the opportunity to study specimens un- der his care and access to archives containing quarry maps and correspondence of past associates of that institution. Thanks are also due Dr. Mary R, Dawson of Carnegie Museum of Natural History for critical reading of the manuscript. One of us (JM) also wishes to acknowledge the many years of helpfulness and generous hospitality extended to him by Drs. Craig C. Black and Mary R. Dawson during visits to the Carnegie Museum of Nat- ural History. LITERATURE CITED Abel, O. 1910. Die Rekonstrucktion des Diplodocus . Abh. K. K. Zool.-Bot. Ges. Vienna, 5:1-60. Cope, E. D. 1877u . On a gigantic saurian from the Dakota epoch of Colorado. Pal. Bull., 25:5-10. . 18776. On Awp/j/roe//fli, a genus of saurians from the Dakota epoch of Colorado. Pal. Bull., 27:2-3. . 1877c. On reptilian remains from the Dakota beds of Colorado. Proc. Amer. Philos. Soc., 17:193-196. . 187&J. The saurians of the Dakota epoch. Amer. Nat., 12:56-57. . 18786. A new species of Amphicoelias . Amer. Nat., 12:563-565. Gilmore, C. W. 1907. The type of the Jurassic reptile Moro- saiirus agilis redescribed, with a note on Camptosaiirus . Proc. U.S. Nat. Mus., 32:151-165. . 1925. A nearly complete articulated skeleton of Cama- rasaurus, a saurischian dinosaur from the Dinosaur National Monument, Utah. Mem. Carnegie Mus., 10:347-384. . 1932. On a newly mounted skeleton oi Diplodocus in the United States National Museum. Proc. U.S. Nat. Mus., 8I(18):1-21. . 1936. Osteology of Apa/o.srt/m/.v, with special reference to specimens in the Carnegie Museum. Mem. Carnegie Mus., 11:175-300. Hatcher, J. B. 1901u. Dip/or/ocw^ (Marsh): its osteology, tax- onomy, and probable habits, with a restoration of the skel- eton. Mem. Carnegie Mus., 1:1-63. . 19016. On the structure of the manus in Brontosaurus . Science, n. s., 14:1015-1017. . 1902. Structure of the fore limb and manus of Bronto- saurus. Ann. Carnegie Mus., 1:365-376. . 1903a. Osteology of //ap/oc«/7r/;o.saz/r«i, with descrip- tion of new species and remarks on probable habits of the Sauropoda, and the age and origin of the Atlantosaurus beds. Mem. Carnegie Mus., 2:1-75. . 19036. Additional remarks on Diplodociis . Mem. Car- negie Mus., 2:72-75. Hay, O. P. 1908. Dr. W. J. Holland on the skull of D/p/oz/ocz/.?. Science, n. s., 28:517-519. Holland, W. J. 1906. The osteology of Diplodociis Marsh. Mem. Carnegie Mus., 2:225-264. . 1915a. Heads and tails; a few notes relating to the struc- ture of the sauropod dinosaurs. Ann. Carnegie Mus., 9:273- 278. . 19156. A new species of Apatosaurus . Ann. Carnegie Mus., 10:143-145. . 1924. The skull of Diplodociis . Mem. Carnegie Mus., 9:397^03. 1978 BERMAN AND MclNTOSH— APATOSAURUS RELATIONSHIPS 35 Huene, F. VON. 1914. Uber der Zweistammigheit der Dinosau- rier, mid BeitragenzurKenntniseinigerSchadel. Neu. Jahrb. Min., Geol. and Pal., 37:557-589. . 1927a. Short review of the present knowledge of the Sau- ropoda. Queensland Mus. Mem., 9:121-126. . \921b. Sichtung der Grundlagen der jetzigen Kenntnis der Sauropoden. Eclogae Geol. Helvetiae, 20:444-470. . 1929. Los Saurisquios y Omitisquios del Cretaceo Ar- gentino. An. Mus. La Plata, ser. 2, vol. III. . 1948. Short review of the lower tetrapods. Pp. 65-106, in Robert Broom commemorative volume (A. L. Du Toit, ed.), South Africa Royal Soc. Spec. Publ., v + 257 pp. Janensch, W. 1914. Uebersicht iiber die Wirbelthierfauna der Tendaguru-Schichten, nebst einer Kurzen Charackterisi- erung der neu aufgefiihrten Arten von Sauropoden. Arch. Biontol., 3:81-1 10. . 1935. Die Schadel der Sauropoden Brac/j/oMMTO.v, Bar- asaurus nnd Dicraeosaimts . Palaeontographica, Suppl. 7, 1 Reihe, 2 Teil, Stuttgart, pp. 147-298. Lydekker, R. 1877. Notice of new and other Vertebrata from Indian Tertiary and Secondary rocks. Rec. Geol. Sur. India, 10:30-43. Marsh, O. C. 1877a. Notice of a new and gigantic dinosaur. Amer. J. Sci., 14:87-88. . 18776. Notice of new dinosaurian reptiles from the Ju- rassic formations. Amer. J. Sci., 14:514-516. . 1878a. Notice of new dinosaurian reptiles. Amer. J. Sci., 15:241-244. . 18786. Principal characters of American Jurassic dino- saurs, Part I. Amer. J. Sci., 16:411—416. . 1879a. Notice of new Jurassic reptiles. Amer. J. Sci., 18:501-505. 18796. Principal characters of American Jurassic dino- saurs, Part II. Amer. J. Sci., 17:86-92. . 1881. Principal characters of American Jurassic dino- saurs, Part V. Amer. J. Sci., 21:417-423. . 1883. Principal characters of American Jurassic dino- saurs, Part VI. Amer. J. Sci., 26:81-85. . 1884. Principal characters of American Jurassic dino- saurs, Part VII, Diplodocidae, a new family of the Saurop- oda. Amer. J. Sci., 27:161-168. . 1890. Description of new dinosaurian reptiles. Amer. J. Sci., 39:81-86. . 1891. Restoration of Brontosaurus . Amer. J. Sci., 41:341-342. . 1896. Dinosaurs of North America. Ann. Rept., U.S. Geol. Surv., Pt. I, 16:133-244. . 1898. On the families of sauropodous Dinosauria. Amer. J. Sci., 6:487-488. McIntosh, J. S., and D. S Berman. 1975. Description of the palate and lower jaw of the sauropod dinosaur Diplodocus (Reptilia: Saurischia) with remarks on the nature of the skull of Apatosaurus . J. Paleontol., 49:187-199. Nopcsa, Baron F. 1930. Zur Systematik und Biologie der Sau- ropoden. Palaeobiologica, 3:40-52. Nowinski, a. 1971 . Nemegtosaurus niongoUensis n. gen. , n. sp. (Sauropoda) from the uppermost Cretaceous of Mongolia. Results Polish-Mongol. Palaeont. Exped., Pt. Ill, Palaeont. Pol., 25:57-85. Osborn, H. F. 1898. Additional characters of the great herbiv- orous dinosaur Ca/aara.vaa/'u.v. Bull. Amer. Mus. Nat. Hist., 10:219-253. . 1899. A skeleton of Diplodocus . Mem. Amer. Mus. Nat. Hist., 1:191-214. . 1904. Manus, sacrum, and caudals of Sauropoda. Bull. Amer. Mus. Nat. Hist., 20:181-190. . 1912. CrmiB. of Tyrannosaurus d,nd Allosaurus . Mem. Amer. Mus. Nat. Hist., n. s., 1:1-30. Osborn, H. F., and C. C. Mook. 1921. Camarasaurus, Am- phicoelias. and other sauropods of Cope. Mem. Amer. Mus. Nat. Hist., n. s., 3:249-387. Ostrom, J. H., AND J. S. McIntosh. 1966. Marsh’s Dinosaurs. Yale Univ. Press, 388 pp. Riggs, E. S. 1903a. The vertebral column of Brontosaurus . Sci- ence, n. s., 17:393-394. . 19036. Structure and relationships of opisthocoelian di- nosaurs. Part 1, Apatosaurus Marsh. Field Columb. Mus., Geo. Ser., 2:165-196. Romer, a. S. 1956. Osteology of the reptiles. Univ. Chicago Press, 772 pp. . 1964. Cope versus Marsh. Systematic Zool., 13:201-207. . 1966. Vertebrate paleontology. Univ. Chicago Press, 3rd ed., 468 pp. White, T. E. 1958. The braincase of Camarasaurus lentus (Marsh). J. Paleontol., 32:477^94. . 1973. Catalogue of the genera of dinosaurs. Ann. Car- negie Mus., 44:117-155. Young, C.C. 1954. On a new sauropod from Yiping, Szechuan, China. Scientia Sinica, 3:491-504. y 'll ' w Copies of the following Bulletins of Carnegie Museum of Natural History may be obtained at the prices listed from the Publications Secretary, Carnegie Museum of Natural History, 4400 Forbes Avenue, Pitts- burgh, Pennsylvania 15213. 1. Krishtalka, L. 1976. Early Tertiary Adapisoricidae and Erinaceidae (Mammalia, Insectivora) of North America. 40 pp., 13 figs $2.50 2. Guilday, J. E., P. W. Parmalee, and H. W. Hamilton. 1977. The Clark’s Cave bone deposit and the late Pleistocene paleoecology of the central Appalachian Mountains of Virginia. 88 pp., 21 figs. $12.00 3. Wetzel, R. M. 1977. The Chacoan peccary, Cc/tugo/u/s' vrugnen (Rusconi). 36 pp., 10 figs. .. $6.00 4. Coombs, M. C. 1978. Reevaluation of early Miocene North American Moropus (Perissodactyla, Chalicotheriidae, Schizotheriinae). 62 pp., 28 figs $5.00 5. Clench, M. H., and R. C. Leberman. 1978. Weights of 151 species of Pennsylvania birds analyzed by month, age, and sex. 87 pp $5.00 6. Schlitter, D. A. (ed.). 1978. Ecology and taxonomy of African small mammals. 214 pp., 48 figs. $15.00 7. Raikow,R.J. 1978. Appendicular myology and relationships of the New World nine-primaried oscines (AvesiPasseriformes). 43 pp., 10 figs $3.50 .’W‘ PALEONTOLOGY AND GEOLOGY OF THE? T BADWATER CREEK AREA, CENTRAL'* WYOMING. PART 16. THE CEDAR RIDGE'LOCAL FAUNA « (LATE OLIGOCENE) .; TAKESHI SETOGUCHI NUMBER 9 PITTSBURGH, 1978 I I ■s * BULLETIN of CARNEGIE MUSEUM OF NATURAL HISTORY PALEONTOLOGY AND GEOLOGY OF THE BADWATER CREEK AREA, CENTRAL WYOMING. PART 16. THE CEDAR RIDGE LOCAL FAUNA (LATE OLIGOCENE) TAKESHI SETOGUCHI Resident Museum Specialist, Section of Vertebrate Fossils (permanent address: Primate Research Institute, Kyoto University, Inuyama City, Aichi 484, Japan) NUMBER 9 PITTSBURGH, 1978 BULLETIN OF CARNEGIE MUSEUM OF NATURAL HISTORY Number 9, pages 1-61, figures 1-32, tables 1-4, appendices A-D Issued 18 December 1978 Price: $4.50 a copy Craig C. Black, Director Editorial Staff: Hugh H. Genoways, Editor, Duane A. Schlitter, Associate Editor', Stephen L, Williams, Associate Editor', Teresa M. Bona, Technical Assistant. © 1978 by the Trustees of Carnegie Institute, all rights reserved. CARNEGIE MUSEUM OF NATURAL HISTORY, 4400 FORBES AVENUE PITTSBURGH, PENNSYLVANIA 15213 CONTENTS Abstract 5 Introduction 5 Acknowledgments 5 Previous Work 6 Collecting Methods 6 General Geology 6 Faunal List 9 Mammalian Taxa Absent from the Cedar Ridge Local Fauna 10 Age of the Strata 11 Correlation of the Strata 12 Faunal Age 12 Paleoecological Setting 13 Systematic Accounts 15 Class Amphibia 15 1 Batrachosauroides sp 15 Class Reptilia 15 Leiocephalus sp 15 Class Mammalia 15 Peratherium sp. cf. P. spindleri 15 Nanodelphys new species 17 Leptictis sp 18 Ankylodon sp. cf. A. annectens 18 Centetodon sp. cf. C. marginalis 19 Domnina sp. cf. D. gradata 20 Proscalops miocaenns 22 Oligoscalopsl sp 23 Micropternodns sp 25 Prosciurus relictus 25 Pelycomys placidus 27 Adjidannio doiiglassi 28 Paradjidaiimo hypsodus , new species 30 Metadjidanino , new genus 32 Metadjidaunio hendryi, new species 32 Eomyidae, genus indet.. Type A 36 Eomyidae, genus indet.. Type B 36 Proheteroniys sp. cf. P. nebraskensis 37 Heliscomys sp. cf. H. veins 39 Affinities of the Oligocene Heteromyids 41 Eiimys parvidens 43 Enniys elegans 43 Euniys brachyodns 46 Enniys sp. cf. E. planidens 46 Palaeolagns bnrkei 48 Palaeolagns sp. cf. P. intermedins 50 Hesperocyon temnodon 51 Miohippns sp 53 Hyracodontidae genus indet 53 Leptomeryx sp. near L. evansi 53 Hypisodns sp. near El. minimus 53 Summary and Conclusions 54 Literature Cited 56 Appendix A 58 Appendix B 59 Appendix C 60 Appendix D 61 ABSTRACT Upper Oligocene strata unconformabiy overlying upper Eocene sediments outcrop along Badwater Creek, in central Wyoming. The fauna from the underlying Eocene sediments has been collected and studied by Carnegie Museum of Natural His- tory since 1962. Late Oligocene mammals were first recognized in the overlying sandy facies during the 1964 field season by field parties of Carnegie Museum. Intensive collecting and use of screen washing techniques have resulted in the recognition of 26 mammal genera. Most of the fossils recovered are fragmentary, and are generally less than 3 mm in size, which indicates considerable transport of the ma- terial. This assemblage of animals is biased by two major factors. First, due to stream sorting, only small teeth and bones were deposited so that larger mammals, which surely lived near the site of deposition at that time, are not represented in the fauna. And second, the local climate was dry and of moderate temper- ature. This is indicated by the presence of calcic feldspars and gypsum crystals throughout the section. Due to the dry climatic conditions, only land mammals, which were adapted to this eco- logical situation, could live there and consequently these animals are represented in the present fauna. On the other hand, the animals, which required a more mesic condition, could not have lived there and consequently they are not represented in the Since 1962, the Section of Vertebrate Fossils of Carnegie Museum of Natural History has been working Tertiary deposits and collecting vertebrate fossils along Badwater Creek in the northeastern part of the Wind River Basin, Natrona and Fremont counties, central Wyoming. The University of Col- orado Museum, the Museum of Natural History of the University of Kansas and The Museum of Texas Tech University cooperated with this work in var- ious ways. A considerable number of vertebrate fossils has been recovered in volcanic rich silts and clay along Badwater Creek. These sediments were thought to be eastern equivalents of the Tepee Trail Formation (Tourtelot, 1957). There is now almost overwhelming evidence that these sediments do not represent the Tepee Trail Formation (Krishtalka and Black, 1975), but until a thorough geologic re- view of the area can be completed, Tourtelot's usage is followed. The vertebrates recovered from these deposits are primarily of the late Eocene age although other faunal levels are also recognized. Much of the fauna has been described since 1966. fauna. Currently sampled upper Oligocene deposits along Bad- water Creek do not preserve the real diversity of late Oligocene mammals. Due to the less favorable ecological conditions, land micro- mammals, which lived there in the late Oligocene, were spe- cialized in having higher crowned and more lophate teeth. The evolution of this type of dentition was the result of the adaptation to a more herbaceous diet in a drier climatic situation. Some rodents represented had more hypsodont teeth than did their middle Oligocene counterparts, but they had not yet developed rootless or ever-growing cheek teeth. Near the Oligocene-Mio- cene boundary, the climate returned to more mesic conditions and these highly specialized rodents, which were adapted to drier conditions, could not have survived and either became ex- tinct or migrated to other areas by the end of the Oligocene. During the early Miocene, a few rodents migrated into North America from Eurasia. These are not direct descendents of the late Oligocene rodents of North America. All the late Oligocene micromammals of North America have their ancestry in the mid- dle or early Oligocene of North America. Based on the micro- mammal assemblages, the faunal gap between the late Oligocene and the early Miocene is greater than the one between the middle and late Oligocene in North America. In 1964, a field party from the Carnegie Museum of Natural History discovered a much later faunal level in the tan silts, which unconformabiy overlie “the Tepee Trail Formation” along Badwater Creek. Recovery of vertebrate fossils from this level was continued by field parties from the Uni- versity of Kansas in 1971 and from Texas Tech University in 1973 and 1974. As in many other early Cenozoic assemblages, most vertebrates in the later faunal level in Badwater Creek area are represented by fragmentary remains, generally consisting of iso- lated or loosely associated teeth. More than fifteen hundred identifiable specimens are at hand. These specimens are described below. The present study is a part of a series of studies of the Bad- water fauna. Most of the materials dealt with were collected by me during the field season of 1974, and some by field parties of Carnegie Museum of Natural History and the University of Kan- sas. The abbreviations used in this paper are as follows; CM, Car- negie Museum of Natural History; KU, University of Kansas; L, length; W, width; AW, anterior width (width of trigonid); PW, posterior width (width of talonid). ACKNOWLEDGMENTS During the course of this study I have become deeply indebted to many individuals, most especially, to my supervisor. Dr. 5 Craig C. Black, for his constant encouragement, diligent sup- port, and critical counsel. 6 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 9 Dr. Black, and Dr. Mary R. Dawson (Carnegie Museum of Natural History, Pittsburgh) originally suggested this study to me as a significant subject for a doctoral dissertation in verte- brate paleontology, and provided general access to and loan of fossil material in their care. I am also grateful to them and to the following people for thoughtful and stimulating discussion concerning Tertiary insec- tivore, rodent, and lagomorph evolution: Dr. John F. Sutton and Dr. Leonard Krishtalka, both at Carnegie Museum of Natural History; Dr. Robert W. Wilson and Dr. Larry D. Martin at the University of Kansas, Lawrence. This study was supported by National Science Foundation grant GB-30840X to Craig C. Black and Mary R. Dawson, and partly by the Institute of Museum Research, Texas Tech Uni- versity, and by the Ministry of Education of the Japanese Gov- ernment. PREVIOUS WORK The presence of mid- or later Tertiary strata along the Badwater Creek area was recognized as early as 1948. In the 1948 Guidebook for the Third An- nual Eield Conference of the Society of Vertebrate Paleontology, Tourtelot (p. 66) briefly mentioned a sequence of soft, tan, ashy siltstone with two beds of vitric tuff which occur along Badwater Creek overlying the late Eocene Tepee Trail Formation in Secs. 23 and 24, T39N R89W, Natrona County, Wyoming. He referred to these rocks as “upper- most Eocene or Oligocene (?) and Younger . . . ,” and said that the siltstone was remarkably similar to Miocene rocks in other parts of Wyoming but there was “little" break between the Tepee Trail Formation and this siltstone sequence. Later (1957) he referred to this tan siltstone sequence as the up- per part of the Hendry Ranch Member of the Tepee Trail Formation. Thus, the Hendry Ranch Member defined by him (1957) includes all the strata, which overlie the Green and Brown Member in the Bad- water area. No fossils were known from these rocks until 1964 when a Carnegie Museum of Natural History field party discovered a vertebrate fauna in the NW '/4, Sec. 24, T39N R89W (locality 19 of Black and Dawson, 1966:303). The tan silt sequence was pros- pected in 1965 and some two tons of matrix were washed from locality 19 in 1966 by Carnegie Mu- seum of Natural History field parties. Vertebrates recovered from locality 19 were then designated the Cedar Ridge local fauna (Black, 1968:51). When identifiable remains are recovered higher in the sec- tion along Badwater Creek, they will be of a later age and therefore the use of the name “Cedar Ridge local fauna" must be restricted to the assemblage found at locality 19. The use of “Badwater local fauna" is restricted to the assemblage found in the Upper Eocene Hendry Ranch Member below the Oligocene strata. Riedel (1969) recognized an unconformity within Tourtelot’s Hendry Ranch Member. Vertebrates at locality 19 are found in the tan tuffaceous siltstones above the unconformity. Prospecting at locality 19 was continued by field parties from the University of Kansas in 1971 and Texas Tech University in 1973 and 1974. I started work on the geology and paleontology of the Ce- dar Ridge local fauna after 1974. Brief accounts of the geology and vertebrate fos- sils of the Cedar Ridge local fauna are to be found in Black (1968, 1969) and Riedel (1969). COLLECTING METHODS All the fossils were collected from one locality. The matrix containing fossils is weakly cemented by carbonate. Normal washing and screening methods are not applicable for this matrix as it is not easily distintegrated because of the cement. I used citric acid to desolve the calcareous cement. After quarrying, all the matrix was dried and broken down, and the concentrates were then soaked in a weak solution of citric acid (3-5 weight percent). Reaction between the calcareous GENERAL The area of the present study is located in the northeastern part of the Wind River Basin along the matter and citric acid lasts almost two days. The matrix was soaked in the citric acid twice to desolve the calcareous cement completely. After dissolving the calcareous cement, normal washing and screening methods were used. When fossils are soaked in a strong solution of citric acid (for example 25 weight percent) for one day, they are damaged by the acid. I used a weak solution to avoid damaging the fossils. GEOLOGY southern edge of the Big Horn Mountains and the southeastern end of the Owl Creek Mountains in 1978 SETOGUCHI— CEDAR RIDGE LOCAL FAUNA 7 Natrona County, Wyoming. All the fossil localities along Badwater Creek of both late Eocene and younger age lie to the south of Badwater Creek be- tween the creek and the Cedar Ridge fault. Love (1939) proposed the name Tepee Trail For- mation for a sequence of volcanic tuffs and flows of late Eocene age in the East Fork Basin near the southeast margin of the Absaroka Range. Tourtelot (1948, 1957) applied this name to a sequnce of vol- canic-rich sedimentary rocks along the southern margins of the Owl Creek and Big Horn Ranges. He subdivided the Tepee Trail Formation into two members, a lower Green and Brown Member and above this the Hendry Ranch Member. The Hendry Ranch Member defined by Tourtelot (1957) includes all the strata, which overlie the Green and Brown Member in the Badwater area. Black (1968, 1969) reported the discovery of late Oligocene vertebrates from a tan siltstone sequence at the top of the Hen- dry Ranch Member. During his study of the geology of the Badwater Creek area, Riedel (1969) recog- nized an unconformity within the Hendry Ranch Member between the gray and buff tuffaceous mud- stones from which late Eocene vertebrates were obtained and the tan tuffaceous siltstones and sand- stones from which late Oligocene vertebrates were obtained. He proposed to restrict usage of the Hen- dry Ranch Member to the gray and buff tuffaceous mudstones, especially excluding the tan tuffaceous siltstones and sandstones, which lie unconformably above them. The vertebrates discovered from the latter sediments are dealt with here. The gray and buff tuffaceous mudstone along Badwater Creek from which late Eocene verte- brates were found has been classified as the Tepee Trail Formation (Tourtelot, 1957). There is now al- most overwhelming evidence that these sediments do not represent eastern equivalents of the Tepee Trail Formation. Rather, these volcanic rich silts and clay were probably deposited at the same time as, or somewhat later than, the volcanic conglom- erates of the Wiggins Formation, which overlies the Tepee Trail Formation in its type area. This prob- lem is outside the scope of the present study. Be- cause a thorough geologic review of the area has not been completed as yet, I shall continue to follow Tourtelot’s usage. Tentatively, I assign the deposit to the Oligocene strata, which include the tan tuffaceous siltstones and sandstone from which late Oligocene vertebrates were obtained. The best exposures are in NE Sec. 24, T39N R89W where more than 200 ft of strata are exposed and an additional 500 ft are concealed by overlying Quaternary gravel and vegetation. The lower contact here is marked by an angular uncon- formity of approximately 2 degrees and erosional relief of up to 40 ft. The top of the unit is truncated by erosion throughout the area. The Oligocene strata and the Hendry Ranch Member of the Tepee Trail Formation are down faulted against the Green and Brown Member, or against the Lost Cabin and Lysite Members of the Wind River Formation along the course of Badwater Creek and the Cedar Ridge. This fault, called the Cedar Ridge fault, can he traced roughly from ESE to WNW throughout the area. A general geologic map of the Badwater Creek area is shown in Fig. 1 . A structural cross sec- tion near the late Oligocene fossil locality is shown in Fig. 3. Below the unconformity are the gray and buff tuffaceous mudstones of the Hendry Ranch Member of the Tepee Trail Formation. The erosional relief ranges up to 40 ft. The bottom of the relief (the point A in the Fig. 3) represents a stream channel. At the point A, conglomerates of medium pebble size are deposited. The maximum thickness of the conglomerates is 3 cm. The trans- verse (perpendicular to the stream course) exten- sion of the conglomerates is 7 m or so, and the layer is lenticular. High on the sides of the relief surface at point B, conglomerates are not observed, and here the fine-grained sandstones of the Oligocene strata lie directly on the tuffaceous clay of the Hen- dry Ranch Member. The relief is lower to the east of the old stream channel. Here the erosional relief is approximately 10 ft or less. At the point C, about 2 ft above the contact with underlying Hendry Ranch Member is a vitric tuff with a thickness of about 1 m. The tuff is traceable laterally so that it is used as a marker bed. Laterally, it thins and is about 0.5 m above point A. Geochron Laboratories, Inc. used potas- sium-argon ratios to determine the age of the vitric tuff at point C (Riedel, 1969). This age was given as 34 million years and falls within the lower Oligocene as delimited by Kulp (1961) and Evernden and oth- ers (1964). If the age determination is valid, it in- dicates that these lowest beds were deposited dur- ing early Oligocene time. Above the vitric tuff are siltstones and fine- grained sandstones about 24 m thick. Cross-bedding and truncate bedding are predominant in the fine- grained sandstones. A calcareously cemented lens of siltstone (Loc. 19) is found about 12 m above the 8 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 9 Fig. I. — Simplified geologic map along Badwater Creek, at Secs. 13, 14, 23, & 24, T39N, R89W, in Natrona County, Wyoming. Scale — 1:24,000. vitric tuff. A number of vertebrate fossils have come from this calcareous lens. Below the lens and above the vitric tuff are predominantly fine-grained sandstones, and the homogeneity of the sequence indicates that no depositional breaks are represent- ed within this sequence of sandstones. The verte- brate fauna is inferred to be of late Oligocene age. The radiometric determination of an age of 34 mil- lion years for the vitric tuff does not correlate well with the postulated age of the fauna. Although they are separated stratigraphically by 12 m, I could find no evidence for the presence of a depositional break, or a hiatus between the tuff and the fossil- iferous lens. I believe that the potassium-argon age determination is in error and is too old. I can not believe that the section preserved north of the Ce- dar Ridge fault consists of a thin sequence repre- senting early Oligocene deposition, a long deposi- tional hiatus and then the resumption of deposition in late Oligocene time. Fig. 2. — Cross section along A-A' shown in the Fig. 1. Horizontal scale — 1:8,000. Vertical scale — 1:6,000. 1978 SETOGUCHI— CEDAR RIDGE LOCAL FAUNA 9 -Erosion surface Oligocene strata Unconformity Fig. 3. — Stnictural cross section near the late Oligocene fossil locality, Loc. 19, at Sec. 24, T39N, R89W, in Natrona County, Wyoming. Horizontal scale — 1:5,000. Vertical scale — 1:2,000. A hard gray tuffaceous sandstone lies 1 m above the siltstones and sandstones and can be used as a marker bed. One and one-half m above the sand- stone there is a pebble to cobble sized conglomer- ate. The thickness is about 30 to 40 cm and its dis- tribution is quite wide. The conglomerate represents depostion on a flood plain. The direction of the cur- rent at the time of deposition of the conglomerates is determined by the direction of axes of elongated cobbles — from SW to NE. The pebble and cobbles consist of Precambrian gneisses and granites. sandstones and limestones of the Paleozoic, and re- worked siltstones of underlying Oligocene strata. The average diameter of the pebbles is 5 cm. Based on the direction of the current, uplift had taken place to the southwest of Badwater Creek. Rapid erosion of Precambrian and Paleozoic rocks and transportation toward the northeast cause the de- position of the conglomerates in Badwater Creek area. All the strata mentioned above dip gently south- ward (about 10°). FAUNAL LIST The following is the list of the amphibians, reptiles, and mam- mals identified in the Cedar Ridge local fauna. Class Amphibia Order Urodela Family Batrachosauroididae ? Batrachosauroides sp. Class Reptilia Order Sauria Family Iguanidae Leiocephalus sp. Class Mammalia Order Marsupicarnivora Family Didelphidae Peratherium sp. cf. P. spindleri Macdonald, 1963 Nanodelphys new species, unnamed Order Insectivora Family Leptictidae Lepiictis sp. Family Adapisoricidae Ankylodon sp. cf. A. annectens Patterson & McGrew, 1937 Family Geolabididae Centetodon sp. cf. C. marghudis (Cope, 1974) Family Soricidae Domnina sp. cf. D. gradata Cope, 1873 Family Talpidae Proscalops miocaenns Matthew, 1901 Oligoscalops ? sp. Family Micropternodontidae Micropternodus sp. Order Rodentia Family Aplodontidae Prosciurus relictus (Cope, 1973) Pelyconiys placidus Galbreath, 1953 Family Eomyidae Adjidaumo douglassi Burke, 1934 Paradjidaunio hypsodns, new species Meladjidaumo hendryi, new genus and new species Eomyidae, genus indet.. Type A Eomyidae, genus indet.. Type B Family Heteromyidae Proheteroniys sp. cf. P. nehraskensis Wood, 1937 Helisconiys sp. cf. H. veins Cope, 1873 10 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 9 Family Cricetidae Eiiniys parvidens Wood, 1937 Eiiniys elegans Leidy, 1856 Einnys brachyodus Wood, 1937 Eiunys sp. cf. E. planidens Wilson, 1949 Order Lagomorpha Family Leporidae Palaeolagus hurkei Wood, 1940 Palaeolagiis sp. cf. P. intermedins Matthew, 1899 Order Carnivora Family Canidae Hesperocyon temnodon (Wortman & Matthew, 1899) Order Perissodactyla Family Equidae Miohippus sp. Family Hyracodontidae Hyracodontidae genus indet. Order Artiodactyla Family Flypertragulidae Leptomeryx sp. near L. evcinsi Leidy, 1853 Hypisodus sp. near//, minimus (Cope, 1873) MAMMALIAN TAXA ABSENT FROM THE CEDAR RIDGE LOCAL FAUNA The Cedar Ridge local fauna is represented by a sufficient number of identifiable specimens to make it worthwhile to consider prominent absences from the known mammalian fauna. A good many forms are absent from the Cedar Ridge local fauna. The reasons for the absence of some taxa, I believe, should be considered from the following two points of view. First, some forms were surely living at the time of deposition in the area where the Cedar Ridge local fauna has been recovered, but for some reasons they were not fossilized there. Second, the other forms could not have lived in the Badwater Creek area so that they are not represented in the present fauna. The remarkable absence of larger forms, such as ungulates and carnivores, could be explained ac- cording to the first reason mentioned above. As I explained in the section of the paleoecological set- ting of the sites, the climate of the area during the late Oligocene was drier than the middle Oligocene and is thought to have been grassland, steppe to semidesert. This is supported by the presence of gypsum crystals throughout the section. Schultz and Falkenbach concluded that the leptauchine ore- odonts (tribe Leptaucheniini) lived in very arid, desert-like regions (Schultz and Falkenbach, 1968:407). They stated (1968:408), “The Leptauch- eniini apparently were well adapted to the unfavor- able climatic conditions of the late Oligocene and were able to survive in great numbers, but most of the other kinds of oreodonts either became extinct or lived in areas where the climate was more hos- pitable. Evidently the hyracodont rhinoceroses also had specialized in such a manner as to live in arid desert areas. Most of the other mammals must have found it difficult to survive, and migrated else- where, or lived along the banks of small streams that existed in the area at that time. The Leptauch- eniini were well adapted for dwelling in deserts. The same was apparently true of the hyracodonts.” I think that surely leptauchine oreodonts and hyra- codont rhinoceroses were living in the Badwater Creek area during the late Oligocene time. No ore- odonts are represented but a few unidentifiable rhi- noceros tooth and bone fragments are found in the fauna. The fossils found at Loc. 19 are mostly iso- lated teeth and jaw fragments. The nature of pres- ervation indicates that teeth and bones were trans- ported some distance after the death of animals and before the time of final burial. Moreover, the size of most teeth and bones found there are less than 3 mm in longest diameter, and specimens over 5 mm are very scarce. This type of preservation in- dicates that the material deposited were well sorted by stream action. The teeth and bones of larger size are not suitable for distance transport and were not deposited there. By this sorting mechanism, no ore- odonts were transported and deposited at the site where the Cedar Ridge local fauna was recovered. The occurrence of Hypisodus in this fauna is mean- ingful. A few teeth referable to Hypisodus are found teeth. The size of these teeth is at the upper limit of the size range of teeth found at Loc. 19. The species of Hypisodus are the smallest of known ar- tiodactyls, indicating that only the teeth of the size of the smallest artiodactyls could be transported. This is true for perissodactyls and carnivores. Gal- breath (1953) reported 10 genera of ungulates and carnivores from the Vista Member of the White River Formation of northeastern Colorado. In the present fauna, only three genera are represented and they are very scarce in comparison with the other smaller forms. The absence of some insectivores and rodents in 1978 SETOGUCHI— CEDAR RIDGE LOCAL EAUNA 11 the present fauna must be explained by unfavorable ecological conditions for them in the Badwater Creek area. Scottimus has been reported from the late Oligocene in the Great Plains (in Nebraska) but this form was not found in the present fauna. Lei- dymys and Paciculus are known from the late Oli- gocene of Oregon but they are not represented in the Badwater Creek area. Interestingly enough, they did not become extinct at the beginning of Miocene time but they continued to survive into the Miocene in the Rocky Mountain region, that is in Montana, Colorado, and Wyoming. Their ancestral forms are known in the middle Oligocene of the Great Plains. These forms all have lower-crowned teeth and probably lived in a more mesic habitat. Most small mammals in the present fauna have high- crowned teeth or teeth with high, thin cross-lophs. The trend towards higher-crowned teeth is perhaps an adaptation for a typical herbaceous diet in a dry environment. The animals, which required a more mesic habitat, are not represented in the present fauna. Martin (1972) clearly described the evolu- tionary pattern of cricetid rodents in relation to the climatic changes; the steppe forms became extinct near the Oligocene-Miocene boundary. This extinc- tion may be related to a return of mesic conditions and the subsequent expansion of the genera Lei- dymus Paciculus. This is the most probable ex- planation as to why Leidymys, Paciculus, and Scol- timus were not represented in the late Oligocene in the Rocky Mountain region but occur there in Mio- cene time. Oregon, from which Leidymus and Pa- ciculus are reported in the late Oligocene, was a AGE OF 1 As noted above, the Oligocene strata lie uncon- formably above the late Eocene clay. The overlying siltstones and sandstones are of the upper Oligo- cene age as indicated by the occurrence of the late Oligocene fauna. We obtained the K-Ar age for the Tuff A as 34.3 ± 1.4 m.y. This age falls into the early Oligocene as delimited by Kulp and Evern- den. The determination of the age for the tuff does not correlate well with the fauna. I believe that the K-Ar age given is too old. If it is correct, no middle Oligocene strata are represented in the Badwater Creek area. In the Wind River Basin and surrounding area, epeirogenic uplift began after Eocene time. For the Oligocene of Wyoming, van Houten (1964:71) stat- ed, “Reduced relief resulted in the slow accumu- costal area and was under a milder climatic regime throughout the Tertiary period. Scottimus is also supposed to have lived near stream banks during the late Oligocene and becomes abundant once the climate returned to a more mesic condition in the Miocene. The ecological requirements of Trimylus and Domnina are not certain. Trimylus has a more bulbous, less high crowned condition of the teeth (Repenning, 1967) than Domnina. The presence of Domnina and the absence of Trimylus in this fauna seems to be analogous to the presence of rodents with higher-crowned and the absence of those with lower-crowned teeth. The relative abundance of Eumys planidens, which has more lophate teeth, may be explained in the same way. All the eomyids and the most of the cricetine E'/z/nyi' species became extinct without leaving any descendants. 1 believe that pseudotheridomyine rodents known from the Miocene of North America are not descended from the North American Oligocene eomyids but repre- sent immigrants from the Old World. The extinction of many eomyids and cricetids in North America by the end of the Oligocene may be related to a return of mesic conditions as Martin (1972) stated. The Cedar Ridge local fauna represents a fauna adapted to a drier environment and is biased due to sorting by stream action. For these reasons, a good many forms are absent from the fauna. This leads to the conclusion that currently sampled Whitneyan deposits in the Badwater Creek area do not pre- serve the real diversity of the late Oligocene mam- mals. : STRATA lation of stream-laid deposits; however, numerous showers of ash from vents in the Yellowstone-Ab- saroka volcanic field contributed a substantial amount of sediment. Altered ash that mantled the uplands supplied most of the mud that was spread as an almost continuous sheet on flood plains over much of Wyoming and the adjacent Great Plains.” These sediments of the Oligocene consist of the White River Formation in Wyoming and Colorado, or the White River Group of the Great Plains. The Oligocene strata along Badwater Creek are separated from the Eocene strata by only a modest angular (about 2°) and erosional unconformity. This indicates that no severe orogenic disturbance was represented in this region during the period of time between the time of deposition of the Hendry 12 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 9 Ranch Member and that of the Oligocene strata. I believe that volcanic ash from vents in the Yellow- stone-Absaroka volcanic field were deposited in this region, but that these sediments were eroded and the erosion acted on the clay of the Hendry Ranch Member. By late Oligocene time, the drain- age system was changed and upper Oligocene strata were deposited along this new drainage system. These sediments were first accumulated as channel fillings and later as flood plains deposits. Along Badwater Creek, no lower and middle Oligocene strata correlative to the White River For- mation are observed. A sequence of sediments of early Oligocene age occurs along Beaver Rim on the southern edge of the Wind River Basin, 50 mi southwest of the Badwater Creek area. The sedi- ments along Beaver Rim are referred to the White CORRELATION In the Beaver Rim area, the White River For- mation is well developed. The thickness of the for- mation varies locally; the formation reaches its maximum exposed thickness of approximately 650 ft along the Beaver Rim in the vicinity of Cameron Springs and may have been as much as 800 ft thick 2.5 mi north of the divide (van Houten, 1964:55- 56). In this area, the White River Formation yields mammalian fossils of early (Chadronian) and pos- sibly of middle (Orellan) Oligocene age. Van Hou- ten (1964:71) concluded that, “There is no clear evi- dence of deposits of late Oligocene age in the southern part of the Wind River Basin.” The Oligocene strata of the upper Oligocene along Badwater Creek are not directly correlative with the White River Formation in the Beaver Rim area. As far as I know the Vista Member of the White River Formation in Logan County, north- eastern Colorado (Galbreath, 1953) and the Whit- ney Member in Nebraska and South Dakota are the only known deposits, which yields vertebrate fos- sils of late Oligocene (Whitneyan) age. These beds are composed of massive, tan silt with a highly cal- careous zone. The thickness is about 100 ft and the areal extent probably is not large (50 to 75 square River Formation (van Houten, 1964). Once the low- er part, at least, of the Oligocene strata of the Bad- water Creek were believed to be of the early Oli- gocene age because of the age of the Tuff A given as 34 m.y. and because the lower Oligocene se- quence occurs along Beaver Rim (Black, 1969:45). As I stated before, I could not find any evidence of a break in deposition, or hiatus between Tuff A and the calcareous lens (Loc. 19 of Carnegie Museum of Natural History) from which the late Oligocene vertebrates are obtained. For these reasons, I consider the potassium-ar- gon age determination is too old. All the sediments were accumulated in the late Oligocene as indicated by the occurrence of vertebrates of the late Oligo- cene age. OF THE STRATA mi). The Vista Member can be distinguished fau- nally and lithologically, but it should be emphasized that the lithologic separation from the (underlying) Cedar Creek Member of the White River Formation is largely arbitrary. Were the fauna not known, the lithologic differences would have no stratigraphic significance. Although the Vista fauna is scanty, the fossils are individually rare, the late Oligocene Badwater fau- na is correlative with the Vista fauna. Lithological- ly, the well laminated appearance of the Oligocene strata in the Badwater Creek area is quite different from the massive nature of the Vista Member in Colorado. It is not certain whether the upper Oli- gocene deposits in the Badwater Creek area and in northeastern Colorado accumulated in different structural basins or not, but if they were deposited in different basins, they should be assigned to dif- ferent formations. This problem is beyond the scope of the present work. Although I believe that the Oligocene strata along Badwater Creek may be assigned to the White River Formation but to a dif- ferent member, I leave it unnamed until a thorough geologic study of this area is completed. FAUNAL AGE The late Oligocene, or Whitneyan, mammalian faunas of North America are not well known. An age determination for a micromammalian assem- blage such as the present fauna is complicated by the fact that we have only the haziest idea as to other late Oligocene small mammal faunas. Of some 1978 SETOGUCHI— CEDAR RIDGE LOCAL EAUNA 13 40 genera of mammals known from the upper part of the Brule Eormation in Nebraska, South Dakota, and Colorado, and considered to be late Oligocene in age, only one-fourth are in the “micro” mammal range. Galbreath (1953) described several mammals of the late Oligocene from the Vista Member of the White River Formation of northeastern Colorado. But the Vista fauna is scanty, and the fossils are individually rare. A late Oligocene age for the present fauna is based upon the following data. (1) Marsupials. — Peratheriinn and Nanodelphys are typically early Tertiary genera ranging from the mid-Eocene to the early Miocene. As far as the spe- cific level is concerned, the species of Peratherium and Nanodelphys are intermediate between the middle Oligocene and the early Miocene forms. Based on marsupials alone, it is fairly safe to con- clude that the age of the fauna is post-middle Oli- gocene and pre-early Miocene. (2) Insectivores. — Ankylodoii, Centetodon, and Micropteniodus are typically Oligocene genera, al- though Centetodon and Micropteniodus are known from the early Miocene of Nebraska and Oregon. Proscalops and Domnina are known from the mid- dle Oligocene for the former and from the late Eocene for the latter into the early Miocene. Of the five genera, one is known only from the Oligocene and four from the mid-Oligocene into the early Mio- cene. At the specific level, all the species of the present fauna are a little advanced over the middle Oligocene species and are not conspecific with any of the known early Miocene species. This is also suggestive of a late Oligocene, or at least pre-Mio- cene, age for the fauna. (3) Lagomorphs. — The occurrence of Palaeola- gus intermedins is not well documented. The other species of Palaeolagus is intermediate between P. hurkei of the middle and late Oligocene and P. liyp- sodus of the earliest Miocene, although it is a little closer to the former. P. hypsodus is known from the Gering Formation in Nebraska and Wyoming and from the Sharps Formation of South Dakota both considered to be basal Miocene in age. The present population is not as advanced as this species and would therefore suggest a latest Oli- gocene age for the fauna. (4) Rodents. — Pely corny s is known from the Oli- gocene. At the specific level, this form is close to the middle Oligocene species. Adjidaumo and Par- adjidaumo are typically Oligocene genera. At the specific level, Paradjidaumo in the present fauna is advanced over the middle Oligocene Paradjidau- mo. It is suggestive of a post mid-Oligocene age for the fauna. The other eomyids differ from the typical Oligocene Adjidaumo and Paradjidaumo genera but are certainly closely related to them and show no resemblance to the Miocene Pseudotlieridomys . Prosciurus, Heliscomys, Proheteromys, 'dod Eurnys are all known from the earliest Miocene but are more typical, abundant, and diverse in the Oligo- cene. The absence of beavers, mylagaulids, and aplodontids (beside Prosciurus) from the present fauna is meaningful. Their absence most probably reflects a different habitat preference. These three rodent families are almost always found in faunas of the early Miocene. They are known from the earliest Miocene of South Dakota, eastern Wyo- ming, and Nebraska but are unknown from the late Oligocene of these same areas. PALEOECOLOGICAL SETTING The Oligocene strata were accumulated as chan- nel-fill. The lower part of the sequence seems to be one of over-filling of the previously eroded valley. Aggradation was greater than erosion. Granites and gneisses, sandstones and limestones, and mud balls from the underlying Hendry Ranch beds are com- mon in the gravel lenses near the base of the unit. As sedimentation proceeded, the previously eroded valley was filled by sediments, and sediments were deposited on a broader flood plain. Well sorted, laminated and truncated sandstones indicate that sedimentation took place in a braided stream situ- ation. Paleosols are not recognized in the Oligocene strata. Andesine is the most common feldspar throughout the section. In a calcareous lens from which the late Oligocene vertebrates are found, by- townite and probably anorthite are present. X-ray diffraction patterns of the sample from the same calcareous lens also indicate that feldspars pre- served are calcic (Hattori, personal communica- tion). The presence of fresh, calcic feldspar indi- cates that the sediments were not heavily weathered, and the paleoenvironment should have been open land or grassland type with moderate temperature. This conclusion is strengthened by the presence of 14 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 9 gypsum crystals throughout the section. The tex- ture of gypsum crystals is fine to medium grained. They occur in beds with disturbed bedding. This seems to be due to expansion during hydration. Gypsum is usually formed at lower temperatures, whereas anhydrite is precipitated at temperatures above 30° C. The presence of evaporites indicates the climate was rather dry at the time of deposition. This is consistent with other evidence that the cli- mate became progressively drier towards the late Oligocene in the Great Plains (Schultz and Falken- bach, 1968). The origin of calcareous lenses is uncertain. The siltstones and sandstones are essentially noncalcar- eous throughout the section. In the Great Plains near the type locality of the Whitney Member, most of the fossils found in the Middle and Upper Whit- ney sediments are enclosed in clay-siltstone nod- ules, which are cemented by calcium carbonate, and the nodules must have been formed by the ac- tion of ground water (Schultz and Falkenbach, 1968:408). This is not the case for the Oligocene sediments along Badwater Creek. There, calcar- eous lenses are rare and only two calcareous lenses have been found so far. Both of them are fossilif- erous. The non-calcareous nature of the siltstone and sandstone may argue against the process of the formation of the calcareous lenses by the action of ground water. The calcium carbonate may have been accumulated in small playa lakes on the flood plain and vertebrate bones and teeth were trans- ported in these lakes. The vertebrate remains are well preserved because the lenses are resistant to weathering or to stream action. Vertebrate fossils are found only in the calcareous lenses, and are not found in the siltstones and sandstones. The sedimentary structure through the whole sec- tion is predominated by the high degree of sorting, cross-bedding, and truncated bedding. All the sand- stones are fine-grained and pebble size grains are observed only in conglomerate layers. The sedi- ments were accumulated under running water with rather low energy. The fossil teeth and bones found at Loc. 19 are isolated and abraded, and the diam- eter of these are mostly less than 3 mm. No bones in articulation are found. The nature of preservation of the fossils indicates that these teeth and bones were transported some distance before final burial by running water, and due to the energy of running water the materials transported were well sorted. Only a few larger bones were deposited near Loc. 19. This sorting mechanism of stream action greatly influenced the kinds of mammals buried and pre- served at Loc. 19. The larger mammals, if they lived near the site of deposition, would have been not easily transported and buried there after their death. During the middle Oligocene, the White River Formation was developed over a vast area east of the ancestral Rocky Mountains in Wyoming, Col- orado, South Dakota, and Nebraska. These depos- its are typically flood-plain sediments accumulated under a climatic regime of relatively high precipi- tation. Toward the end of Oligocene time, appar- ently precipitation became reduced and the aerial development of the White River Formation also be- came reduced. The aerial distribution of the late Oligocene sediments is greatly restricted and rep- resented only in a few areas — the Vista Member in Colorado and the Whitney Member in Nebraska and South Dakota of the White River Formation. This restriction was caused by the reduced precip- itation and subsequent reduced drainage systems in the Great Plains region. This interpretation is con- sistent with the climate becoming progressively drier in the late Oligocene (Schultz and Falkenbach, 1968; van Houten, 1964). The vegetation in the late Oligocene must have been of a steppe-type. In re- lation to the restricted distribution of the late Oli- gocene sediments, the late Oligocene mammal fau- nas are not well known except for the one from the type area of the Whitney Member of the White Riv- er Formation in Nebraska. The latter fauna is com- posed mostly of oreodonts. In the early Miocene, climatic conditions appar- ently changed again to a period of more precipita- tion. The Arikaree group in Nebraska was depos- ited as channels cut through to the underlying middle Oligocene series in most areas. These sedi- ments are mostly of the channel-filling type initially. The extensive development of the lower Miocene series reflects the return of the climate to a more mesic condition with greater precipitation and re- sultant development, or rejuvenation of drainage systems. The early Miocene must have been con- siderably more humid and the vegetation more lux- uriant. The climatic and environmental changes near the Oligocene-Miocene boundary caused a great differ- ence in the composition of faunas between the late Oligocene and the early Miocene. 1978 SETOGUCHI— CEDAR RIDGE LOCAL EAUNA 15 SYSTEMATIC ACCOUNTS Class Amphibia Order Urodela Family Batrachosauroididae Auffenberg, 1958 ? Batrachosauroides sp. (Fig. 4) Referred specimens. — Vertebrae, CM 33649 and uncatalogued specimens. Locality. — Loc. 19, Badwater Creek, Wyoming. Age. -Late Oligocene. Discussion. — The vertebra is amphicoelous. When Taylor and Geese (1943) established a new salamander genus Batrachosauroides from the Mio- cene of Texas, they determined that the vertebrae were amphicoelous and not opisthocoelous. The present specimen may be referred to this genus. Auffenberg (1958:170-171) argued about the genus Batrachosauroides and stated that the vertebrae re- ferred to B. dissimulans are all strongly opisthocoe- lous. At present, I cannot evaluate which statement should be more reasonable. Tentatively I followed Taylor and Hesse and I refer the present specimen to that genus. Class Reptilia Order Sauria Family Iguanidae Leiocephalus sp. (Fig. 5) Referred specimens. — Jaws; CM 33650 and uncatalogued specimens. Locality. — Loc. 19, Badwater Creek, Wyoming. Age. — Late Oligocene. Description and discussion. — The pleurodont teeth have tall, slim, and straight-sided shafts, ex- cept for a slight flaring toward the crown. The crown is flattened linguobuccally into a narrow tri- cuspid fan-shaped structure, the central cusp larg- est. Each side cusp is prominently separated from the main cusp by a wide groove, which fades out at the base of the crown. These grooves lack an as- sociated ridge. The present form may well be compared with Leiocephalus sp. described by Estes (1963:239) from the early Miocene Thomas Farm local fauna of Florida. As in Florida specimens, the grooves on the crown which separate side cusps from the main cusp lack an associated ridge, seen in many such lizard teeth, which extends from the apex of the lateral cusp to the base of the crown. Class Mammalia Infraclass Metatheria Order Marsupicarnivora Family Didelphidae Gray, 1821 Peratherium sp. cf. P. spindleri Macdonald, 1963 (Fig. 6, Table 1) Referred specimens. — P'“-M'; CM 17085; M'-M'*: CM 33404; M--M^: CM 33439; M': CM 33404-33438; M^: CM 33440-33462; M^: CM 33463-33479; M^; CM 33541-33544; M.-M^: CM 33480; Mj-M,: CM 17080, CM 33481; M,; CM 33482-33500; M,; CM 33501-33520; M,.,; CM 33521-33536; M^: CM 17082, CM 33537- 33540. Locality. — Loc. 19, Badwater Creek, Wyoming. Age. — Late Oligocene. Description . — The size of teeth is smaller than that of Pera- therium ftiga.x. Except for size, the morphology of the present form agrees exactly with that seen \n P . fitga.x . On M', the meta- conule is not well defined. The posterior walls of the protocone and the metacone form a continuous posterior border of the tooth. Stylar cusp C is prominent and relatively high. Unlike P. knighii, the buccal border of the crown is concave as in P. fugax. In relation to this, stylar cusp C is situated more lingually in the present form and in P. fugax than in P. knighti. On MC the metaconule is prominent and expanded posteriorly. The poste- rior walls of the protocone and the metacone meet at an angle. The buccal border of the stylar shelf between stylar cusps B and C is concave as in P. fugax. On M^, the concavity of the buccal border of the stylar shelf is more exaggerated than on M' and M^. The metaconule is less prominent than on M^. The morphology of lower molars is very similar to that seen in P. fugax and P. knighti. The buccal cingulum is slightly more prominent than in P. fugax. but this character is variable. I can- not find any morphological characters other than size on the lower molars to separate this species from other species of Per- atherium. On M3, the paraconid is elongate transversely and the lingual end of it is situated more lingually than in P. fugax. Be- cause of this character, the width of M3 is slightly greater than the corresponding tooth of P. fugax. But this character is not always true for all specimens of M3; some specimens have a normal paraconid. Moreover, this character is seen also in a few specimens of P. fugax from the middle Oligocene. Discussion. — Macdonald (1963) described P. spindleri from the lower Miocene of the Wounded Knee Area, South Dakota. The diagnosis of this new species oi Peratherium given by him ( 1963: 164) is as follows: “of medium size; strongly developed anterior and posterior cingula; labial cingulum con- tinuous.” The holotype of P. spindleri is signifi- cantly smaller than P. fugax. The size of the Bad- water specimens is intermediate between P. fugax and P. spindleri. Badwater specimens show a wide range of morphological variation of the develop- ment of cingula. On some specimens, the develop- ment of cingula is weak so that the anterior cingu- 16 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 9 1978 SETOGUCHI— CEDAR RIDGE LOCAL EAUNA 17 Table 1. — Dimensions of teeth of Peratherium sp. cf. P. spindleri and unnamed new species of Nanodelphys. M> Mb' M'* M‘ M'' Statistics L W L W L W L W L w L w N 36 36 25 Peratherium sp. cf. P. spindleri 25 19 19 20 20 23 23 17 17 OR 1.62- 1.73- 1.66- 1.81- 1.68- 2.21- 1.66- 0.88- 1.78- 1.00- 1.74- 0.99- 2.00 2.22 2.08 2.43 2.09 2.65 1.91 1.12 2.09 1.23 2.07 1.27 Mean 1.788 1.944 1.836 2.098 1.887 2.433 1.798 1.016 1.925 1.119 2.005 1.181 SD 0.1226 0.1279 0.1175 0.1781 0.1133 0.1403 0.0700 0.0691 0.0876 0.0675 0.1004 0.0870 CV 6.86 6.58 6.40 8.49 6.01 5.77 3.89 6.80 4.55 6.03 5.01 7.37 Nanodelphys new species N 16 16 17 17 18 18 8 8 25 25 24 24 OR 1.18- 1.30- 1.18- 1.48- 1.12- 1.52- 1.20- 0.62- 1.23- 0.67- 1.24- 0.70- 1.46 1.77 1.52 1.77 1.32 1.74 1.64 0.81 1.53 0.90 1.47 0.87 Mean 1.319 1.444 1.355 1.604 1.243 1.612 1.355 0.735 1.348 0.768 1.375 0.796 SD 0.0620 0.1202 0.0787 0.0910 0.0548 0.0683 0.1496 0.0659 0.0679 0.0537 0.0541 0.0456 CV 4.70 8.33 5.81 5.68 4.41 4.24 11.04 8.97 5.04 7.00 3.94 5.73 lum does not unite with the buccal cingulum. On others, the anterior cingulum continues to run pos- teriorly along the buccal base of the protoconid and unite with the buccal cingulum. The latter character is also seen in a few specimens ofP.fugax from the middle Oligocene. I do not agree that the greater development of cingula is a good criterion to sep- arate species of Peratherium. Size is the only cri- terion, which separates P. spindleri from P. fitgax. Nanodelphys new species (Fig 1., Table 7) Referred specimens. — CM 33549, and broken My CM 33600; M^^-My CM 33601; Mb CM 33602-33616; My CM 33617-33632; My CM 33628, CM 33633-33647; My CM 33648; P2-P3: CM 19806; M^-M.,: CM 19805, CM 33553; M2-M4: CM 33554; M3-M4: CM 17081, CM 19803, CM 21697; M,: CM 33545-33552; Mj: CM 19804, CM 33555-33575; M3; CM 33576- 33594; M^: CM 19802, CM 33595-33598. Locality. — Loc. 19, Badwater Creek, Wyoming. Age. — Late Oligocene. Description. — The general morphology of upper molars re- sembles that of N. minutus. The paracone and the metacone are somewhat more appressed anteroposteriorly. The posterior ridge of the paracone and the anterior ridge of the metacone are more sharp and well defined than in N. minutus. These two ridges are as strong as the parastylar or metastylar crests in the present form, whereas in N. minutus these ridges are clearly weaker than the parastylar and metastylar crests. This condition indicates that the upper molars of this species of Nanodelphys approach the dilambdodont tooth pattern. In the lower molars, the postcristid and the hypoconulid block the talonid basin posteriorly. The crista obliqua is straight, not concave as in N . minutus and the hypoflexid does not excavate the talonid basin bucally as in N. minutus. The hypoconulid is displaced more lingually than in N. minutus and situated pos- terior ti he entoconid. The postcristid is almost transverse and forms an acute ridge. The hypoconulid is lower than the ento- conid but still higher than the talonid basin. The posterior cingu- lum is very weak and terminates lingually at the base of the hypoconulid. The posterior cingulum does not unite with the occlusal surface of the hypoconulid. Discussion. — Upper molars of the present form approach the dilambdodont condition. Nanodel- phys minutus has upper molars with less well-de- veloped dilambdodont tooth pattern, as I discussed elsewhere (Setoguchi, 1975). In animals with di- lambdodont upper teeth, the lower molars have a hypoconid with a very sharp buccal angle that oc- cludes with the ectoloph of the upper teeth. These teeth either lack a hypoconulid or have this cusp displaced (Robinson, 1968). In Nanodelphys minu- tus the displacement of the hypoconulid is less em- phasized in the lower dentition. The hypoconulid is situated posterobuccal to the entoconid in N. mi- nutus. In the present form, the displacement of the hypoconulid is emphasized; this cusp is situated posterior to the entoconid. This condition is very Fig- 4. — IBatrachosauroides sp. CM 33649, vertebra. x8. Fig. 5. — Leiocephalus sp. CM 33650, jaw with teeth, x 10. Fig. 6. — Peratherium sp. cf. P. spindleri. a; CM 33439, left M^-M^. b: CM 33481, right M2-M3. x8. Fig. 7. — Nanodelphys new species, unnamed, a; CM 33600, left M^-M®. b: CM 19805, right M2-M3. x 10. Fig. 8. — Leptictis sp. CM 21676, right M2. x8. Fig. 9a. — Ankylodon sp. cf. A. annectens. CM 33658, right M^. x8. 18 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 9 Table 2. — Dimensions of teeth <>/ Leptictis i/7. and Ankylodon cf. A. annectens. Statistics M' M^ F ’4 M, M3 L AW PW L AW PW L W L AW PW L AW PW Leptictis sp. N 1 1 1 OR 3.46 2.77 2.34 Ankylodon sp. cf. A. 1 annectens N 1 1 2 1 1 2 2 2 2 2 1 1 1 OR 1.72 3.11 2.57- 0.98 2.08 1.76- 1.26- 1.74- 1.53- 1.47- 1.87 1.57 1.29 2.66 1.77 1.32 1.85 1.66 1.65 similar to the talonid structure of Peratheriiim. But in the lower dentition of Peratheriiim, the hypocon- ulid forms a horizontal, posteriorly directed ledge directly posterior to the entoconid, and the talonid basin opens posteriorly across the flattened hypo- conulid. Moreover, in Peratheriiim, the posterior cingulum is well developed; this cingulum unites lingually to the buccal face of the hypoconulid so that the hypoconulid and the posterior cingulum are a continuous structure. On the other hand, in the present form of Nanodelphys, the hypoconulid and the postcristid block the talonid basin posteriorly, and the posterior cingulum does not form a contin- uous structure with the hypoconulid. Dr. Larry D. Martin at the University of Kansas (personal communication) informed me that he has several specimens of Nanodelphys from the lower Miocene Gering Eormation, Nebraska, that are not separable from the present form at the specific level. He has better specimens and he will give the diagnosis of the new species so that I leave the new species unnamed. Infraclass Eutheria Order Insectivora Eamily Leptictidae Gill, 1872 Leptictis sp. (Fig. 8, Table 2) Referred specimens. — Mj: CM 21676. Locality. — Loc. 19, Badwater Creek, Wyoming. Age. — Late Oligocene. Description: — The trigonid is wider transversely than the tal- onid. The specimen available is worn. Apparently, the paraconid is reduced to crest-like shape and it gives the tooth a more quad- rate outline in occlusal view. The apices of the protoconid and metaconid are nearly opposite each other. The hypoconid is stout and the crista obliqua joins to the posterior wall of the trigonid a little buccal to the midpoint of it. The entoconid is small and is situated a little posterior to the hypoconid. Between the metaconid and the entoconid a wide and deep valley opens lingually. On the lingual margin of the floor of the valley, a small cuspule is present. Discussion. — Only one specimen referable to Leptictis is available in the present fauna. Leptictis i—Ictops) is the most varied and individually the most abundant genus of the family in the White River Oligocene (Scott and Jepsen, 1936:13). Lep- tictis in the present fauna is the youngest occur- rence of the genus. But, it is difficult to give the specific identification for this material because the only specimen available is worn. Family Adapisoricidae (Schlosser, 1887) Ankylodon sp. cf. A. annectens Patterson and McGrew, 1937 (Fig. 9, Table 2) Referred specimens. — M': CM 21675; M^: CM 17095, CM 33651, CM 33652, CM 33653; M^; CM 21699; P4: CM 33654, CM 33655; M,: CM 33656, CM 33657; M2; CM 33658; M3; CM 33659. Locality. — Loc. 19, Badwater Creek, Wyoming. Age. — Late Oligocene. Description. — Materia! referable to Ankylodon is known from the late Eocene to the middle Oligocene. The late Eocene An- kylodon is known only from a single M', a single Mj, and some fragmentary materials (Krishtalka and Setoguchi, 1975). An ex- cellent, nearly complete palate with complete jaws of both sides is known from the Chadronian. Ankylodon annectens (including A. progressns; see Krishtalka and Setoguchi, 1975) is known only from the lower dentition. Like the Chadronian Ankylodon, M^ of the present form is transverse and has a strong hypocone. The paracone is tall and transverse. The metacone is reduced in size and in height when compared with the Chadronian species. The postmetacrista and the metastylar area are also weaker than in the Chadronian form. In relation to these, the posterior wing of the metaconule ter- minates nearly at the posterior base of the metacone, whereas in the Chadronian specimen the posterior metaconule wing ex- tends along the posterior base of the matacone and joins the posterolingua! base of the postmetacrista. The parastylar area expands buccally but it is narrower anteroposteriorly than in the 1978 SETOGUCHI— CEDAR RIDGE LOCAL FAUNA 19 Table 3. — Dimensions of teeth <>/ Centetodon sp. cf. C. marginalis and Domnina cf. D. gradata. Sta- p4 M' M2 P4 M, M2 M, tistics L W L AW PW L AW PW L W L AW PW L AW PW L AW PW Centetodon sp. cf. C. marginalis N 1 1 1 1 1 1 2 2 1 1 1 2 2 2 OR 1.69 2.18 — 2.22 — 1.38 2.47 2.27 1 .55- 0.88- 1.86 1.21 1.13 1.54- 0.98- 0.80- 1 .64 1.01 1.66 1.06 0.81 Domnina sp. cf. D. gradata N 2 2 3 3 3 3 3 3 4 4 4 2 2 2 2 2 ') OR 1.97- - 1.98- 1.99^ 2.05- 2.15- 1.89- 2.04- 1.80- 2.12- 1.21- 1.31- 1.81- 1.14- 1.16- 1.49- 0.94- 0.80- 1.99 2.00 2.08 2.17 2.22 1.97 2.07 1.92 2.20 1.47 1.58 1.96 1.38 1.42 1.52 0.96 0.85 Mean 2.043 2.097 2.193 1.920 2.057 1.860 2.170 1.353 1.490 Chadronian form. In this regard the stylar shelf area is narrower anteroposteriorly than the Chadronian species. CM 21699 is identified as M^. The metacone is the only prom- inent cusp. It is conical and is situated on the middle of the buccal half of the crown. The parastylar area expands antero- buccally but the preparacrista is represented by a weak ridge on the anterobuccal face of the paracone. The stylar shelf is trun- cated and narrow posteriorly. The metacone is greatly reduced. The protocone forms a triangle with the acute angle lingually. The hypocone is greatly reduced and represented by an enamel crenulation. P4 is molariform. It is basially similar in construction with its homologue in Geolabis, but is proportionately much shorter and broader. In the Chadronian species of Ankylodon, the paraconid is a forward-leaning, short transverse ridge, but it is still a well- defined cusp. In the present form, the paraconid is not a distinct cusp but just an anterior continuation of the anterolingual ridge of the protoconid, which lingually slopes sharply ventrad. The protoconid is taller than the metaconid. The metaconid is elon- gated lingually so that the protoconid and the metaconid are more widely separated from each other than in the Chadronian species and the Orellan Ankylodon annectens. A short weak anterior cingulum is present on the anterior face of the proto- conid. The talonid structure agrees exactly with that of the Chad- ronian species and the Orellan A. annectens. On M,, the trigonid is a little narrower transversely than the talonid as in the Orellan A. annectens. whereas the former is clearly narrower than the latter in the Chadronian form. No cin- gula are present at all. The metaconid is taller than the proto- conid as in the other forms of Ankylodon. The talonid is shorter than that in the Orellan A. annectens and more so than the Chad- ronian form. No trace of the hypoconulid is present whereas a rudimentary hypoconulid is clearly present in the Chadronian and the Orellan forms. The entoconid is taller than the hypoconid as in the other forms of Ankylodon. but the difference in height between the metaconid and the entoconid is more exaggerated in the present form than in the Orellan form and more so than in the Chadronian form. The notch between the metaconid and the entoconid is deeper than in any other forms of Ankylodon. CM33658 is identified as M2 because the trigonid is slightly wider transversely than the talonid. No cingula are present. The talonid is narrower anteroposteriorly than in the other members of this genus, and the crista obligua is concave. No trace of the hypoconulid is seen, whereas in the late Eocene form there is a distinct hypoconulid. The notch between the metaconid and the entoconid is also deep, whereas in the late Eocene species the entocristid joins these two cusps and closes the talonid basin lingually. On M;j, the talonid is narrower transversely than the trigonid. The hypoconulid is present although rudimentary. Discussion. — The present form is clearly differ- ent from the Chadronian species in having a more reduced parastyle and narrower stylar area on the upper molars, more reduced paraconid, no cingu- lum, deeply separated metaconid and entoconid with the former clearly taller than the latter, and transversely narrower talonid. In these respects, the Orellan Ankylodon annectens is intermediate between these two forms. The differences between the present form and the Chadronian species cited above are more clearly exaggerated if my material is compared with the late Eocene form. The present material is more advanced or specialized than either the late Eocene or the Chadronian species. But the distinguishing features listed above, or the morpho- logical differences between the present form and the Orellan A. annectens, seem minor and possibly are within the range of usual variation within a species. Family Geolabididae (McKenna, 1960) Centetodon sp. cf. C. marginalis (Cope, 1874) (Fig. 10, Table 3) Referred specimens. — P"*: KU 16606; M‘: CM 33661; M^; CM 33662; P.,; CM 21673; P4-M2; CM 33663; P^: CM 21674, CM 33664; M2-M,,; CM 21672, M..,: CM 33665, CM 33666. Locality. — Loc. 19, Badwater Creek Wyoming. Age. — Late Oligocene. Description. — The structure of P^ closely resembles that in Centetodon marginalis and C. chadronensis (Lillegraven and McKenna, manuscript). A small anterior lingual cingulum is present on the anterolingual base of the protocone. The width 20 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 9 of this cingulum is almost half that of posterior cingulum. The anterior lingual cingulum is not present on the Chadronian Cen- tentodon chadronensis, but on C. marginalis a small anterior cingulum is present on the base of the anterior face of the pro- tocone. It is widely separated from the posterior cingulum. On the present form, the anterior lingual cingulum is shifted more lingually and almost united with the posterior cingulum along the base of the lingual face of the protoconid. In this connection, the lingual base of the crown of P* is wider anteroposteriorly than in Centelodon chadronensis and C. marginalis. The pos- terior cingulum has a slight elevation on the posterolingual base of the protocone and a low, small cingulum runs anteriorly along the lingual base of the protocone. That elevation is more buccal on C. marginalis and directly posterior to the protocone on C. chadronensis. The lingual root is not bifurcated on the material at hand. Distinctions in the upper molars between Centelodon margin- alis and my specimens are minor. A tendency toward a deeper labial emargination of the stylar shelf on is observed in this form, but the general morphology is essentially the same as in C. marginalis. P4 is a large semimolariform tooth and its structure is exactly the same as the corresponding tooth of Centelodon marginalis. Moreover, there are no really obvious morphological differences on lower molars between C. marginalis and the present speci- mens). The mental foramen is below the posterior root of P3. Discussion. — The general morphology of my ma- terial is essentially the same as that of Centelodon marginalis. The only obvious difference is in the degree of the development of the anterior lingual cingulum on P. The anterior lingual cingulum is more developed in the present form than in C. mar- ginalis. Lillegraven and McKenna (Manuscript) clearly describe the evolutionary sequence of Cen- tetodon chadronensis -C. marginalis. In C. chad- ronensis there is no anterior cingulum on P, but a small anterior cingulum is usually present on the base of the anterior face of the protocone in C. mar- ginalis. Even in the latter form, the anterior cingu- lum is widely separated from the posterior cingu- lum at the lingual margin of the protoconid. In the present form, the anterior cingulum is shifted more lingually and closer to the posterior cingulum. This is just a continuation of the sequence toward better development of the anterior cingulum on P in the Centelodon chadronensis -C. marginalis lineage. The present form is surely in this lineage and is descended from the Orellan C. marginalis. A new species of Centelodon terminalis will be described by Lillegraven and McKenna (manu- script). The materials referred to the new species were originally discussed by Martin (1972) from the lower Miocene Gering Formation of Nebraska. This form is known only from P4 and Mj. Morphologi- cally, however, it shows no obvious differences from homologous teeth of C. marginalis except for their size. The size of the tooth suggests an animal significantly larger than Centelodon marginalis yet smaller than C. wolffi. The trends in proportional changes seen in the transition from C. chadronensis to C. marginalis seem to continue into C. germin- alis. The size of the present form, which is inter- mediate between C. marginalis and C. terminalis, supports this hypothesis. Family Soricidae (Fischer von Waldheim, 1817) Domnina sp. cf. D. gradata Cope, 1873 (Fig. 11, Table 3) Referred specimens:— CM 33667; CM 21662, CM 33668, CM 33669; M': CM 21663, DM 33670, DM 33671, DM 33672; M^; CM 33673, CM 33674, CM 33675, CM 33676, CM 33677; P,-P3, M,; CM 33678; M.-Mj; CM 33679; M,: CM 21664, CM 33680, CM 33681, CM 33682, CM 33683, CM 33684; M2-M3: CM 33685; Mj: CM 21665, CM 33686; M3: CM 33687. Locality. — Loc. 19, Badwater Creek, Wyoming. Age. — Late Oligocene. Description. — P has a tall paracone that tapers posterobuc- cally to the metastylar tip of the crown. The anterior wall of the paracone is very steep as in the other species of Domnina. A low parastyle runs anteriorly from the base of the paracone. The protocone is weak. The hypoconal shelf expands posteriorly, unlike the condition in Trimylus. M‘ lacks an emargination of the posterior border of the crown and the hypoconal shelf expands posteriorly. Accordingly, the crown is quadrate in occlusal view; rather longer than wide, whereas in Trimylus M' is wider than long. M^ approaches M‘ in general morphology except for a wider paracone and smaller hypocone. M^ lacks an emargination of the posterior border of the crown. P, is longer and wider than either P2 or P,. Low r molars have a high entocristid, which joins the ento onid to the posterior face of the metaconid and closes the talonid basin lingually. On M|, the anterior cingulum runs along the ant rior base of the paraconid and along the anterobuccal base of the protoconid. The buccal cingulum runs posteriorly from the buccal base of Fig. 9b-e. — Ankylodon sp. cf. A. annectens (continued), b; CM 21699, left M^. c: CM 33654, left P4. d: CM 33658, right Mg. e; CM 33659, left M3. x8. Fig. 10. — Centelodon sp. cf. C. marginalis. a: KU 16606, left P‘,x8. b: CM 33662, right M^,x8. c: CM 33663, left P4-Mj,x 10. d: CM 21672, left M2-M3, xlO. Fig. 11. — Domnina sp. cf. D. gradata. a: CM 33667, left P-M’. b: CM 33673, left M^. c: CM 33678, left P.-Pj, M,. d: CM 33685, left M2-M3. x8. Fig. 12.— Proscalops miocaenus. a: CM 33688, left M'. b: CM 33690, left M^ c: CM 33694, left M,. d: CM 33696, left Mj. x8. 1978 SETOGUCHI— CEDAR RIDGE LOCAL FAUNA 21 22 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 9 Table 4. — Dimensions of teeth o/ Proscalops miocaenus, Oligoscalops ? sp., and Micropternodus ip. M' M" M" M, M, M, Statistics L AW PW L AW PW L AW PW L AW PW L AW PW L AW PW Proscalops miocaenus N 2 2 2 2 2 1 1 4 4 4 1 1 1 OR 2.02- 1.99- 2.67- 1.76- 1 ri 1.41 1.77 _ 1.79- 1.02- 1.34- 2.18 1.41 1.30 2.37 2.05 2.78 1.96 2.53^^ 2.06 1.10 1.46 1.905 1 .075 1.405 Oligoscalops ? sp. N 1 1 1 OR Micropternodus sp. 1.62 1.17 1.03 N 1 OR — ' 1.23 - * Width at the mesostyle. the protoconid. On five of the six M,s, the anterior cingulum does not unite with the buccal cingulum although they are very close together. On one specimen, these cingula are united along the buccal base of the protoconid, although the cingula are very weak there. The labial cingulum is very weak along the buccal base of the hypoconid. The mental foramen is below the middle of Mj. Discussion. — Patterson and McGrew (1937) and Repenning (1967) have thoroughly described Dom- nina gradata from the Orellan of Colorado, South Dakota, and Nebraska. In the absence of preserved mandibular condyles or the antemolar dentition, Domnina, as well as all other heterosoricines, is best defined by P"* and M' that lack an emargination of the posterior border of the crown and the resul- tant posterior expansion of the hypoconal shelf. Istead, the posterior margin of P** and M^ of Dom- nina is nearly straight or expands a little posteriorly and the crown is longer than wide, especially on Mb In relation to the non-bulbous feature of P4 and Ml in Domnina (in Trimylus, lower molars are bul- bous), P^ has a tall paracone and the anterior wall of the cusp is very steep, whereas that in Trimylus forms a more gentle slope. Such is the case for P"* and M’ in this sample. The lower molars referred here are very similar to those of Domnina and differ from those of Tri- niylus in that a high crest joins the ntoconid to the posterior face of the metaconid and closes the tal- onid basin lingually. In Trimylus a deep notch iso- lates the entoconid from the metaconid (Repenning, 1967). The anterior cingulum anterobuccal to the protoconid on Mj is a little more developed than in the Orellan Domnina gradata. In the latter form. the buccal cingulum on M, is not continuous around the base of the protoconid. In the present form, this feature of the cingula approaches being continuous around the base of the protoconid. The early Miocene D. greeni is not well known. In the original description, Macdonald stated ( 1963:168), “Labial cingulum on anterior labial face of trigonid only.’’ In the present form, the buccal cingulum buccal to the hypoconid is reduced. The present form may have given rise to D. greeni. Family Talpidae Gray, 1825 Proscalops miocaenus Matthew, 1901 (Fig. 12, Table 4) Referred specimens. — M': CM 33688, CM 33689; M^: CM 33690, CM 33691; M^; CM 33692; M,; CM 21668, CM 33693, CM 33694, CM 33695; M,; CM 33696. Locality. — Loc. 19, Badwater Creek, Wyoming. Age. — Late Oligocene. Description .—The size of the teeth is close to that of the ho- lotype of Proscalops miocaenus. On M‘, the protocone is V-shaped with an apex which is oriented anterolingually. The anterior arm of the protocone runs anterobuccally and soon turns buccally. In continues to run along the anterior face of the para- cone and connects to the parastyle. The anterior face of the anterior protocone arm has a small anterior projection at the midpoint between the protocone and the paracone. This small projection is what Reed called the protostyle (Reed, 1961;487). A small, but distinct hypoc ne is present posterior to the pro- tocone. Thus, the lingual portion of the tooth is somewhat broad- er. The paracone is smaller than the metacone. The former is blade-like running rather posterobuccally, whereas the latter is V-shaped. The posterior arm of the metacone is longer than the anterior arm. The former extends posterobuccally and connects to the metastyle which forms the posterobuccal corner of the crown. Thus, the buccal portion of the crown is truncated an- 1978 SETOGUCHI— CEDAR RIDGE LOCAL FAUNA 23 teriorly so that the anterior portion is narrower transversely than the posterior one. A small mesostyle is present near the midline on the buccal border of the crown. On M^, the anterior arm of the protocone terminates at the base of the anterior face of the paracone. It does not extend buccally beyond the point of the apex of the paracone so that it appears to be shorter anteroposteriorly. As in M‘ a small ‘‘pro- tostyle” is present on the anterior arm of the protocone, and a small hypocone is also present posterior to the protocone. The protocone is more acutely V-shaped than in M', and the apex is oriented more mesially. The paracone and the metacone are sub- equal, forming acute V’s. The parastyle and the metastyle are also subequal. A robust mesostyle lies on the middle of the buc- cal border of the crown. Thus, the buccal part of the crown is symmetrical. is reduced. The general morphology agrees with that of although is smaller than M^. The portion posterobuccal to the anterior arm of the metacone is completely truncated so that the posterior arm of the metacone and the metastyle are not present. The mesostyle is also reduced in size. The paracone is wider buccally than in M'. On M,, the trigonid is narrower transversely than the talonid. The protoconid is the tallest of the cusps on the trigonid. The protoconid is elongated transversely and forms a V with the apex buccal. The paraconid is the lowest of the trigonid cusps and is situated anterior to the anterior face of the protoconid. The metaconid is situated posterior to the posterior face of the pro- toconid. Thus, the lingual side of the trigonid is wide antero- posterioriy while the protoconid itself is compressed anteropos- teriorly. A rudimentary anterior cingulum is present on the anterobuccal base of the paraconid. The hypoconid is elongated transversely. The crista obliqua extends anterolingually and con- nects to the posterobuccal corner of the metaconid. The buccal face of the crista obliqua is very steep and the hypoflexid is very deep. The median cingulum is not present between the proto- conid and the hypoconid, as in Mesoscalops. The entoconid is conical. A transverse ridge unites the hypoconid and entoconid. No entocristid is present between the entoconid and the meta- conid so that the talonid basin opens lingually. A rather well- developed posterior cingulum is present on the posterior base of the talonid. CM 33969 is identified as M2 of this species. The trigonid is a little wider transversely than the talonid. The lingual part of the trigonid is narrower anteroposteriorly than in M,. The pro- toconid is also compressed anteroposteriorly as in M,, but the paraconid is merely a lingual extension of the anterior arm of the protoconid. The anterior cingulum is wide lingually and de- scends buccally to the base of the anterior face of the protoconid. The metaconid is elongated anteroposteriorly and has a small notch on its buccal face. The almost transverse posterior arm of the protoconid connects to the midpoint of the elongated meta- conid, forming an anterior wall with a notch on the buccal face of the metaconid. The hypoconid is compressed anteroposte- riorly. The crista obliqua reaches to the posterobuccal corner of the metaconid, posterior to the notch of the metaconid men- tioned above. Thus, the crista obliqua does not connect to the trigonid proper but to the posterior extention of the metaconid. The remainder of the features are almost identical to those of M,. Discussion. — The lower molars of the present form do not have a median cingulum between the protoconid and the hypoconid, as do those of Meso- scalops. Oligoscalops is best defined by that has a large parastylar area, whereas the corresponding tooth of Proscalops lacks a parastylar area. In the present fauna, no materials referable to are avail- able. In her original description of Oligoscalops, Reed stated (Reed, 1961:486-487), “M' in Oligo- scalops is triangular in general outline, with the pro- tocone directed anteriorly. The hypocone is rudi- mentary, a mere protuberance labial and posterior to the protocone .... In Proscalops miocaenus the tooth is generally similar, although the hypocone is somewhat better developed and the lingual portion of the tooth therefore somewhat broader. A rudi- mentary protostyle is present." M' of the present form has the small "protostyle," better developed hypocone, and the broader lingual portion of the crown. This form does not belong within the genus Oligoscalops. According to Reed (Reed, 1961:487), in Proscalops tertiiis and P. seciindiis, the hypo- cone is better developed on M' than in P. niiocaen- us. The degree of development of the hypocone on M' of the present form is close to that in P. niio- caeniis. Oligoscalops ?sp. (Fig. 13, Table 4) Referred specimen. — CM 17441, a left ramus with M.2. Locality. — Loc. 19, Badwater Creek Wyoming. Age. — Late Oligocene. Description. — M.2 is small. The size of the tooth is smaller than that of Proscalops miocaenus. The trigonid is narrow antero- posteriorly. The protoconid is elongated transversely and the paraconid forms a transverse ridge at a slightly worn stage. The paraconid does not reach to the lingual extremity of the base of the trigonid, leaving a shelf lingual to it. On this specimen, the metaconid is broken and only the base of it remains. The base of the metaconid is longer transversely than that of the paraconid reaching the lingual border of the trigonid. A rather broad an- terior cingulum is present at the base of the anterior face of the trigonid, and the lingual extension of the cingulum surrounds the anterolingual base of the paraconid and continues to run poste- riorly to connect to the anterolingual base of the metaconid. The talonid is a little narrower than the trigonid. The hypoconid is compressed anteroposteriorly. The crista obliqua runs anterolin- gually and connects to the trigonid at a point one-third of the way from the lingual side on the posterior wall of the trigonid. The hypoflexid is deep. The entoconid is conical and situated on the posterolingual corner of the tooth. A transverse ridge unites the posterior corners of the hypoconid and the entoconid. A small anteroposteriorly elongated metastylid is situated on the lingual border between the metaconid and the entoconid. The metastylid conpletely blocks the talonid basin lingually. The hy- poconulid is very low, situated just posterior to the entoconid. The posterior cingulum is rudimentary. 24 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 9 1978 SETOGUCHI— CEDAR RIDGE LOCAL FAUNA 25 Discussion. — Oligoscalops is smaller than Pro- scalops miocaeniis. Oligoscalops was established by Reed (1961), and a new species O. whitmanensis was regarded as the smallest known member of the Proscalopinae. She gave the diagnosis for the upper dentition of this new form but did not describe the morphology of the lower dentition. The holotype is CM P 25800, partial skull and jaws, but a lower jaw with P2-M3, KU 8143 is included in the hypodigm by her. In her discussion, she stated (Reed, 1961:488), “(In the Kansas specimen) In Mj the tal- onid is wider than the trigonid, judging from the fragments that remain, and the opposite is true of M2.’’ In the measurements of the new form, she gave the trigonid and talonid width of M2 of the Kansas specimen as 1.7 and 2.1 mm, respectively. According to the measurements given by her, the talonid is wider than the trigonid. Family Micropternodontidae Stirton and Rensberger, 1964 Micropternodus sp. (Fig. 14, Table 4) Referred specimen. — Mandible with M3. Locality. — Loc. 19, Badwater Creek, Wyoming. Age. — Late Oligocene. Description and discussion. — Only one lower jaw with M3 and the posterior part of the ramus is avail- able. On M3, the tip of the paraconid and the lingual half of the talonid are broken off. The tooth is some- what worn. The trigonid is transverse and com- pressed anteroposteriorly. The metaconid is taller than the protoconid. The protoconid is acutely V-shaped with its apex buccally. The anterior cingu- lum is developed along the anterior base of the trigonid. This cingulum is narrower buccally, ter- minates at a point anterior to the protoconid, and does not reach to the buccal face of this cusp. Lin- gually, the anterior cingulum becomes broader and its lingual extremity reaches to the lingual base of the paraconid. Thus, the cingulum surrounds the anterolingual base of the paraconid. The hypoconid is also V-shaped with its apex buccal. The hypo- conid is lower than the protoconid. The crista ob- liqua does not connect to the protoconid nor to the protolophid; instead, it runs directly anterolingually to connect to the metaconid. This characteristic fea- ture is also seen in M3 referable to Micropternodus borealis where the hypo onid is connected to the metaconid by a crest (Russell, 1960:945). Because of the lingually extended crista obliqua, the hypo- flexid is long transversely. The hypoflexid slopes down buccally. Because the lingual portion of the talonid is broken, it is impossible to tell whether or not the hypoconulid is present on M3. The coronoid process of the mandible is rather slender. The anterior face of the coronoid process is almost perpendicular to the occlusal plane or leans somewhat anteriorly. The present form is identified as Micropternodus sp. because of the similarity of the talonid structure to that of M. borealis of the early Oligocene. The present sample is not sufficient to warrant specific identification. Order Rodentia Family Aplodontidae Trouessart, 1897 Prosciurus relictus (Cope, 1873) (Fig. 15, Table 5) Referred specimens CM 33262. P-'-M^; CM 33261; P^- M‘: CM 19714, CM 33266; P^-M^: CM 33263-33265; P^-M^; CM 17078; F: CM 33276; DP'*: CM 33268-33275; F; CM 17405, CM 17408, CM 17410, CM 33277-33305; M'-M^: CM 33267; M> or M'*; CM 17406, CM 19793, CM 33306-33379; M^: CM 17089, CM 17094, CM 33380-33403; P^-M,; CM 19791, CM 33122-33124; P^-Mj: CM 33121; DP^: CM 33133-33137; P^; CM 17076, CM 17409, CM 33138-33158; M.-M,,: CM 33125-33128; MjM.,: CM 33129-33132; M, or M.: CM 17403, CM 33159-33220; M;,: CM 17077, CM 17404, CM 19794, CM 33221-33260. Locality. — Loc. 19, Badwater Creek, Wyoming. Age. — Late Oligocene. Description. — P-’ is peglike. The apex of the cusp is shifted slightly anteriorly and a small ridge descends posteriorly from the apex. DP** is identified on the basis of its tiny size and the similar tooth structure to that of F referable to this species. In both the parastylar lobe extends anterobuccally and a small, but distinct, cusp is formed on its anterior border. This cusp is lower than the protoconule. The cusp lies lingual to the metacone-paracone line and the distance between the parastylar cusp and the para- cone is nearly the same as that between the paracone and the metacone. The transverse valley between the parastylar cusp and the protoloph opens more widely in DP'* than in F. The Fig. 13. — Oligoscalops ? sp. CM 17441, left M3. xIO. Fig. 14. — Micropternodus sp. CM 17442, left M-,. x 10. Fig. 15. — Prosciurus relictus. a; CM 33263, left P‘*-M’. b: CM 19791 , left P4-M,. c: CM 33126, right M,-M;,. x7. Fig. 16. — Pelycomys placidus. a: CM 33101 , right F-MT b: CM 19792, left P^-Mj. x5. 26 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 9 Table 5. — Dimensions of teeth of Prosciurus relictus. DP^ M' M^ M' or M^ Statistics L W L W L W L W L W N 8 8 25 25 7 7 6 6 43 43 OR 1.48- 1.42- 1.36- 1.63- 1.39- 1.91- 1.44- 1.82- 1.33- 1.64- 1.66 1.73 1.95 2.30 1.62 2.17 1.69 2.14 1.83 2.32 Mean 1.554 1.623 1.728 1.943 1.543 2.021 1.567 1.982 1.610 2.000 SD 0.06! 0.096 0.150 0.219 0.088 0.111 0.116 0.135 0.114 0.154 CV 3.29 5.92 8.70 11.28 5.72 5.47 7.41 6.83 7.06 7.70 DP4 P4 M. M2 M| or M2 L AW PW L AW PW L AW PW L AW PW L AW PW N 5 5 5 20 19 20 7 6 7 7 7 7 47 47 47 OR 1.38- 0.85- 1.31- 1.33- 0.88- 1.30- 1.50- 1.36- 1.54- 1.63- 1.46- 1.60- 1.38- 1.25- 1.50- 1.54 0.98 1.48 1.91 1.49 1.97 1.83 1.61 1.80 1.97 1.80 1.95 1.94 1.74 2.07 Mean 1.464 0.914 1.370 1.631 1.186 1.638 1.703 1.480 1.697 1.794 1.624 1.796 1.739 1.514 1.740 SD 0.074 0.054 0.076 0.162 0.165 0.157 0.116 0.097 0.102 0.151 0.132 0.143 0.117 0.118 0.129 CV 5.08 5.92 5.56 9.94 13.88 9.57 6.84 6.58 6.03 8.43 8.14 7.95 6.73 7.77 7.41 floor of the lingual half of the valley is much lower than that between the protoloph and the metaloph. A short ridge descends posterolinguad from the parastylar cusp but no connecting ridge is seen between the parastylar cusp and the protocone so that the valley anterior to the protoloph opens lingually unlike P. The structure of the protocone is similar to that of P. The pro- toconule is smaller and lower than the metaconule, and both of them are smaller and lower than the subequal paracone and metacone. The protoloph is low in position and extends almost transversely, and the protoconule is formed on its anterior face between the protocone and the paracone. The paracone is elon- gated transversely. The metaloph is also low in position and extends posterobuccally so that the valley between the protoloph and the metaloph becomes wider buccally. This valley is rather wide and U-shaped, not V-shaped as in ?■*. The metaconule is formed on the posterior face of the metaloph. Both the proto- conule and the metaconule are connected to the protocone by weak ridges. No hypocone is formed and the posterior cingulum is weak. The tooth is three rooted; the root underneath the pro- tocone is the largest. This root is elongated trasversely and its buccal margin reaches to the protoconule-metaconule line. The root extends slightly lingually. Two roots are on the buccal base of the tooth. The anterior root is larger and occupies the bases of the parastylar lobe and the anterior half of the paracone. This root is elongated anteroposteriorly and extends anterobuccally. This third root is just underneath the metacone and extends slightly buccally. The cross section of the root is rounded. These three roots are not close together but open widely. This is another reason why these teeth are identified as 0?“*. P”* and molar structures are almost the same as those described by Galbreath (1953). On all cheek teeth, a single metaconule is present. The lower jaw is relatively longer than that of Pelycomys. The anterior face of the incisor is flat rather than rounded as in Pe- lycomys and the external face is flat as well. The posterior face is rounded and narrow so that the tooth is narrower posteriorly. DP4 is also identified on the basis of its tiny size, its similarity to P4, and widely open roots. The protoconid is lower than the metaconid but is wider anteroposteriorly. Both metalophis I and II are complete although they are low in position, and block the trigonid basin anteriorly and posteriorly. The mesoconid is dis- tinct and as high as the entoconid. The ectolophid is not con- spicuous but forms an acute edge between the talonid basin and the valley between the protoconid and the hypoconid. No meso- iophid is seen. The hypoconid is at the posterobuccal corner of the tooth and is greatly compressed anteroposteriorly; the anterior face is vertical. The hypoconid is separated from the posterolophid by a deep notch. The posterolophid is a tall, trans- verse blade high above the bottom of the talonid basin. The entoconid is separated from the posterolophid by a small notch. The hypolophid runs buccally from the entoconid and soon turns posteriorly to unite to the anterior face of the posterolophid. The hypolophid does not extend to the ectolophid. The hypolophid, entoconid, and posterolophid share a common base, which raises high above the bottom of the talonid basin. The mesostylid is inconspicuous, and between it and the entoconid is a deep notch, which runs transversely from the talonid basin. The metastylid crest is not present. The other cheek teeth are very close to those described by Wood (1937) and discussed by Galbreath (1953). Discussion. — The Badwater specimens are al- most identical to Orellan Prosciurus relictus. Pro- sciurus is the best represented of several genera of closely related rodents of the Prosciurinae. The subfamily Prosciurinae has been regarded as a member of the family Ischyromyidae (as that group is defined by Black, 1971:181) or the Paramyidae (Wood, 1955:171; Wood, 1962:226). A slight varia- tion in the classification is presented by Wood (1973), in which the Prosciuridae is considered a separate family. 1978 SETOGUCHI— CEDAR RIDGE LOCAL FAUNA 27 Recently, Rensberger (1975) transferred the Pro- sciurinae from the Ischyromyidae or Paramyidae to the Aplodontidae. The aplodontid taxa are segre- gated as Prosciurinae, Allomynae, and Aplodontin- ae. I agree with him. Here, Prosciums and Pely- comys are treated as members of the Aplodontidae. Pelyconiys placidus Galbreath, 1953 (Fig. 16, Table 6) Referred specimens. — CM 33101; P"': CM 33102, CM 33103; M' or M^: CM 33104, CM 33105; M^: CM 33106, CM 33120; P4-M2: CM 19792; M^; CM 17042; M,,: CM 33107, CM 33108. Locality. — Loc. 19, Badwater Creek, Wyoming. Age. — Late Oligocene. Description. — The lower jaw is relatively deeper and shorter than that of Prosciums relict us. The incisor is characterized by greater compression laterally than in Prosciums. The internal face of the incisor is flat, the anterior face is rounded, and the external face is gently curved toward the posterior border of the internal face, as Galbreath ( 1953) described. The lower cheek teeth are subtriangular to subrhombic in shape. On P4, the principal cusps are rounded and large. The trigonid is elongated transversely. The protoconid is tall and somewhat compressed anteroposteriorly. The metaconid is the tallest of the principal cusps. It lies on the anterolingual corner of the tooth. Metalophulid I is absent and the protoconid and the metaconid are separated by a deep notch, which runs ante- robuccal from the posterobuccal base of the metaconid. Meta- lophulid II is very weak but present and unites the protoconid and the metaconid posteriorly. The mesostylid is tiny and low in position. The metastylid crest runs downward and pos- teriorly from the metaconid, and unites with the mesostylid. The mesoconid is small and rounded. The buccal mesolophid is ab- sent so that a wide valley opens buccally between the protoconid and the hypoconid. The entolophid is a rather wide ridge and descends posteriorly from the posterolingual corner of the pro- toconid. The ectolophid, posterior to the mesoconid, turns pos- terobuccal and unites with the lingual corner of the transeversely elongated hypoconid. The hypoconid is stout forming the pos- terobuccal corner of the tooth. The posterolophid is prominent. It unites with the posterolingual corner of the hypoconid but does not unite with the entoconid. The entoconid is rather small but is a distinct cusp, which is higher than the mesoconid. The hypolophid runs buccally from the entoconid and turns slightly posteriorly and joins the ectolophid between the hypoconid and the mesoconid. The lingua! half of the hypolophid is wide and the anterior slope is more gentle than the posterior one so that the basin between the hypolophid and the trigonid becomes nar- rower transversely. The buccal half of the hypolophid is thin and low. A distinct transverse valley separates the hypolophid from the posterolophid. The entoconid is separated from the meso- stylid by a deep notch. Mj has a complete metalophulid I, which runs anterad from the protoconid and soon turns linguad. The metalophulid II is also complete but lower than the metalophulid I. A basin is formed on the trigonid between the metalophulids, and the protoconid and the metaconid. The mesoconid has a tiny buccal mesolophid and tends to divide the valley between the protoconid and the hypo- conid into two parts. The hypolophid is wider than on P4 and runs buccally. The buccal ends connects to the ectolophid posterior to the mesoconid; near the buccal end it becomes narrower. The basin between the hypolophid and the trigonid is deeper than on P4 and the notch which separates the entoconid from the meso- stylid is also deeper than in P4. M.> agrees with M, in general morphology. The mesoconid is more prominent than in P4 and Mj. The mesoconid itself divides the valley between the protoconid and the hypoconid into two parts; the posterior part is wider than the anterior part. The ec- tolophid posterior to the mesoconid runs posteriorly and slightly lingually, and turns posterobuccally . The buccal slope is gentle so that the valley between the mesoconid and the hypoconid is long transversely in an unworn stage but becomes shorter with wear. The posterolophid is also prominent and tends to have a distinct cusp near its union with the hypoconid. The hypolophis is taller than in M( and connects to the ectolophid more posteriorly than in M,. The basin between the hypolophid and the trigonid is more widely open than in P4 and M,. On M3, the protonid is low but the metaconid is high. Although the metalophulid I is complete, the metalophulid II is incomplete on the metaconid side. The basin between the metalophulids is wider than in M, and M.^, and opens posterolinguad. The talonid basin becomes wider and the difference in height between the tal- onid basin and the trigonid is reduced. The mesoconid is large and divides the valley between the protoconid and the hypoconid into two parts. The hypoconid is massive and its buccal arm extends anteriorly to the protoconid enclosing the large mesoconid buc- cally. The entoconid is distinct but the hypolophid is weakened. The hypolophid runs posterobuccally making the talonid basin wider. The valley between the posterolophid and the hypolophid is wider and deeper than in the rest of the cheek teeth. A maxilla and a few isolated teeth are tentatively assigned to this species. The size of these agrees with that of the lower teeth of the species. They are not associated with any lower jaws so that whether or not they are referable to Pelyconiys remains un- certain, as only lower cheek teeth of Pelyconiys are known. On the parastyle is prominent. The protocone is elongated antero- posteriorly. The protoconule and the metaconule are subequal, but only the former connects to the the protocone by the thin pro- toloph. After heavy wear, the metaconule will unite with the pro- tocone. The posterior crest of the protocone runs posterad and continues to run transversely as the posterior cingulum. Near the union of the posterior crest with the posterior cingulum, a small hypocone is present. The metaloph unites the metacone and the metaconule on their anterior side. A tiny but distinct mesostyle is present between the paracone and the metacone on the buccal border of the tooth. All the specimens referable to M' and M'^ are heavily worn. A parastyle seems to be present on the buccal end of the anterior cingulum. The posterolingual corner of the tooth extends linguad indicting the presence of the hypocone. M^ is triangular in shape. The anterior crest of the protocone is heavy . The buccal end of the anterior cingulum has a wide base. The paracone is elongated anteroposteriorly. The protoconule is tiny and situated just between the protocone and the paracone. The protoconule connects to the protocone by a ridge. The basin between the protoloph and the anterior cingulum is wide. Poste- rior to the protoloph is a broad basin. Only a tiny metaconule is present and it is completely isolated. The posterior crest of the protocone, the posterior cingulum, and the buccal cingulum sur- round the basin. 28 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 9 Discussion. — The present form differs from Pro- sciurus in having the following morphology: incisors are laterally compressed; on lower cheek teeth, metalophulid II is essentially complete; hypolophu- lid is well developed and separated from the poste- rolophid. These are the diagnostic features ofPely- co/nys. Galbreath (1953) recognized two species of Pe- lycomys — P. rugosus, the type species, andF. plan- idus. P. planidus differs from P. rugosus in having narrower trigonids, better developed mesoconids, and weaker and lower metalophulid II than meta- lophulids I. The Badwater specimens are almost identical to Pely corny s planidus. Eamily Eomyidae Deperet and Douxami, 1902 Adjidaumo douglassi Burke, 1934 (Fig. 17, Table 6) Referred specimens. — M‘ or M'^; CM 33702, CM 33703; P^- M,: CM 33697; P,; CM 33698; M.-M^; CM 33699; M,; CM 33700; M^; CM 33701. Locality. — Loc. 19, Badwater Creek, Wyoming. Age. — Late Oligocene. Description and discussion. — The morphology of the mandible is very similar to that of Adjidaumo minimus. The mandible is long and slender. The mental foramen lies anterior to P4 and almost on the dorsal surface of the mandible. The masseteric fos- sa ends rather acutely below the talonid of P4. The doral masseteric ridge is stronger than the ventral and rises gently to the ascending ramus, which orig- inates opposite the posterior half of M.2. The crown pattern of the lower cheek teeth of A. douglassi differs from that of A. minimum and A. minutus. In A. minimus, the protoconid and the hy- poconid are on the buccal side of the tooth, and the ectolophid also lies on the buccal margin to the midline of the crown. But in the type of A. doug- lassi, the protoconid and the hypoconid lean lin- guad making the buccal wall of each cusp more gentle and pushing the ectolophid towards the mid- line of the crown. The difference in morphology of these two types is clearly seen on worn specimens. When worn, the ectolophid in A. minimus lies buc- cally, whereas it in A. douglassi lies near the center of the crown. In A. minimus, the protoconid and the hypoconid are more stout than in A. minutus and A. douglassi. M2 of A. minutus is wider than long, whereas M2 of both A. minimus and A. doug- lassi is clearly longer than wide. A. minutus is larger than both A. minimus and A. douglassi, which are essentially of the same size. A. douglassi is a direct descendent from A. minimus but not via A. minu- tus. All the specimens in the Badwater fauna are well worn. The P4 has a narrower trigonid, almost half as wide as the talonid. The ectolophid lies near the midline of the tooth. No indication of the mesolo- phid is seen. M, and M2 have essentially the same morpholo- gy. M, is clearly longer than wide. One specimen of M, has a little narrower anterior half than the posterior but another specimen has the same ante- rior and the posterior width. On M2, the anterior half is wider than the posterior half on the type of A. douglassi and CM 33701. The anterior cingulum is joined to the base of the metaconid and by a short crest to the metalophid where the latter leaves the protoconid. The buccal end of the cingulum is ap- parently free. These features are also found in A. minimus. The mesolophid is short on all the speci- mens passing half way to the lingual border on Mj and M2. The ectolophid lies near the midline of the crown. No M3S are present in the sample. The upper dentition of Adjidaumo has not been adequately described and figured. Wood (1937:237- 238) briefly discussed the morphology of upper mo- lars of this genus but did not figure them. The upper cheek teeth of Adjidaumo in the present fauna are identified based primarily on size and the mirror imaged structure of the lower molars referable to A. douglassi. Only two heavily worn specimens are identified. Both of them are M^ or M^. The anterior Fig. 17. — Adjidaumo douglassi. a: CM 33699, right M^Ma. b: CM 33703, left M' or M'-*. x8. Fig. 18. — Paradjidaumo hypsodus, new species, a: CM 33707, right P. b; CM 33704, holotype, right M>. c: CM 33723, right M^ d; CM 33732, right P4. e: CM 33743, right M,. f; CM 33752, right M-2. x8. Fig. 19. — Metadjidaumo hendryi, new genus and new species, a: CM 33775, right P**, x8. b; CM 33780, left M‘, x8. c; CM 33783, left M^, x 15. d; CM 33784, left P4, x8. e: CM 33786, holotype, left M,, x8. f; CM 33808, left Mj, x8. Fig. 20. — Eomyidae, genus indet.. Type A. a: CM 3381 1, right M^. b; CM 33812, right M,. x8. Fig. 21.- — Eomyidae, genus indet.. Type B. a: CM 33813, right P^M'. b: CM 33815, right P'*. c: CM 33816, lower molar? x8. Pig. 22. — Proheteromys sp. cf. P. nehraskensis. a: CM 33818, left M', x8. b; CM 33846, left M^, x8. c: CM 33857, right M>, xl5. d: CM 33884, left M.^, x8. 1978 SETOGUCHI— CEDAR RIDGE LOCAL FAUNA 29 30 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 9 Table 6. — Dimensions of teeth of Pelycomys ptacidus and Adjidaumo douglussi. Teeth and Pelycomys placidus Adjidaumo douglassi measurements N OR Mean N OR Mean p4 Length 3 2.13-2.57 2.353 Width 3 2.30-2.96 2.600 M' Length 1 2.41 Width 1 3.09 M2 Length 1 2.49 Width 1 3.27 M' or M2 Length 2 0.87-0.91 Width 2 0.94-0.97 M^ Length 2 2.62-2.78 Width 2 2.75-2.88 P4 Length 1 2 72 2 0.78-1.02 Width 2 0.76-1.05 Anterior 1 1.90 Posterior 1 2.56 M, Length 1 2.47 3 0.91-0.93 0.920 Width 3 0.79-0.88 0.847 Anterior 1 2.21 Posterior 1 2.86 M, Length 2 2.51-2.72 2 0.92 Width 2 0.91-0.95 Anterior 1 2.10 Posterior 2 2.66-2.93 M3 Length Width 2 3.19-3.48 Anterior 2 2.64-2.73 cingulum is restricted on the buccal half of the tooth, whereas the posterior cingulum is well de- veloped. The mesoloph is short and extends only half way to the buccal border. Union of the meso- loph to the metaloph is at almost the midline of the crown. The mesoloph is connected to the proto- cone. Paradjidaumo hypsodus, new species (Fig. 18, Table 7) Holotype. — CM 33704, isolated right M*. Hypodigm.— Type and P: CM 33705-33708; M': CM 17097, CM 33709-33718; M^: CM 33719-33730; M^: CM 33731; P^; CM 17436, CM 19808, CM 33732-33742; M,; CM 17437, CM 17440, CM 33743-33743; Ma-M..,: CM 33749; M.,; CM 17433, CM 33750- 33762; M3; CM 33763-33769. Etymology . — Hypsodont Paradjidaumo. Locality. — Loc. 19, Badwater Creek, Wyoming. Age. — Late Oligocene. Diagnosis. — Higher crowned than P. trilophus — cusps and lophs (lophids) are higher and valleys are deeper than in any other known species of Parad- jidaumo. Description. — The cheek teeth of Paradjidaumo are meso- dont, showing in relation to other eomyids an increase in the height of the cross lophs and lophids as well as increase in height of cusps (Black, 1965:26). The teeth of the present species are more advanced in this character than are those of P. trilophus. ?■* is essentially molariform. It differs from M' and M^ pri- marily in the absence of the anterior cingulum. On one specimen (CM 33705), however, a very narrow and shallow scar is present below the lingual end of the paracone on the anterior face of the tooth and the scar is covered anteriorly by a thin, short ridge. This ridge may be a remnant of the anterior cingulum and the scar, a remnant of the anterior valley. The scar is low in position so that it would be obliterated only after considerable wear. On the other specimen (CM 33707), no trace of an anterior cingulum and an anterior valley are seen. The protocone and the paracone are elongated transversely forming elongated blades. The buccal part of the protocone and the lingual part of the paracone run slightly anterad and join together at an angle below the tips of both cusps which are of the same height. The mesoloph reaches across the crown to a tiny mesostyle. The mesocone is well defined situating closer to the protocone than to the hypocone. The anterobuccal corner of the mesocone is very close to the posterobuccal corner of the protocone but apparently they are not connected to each other. This is a situation of the incipient separation of the protocone and the protoloph from the mesoloph as seen in some of the European forms, Vike Pseudotheridomys. The posterior cingulum is well developed but the valley between the metaloph and the posterior cingulum is not deep so that these elements become fused into a single loph with further wear. The base of each cusp is broad as in P. trilophus but the tips of these cusps near the unworn crown surface are high above the base and become compressed anteroposteriorly leaving deep valleys between the cusps and the mesoloph. These valleys are deeper than in P. trilophus. M‘ differs slightly from M- in crown pattern. The anterior cingulum is distinct on both M' and M'^ when the teeth are un- worn but restricted to the lingual half of the crown. It quickly merges with the protoloph as wear proce ds. The posterior val- ley is deeper than the anterior valley so that the former remains distinct somewhat longer but it also eventually fuses with the metaloph, producing the “Omega” pattern as in P. minor and P. trilophus. The protocone is pushed posterolinguad so that the anterior half of the crown is narrower transversely than the pos- terior half on M‘. On M^ both halves are of the same width or the anterior half is a little wider. The mesostyle is tiny but dis- tinct. It is lower than both the paracone and the metacone, and separated from them by notches on M'. The mesoloph reaches to the mesostyle across the crown surface. On M'T the mesostyle is more prominent and higher than in M‘, and united to both the paracone and the metacone forming the lingual wall to block the 1978 SETOGUCHI— CEDAR RIDGE LOCAL FAUNA 31 Table 7. — Dimensions of teeth of Paradjidaumo hypsodus, new species, and Metadjidaumo hendryi, new genus and new species. Statistics F M M‘ M 2 P4 M, ^ _ Mo L W L W L W L W L W L W Paradjidaumo hypsodus , new species N 4 4 10 10 12 12 10 10 6 6 11 11 OR 1.28- 1.32- 1.17- 1.29- 1.20- 1.34- 1.21- 0.93- 1.29- 1.34- 1.30- 1.40- 1.58 1.50 1.54 1.88 1.62 1.72 1.60 1.36 1.44 1.51 1.49 1.62 Mean 1 .432 1 .405 1.384 1.519 1.389 1.542 1.387 1.164 1.352 1.415 1.366 1.482 SD 0.140 0.174 0.139 0.120 0.115 0.159 0.050 0.071 0.062 0.066 CV 10.14 11.48 9.98 7.77 8.29 13.70 3.73 5.01 4.53 4.42 Metadjidaumo hendry ■(, new genus and new species N 8 8 6 6 2 2 1 1 12 12 13 13 OR 0.95- 0.96- 0.97- 1.03- 1.04- 1.11- 0.98 0.95 0.96- 0.94- 0.90- 0.94- 1.12 1.21 1.07 1.27 1.09 1.13 1.13 1.15 1.04 1.12 Mean 0.970 1.061 1.018 1.175 1 .034 1.074 0.991 1 .066 SD 0.070 0.084 0.034 0.088 0.066 0.067 0.038 0.053 CV 7.27 7.87 3.31 7.53 6.39 6.22 3.85 5.00 central valley buccally. The buccal corner of the metacone is elongated anteriorly to join the mesostyle so that the metacone and the paracone are closer together on than on M'. The central valley is deeper on both M‘ and than in P. trilophiis and P. minor. The niesocone is not well defined on either M' or M^. The anterior extremity of the anterior arm of the hypocone is very close to the protocone. They are united by a short, thin ridge on M', and by a thicker ridge on M^. This condition, es- pecially of M‘ is close to that of P^, but no specimens available of M' show closer separation of them. is the smallest of the upper cheek teeth but the crown elements are not as greatly reduced as in P. minor. The proto- cone and the paracone are the largest cusps and they are joined by a strong protoloph as in P. minor, but the protocone lies more anteroposteriorly than in P. minor, and the anterior extremity of the protoconal ridge ends more lingually. The anterior cingu- lum is longer than in P. minor. All lophs rise to the same level as the protocone leaving rather deep valleys. On one specimen (CM 33731), the posterior valley is shallow but distinct. On the other, it is represented by a small pit. The mesoloph reaches to the lingual wall. No mesostyle is visible. The central valley an- terior to the mesoloph is deeper than the posterior one which, in turn, is deeper than the anterior valley. This molariform mor- phology of in the present species is unusual. P4 is longer than wide. The crown height is almost the same as in P. trilophus. The trigonid is higher than the talonid, but less prominent than \nP. trilophus. The protocone and the meta- cone are closely appressed and bounded posteriorly by the thin metalophid. A short anterior cingulum descends sharply from the anterior face of the protoconid to merge into the base of the metaconid. The posterior arm of the protoconid extends posteri- ad and connects to the mesolophid at right angle. The mesoio- phid extends transversely to the buccal border of the tooth and there turns anterad at a right angle to connect to the posterior base of the metaconid. The mesoloph is a thin ridge and slightly higher than the lingual wall of the tooth. The hypoconid forms a ridge, which extends more anteroposteriorly than in P. trilo- phiis. The talonid is narrower in relation to the crown length to make the tooth longer than in P. trilophus. In P. minor and P. trilophus, a rather broad and deep excavation is present on the buccal wall of the tooth between the protoconid and the hypo- conid. In the present species, this pit becomes shallower and narrower, and the excavated rather posteriad on the anterior base of the hypoconid. On some specimens (CM 33732), the central valley posterior to the mesolophid is deeper than the anterior one, but on one specimen (CM 33738) these two are essentially of the same depth. M, is slightly longer than wide, whereas M.2 is rather wider than long. Both M, and Mj are higher crowned than in P. minor and P. trilophus. Cusps become less prominent than in other species of Paradjidaumo. The anterior and posterior cingula, the metalophid, and the hypolophid lie essentially on the same level as the protoconid and the hypoconid, which are no more distinct cones hut rather thin, elongated ridges. The metaconid is a sharp-pointed cusp slightly above the level of lophids. The meta- lophid descends transversely from the peak of the metaconid. Only the metalophid is slightly higher than the remainder of the cross lophids in an unworn stage. The entoconid is a small knob. On M,, the mesolophid reaches to the lingual border of the tooth on all specimens available but on Mj, some specimens show that the mesolophid ends in the center of the crown. The central valley is deep. On some specimens of Mo, the trigonid is higher than talonid. The anterior cingulum of M, is not as closely ap- pressed to the metalophid as in P. minor, but on some (CM 33752) the former is appressed to the latter. The posterior cingu- lum is short and is confined to the lingual quarter of the pos- terior face of the tooth on M, and Mo. M3 differs from M, and M2 primarily in the absence of the posterior cingulum. M3 in P. minor and P. trilophus have essen- tially the same crown pattern as M, and Mo. But in M^ of these species of Paradjidaumo the crown elements are reduced. In the present species, as far as M3 is concerned, the same trend is retained. The mesolophid reaches to the entoconid across the crown surface, instead of reaching to the lingual wall between 32 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 9 the metaconid and the entoconid as in M, and Mj. The hypolo- phid forms a posterior wall and is convexed posteriorly. The crown is also higher than that of P. minor and P. trilophus. Discussion. — P. hypsodiis differs from P. minor and P. trilophus in having more hypsodont teeth. In P. minor, the che k teeth are not high crowned, and the paracone and the metacone on the one hand, and the metaconid and the entoconid on the other are elevated high above the cross lophs and lophids. In P. hypsodus, however, the lophs, the lophids, and the cusps are elevated to a nearly com- mon plane. The paracone and the metacone on the upper cheek teeth, and the metaconid and the en- toconid on the lower teeth rise only slightly above the lophs and the lophid. In relation to the high lophs and lophids, the protocone and the hypocone on the upper teeth become higher, forming elon- gated ridges elevated to the same plane as the cross lophs. The lower cheek teeth show exactly the same trend. The cheek teeth of P. hypsodus are charac- terized by strong lophs and lophids rather than being cuspidated. This situation is analogous to the tooth character of Eumys planidens discussed else- where in this article. The cheek teeth of Paradjidaumo hypsodus be- come higher crowned. Not only the lophs and lo- phids, but also the cusps become elevated as if elon- gated ridges were developed on the summit of the original cusps. The base of each elongated cusp is thick and broad showing the cuspidate condition of the ancestral stock to P. hypsodus. This condition could easily be derived from the tooth pattern of P. trilophus. As for the height of the crown of the cheek teeth, P. trilophus is intermediate between P. minor and P. hypsodus. The crown patterns of these three species of Paradjidaumo are essentially the same. The essentially unreduced morphology of M® is worth discussing. Although the posterior cingulum is reduced, M^ of P. hypsodus has all other crown elements, whereas on M^ of P. minor the crown elements are reduced. In the latter, the size is not greatly reduced although it is a little smaller than M' and M-. M^ and M~ of P. hypsodus have high lophs, and M® also follows this trend so that lophs and elongated cusps become emphasized in this species. The reduction of M'* is a trend in the central stock of Paradjidaumo with increased hypsodonty; however, M® is modified to emphasize all the molar elements. This kind of rejuvenation is not common in rodent evolution. Metadjidaumo, new genus Type species. — Metadjidaumo hendryi, new species. Etymology. — From Greek metd, met- after, descent of Adji- daumo. Diagnosis. — Near size of Adjidaumo minimus and A. douglassi; molars higher crowned than in A. minimus and A. douglassi with high, thin lophs (lo- phids); trigonid higher than talonid on M, and M^; no posterior cingulum on Mj. Metadjidaumo hendryi, new species (Fig. 19, Table 7) Holotype.— CM 33786, isolated left M,. Hypodigm.— Type and P"-M': CM 33770, CM 33771; P^: CM 33772-33777; M'; CM 17435, CM 33778-33780; M^; CM 33781- 33783; P,; CM 33784; M,-M.: CM 33785; M,: CM 17439, CM 33787-33796; M.; CM 17438, CM 33797-33808. Etymology. — For Mr. Jim Hendry, who provided his cabin during field seasons. Locality. — Loc. 19, Badwater Creek, Wyoming. Age. — Late Oligocene. Diagnosis. — Only known species of genus. Description. — The fourth upper premolar is square-shaped. All the cusps are well developed. Among them the metacone and the hypocone are taller than the paracone and protocone. The anterior half is narrower transversely than the posterior half. There is considerable variation in the morphology of the anterior half of the crown. The protocone is stout and extends antero- buccally leaving a narrow valley between it and the hypocone. The valley runs slightly posterad on CM 33775 as in M' or M-, whereas the protocone is shifted more buccally from the usual position so that the valley between the protocone and the hy- pocone runs rather anterad on CM 33777. The protoloph is a thick ridge extending anterobuccally from the protocone and joining the paracone at its anterolingual corner. On two speci- mens, no anterior cingulum is seen. On other specimens, how- ever, a very shallow, compressed pocket is present on the an- terior face of the paracone which is bounded anteriorly by what is probably a short anterior cingulum that fades into the anterior face of the paracone. This pit would be obliterated with further wear. 7 he mesocone is well defined on CM 33775. It is half as large as the protocone and is a little lower than the latter. The mesoloph is essentially not present on this specimen. On the other specimens, the mesocone is merely a buccal extension of the anterior arm of the hypocone and quickly merges with the floor of the crown near the posterolingual base of the paracone. On three specimens, the mesocone or a thin ridge, the mesoloph, is connected to the protocone by what may be called the pos- terior arm of the protocone. On one specimen, the mesocone and the protocone are close together but no connection is seen between these cusps, and on one other specimen, these two cusps are clearly separated from each other. The separation of the protocone from the mesoloph is not common in North Amer- ican eomyids, but this characteristic feature is commonly seen 1978 SETOGUCHI— CEDAR RIDGE LOCAL FAUNA 33 in European eomyicls. In four specimens, no metastyle is pres- ent, but on others the metastyle is present on the buccal border at the posterior base of the paracone. The posterior cingulum is well developed but the posterior valley is shallower than the central valley. Only two upper jaw fragments, both of which bear P* and M', are available and M' on both is heavily worn. M' is clearly wider than long on both specimens. One specimen (CM 33771) has a wider posterior half than anterior, and the other (CM 33770) is nearly square. The valley between the protocone and the hy- pocone is narrow and long reaching to nearly the center of the crown. The valley runs almost transversely but a little posteriad. The ectoloph or mure lies at the center of the crown on CM 33770 and a little buccal to the midline of the tooth on CM 3377 1 . The mesoloph is short. The buccal halves of the paracone and the metacone are curved posterad and anterad, respectively, so that the central valley becomes narrower buccally. On worn teeth, the lingual three-fourths of the crown is flattened, whereas the buccal one fourth remains a little higher. On CM 33771, the buccal portion of M' is not completely worn with the buccal wall between the paracone and the metacone unworn although both cusps are worn. This indicates that on M’ the buccal wall be- tween these two cusps is low and not elevated to block the central valley buccally. This condition is also seen on M‘ of Paradjidaumo. On M^ of Paradjidaumo, however, the buccal wall between the paracone and the metastyle, and between the latter and the metacone is elevated nearly to the level of these three cusps so that the central valley is blocked buccally by this wall. No unworn definite M* or unworn M^ associated with P* on the same jaws are available in the present fauna. This makes it difficult to identify M' and M''^ of the new genus and species. Isolated M' and M^ are identified on the basis of size and similar morphology with the low buccal wall between the paracone and the metacone for the former and with the high wall between them for the latter. M‘ and M^ are of similar size and the crown pattern is essen- tially the same. The size is very close to Adjidaumo minimus, but a little larger. The crown is definitely higher than in A. min- imus and A. doiiglassi. The ratio of the crown height to the width is nearly 1 .0 in the present form, whereas it is clearly below 1 .0 in Adjidaumo doiiglassi and is definitely over 1 .0 in Paradjidau- mo hypsodus. The molars emphasize ridges rather than stout cusps. The protocone is elongated sending a thick anterior arm anterobuccally. From the anterobuccal corner of the arm the anterior cingulum extends buccally and reaches to the anterior base of the paracone. The anterior cingulum is restricted to the buccal half of the crown. A short, transverse protoloph unites the anterior arm of the protocone and the transversely elongated paracone. The hypocone projects more lingually than the pro- tocone. The buc al tip of the hypocone is excavated making the posterior valley longer transversely. The anterior arm of the hy- pocone extends to the center of the crown, and there it turns anterad making what may be called the mur . A short mesoloph runs buccally; it does not reach to the buccal border of the crown. The posterior cingulum is well developed from the tip of the hypocone to the posterior base of the metacone. On some specimens, the union of the protocone and the mesoloph is seen by way of the posterior arm of the protocone, but on some, the connection is by a thin ridge (CM 33780, M' and CM 33783, M^), and almost no connection is seen (CM 33782, M-), where the mesoloph connected to the anterior arm of the hypocone. No metastyle is seen on any of the specimens. No lower jaw materials associated with upper jaws are avail- able. The lower molars are identified primarily based on size and the mirror image morphology of the upper molars. is also identified based on size and the similar morphology to M, and M,. One lower jaw fragment (CM 33784) has P4. The diastema is not deep and essentially the same as that of Adjidaumo. Pj is larger than that of Adjidaumo doiiglassi and wider than M, of A. doiiglassi. The trigonid is narrower than the talonid. The protoconid and the metaconid are of equal size and are joined posteriorly by a short crest. The anterior valley between the cusps is open. The mesolophid is short and ends half way across the crown. The mesoloph is lower than the ectolophid. No pos- terior cingulum is present on one specimen (CM 33784) but on the other specimen a small pit is present just posterobuccal to the entoconid and the pit is bounded posteriorly by what is prob- ably a short posterior cingulum that fades into the posterior face of the entoconid. One lower jaw (CM 33785) has M, and M2. These teeth are well worn. M, is clearly longer than wide and a short posterior cingulum is present. Mo is rather wider than long and has no posterior cingulum. Isolated M, and Mo are identified solely on the basis of this morphology — longer teeth with the posterior cingulum for M, and wider teeth without the posterior cingulum for Mo. The following morphology is described based on isolated teeth referable to M, and M.^. M, differs only slightly from Mj in crown pattern. The anterior cingulum is distinct on both M, and M2 but unlike Adjidaumo it does not connect to the meta- lophid. The anterior cingulum extends from near the anterior base of the metaconid to the anterobuccal corner of the proto- conid. On one specimen (CM 33793, M,). the anterior cingulum is separated from the protoconid by a small notch, and nearly so on the holotype, CM 33786, also M,. On the other specimen (CM 33795, M,), the lingual half of the protoconid is excavated and the anterior cingulum is connected to the protoconid on its an- terior side. All the specimens except for one referable to Mj are worn at least on the trigonid. On this unworn specimen (CM 33808), the morphology of the anterior cingulum is close to CM 33795; connection to the protoconid is as in Paradjidaumo. The trigonid is clearly higher than the talonid. The metaconid is the most prominent cusp and a little taller than the protoconid. The entoconid is conical and higher than the hypoconid, but lower than the protoconid. From the protoconid a short ridge descends linguad and meets a short ridge, which also descends buccally from the transversely elongated metaconid. These two ridges are arranged transversely and represent what is called the metalo- phid. The posterior arm of the protoconid runs posterolinguad and meets the short anterior arm of the hypoconid. The junction lies near the center of the crown. From this junction, the me- solophid extends linguad only half way across the crown. The mesolophid is distinct and higher than in Adjidaumo minimus and A. doiiglassi. These three ridges are really lower than the trigonid and a little lower than the hypoconid. The posterior wall of the metaconid is vertical and the central valley is deeper than the anterior valley. A metastylid is not present. The metaconid is rather widely separated from the entoconid. The hypoconid is elongated and the posterior arm of it extends posterolinguad. A short ridge runs posterobuccally from the buccal corner of the entoconid and meets the posterior hypoconid arm at a right an- 34 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 9 gle. From this junction, the posterior cingulum runs posterolin- guad but it does not connect to the entoconid so that the pos- terior valley opens lingually on M,. The posterior valley is essentially of the same depth as the central valley. On M2, the hypolophid marks the posterior margin of the crown and is con- vex posteriorly. Discussion. — The present form differs from Ad- jidaumo in having higher crowned teeth and from Paradjidanmo in having weaker development of the mesoloph and mesolophid. The higher trigonid in this new genus is quite unique among the North American eomyids. I cannot find any form having the higher trigonid on the lower cheek teeth even among the European forms. Except for the higher trigonid, primitive features characterize the tooth structure of this new form. The weakness of the mesolophs and the mesolo- phids and the emphasis on the separate cusps are primitive among eomyids, as discussed by Wood (1973). In these respects, Parndjidnniuo and Cen- timanomys are more specialized in having long me- solophs and mesolophids. Meliakroitniomys is also specialized in having bilophate cheek teeth of which each loph is formed of two cusps and accessory structures are very small. Namatomys has a pecul- iar feature having the posterior protoconid arm fused to the base of the metaconid. Viejadjidainno has Ml with only the hypolophid but without the posterior cingulum. Yoderimys has a long lophid, which passes from the center of the anterior cingu- lum to the center of the metalophid on M, and M2. This removes Yoderimys from any close relation- ship with other eomyids. The narrow trigonid with near fusion of the protoconid and the metaconid on P4 in Aidolithomys is also unique and suggests its isolated position within the family. These rather specialized features make it difficult to believe that any of these genera, Namatomys, Viejadjidanmo, Meliakroimiomys, Yoderimys, Aidolithomys, Cen- timanomys, and Paradjidanmo, could give rise to Metadjidanmo. Adjidanmo retains the most gener- alized features of che k tooth structure among North American eomyids. The most remarkable modification seen in the present form from the original stock is the higher crowned cheek teeth. Although the posterior half becomes a little more highly elevated than in Adji- danmo, the anterior half becomes elevated above the posterior half, and the tendency is more em- phasized on lower cheek teeth than upper. This gives us the impression that the trigonid becomes higher than the talonid as if these teeth might have been rejuvenated to nearly the tribosphenic condi- tion. This trend took place in correlation with the development of hypsodonty of cheek teeth; the rate of increase in height is higher on the anterior half than the posterior half making the trigonid taller than the talonid on lower cheek teeth. The trend towards hypsodonty is seen also in Paradjidanmo hypsodns in the present fauna, but a differ ntial rate for height increase of different parts of a single crown is highly unusual. The talonid is not as high as the trigonid in the present form. This could mean that the degree of modification on the talonid might not be as great as on the trigonid in the original form. Compared with the talonid structure of Adjidanmo donglassi, a great resemblance is seen between the talonid struc- ture in the present form and A. donglassi. The hy- poconid is elongated sending its posterior arm pos- terolinguad. A short ridge originates from the buccal corner of the entoconid running postero- buccally and meets the posterior hypoconid arm rather nearer to its posterior end than to the hy- pocone. From the posterior end of the posterior hypoconid arm, the posterior cingulum runs linguad on Mj. On M.,, the posterior cingulum is reduced on A. donglassi, whereas it is gone on the present form. A short posterior protoconid and a short an- terior hypoconid arm, and the transverse mesolo- phid make a triple junction near the center of the crown. The mesolophid is rather short and does not connect to either the metaconid or the entoconid so that two valleys anterior and posterior to the mesoloph join together and run further linguad. The hypoconid itself leans linguad so that the tip of it is situated rather closer to the midline of the crown. Adjidanmo donglassi and the present form share the common features listed above. This suggests that they have a common ancestry sometime in the mid-Oligocene. As I show elsewhere in this article, A. donglassi is a direct descendent from A. minimus of the early Oligocene but not via A. minntns of the middle Oligocene. The present form may stand on a side branch from A. minimus. The trigonid structure of Metadjidanmo differs from that of A. donglassi, even though some simi- larities are seen. The protoconid is taller than the mesolophid and the ectolophid so that the latter connects to the protoconid on its posterior wall be- low the tip of this cusp on M, of A. donglassi. This is also true for M, of Metadjidanmo although the connection is well below the tip of the protoconid 1978 SETOGUCHI— CEDAR RIDGE LOCAL FAUNA 35 Table 8. — Dimensions of teeth of EomyiJae , genera indet., Type A and Type B. p4 M' M" M' or NT' M, Statistics L w L w L w L W L W Eomyidae, genus indet.. Type A N 1 1 2 2 1 1 OR 1.27 1.36 1.16- 1.29^ 1 .07 1 .07 1.21 1.33 Eomyidae, genus indet.. Type B N 3 3 1 I 2 2 OR 0.78- 0.92- 0.98 1.08 1.23- 1.23- 1.08 1.08 1.24 1.31 Mean 0.927 0.983 because it is a greatly elevated cusp. On Mj of A. douglassi, the metaconid has a wider base lingually and the base becomes narrower buccally. A ridge descends buccally but slightly anterad towards the buccal base of the metaconid and meets a short ridge descending linguad from the tip of the proto- conid. The junction is well below the tips of the protoconid and the metaconid, and even below the anterior cingulum. These two ridges combined to- gether are what is called the metalophid. This kind of condition may indicate the origin of the metalo- phid. The same morphology is seen on Mj of Metadji- danmo, although the metaconid and the protoconid are greatly elevated and share a common base. The elevation takes place almost vertically so that the posterior wall of the metaconid and the protoconid are vertical. The Junction of two ridges descending from the tips of the protoconid and the metaconid are a little below the anterior cingulum. The ante- rior valley is not deep and its floor lies a little below the junction between two ridges just mentioned above. The presence of the rather shallow anterior valley indicates that the whole trigonid becomes elevated from the base of the crown. The anterior cingulum is well developed and ex- tends along the whole anterior face of molars on both Adjidaumo douglassi and Metadjidoumo . But here a major difference takes place between these forms. In the holotype of the A. douglassi, the an- terior cingulum is joined to the anterolingual base of the protoconid, and the buccal part of the former is free but descends buccally joining to the proto- conid far below on its anterobuccal base. In the present form, the anterior cingulum does not con- nect to the metalophid but connects to the proto- conid making the anterior valley longer transverse- ly. The protoconid in A. douglassi is not as large and prominent when compared with the hypoconid, and on Mj, the trigonid is narrower than the talonid. On the other hand, in the present form, the proto- conid is more prominent and stout than the hypo- conid, and the trigonid is rather wider than the tal- onid on M, although the tooth itself is longer than wide. The wider trigonid could be associated with the heightening of this part. Except for the union of the anterior cingulum with the protoconid and the wider trigonid, an over- all similarity of crown structure is seen between A. douglassi and the present form. The absence of the posterior cingulum on M2 in the present form should be mentioned. On M2, the hypolophid forms the posterior border of the crown and the posterior cingulum is not present. This feature is highly spe- cialized for M2 and not seen commonly in the North America eomyids. Viejadjidaumo shows this fea- ture even on M,. On the holotype of Adjidaumo douglassi, the posterior cingulum is more reduced on M2 than on M, because the posterior hypoconid arm extends more linguad on the posterior border of the crown on Mo. The short posterior cingulum runs linguad along the posterior margin of the crown and ends posterior to the entoconid, but does not connect to the latter. The posterior valley is represented by a shallow pocket. If the reduction proceeds, the situation without the posterior cin- gulum on M2 will take place. 36 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 9 Eomyidae, genus indet., Type A (Fig. 20, Table 8) Referred specimens. — M‘: CM 33809; M-: CM 33810, CM 33811; M,; CM 33812. Locality. — Loc. 19, Badwater Creek, Wyoming. Age. — Late Oligocene. Description. — Three isolated upper molars and one lower mo- lar are available. They are low-crowned and of equal size. The determination of upper and lower molars are primarily based on the position of roots; both are three rooted; in upper molars, one anteroposteriorly elongated root is underneath the protocone and the hypocone, and two roots are underneath the paracone and the metacone, respectively; in lower teeth, one transversely elongated root is underneath the hypoconid and the entoconid, or underneath the talonid, and two roots are underneath the protoconid and the metaconid, respectively. One upper tooth (CM 33809) is longer than wide, probably M\ and this tooth is heavily worn. All the cusps are well defined and connected by rather thin ridges. The anterior cingulum is rep- resented by a thin ridge, which is thinner than the posterior cingulum and extends along the whole anterior border of the crown. The buccal end of the anterior cingulum is connected to the anterior base of the paracone. A rather thick ridge connects the anterior cingulum with the anterobuccal corner of the pro- tocone. The lingual portion of the anterior cingulum is free. The floor of the anterior valley is unworn as are the floors of the central and the posterior valleys. All three valleys are of equal depth. The base of the protocone is wider anteroposteriorly than the remainder of cusps. The protoloph is curved anteriorly mak- ing the anterior valley narrow. The central valley is wide and opens buccally. The mesoloph is not defined clearly. A short and rather thick mure connects the posterobuccal corner of the protocone with the metaloph. The posterior cingulum extends only half way to the buccal border along the posterior margin of the tooth. The posterior valley opens buccally. Two upper molars are wider than long and are probably M'^s. One of them (CM 33811) is almost unworn. The crown structure is almost the same as the one described just above. The four cusps are stout and rather tall, and all the ridges are thin and low in position; thus the crown pattern emphasizes cusps rather than ridges. The anterior arm of the protocone with a thick base descends from the stout protocone anterobuccally and its ante- rior extremity connects to the anterior cingulum. A ridge also with a thick base descends almost transversely from the para- cone and joins the anterior protocone arm forming the protoloph . The posterior arm of the protocone descends linguad hut slightly posterad and near the center of the crown turns posterad to join the anterior arm of the hypocone. The mesoloph is not well defined; two short, thin and low ridges are slightly elevated from the floor of the central valley just buccal to the mure — one is near the anterior end of the mure, and the other is near the middle of it. The buccal border between the paracone and the metacone is slightly elevated but well below the tips of these cusps. The posterior cingulum is rather long and connects to the metacone on its posterior base. One lower molar (CM 33812) is longer than wide and the tri- gonid is narrower than the talonid. It is probably M,. This tooth also emphasizes cusps rather than ridges. The anterior cingulum is well developed extending along the whole width of the tooth. The base of it is rather wide and it is not as close appressed to metalophid. The lingual end of the anterior cingulum is fused to the metaconid at its anterior base. The anterior cingulum does not connect to either the metalophid nor the protoconid so the V-shaped valley extends between the anterior cingulum and the metalophid, and opens buccally. The metaconid and the ento- conid are conical and taller than the buccal cusps. The size of the entoconid is reduced compared with that in other eomyids. The metaconid and the entoconid are widely separated. No meta- stylid is seen. The hypoconid is more stout than the protoco- nid. The hypolophid is the most prominent of the ridges. The rather weak metalophid connects the protoconid with the meta- conid. The ectolophid is also weak and lower in position than both the metalophid and the hypolophid. The mesolophid is short and as weak as the ectolophid. The mesolophid extends linguad only half way to the lingual border and merges to the floor of the central valley posterior to the metaconid. The posterior cin- gulum is short and restricted to only the lingual quarter of the posterior margin. The posterior valley is represented by a small pocket situated posterolingual to the entoconid. Discussion. — In size this form is close to Adji- daumo douglassi but is a little larger than it. The well developed cusps and weakness of lophs and lophids in the present form are unusual among eomyids. The reduction of the entoconid is seen on M2 in Adjidoumo douglassi as compared to this cusp on M, of this species. This, together with the well-developed anterior cingulum indicates the close relationship of this form to Adjidaumo. In Adjidauino, lophs and lophids are better developed than in the present form. At present, I cannot find any close relation of this form to any other eomyids. Eomyidae, genus indet.. Type B (Fig. 21, Table 8) Referred specimens. — P‘*-M‘; CM 33813; P'': CM 33814, CM 33815; M, or M^; CM 17434, CM 33816. Locality. — Loc. 19, Badwater Creek, Wyoming. Age. — Late Oligocene. Description. — An upper jaw fragment (CM 33813) with P'* and M‘ is available. Size is close to Adjidaumo douglassi. Both teeth are heavily worn but they are higher crowned than in A. doug- lassi. Unlike Adjidaumo. the mesoloph is long and reaches across the crown surface to the mesostyle, which is thin and elongated transversely on M'. Both the paracone and the meta- cone are elongated transversely and of equal size. The anterior and the posterior cingula are well developed and both are re- stricted only on the buccal half of the tooth. The anterior cin- gulum is closely appressed to the protoloph. No lingual portion of the anterior cingulum is seen. The protocone and the hypo- cone are close together, and the valley between them is narrow running linguad and posterad as well. The fourth premolar associated with M‘ on the same upper jaw is heavily worn. Two isolated Ps (CM 33814, CM 33815) are present. A root beneath the paracone of these teeth extends 1978 SETOGUCHI— CEDAR RIDGE LOCAL EAUNA 37 anterobuccally indicating that the tooth is the first tooth of the cheek tooth series. All the cusps are well defined. The hypocone is the largest and extends more linguad than the protocone so that the posterior half is wider than the anterior half. A short anterior cingulum is present on the buccal quarter of the anterior margin of the tooth. This cingulum does not connect to the para- cone. The mesoloph extends only half way across the crown to the buccal border. No mesostyle is present but the buccal border between the paracone and the metacone is slightly elevated. The posterior cingulum is long and connects to the hypocone on its posterobuccal corner making the posterior valley longer trans- versely. The lingual half is narrower anteroposteriorly than the buccal half. Two isolated lower molars (CM 17434, CM 33816) are at hand. On both, the buccal half is narrower anteroposteriorly than the lingual half. Both specimens have the well-developed anterior cingulum connecting to the tip of the protoconid. The central valley is deeper than the anterior valley. The lingual wall is el- evated with a thick base, but it is below the metaconid and the entoconid. A mesostylid is present on one specimen but not on the other. The mesolophid is well developed reaching to the buccal wall but the connection is below the mesostylid. The posterior cingulum is not present. Discussion. — The above described specimens are characterized by small size, long mesolophs and mesolophids, narrower lingual half on upper and buccal half on lower teeth, and high-crowned teeth. Some similarities between this form and Par- adjidaumo are seen — higher-crowned teeth, long mesolophs and mesolophids, and anterior cingulum connecting to the protoconid on the lower molars. The present form is smaller than Paradjidaumo, and has a narrower lingual length on the upper and buccal length on the lower teeth. I cannot find any close relatives of this form. Eamily Heteromyidae Allen and Chapman, 1893 Heteromyid specimens are reasonably common in the Badwater assemblage, but consist almost en- tirely of isolated teeth. Only in 13 specimens are two or more cheek-teeth in association. Two species are present. Specimens referable to M* are clearly able to be separated into two size categories. One, having greater length and width, emphasizes stronger lophs and deeper transverse valley, and the other, smaller type, is characterized by inde- pendent cusps and a shallow transverse valley being essentially the same depth as the anteroposterior valley. The second upper molars are also divisible into two size categories, but the separation is not as clearcut. In the first lower molars, the larger size group usually has stronger lophs and lophids than the smaller. Proheteromys sp. cf. P. nebraskensis Wood, 1937 (Fig. 22, Table 9) Referred specimens. — M‘: CM 33817-33842; M^: CM 17423, CM 17426, CM 17427, CM 33843-33856; M,; CM 17425, CM 33857-33879; Ma; CM 17429, CM 17431, CM 33880-33885. Locality. — Loc. 19, Badwater Creek, Wyoming. Age. — Late Oligocene. Description. — M‘ of this species is considerably larger than that of Heliscomys in this fauna. The size of the former is very close to that of Heliscomys schlaikjeri, and a little larger than that of H . teiuiiceps. Morphologically, the present form is very close to H. schlaikjeri. As in H. schlaikjeri. M' is more lophate than are those of the middle Oligocene species of Heliscomys, although the principal cusps are still prominent. On most spec- imens, the protocone and the metacone are of equal size, but the protocone is lower than the paracone even on unworn stage. Several specimens show that whereas the protocone and the hypocone are worn a little, the paracone and the metacone are not worn at all, consequently the paracone becomes much higher than the protocone. The paracone is situated anterobuccal to the protocone so that the tooth is longer and the anterior half of the tooth is considerably wider than the posterior half. The char- acteristic feature of the anterior half of the tooth being wider is also seen in M, of Heliscomys schlaikjeri. although Black ( 1961) did not mention this in his description of his new species. On most specimens, the protocone unites with the paracone only at its base, and these two cusps share a common base, which is raised above the floor of the transverse valley. The notch be- tween the protocone and the paracone is lower than the notch between the hypocone and the metacone. The anterior cingulum is strong and runs linguad from the anterior base of the paracone, but well below the paracone and the protocone. The lingual cin- gulum exhibits only one large cusp opposite the lingual end of the transverse valley, and shows no evidence of having been divided into two cusps, as in H. tenuiceps. This cingular cusp is smaller and lower than the hypocone, but much higher than the anterior cingulum, which unites with the lingual cingular cusp at its an- terior base. The cingular cusp connects to the hypocone near its base and the anterior cingulum. The transverse valley is con- fluent with the valley between the anterior cingulum and the protocone, although the former valley is deeper than the latter, but is blocked by the cingular cusp and the hypocone postero- lingually. The hypocone and the metacone are subequal and share a common base, which is high above the floor of the trans- verse valley. The valley, which separates the metacone from the hypocone, is elongated anteroposteriorly and is blocked poste- riorly by the posterior cingulum. The short posterior cingulum runs buccally from the posterobuccal corner of the hypocone to the posterior base of the metacone. With wear, the posterior cingulum is obliterated as in Heliscomys tenuiceps and H. schlaikjeri. M^ of this form is identified based on larger size and more lophate morphology. The morphology of M^ almost agrees with that of M'. On M^, the crown is transverse but the anterior half of the tooth is not greatly wider than the posterior half, even though the former is a little wider than the latter. The paracone and the metacone are closer together than in M'. The floor of the transverse valley between these buccal cusps is raised so that the valley becomes shallower buccally. The protocone is 38 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 9 Table 9. — Dimensions of teeth o/ Proheteromys sp. cf. P. nebraskensis. Statistics M‘ M^ M. M2 L AW PW L AW PW L W L W N 26 26 26 17 17 17 24 24 8 8 OR 1.08- 1.21- 1.09- 0.87- 1.12- 1.08- 1.11- 1.01- 0.96- 1.07- 1.27 1.51 1.38 1.00 1.34 1.25 1.30 1.19 1.06 1.20 Mean 1.149 1.416 1.255 0.945 1.210 1.154 1.183 1.117 1.026 1.136 SD 0.050 0,071 0.075 0.038 0.056 0.050 0.047 0.047 0.035 0.047 CV 4.34 5.03 6.01 3.98 4.61 4.35 3.94 4.22 3.37 4.10 lower than the paracone. With wear, the protocone and the hy- pocone are lowered faster than the paracone and the metacone are. The anterior cingulum is also prominent but lower than the protocone. The lingual cingulum is very thick and tall. No dis- tinct cusps are seen on the lingual cingulum. The lingual cingu- lum is highest at the point opposite the lingual end of the trans- verse valley. From there, a ridge descends anterobuccally and is confluent with the anterior cingulum. Posteriorly, the lingual cingulum connects to the hypocone at its lingual base. On one specimen (CM 33844), the lingual cingulum tends to divide into two parts by a small notch, and on one specimen (CM 33846), this cingulum is completely divided into two parts; both parts are conical; the anterior one is small and is anterolingual to the protocone, and the posterior one is larger and is anterolingual to the hypocone. On both specimens (CM 33844, CM 33846), all six cusps are not united by lophs, and the transverse valley and the anteroposterior valley are of the same depth. This situation is very close to that seen in Heliscomys. Lower fourth premolars are not available. M, is square-shaped and somewhat longer than wide. The principal cusps are prominent. Among them, the protoconid and the entoconid are subequal and smaller than both the metaconid and hypoconid. The metaconid is the highest cusp, and the other three cusps are of the same height. After a little wear, the tips of the protoconid and the hypoconid are truncated, whereas the metaconid and the entoconid remain unworn. Then, the differ- ence in height between the buccal cusps and the lingual cusps becomes more emphasized. The differential height in cusps of M, is just the mirror image of that of M’; the buccal cusps are higher than the lingual ones in M,, whereas the opposite con- dition holds in M‘. The protoconid and the metaconid share a common base as do the hypoconid and the entoconid. These bases are elevated high above the floor of the transverse valley so that the transverse valley is deeper than the anteroposterior valley. The notch between the protoconid and the metaconid is a little higher than the valley between the anterior cingulum and the protoconid. On most specimens, the anterior cusps do not form a distinct loph. But on several specimens, especially on CM 33859 and CM 33858, the protoconid and the metaconid connect broadly to each other at their base forming the meta- lophid, although two cusps are still prominent. The anterior cin- gulum is strong although it is low leaving a transverse valley be- tween it on the one hand and the protoconid and the metaconid on the other. The anterior cingulum connects to the buccal cin- gulum at the anterobuccal corner of the tooth. No connection between cingula and the protoconid is seen. The protostylid is represented by a thick and elevated ridge, which is merely the posterior continuation of the buccal cingulum. The protostylid is situated posterobuccal to the protoconid. The hypostylid is more conical and of the same height as the protostylid. The hypostylid connects to the hypoconid at its base. Both the proto- stylid and the hypostylid are significantly lower than the proto- conid and the hypoconid. These stylids are widely separated by a valley, which is confluent with the transverse valley. The de- gree of development of the posterior cingulum varies. On most specimens, the short posterior cingulum runs buccally from the post robuccal corner of the entoconid to the posterior base of the hypoconid. A small cingulum connects the hypostylid with the hypoconid at the posterobuccal corner of the tooth, but this cingulum does not connect to the posterior cingulum between the hypoconid and the entoconid, on most specimens. On some (for example, CM 33874), these two cingula unite with each other and surround the posterobuccal base of the hypoconid. M2 of this form is identified based on more lophate morphol- ogy. The size of M, is not significantly larger than that of Hel- iscomys in this fauna so that the difference in size is not appli- cable for separation into two groups. Mj is wider than long. The protoconid is subequal to the metaconid and the hypoconid. The entoconid is the smallest among the principal cusps. The pro- toconid and the metaconid are transversely elongated and united broadly with each other at their base forming the prominent metalophid. The low entoconid unites with the hypoconid at its base, but the hypolophid is not so strongly lophate as in the metalophid. The anterior cingulum is as in M[. The protostylid is more conical than on Mj. The hypostylid is significantly lower and smaller than the protostylid. On most specimens, the pos- terior cingulum is absent, but on one specimen (CM 33884), a rudimentary cingulum runs along the posterior bases of the hy- poconid and the entoconid. The morphology of M,, varies. The protoconid and the meta- conid are subequal but the former is a little higher than the latter. The transverse valley is deep separating the metalophid from the hypolophid. The entoconid is reduced in size. The hypoconid is lower than the anterior cusps. The anterior cingulum is present on most specimens, but weak. The protostylid is a small cusp on the buccal cingulum or absent. The hypostylid is greatly re- duced if present, or absent. The transverse valley opens buc- cally. No posterior cingulum is present on all specimens avail- able. Discussion. — The upper dentition of this form is very close to that of Heliscomys schlaikjeri. The holotype of H. schlaikjeri is worn a little so that it is rather difficult to compare it precisely with the present form. The size is almost the same. That the molars are more lophate and the anterior half of M* 1978 SETOGUCHI— CEDAR RIDGE LOCAL FAUNA 39 Table 10. — Dimensions of teeth o/ Heliscomys ip. cf. H. vetus. Statistics P' M‘ M^ "4 M, M., L W L AW PW L AW PW L W L W L W N 27 27 27 27 27 15 15 15 9 9 17 17 15 15 OR 0.64- 0.66- 0.74- 0.96- 0.92- 0.72- 0.89- 0.86- 0.48- 0.52- 0.88- 0.89- 0.85- 0.94- 0.91 0.93 0.94 1.16 1.11 0.92 1.10 1.07 0.73 0.72 1.07 1.08 1.00 1.14 Mean 0.739 0.807 0.859 1.046 0.990 0.806 0.994 0.961 0.623 0.634 0.970 0.967 0.923 1.004 SD 0.068 0.078 0.053 0.061 0.055 0.062 0.063 0.057 0.073 0.069 0.062 0.062 0.055 0.056 CV 9.17 9.69 6.21 5.86 5.53 7.66 6.29 5.93 11.79 10.87 6.36 6.39 5.94 5.62 is wider than the latter is clearly shared in both forms. I have not found any reason to separate the present form from H. schlaikjeh. The present form and H. schlaikjeh differ from H. tenuiceps of the middle Oligocene in having a wider anterior half of M'. The size of//, tenuiceps is a little smaller than that of both the present form and H. schlaikjeh. Heliscomys tenuiceps and H. schlaikjeh, which are known only from the upper dentition may eventu- ally be removed from the genus. This problem will be discussed below. The most remarkable feature seen in the upper dentition of the present form is the presence of the smaller protocone than the paracone. The holotype of H. schlaikjeh is worn and the protocone unites with the paracone with wear on this specimen. It is impossible to tell whether H. schlaikjeh might have had the smaller protocone on M* or not. The pres- ence of a small protocone on the upper molars is unusual among heteromyids. The morphology of the smaller protocone on M* is reflected in the mor- phology of ML too. On Ml, the protoconid is smaller than the para- conid. The size and general morphology except for the smaller protoconid are very close to Prohetero- inys nebraskensis. The holotype ofP. nehraskensis is worn. The paratype established by Wood (1937) is an almost unworn specimen. Wood stated ( 1937; 215) that “(InP. nebraskensis ) The protostylid of the molars is separate from the cingulum when unworn, and it far to the rear. The protoconid and metaconid are connected by a cingulum along their anterior margin.” All the specimens at hand referable to M, have the anterior cingulum, which is clearly sepa- rated from the protoconid and the metaconid. On most specimens at hand, the protostylid is con- nected to the buccal cingulu, but on one specimen (CM 33876), the protostylid is clearly separated from the cingulum. The degree of the development of the anterior cingulum usually varies among het- eromyids. I believe the strong or weak development of the anterior cingulum is not a good criterion for separating species. Although the present form has a small protoconid on Ml, I believe the morphology of the present form is not clearly separable from Proheteromys nebras- kensis. The present form is best described as Pro- heteromys sp. cf. P. nehraskensis. And, I consider Heliscomys schlaikjeh to be conspecific with the present form. Heliscomys sp. cf. H. vetus Cope, 1873 (Fig. 23, Table 10) Referred specimens.— CM 17421, CM 33886-33889; P- M^: CM 17091, CM 19879, KU 16626; P-M^: CM 19787; PT CM 19790, CM 33890-33906; M'-M^; CM 33907; M‘: CM 17092, CM 33908-33923; M^: CM 33924-33935; P4-M,; CM 19786; P^-M,: CM 33936; P4; KU 16627, CM 17432, CM 19788, CM 33937- 33940; M,; CM 17424, CM 33941-33954; M.^-M.,: CM 33955; M.; CM 17430, CM 33956-33966. Locality. — Loc. 19, Badwater Creek, Wyoming. Age. — Late Oligocene. Description. — Nine upper jaw fragments with P'* and M‘ are referable to Heliscomys. The size, both length and width, varies, even among these specimens. Materials referable to P^ are not able to be separated into two groups on the basis of size. All the specimens referable to heteromyid P are described below as Heliscomys, but surely some of them must be referred to Pro- heteromys. At present, I cannot tell morphological differences between the upper fourth premolars of Heliscomys and Pro- heteromys in the Oligocene. P shows a pattern of a large, anteriorly placed protocone and a three-cusped metaloph. The protocone and the hypocone are subequal and conical on most specimens. The metacone is a little smaller than the hypocone. On several specimens, the pro- tocone is smaller than the hypocone and subequal to the meta- cone. The degree of the development of the entostyle varies; on some specimens, the entostyle is a small, low cusp; on some, it is a rudimentary cusp on the lingual base of the hypocone; on two specimens (CM 33891, CM 33897), the entostyle is not seen at all and the tooth has three cusps, one on the anterior loph and two cusps on the posterior loph. Galbreath stated (I953;63) that there seems to be a definite correlation between the size of the premolar and the amount of reduction of the entostyle. In the present fauna, no such correlation is seen; the cuspidate ento- 40 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 9 Fig. 23a-e. — Heliscomys sp. cf. H. vetus. a: KU 16626, skull with left lateral view, b: same, ventral view, c: same, crown view of teeth, d: KU 16627, left lower jaw with P4, lateral view, e: same, crown view. style is seen on both large and small teeth, and the rudimentary entostyle is also observed on both types. Most specimens show a tendency towards the formation of transverse lophs. The hy- pocone and the metacone share a common base and the valley separating these cusps is not deep. These two cusps are not discrete and with further wear they lose their individual identity. On one specimen (CM 17091), however, no tendency towards the formation of a loph is seen and the cusps are discrete. Even on this specimen, the entostyle is situated anterolingual to the hypocone as in another specimen. With wear, the metaloph and the entostyle give rise to an incipient J-pattern. The entostyle, if present, does not connect to the protocone. On some speci- mens, there is a small cusp at the anterobuccal base of the pro- tocone. On others, no cusp is seen there. The size of M‘ is considerably smaller than that of Prohelero- inys. The morphology of M* greatly varies. Some of them are not separable morphologically except by size from the corre- sponding tooth of Proheteromys. The principal cusps are sub- equal. The transverse valley is a little deeper than the antero- posterior valley. Great morphological variation is seen in the structure of the lingual cingulum. On some specimens (for ex- ample, CM 33907), the small but distinct protostyle is present just lingual to the protocone. It is separated from the anterior cingulum by a small notch, and from the entostyle by the trans- verse valley. In this form, the transverse valley is opened lin- gually. The entostyle is lingual and a little anterad to the hypo- cone, and only slightly smaller than the hypocone and the metacone. On one specimen (CM 33918), a small cusp is present at the anterolingual corner of the tooth and posterior to it another smaller cusp is present just linguad to the protocone. The ento- style is rather prominent. These three cusps form the lingual border of the tooth so that the transverse valley is not opened lingually. One specimen (CM 33917) shows the entostyle shifted more anteriorly closer to the small protostyle and it tends to block the transverse valley lingually. On this specimen, the ento- style is separated from the protostyle by a small notch, which lies little above the floor of the transverse valley. On other spec- imens (CM 33910, CM 33914), the entostyle is shifted further anteriorly just lingual to the transverse valley. From the ento- style, a thick ridge descends anteriorly and is confluent with the anterior cingulum. In this situation, it is impossible to distinguish either a protostyle or entostyle in this ridge. Thus the central valley is completely blocked by this lingual cusp. This mor- phology agrees exactly with that seen in Proheteromys. The pro- tocone is of the same size as the paracone on M'. Otherwise, morphologically this tooth is not easily separated from the cor- responding tooth of Proheteromys. M^ agrees in most respects with the patterns seen on M', and the morphological variation among individuals seems to be re- duced. The morphological differences from the pattern seen in M’ are in the construction of the lingual cingulum and the styles. On M^ there is a high ridge which closes the transverse valley at the lingual margin. In this ridge it is impossible to distinguish either a protostyle or entostyle. On several specimens (CM 33930, CM 3393 1 , CM 33933), a small notch lingual to the trans- verse valley divides the lingual cingulum into two portions. The notch is not deep and both parts of the cingulum do not form any cusp. On all specimens, the connection of the lingual cin- gulum to the hypocone is stronger and at a higher level than is the connection to the anterior cingulum. The mandible is rather slender. The diastema is short and the diastemal depression is shallow. The anterior end of the mas- seteric fossa is swollen and makes the jaw appear massive at this point. The mental foramen lies anterior to P4, and almost on the dorsal surface of the mandible. Nine specimens referable to heteromyid P4 are available. They cannot be separated into two groups only based on size. Two jaw fragments with at least P4 and M, are available, of which lower first molars are referable to Heliscomys. All the materials 1978 SETOGUCHI— CEDAR RIDGE LOCAL EAUNA 41 of P4 in this fauna will be described below as Heliscomys but the possibility that some of them should be referred to Proheteroinys is not ruled out. Eight specimens of P4 out of nine are quadri- cuspate. On most specimens, the m taconid, hypoconid, and entoconid are of equal size and height. On some specimens (CM 17432, CM 33938), the metaconid is the smallest and the hypo- conid is the largest among these three cusps. Seven specimens out of eight quadriscuspate lower fourth premolars have the tiny protoconid, which is significantly smaller and lower than the metaconid and situated on the anterobuccal corner of the tooth. On six specimens out of these seven, the protoconid is separated by small notches from both the metaconid and the hypoconid. On one specimen (CM 33936), the protoconid is united with the hypoconid by a short ridge, but separated by the anteroposterior valley from the metaconid. On one specimen (CM 33903), the protoconid is much larger than on the other specimens although it is a little smaller than the metaconid. On this specimen, the protoconid is separated from both the metaconid and hypoconid. On all eight specimens there is an indication of a hypoconulid between the hypoconid and entoconid at the posterior margin of the tooth. One specimen (CM 33937) is triscuspate. On this, the metaconid is situated just anterad to the midline between the hypoconid and entoconid. The hypoconid is the largest and the other two cusps are of equal size. No shelf is seen on the an- teroexternal corner of the tooth. No indication of a hypoconulid is seen. The structure of M, and M2 is very close to those of Helis- comys veins described by Galbreath (1953) from the middle Oligocene of Colorado. The pattern of lower molars is that of four well-developed primary cusps more or less bordered on three sides by low cingula, which develop cusps. The teeth are cuspate rather than lophate, but with wear, lophs tend to form. The protostylid is larger than the hypostylid. but smaller and lower than the protoconid. On most specimens, the anterior cingulum is united to the anterolingual angle of the protoconid by a weak crest. The connection of the anterior part of the buccal cingulum to the protostylid varies; these two structures are sep- arated by a notch, or where the cingulum is weak, it may unite to the protostylid without any notch between them. The trans- verse valley separates the protostylid from the hypostylid. Pos- terior to the hypostylid the cingulum varies from strong to weak, on some specimens extending across the posterior face of the entoconid and in others fading out on the posterior face of the hypoconid. M3 is composed of four well-developed primary cusps. Stylids are reduced. The hypoconid is somewhat reduced and is the smallest of the primary cusps. The anterior cingulum is weak on the face of the metaconid, absent at the midline of the anterior border of the tooth, and stronger on the anterior and buccal faces of the protoconid. The posterior cingulum is greatly reduced; on two specimens, it is absent completely, but on one specimen a small shelf lies on the posterior border between the hypoconid and the entoconid. Discussion. — In size these specimens appear to be slightly larger than the mean for the middle Oli- gocene populations from Colorado discussed by Galbreath (1953:65). However, most of them fall well within the size range given for the Colorado specimens. Also, the structure of the cheek teeth and the variation seen in the present material co- incide well with those which are observed in the earlier Colorado populations. The same situation was reported by Black ( 1965:45) for the early Oiigo- cene population. It must be noted that the morphological variation observed in the middle Oligocene populations is duplicated in the late Oligocene Badwater fauna, and also in the early Oligocene Pipestone material (Black, 1965); in the late Oligocene Badwater pop- ulation, the three-CLisped condition of P4 as well as the four-cusped condition are both represented. Both the three-cusped and four-cusped condition persisted from the early Oligocene through the late Oligocene. Wood (1939:560), Wilson (19496:115), Galbreath (1953:65) and Black ( 1965:45) have all suggested that the four-cusped condition of P4 was primitive and that the three-cusped condition rep- resented reduction from that more primitive stage. This would certainly seem the most probable evo- lutionary pattern in Heliscoinys. In the late Oligo- cene population, the three-cusped condition is seen on only one specimen out of nine of P4. Although the sample size is not big enough to determine def- initely, it would appear that the selection was fa- voring the four-cusped condition and the reduction or eventual loss of the protoconid on P4 was under somewhat strong selective pressure through the Oligocene Epoch. If this were the case, this varia- tion suggests that Heliscoinys was ancestral to Pro- heteromys. Affinities of the Oligocene Heteromyids The Oligocene and the early Miocene heteromy- ids having the three-cusped P4 and more cuspidate molars tend to be assigned to Heliscoinys , and those having the four-cusped P4 and more lophate molars to Proheteroinys. The morphological varia- tion seen in P4 of Heliscoinys vetns creates a prob- lems for the taxonomic assignment. Heliscoinys veins has the three-cusped as well as the four-cusped P4. The difference in morphology of P4 is not a good criterion any more for the sep- aration of Heliscoinys from Proheteroinys. It seems to me that the best criterion for separation of these genera is the degree of development of lophs and lophids on molars; Proheteroinys is more lophate and Heliscoinys is more cuspidate. If this is true, Heliscoinys teniiiceps and H. schlaikjeri must be removed from the genus Heliscoinys and referred to Proheteroinys, because both forms have clearly lophate crown patterns. 42 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 9 1978 SETOGUCHI— CEDAR RIDGE LOCAL FAUNA 43 On Ml of both forms just mentioned, the lingual cingulum is high enough to block the transverse val- ley lingually. In Heliscomys vetiis, Mi has a mod- erate lingual cingulum having two styles, which are separated from each other by the continuation of the transverse valley. So the transverse valley is open lingually on this form. But the morphology of M, of H. veins shows a considerable range of vari- ation as stated in the description. From the “nor- mal” situation of the lingual cingulum, it tends to unite two styles into single style and form a high ridge to block the transverse valley lingually. In an undescribed collection many specimens referable to Heliscomys veins have exactly the same morpho- logical variation. The variant having the strong lin- gual cingulum with a single style blocking the trans- verse valley is not easily separated from Heliscomys ienniceps morphologically other than size. H. len- niceps and H. schloikjeri are larger than H. veins. This suggests that H. veins should have given rise to H. schlaikjeri via Proheleromys nehraskensis- stage in the late Oligocene. Family Cricetidae Rochebrune, 1883 Eumys parvidens Wood, 1937 (Fig. 24, Table 11) Referred specimen. — M,: CM 32939. Localily. — Loc. 19, Badwater Creek, Wyoming. Age. — Late Oligocene. Description. — M, is square-shaped, except for the anterocone. The tooth is smaller than that of Eumys elegans. The anterocone is large and situated on the buccal side of the tooth. The anterior cingulum is very short and curves posterobuccally; the poste- robuccal end connects to the anterior arm of the protocone. A short ridge runs posterobuccally from the summit of the ante- rocone but does not reach the buccal border of the tooth. No connection between the anterocone and the paracone is seen so that a narrow valley runs buccally between these cusps. The protocone is stout and somewhat elongated anteroposteriorly . The anterior arm of the protocone descends anterobuccally from the anterobuccai corner of the protocone and turns anterolin- gually connecting to the anterior cingulum. A small valley is seen in front of the protocone. This valley is blocked anterolingually by a thin ridge forming the anterolingual corner of the tooth. The paracone is round and as high as the protocone, which is slightly higher than the anterocone. The paracone is situated buccal to the posterior half of the protocone. The mure is a thin ridge descending posterobuccally from the posterobuccal corner of the protocone turning posterolingually from the lingual extremity of the mesoloph. The protolophule II originates at the posterolin- gual corner of the paracone and runs lingually but slightly pos- teriorly. This connects to the mure at the middle of it between the posterobuccal corner of the protocone and the mesoloph. The mesoloph is short but distinct. This merges to the anterior border of the metacone at the midpoint of it so that a narrow valley runs buccally between the paracone and the mesoloph. The hypocone is stout and as high as the protocone forming the posterolingual corner of the tooth. A deep valley runs lingually between the protocone and the hypocone but is blocked lingually by a tiny entostyle. The metacone is somewhat compressed anteroposteriorly, and is situated just posterior to the paracone and buccal to the hypocone. It is almost as high as the paracone. The metalophule II runs almost lingually from the posterolingual corner of the metacone and connects to the body of the stout hypocone. A valley is present between the metacone and the hypocone but blocked by both the mesoloph anteriorly and the metalophule II posteriorly. The posterior cingulum runs buccally behind the metacone and connects to the posterobuccal corner of the metacone so that the valley just behind the metacone does not open buccally. All the ridges are very low. Discussion. — This taxon is represented by a sin- gle tooth. The tooth is characterized by well-de- fined cusps and low ridges. The structure of the tooth differs from that seen in Eumys elegans in having the protocone and the hypocone essentially as high as the paracone and the metacone, antero- posteriorly elongated protocone, and low ridges. These characteristic features are less specialized than those of Enniys elegans, which shows more developed ridges. I agree with Wood (1937) who stated, “In general, this form {E. parvidens) has a primitive Enmys pattern on a small scale.” Eumys elegans Leidy, 1856 (Fig, 25, Table 1 1) Referred specimens. — M‘: CM 32918-32921. CM 32940; M-: CM 32922, CM 32923, CM 33109, CM 33110, CM 33112; M^; CM 32926-32934; M.-M.: CM 17086; M,: CM 32900-32905, CM 32907, CM 32908, CM 32910, CM 32911, CM 32935, CM 32936; M.; CM 32912-32917, CM 32937, CM 33113; M,; CM 32938. Localily. — Loc. 19, Badwater Creek, Wyoming. Age. — Late Oligocene. Description. — On M', the anterocone is prominent and trian- gular in shape, with the apex leaning anteriorly. The anteroloph, or the anterior cingulum forming the base of the trigon of the anterocone, is transverse. The anterior arm of the protocone connects to the anterocone at the midpoint of the base of the triangle in most specimens. The paracone and the metacone are Fig. 23f. — Heliscomys sp. cf. H. vetiis (continued), f; CM 33936, left P4-M3. xl5. Fig. 24. — Eumys parvidens. CM 32939, left M'. x8. Fig. 25. — Eumys elegans. a; CM 32919, left M'. x8. b; CM 17086, right M,-M2. x6. c: CM 32922, right M^. x8. Fig. 26. — Eumys brachyodus . a; CM 17412, left Mh b: CM 32925, left M^. c; CM 32909, right M,. x8. Fig. 27. — Eumys. sp. cf. E. planidens. a; CM 32941, right M'-M^ b; CM 19799, right Mj-Mj. x8. 44 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 9 Table II. — Diniensions of teeth o/ Eumys parvidens, Eumys elegans, and Eumys brachyodus. Statistics M> M^ M" M, Mj M, L W L W L W L W L W L W Eumys parvidens N 1 1 OR 2.40 1.45 Eumys brachyodus N 1 1 3 3 2 2 OR 2.89 1.89 1.82- 1.83- 1.86- 1.41- 1.91 1.93 2.13 1.68 Mean 1.860 1.880 Eumys elega^ ns N 2 2 3 3 5 5 10 10 7 7 I 1 OR 2.84- 1.70- 2.03- 1.77- 1.44- 1.51- 2.09- 1.51- 2.07- 1.71- 1.96 1.78 2.98 2.02 2.07 2.03 1.67 1.80 2.70 2.07 2.31 2.01 Mean 2.050 1.880 1.568 1.660 2.293 1.715 2.186 1.871 SD 0.171 0.178 0.101 0.111 CV 7.45 10.37 4.60 5.91 subequal but the latter is somewhat elongated transversely. Both are slightly taller than the protocone and the hypocone. The protocone is compressed posteriorly so that the posterior wall is vertical, whereas the anterior wall descends gently. Erom the summit of the protocone two ridges originate — one runs antero- buccally forming the anterior arm of the protocone, and the other runs almost transversely forming the protolophule II and con- nects to the paracone. The paracone is situated just buccal to the protocone. The mure is very short and runs anteroposte- riorly. This connects to the protolophule II near the lingual cor- ner of the paracone. The metaloph is very short but distinct. On most specimens the metaloph is situated between the paracone and the metacone, but on some (for example, CM 32920) it is closely related to the metacone though it does not unite to the latter. The hypocone is stout forming the posterolingual corner of the tooth. The anterior wall of the hypocone descends gently towards the posterior base of the protocone so that the valley between the protocone and the hypocone is very narrow in con- trast to the wide valley in Eumys pUinidens. Erom the summit of the hypocone two ridges originate — one passes anterobuccally forming the anterior arm of the hypocone and the other runs buccally forming the posterior cingulum. The metalophule unites to the anterior arm of the hypocone just anterior to the hypo- cone. Because of the transversely elongated metacone, the meta- lophule is very short. All the ridges are rather thick and high above the crown of the tooth but they do not rise to the summits of the cusps. is longer than wide. All the cones are compressed antero- posteriorly but are still well defined. The paracone and the meta- cone are taller than both the protocone and the hypocone. The protocone is elongated posterolingually. The anterior arm of the protocone forms the anterior cingulum so that no lingual part of the anterior arm of the protocone runs posterobuccally. The pro- tolophule II is short and the protolophule I is not present. A narrow valley is present between the paracone and the anterior cingulum. The mure runs anteroposteriorly but is very short. The hypocone is smaller than in M'. The anterior base of the hypocone extends anterobuccally forming the base of the ante- rior arm of the hypocone. The valley between the protocone and the hypocone runs posterolingually and is very narrow. This valley becomes deeper lingually. A short mesoloph is present between the paracone and the metacone. The metacone is elon- gated transversely and the metalophule is but a lingual extention of the metacone. The summit of the hypocone is shifted more lingually than in M‘ so that the valleys between the posterior cingulum and the metacone and the metacone and the mesoloph are longer transversely than in M'. is very small. The protocone extends posteriorly forming the lingual border of the tooth. The paracone is the tallest and the metacone is as high as the protocone. All the ridges rise to almost the same level of the protocone and the paracone. The bases of all the ridges and cusps are broad so that all the valley s between ridges and cusps are rather narrow. The protolophule II runs posteriolingually from the lingual corner of the paracone. M, is long and narrows anteriorly. The anteroconid is elon- gated transversely and situated on the front of the tooth. The anterior arm of the protoconid runs anterolingually from the summit of the protoconid but soon turns anteriorly and joins the anteroconid just buccal to the midpoint of the tooth. No direct connection between the anteroconid and the metaconid is pres- ent on most specimens, but there is considerable variation in morphology; some have no connection of the anterior arm of the protoconid with the anteroconid (CM 32901); some have a con- nection between the anteroconid and the metaconid (CM 32900). The metaconid and the entoconid are subequal and taller than both the protoconid and the hypoconid. The lingual wall of the metaconid is curved lingually. The anterior and the posterior walls of the metaconid are very steep leaving a sharp crest be- tween them, which forms the metalophulid I. The posterior pro- toconid arm runs posterolingually and then turns buccally. It connects with the posterior corner of the metaconid so that a basin without outlet is formed between the protoconid and the 1978 SETOGUCHI— CEDAR RIDGE LOCAL EAUNA 45 metaconid on most specimens. On some specimens (CM 32900), the posterior arm of the protoconid is closely related to the meta- conid but does not connect to it. The ectolophid runs almost anteroposteriorly and is lower in position than the posterior arm of the protoconid. The mesolophid is distinct but shorter than the posterior arm of the protoconid on most specimens. On one specimen (CM 32935), the mesolophid is clearly longer than the posterior arm of the protoconid but does not reach the lingual border of the tooth . The hypoconid is stout. The anterior arm of the hypoconid is short. The valley between the protoconid and the hypoconid is broad lingually but becomes narrower buc- cally. The posterior cingulum does not connect to the entoconid. Mj is slightly longer than wide. All the cusps are somewhat compressed anteroposteriorly. The metaconid and the entoconid are taller than the protoconid and the hypoconid. All the ridges rise to almost the same level as the protoconid and the hypo- conid. The anterior cingulum is complete buccally and lingually. The protoconid and the metaconid unite to the anterior cingulum separately. The posterior arm of the protoconid extends postero- lingually. On some specimens (CM 32937), it runs between the metaconid and the entoconid, on some (CM 32912) it is closely related to the metaconid but does not reach the lingual border of the tooth, and on some (CM 32914) it reaches nearly the lingual border of the tooth and connects to the metaconid on its posterior base. The lingual portion of the mesolophid is not pres- ent. The buccal portion of it is clearly defined but short on most specimens. Some specimens (CM 17086) do not have the buccal portion of the mesolophid. The valley between the protoconid and the hypoconid is excavated posteriorly and narrow buccally. The ectolophid runs almost anteroposteriorly. The hypolophulid 1 runs transversely but slightly anteriorly. The anterior arm of the hypoconid runs anterolingually from the summit of the hy- poconid. The posterior cingulum is strong. On M3, the general morphology agrees with that of M2. The posterior protoconid arm reaches the lingual border of the tooth. The entoconid is greatly reduced. It forms a thin ridge like the posterior arm of the protoconid and runs parallel with the latter. The hypoconid is reduced and the valley between the protoconid and the hypoconid is deep. The posterior cingulum is not as strong as in M2. The reduced hypoconid and entoconid and the weak posterior cingulum make the tooth narrower posteriorly. Discussion. — The teeth of this species show a wide range of morphological variation. Based on the variable morphology, I believe, too many species of Eumys have been described. The type species of Eumys is E. elegans. The characteristic features of this species as listed by Wood ( 1937) are subequal buccal and lingual portions of the anterior cingulum, long posterior arm of the protoconid being free from both the metaconid and the ento- conid, no lingual portion and weak buccal portion of the mesolophid, and reduced hypoconulid. Most of the present specimens referable to Mg show ex- actly the same features. They surely belong to E. elegans. One specimen (M,, CM 32935) has a longer me- solophid than the posterior arm of the protoconid. This characteristic feature is seen in the European cricetid, Cricetodon. Wood (1937) described a sim- ilar form from the Upper Oreodon Beds of Nebras- ka under the name of Cricetodon nebraskensis . The present form is very close to the holotype of C. nebraskensis but all the morphology except the long mesolophid agrees with that of Eumys elegans. I am not confident to separate this form from E. ele- gans. Martin (1972) placed Eumys obliquidens, E. cri- cetodontoides, E. latidens, E. spokanensis and Cri- cetodon nebraskensis into the synonomy of E. ele- gans. Specific characters, especially of Mo, of each named species are summarized as follows: Eumys obliquidens — the posterior arm of the protoconid runs posteriomesiad and unites with the entoconid; E. cricetodontoides — lingual part of the anterior cingulum is long, posterior arm of the protoconid long but not united with metaconid, mesolophid short but distinct; E. latidens — lingual part of the anterior cingulum is half as long as the buccal part, posterior arm of the protoconid is not long and not united with the metaconid, no mesolophid; E. spo- kanensis— lingual portion of the anterior cingulum obsolete, posterior arm of the protoconid closely applied to entoconid, no mesolophid; Cricetodon nebraskensis — longer mesolophid than the poste- rior arm of the protoconid. When Galbreath (1953) discussed the variation among the eumyine rodents, he stated that presence or absence of cingula are good but strength of de- velopment of cingula are poor criteria to evaluate the characters of the teeth. 1 agree with him. As for the present specimens referable to M2, the buccal part of anterior cingulum is almost always present and extends along the buccal half of the tooth. Al- though the lingual part of it is usually present, the degree of development varies; it is long and extends to the lingual margin of the front of the tooth on some specimens, but it is short and half or less as long as the buccal part on the others. The degree of development of the lingual portion of the anterior cingulum used as the key to separate E. latidens from E. cricetodontoides by White (1954) is of no value. Various degrees of development of the cin- gulum are seen within the specific variation of E. elegans. On most of the present specimens of M2, a short buccal portion of the mesolophid is present. On one specimen (CM 17086), a tiny mesoconid is seen on the middle of the ectolophid but the buccal portion of the mesolophid is totally absent. The presence 46 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 9 or absence of short buccal portion of the mesolo- phid seems to me to be of no great value for taxo- nomic evaluation for species of Eumys. I do not find any reasons to separate E. criceto- dontoides from E. elegans. As I described above, CM 32912 is anE. latidens-iy^o form and CM 32935 is a Cricetodon iiebraskensis-type form. I believe they are variants of E. elegans and I cannot find any reasons to separate them from E. elegans. E. obliqiddens and E. spokanensis are characterized by having the posterior protoconid arm closely ap- plied to or united with the entoconid. No specimens in the present fauna show this characteristic fea- ture. Based on the present Badwater Oligocene fau- na, I agree with Martin (1972) in part to place E. cricetodontoides, E. latidens, and Cricetodon ne- braskensis into the synonymy of E. elegans. Eumys brachyodus Wood, 1937 (Eig. 26, Table 11) Referred specimens . — M‘: CM 17412; M^; CM 32924, CM 32925, CM 33096-33098; CM 33111; M^: CM 33099, CM 33100; M,: CM 32906, CM 32909. Locality. — Loc. 19, Badwater Creek, Wyoming. Age. — Late Oligocene. Description. — The general morphology of M‘ almost agrees with the corresponding tooth of Eumys elegans. The present form differs from E. elegans in having a wider crown and the reduced anterocone. The crown is compressed anteroposterior- ly. The anterocone is reduced. The anterior arm of the protocone unites with the anteroloph near its lingual margin, and from the summit of the anterocone one short ridge descends postero- buccally but this ridge does not connect to the paracone. The protocone leans buccally so that the protocone occupies the lin- gual half of the anterior part of the crown. The valley between the protocone and the hypocone is lying transversely and its anterobuccal extremity is very close to the lingual tip of the paracone. That valley extends buccal to the middle of the crown. M^ is wider than that of Eumys elegans. The width is almost subequal to the length, but on most specimens the length is little greater than the width. On one specimen (CM 32925), the crown is clearly wider than long. The lingual half of the anterior cin- gulum is not present on all specimens available. The valley be- tween the protocone and the hypocone is long transversely. The mesoloph is very short. Discussion. — The size of the present form is very close to that of Etiniys elegans. The teeth of E. brachyodus is short and wider than E. elegans. Moreover, in E. brachyodus, the lingual half of the anterior cingulum is not present and the mesoloph is extremely short on M^. The present form shows these morphology. The remainder of morphology of E. brachyodus is almost exactly the same as E. ele- gans. I believe that Eumys brachyodus has a com- mon ancestry with Eumys elegans. Eumys sp. cf. E. planidens Wilson, 1949u (Eig. 17, Table 12) Referred specimens. — M'-M'-; CM 19795, CM 32943, CM 32944; M'-M^: CM 17088, CM 32941; M‘: CM 17413, CM 32946- 32980, CM 33061, CM 33093, CM 33095, CM 33114; M^-M^: CM 19713, CM 32942, CM 32945; M'"; 33003, CM 33062, CM 33094, CM 33115, CM 33116; M^: CM 17411, CM 17414, CM 33004- 33026; MrMj: CM 19798, CM 1979 ; M,: CM 17416-17419, CM 33028-33037, CM 33063-33078, CM 33117, CM 33118; M^; CM 17420, CM 33038-33060, CM 33079-33091, CM 33119. Locality. — Loc. 19, Badwater Creek, Wyoming. Age. — Late Oligocene. Description. — On M', the anterocone is large. It is nearly two- thirds as wide as the tooth at the paracone-protocone line. All the cones except the anterocone are compressed anteroposte- riorly and together with all the ridges rise to a nearly common plane. All the ridges are thin. The anterior protocone arm runs anterobuccally to reach the anterocone buccal at the midline. Between the anterior protocone arm and the anterior cingulum, a deep valley opens lingually. The posterior protocone arm runs posterobuccally but is short. It connects to the mure and the protolophule buccal to the paracone. Between the paracone and the anterior protocone arm, a valley opens deeply and runs anterobuccally. The protolophule is just a lingual extension of the paracone. No mesocone is seen, but a tiny buccal projection is present on the middle of the mure. The metalophule is also just a lingual extension of the metacone. The protolophule and the metalophule run parallel to each other and open into a wide valley lingually. This valley becomes narrower buccally because of the widened bases of the paracone and the metacone. The hypocone is also compressed anteroposteriorly so that the valley between the protocone and the hypocone is wide and opens lingually. The posterior cingulum located buccally from the hy- pocone forms a deep valley between it and the metalophule. This valley opens buccally. The tooth is characterized by deep valleys and thin ridges. On M^, the protocone is greatly compressed anteroposteriorly forming a high, thin ridge. The length of the protocone is almost half the width of the tooth. The anterior arm is short, half as long as the posterior protocone arm. The anterior arm merges into the buccal part of the anterior cingulum, which reaches to the anterobuccal margin of the tooth. No lingual anterior cin- gulum is seen. The paracone and the metacone are more com- pressed anteroposteriorly than on M' forming thin ridges running transversely. The valley between the paracone and the metacone opens more widely than on M’, but is a little narrower buccally. The mure runs posterolingually and is situated in the center of the tooth. The hypocone is also more compressed than on M'. The valley between the protocone and the hypocone is wide and opens posterolingually. The posterior cingulum runs from the hypocone buccally and slightly posteriorly. This cingulum is short, almost half as long as the metacone-metalophule. This cingulum does not form a wall on a posterobuccal corner of the tooth so that the valley between it and the metacone is wider than on M‘. On M^, the lingual portion of the anterior cingulum and the protocone form the anterior border of the tooth. The lingual margin of the protocone extends posteriorly and reaches to near- ly the posterolingual corner of the tooth. A hypocone does not occur on this tooth but the ridge correlated to the anterior arm 1978 SETOGUCHI— CEDAR RIDGE LOCAL FAUNA 47 Table 12. — Dimensions of teeth o/Eumysip. cf. E. planidens. Statistics M' M^ M" M, M.2 M,3 L w L W L W L w L W L w N 16 16 11 1 13 13 16 16 12 12 4 4 OR 2 27- 1.57- 1.86- 1.64- 1.53- 1.68- 1.87- 1.30- 1.85- 1.75- 2.24— 1.74- 2.96 2.09 2.18 2.18 1.88 1.93 2.35 1.78 2.27 2.08 2.40 2.11 Mean 2.734 1.884 1.989 1.957 1 .756 1.822 2.156 1.568 2.084 1.942 2.338 1.920 SD 0.176 0.126 0.107 0.132 0.102 0.070 0.153 0.131 0.135 0.105 CV 6.42 6.69 5.37 6.74 5.80 3.83 7.10 8.33 6.50 5.43 of the hypocone remains forming a posterior extension of the mure. The valley between the protocone on the one hand, and the mure and the anterior arm of the hypocone on the other, forms a wider, anteroposterior trench. No posterior cingulum is seen. The paracone and the metacone are thinner transversely than on M^. Mi is longer than wide. The anteroconid is elongated trans- versely. The connections of the anteroconid with the anterior arms of the protoconid and the metaconid vary. On most spec- imens, the anterior arm of the protoconid runs anterolingually and unites with the anterior arm of the metaconid, which runs almost transversely. Then the former turns anteriorly and con- nects to the anteroconid at its midpoint. The buccal part of the anterior cingulum is as long as the lingual part. On some speci- mens (CM 33029), the anterior arm of the protoconid joins the anteroconid more lingually so that the lingual part of the anterior cingulum is much shorter. Some specimens (CM 19799) show no connection between the anterior arms of the protoconid and the metaconid on an unworn stage, and only the anterior arm of the metaconid connects to the anteroconid on its lingual side so that the lingual part of the anterior cingulum is short. One spec- imen (CM 17416) shows peculiar features— the anteroconid on its buccal side; the lingual part of the anterior cingulum extends posteriorly and joins the anterior arm of the metaconid; no con- nection between the anterior arms of the protoconid and the metaconid at all. All the ridges and the anteroconid rise to the same level as the protoconid and the hypoconid. The metaconid and the entoconid are slightly taller than both the protoconid and the hypoconid. The posterior arm of the protoconid extends lingually but does not reach the lingual border of the tooth. This arm connects to the posterior margin of the metaconid at the base. A mesoconid is tiny and has a small buccal projection. No mesolophid is present. The entoconid is triangular in shape. The metalophid II is transverse. The valley between the entoconid and the posterior arm of the protoconid becomes narrower lin- gually but opens there. The hypoconid is compressed antero- posteriorly and forms a wide valley between it and the proto- conid. The valley between them becomes narrower buccally because the buccal margin of the protoconid extends slightly posteriorly. The valley opens buccally. The posterior cingulum runs posterolingually to near the lingual border of the tooth form- ing a long valley between it and the entoconid. This cingulum does not connect to the entoconid. M2 has no lingual part of the anterior cingulum. The anterior wall of the metaconid has a small excavation indicating the orig- inal presence of a valley between the metaconid and as ancestral lingual anterior cingulum. The posterior cingulum is prominent. The metaconid is more stout than on Mi but more compressed anteroposteriorly . The protoconid is also compressed. The pos- terior arm of the protoconid runs almost buccally. The base of that arm reaches to the lingual border of the tooth but does not connect to the metaconid so that the valley between it and the metaconid opens lingually. The mesoconid is small and has a buccal projection as in M,. No mesolophid is present. The meta- lophid II is also transverse. The valley between the hypoconid and the protoconid is wide, but is blocked buccally by a small, thin ridge. In M3, the buccal part of the anterior cingulum is shorter than in M2. No lingual part of the anterior cingulum is present. The posterior arm of the protoconid is short and the hypoconid is smaller than in Mj. No buccal projection on the mesoconid is seen. The valley between the protoconid and the hypoconid is wide and opens buccally. The posterior cingulum is prominent. Discussion. — A new species of Eninys, E. plani- dens, was established on the basis of a single spec- imen (Univ. Colo. No 19810, a left ramus of man- dible with M2-M3) by Wilson in 1949. Galbreath (1953) reported two additional lower jaws with M,- M;j. The diagnostic features seen in the molars are ridges and cusps of grinding surface of cheek teeth rising to a nearly common plane. The specimens at hand show exactly the same characteristic features as the holotype of E. planidens. The anteroposteriorly compressed cusps and thin ridges are characteristic of E. planidens and are not seen in any other species of Eumys described. Gal- breath once stated thatE. planidens may eventually be removed from the genus (Galbreath, 1953:74). Martin (1972) followed this argument and estab- lished a new genus for ""Eninys" planidens. This should be published in the near future. Three specimens of Eninys planidens have been reported. Galbreath (1953) described the morpho- logical variation seen in the anteroconid of Mj. The present Badwater specimens show a greater varia- tion than he recognized. He also mentioned an in- teresting variation seen in M3 as follows: “In them, the posterior protoconid arm does not extend trans- versely beyond the mesoconid crest, whereas the type specimen has this arm extending to the internal 48 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 9 border." (His mesoconid crest must be called the entoconid crest, because no mesoconid or mesolo- phid is present at all on Mg.) On Mg of the holotype of this species, the internal border between the in- ternal extremities of the metaconid and the ento- conid is concave internally making the posterior arm of the protoconid shorter than the entoconid crest even if this posterior arm extends to the in- ternal border. But on the holotype, the grinding sur- face of this posterior arm is shorter than that of the entoconid crest so that this arm does not extend beyond the mesoconid crest. The base of the pro- toconid posterior arm extends to the lingual border so that the base of it is longer than the grinding surface. This situation is rather close to that in Gal- breath’s specimens. Wilson (1949) stated in the description of the ho- lotype that on Mo and on Mg the posterior arm of the protoconid is closely related to the metaconid. On Mg, the posterior arm of the protoconid runs linguad between the metaconid and the entoconid but slightly closer to the metaconid so that the val- ley betwe n the metaconid and the protoconid pos- terior arm is narro er than the valley between the latter and the entoconid. Even so the posterior arm of the protoconid does not unite with the metaco- nid. On Mg of the holotype, the posterior arm of the protoconid is connected to the metaconid and the basin between the metalophid and the anterior arm of the protoconid on the one hand and the posterior arm of it on the other hand is blocked completely. Galbreath (1953) gave the description of the poste- rior arm of the protoconid as follows: the posterior protoconid arm is long and free, but closer to the metaconid than the entoconid. The Badwater spec- imens of M.g show variation in length of the poste- rior protoconid arm; on more specimens, it runs between the metaconid and the entoconid, closer to the former but free and does not reach to the lingual border of the to th; on some (CM 33045 and 33053) it nearly reaches to the lingual border. Most spec- imens have no metastylid but on one specimen (CM 33048) a tiny but distinct metastylid is seen on the lingual border just linguad to the posterior proto- conid arm, which does not extend to the lingual border of the tooth. On Mg of the holotype, the entoconid and the hypolophulid lingual to its union with the ectolophid form a straight transverse ridge but the hypolophu- lid buccal to the ectolophid (= the anterior arm of the hypoconid) runs posterobuccally forming a thin ridge. So these two ridges join with an angle at their union with the ectolophid. On the Badwater speci- mens, these two ridges form a straight line without an angle making a straight transverse hypolophulid. Carnegie Museum of Natural History has a few undescribed specimens referable to Eumys plani- dens from Toadstool Park, Nebraska, Orellan, or the middle Oligocene in age. On some of the refer- able MgS, the posterior arm of the protoconid does not connect to the metaconid even if the former is close to the latter. This condition is rather close to the Badwater specimens. On all teeth from the Toadstool Park referable to Mg, the hypolophulid lingual to the ectolophid joins the anterior arm of hypoconid at an angle as seen in the holotype. In all the species of Eumys except for “Eumys" plan- idens, the hypolophulid lingual to the ectolophid joins to the hypoconid anterior arm at an angle so that the situation seen in the holotype of E. plani- dens is close to the generalized forms of Eumys species. Although there is a considerable time span between the middle to the late Oligocene, the pres- ent forms are best described as Eumys sp. cf. Eu- mys planidens. Order Lagomorpha Family Leporidae Gray, 1821 Palaeolagus burkei Wood, 1940 (Fig. 28, Table 13) Referred specimens. — skull: CM 33967; DP-: CM 33968- 33977; DF: CM 17075, CM 33978-33998; P^-P": CM 34035, CM 34036; P’-M^: CM 34037; P^: CM 33999-34008; DP" and F: KU 16631; DP": CM 34009-34029; P": CM 19712, CM 34030-34034; M": CM 18267, CM 34038-34042; M-: CM 34043-34050; DPg: CM 34051-34059; P.,: CM 18269, CM 34060-34077; DP^: CM 34078-34093; P4: CM 34094-34100; M,: CM 17075, CM 34101- 34103. Fig. 2%.— Palaeolagus hurkei. a: CM 39037, left P^-MC x5. b: CM 38968, right DP^ x8. c: CM 38980, right DP^ x8. d: CM 39011, right DP", x8. e: CM 39051, right DP3, x8. f: CM 39081, right DP4, x8. g: CM 39101, right M,, x8. Fig. 29. — Palaeolagus sp. cf. P. intermedins, a: CM 39105, right DP^ b: CM 39107, left DP", c: CM 39108, right M'. d: CM 39109, right DP.,, e: CM 39110, right M,. x8. 1978 SETOGUCHI— CEDAR RIDGE LOCAL FAUNA 49 50 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 9 Locality. — Loc. 19, Badwater Creek, Wyoming. Age. — Late Oligocene. Description. — The skull is flat. The braincase is not inflated. The plane of the palate is not bent on the basicranial axis, and the angle between the plane of the palate and the basicranial axis is small. The snout is very narrow and slender. On DP'% the tooth has two anterior re-entrants; one is on the anterobuccal corner of the tooth and crosses one-third or one- fourth of the occlusal surface; the other is situated lingual to the midline of the tooth and crosses almost two-thirds of the occlusal surface. The latter re-entrant persists to the base of the tooth whereas the former does not extend to the base of the tooth and will be obliterated in an earlier stage of wear. A small and shal- low posterolingual re-entrant is observed, which will also be obliterated with wear. The enamel is well developed lingually and is reduced buccally. Several specimens are identified as DP^. In these, the tooth has buccal roots. An isolated crescent is present between the central and lingual lobes. The anterior extension of the crescent is longer transversely than the posterior one. The enamel is thicker anteriorly than posteriorly. Cement is weakly developed. The anterior loph of P-* is narrower transversely than the pos- terior loph. Following moderate wear, the tooth has an internal, straight walled hypostria crossing about one-third of the occlusal surface. A J-shaped crescent has a connection to the antero- buccal side of the tooth. Following further wear, the hypostria becomes shortened and the crescent completely worn away. Several deciduous upper fourth premolars are recognized. One of them (KU 16631) is a DP'* with unerupted permanent P* un- derneath. These two teeth were carefully separated. DP* has buccal roots, but they are short and weak. The anterior loph is a little narrower transversely than the posterior loph. An isolated crescent which is concave buccally is retained between the cen- tral and lingual lobes. Another small, circular crescent is present just inside and a little buccal to the convex crescent just men- tioned above. These crescents will be obliterated with further wear. On most specimens, the hypostria are straight-walled, but on some specimens (for example KU 16631) the posterior wall is crenulated. The enamel is well developed on its anterior and lingual sides, reduced on the buccal side, and absent on the posterior side and on the posterobuccal corner of the tooth. On an unworn specimen of P* (KU 16631), the anterior loph is slightly narrower than the posterior loph. The crown is square- shaped but will be elongated transversely with wear. The central lake is elongated anteroposteriorly but does not have a connec- tion to the anterobuccal side of the tooth. The hypostria is straight-walled but at an unworn stage the anterior and pos- terior walls meet with an angle nearly 90°. With wear the hy- postria becomes narrower. The enamel is thicker anteriorly than posteriorly. The buccal wall is formed by thin enamel, but with further wear the enamel becomes absent. When worn, P*, M*, and resemble one another in pattern. The crescent is completely worn away. In occlusal view each of these teeth has an internal, straight-walled hypostria crossing nearly one-half of the occlusal surface. The teeth become nar- rower transversely from P* to M^. On DP^, there are three main lobes. The central lobe is the tallest when unworn. The posterior lobe is the widest transverse- ly. The enamel is developed on the anterior face of the anterior lobe and the posterior face of the central lobe. No separation by an enamel and between the anterior and the central lobes is seen even when unworn. The posterior lobe is completely surrounded by enamel when unworn. With slight wear, the anterior and the central lobes unite together forming a single lobe clearly sepa- rated from the posterior lobe. With further wear, anterior lobe and the posterior lobe unite, first in the middle of the tooth and next on the lingual side, leaving a small enamel lake on the lingual side of the tooth. The small lake will be lost with further wear. Pg has only two lobes. The talonid is wider transversely than the trigonid. After wear the internal re-entrant between the tri- gonid and talonid is retained. A shallow groove is present on the anterobuccal corner of the trigonid. This groove runs all the way down to the base of the tooth. On DP4, the tooth has two main lobes and a small accessory one. The anterior lobe is the trigonid and the second one is the talonid. The small accessory lobe is the hypoconulid. On unworn teeth, the hypoconulid is separated by enamel from the trigonid but soon they unite and the hypoconulid will be obliterated. The union of the trigonid and the talonid is solely by cement as in P4. P4, M], and M2 resemble one another in pattern. No unworn specimens referable to P4 are available. Worn specimens have only two lobes. One specimen (KU 16630) has an almost unworn DP4 and unworn M; in situ in the same individual. The unworn specimen of M, has a tiny hypoconulid posterior to the talonid. The hypoconulid is smaller than the corresponding cusp on DP4 and is especially narrower transversely. This cusp on M, will be obliterated with further wear. Although no unworn materials of P4 are known, I assume that P4 has the hypoconulid posterior to the talonid when unworn, because DP4 and Mj have a clear hypoconulid before wear. M3 consists of two small lobes of which the posterior one is the smallest. Discussion. — The present form is directly com- parable to the materials described by Wood (1940) and Dawson (1958). On P2, although shallow, the anterior re-entrant is persistent to the base of the tooth. This characteristic feature is seen in Palaeo- lagiis biirkei, but not seen in P. hypsodus. In this respect, the present form is closer toP. burkei than to P. hypsodus. Palaeolagus sp. cf. P. intermedius Matthew, 1899 (Fig. 29, Table 13) Referred specimens. — DP*; CM 34104-34106; DP*; CM 34107; M>; CM 34108; DP"; CM 34109; M2; CM 34110. Locality. — Loc. 19, Badwater Creek, Wyoming. Age. — Late Oligocene. Description. — These teeth are larger than those of Palaeola- gus burkei. A DP" is tentatively referred to this species. The tooth has two anterior re-entrants; one is on the anterobuccal corner of the tooth, and the other is just lingual to the midline of the tooth and crosses almost two-thirds of the occlusal sur- face. The latter re-entrant is persistent to the base of the tooth. The enamel is well developed posteriorly and lingually but re- duced buccally. The material referable to DP* is larger than the corresponding 1978 SETOGUCHI— CEDAR RIDGE LOCAL FAUNA 51 Table 13. — Dimensions of teeth o/ Palaeolagus burki Palaeolagus sp. cf. P. intermedius. Palaeolagus burkei DP" Dp3 p3 Dp4 P^ M> M" L W L AW PW L AW PW L AW PW L AW PW L AW PW L AW PW N 4 4 5 5 5 5 5 5 6 6 6 4 4 4 4 4 4 4 4 4 OR 1, 40-0.98- 1.23- 1.52- 1.98- \A9- 1.50- 1.99- 1.30- 1.86- 1.74- \,39- 1,74- 1 .87- 1.46- 2.43- 2.67- 1.21- 2.14- 1.93- 1.60 1.13 1.50 1.76 2.30 1.67 1.93 2.85 1.56 2.49 2.43 1.82 2.96 3.09 1.65 2.83 3.03 1.41 2.53 2.24 Mean 1.540 1.035 1.402 1.658 2.116 1.568 1.776 2.432 1.400 2.103 2.067 1.600 2.353 2.450 1.550 2.610 2.813 1.303 2.273 2.010 DPa P3 DP4 P4 Ml M^ L W L AW PW L AW PW L AW PW L AW PW L AW PW N 4 4 4 4 4 4 4 4 4 4 4 4 4 4 OR 2.08- 1.30- 1.85- 1.25- 1.56- 1.63- 1.44- 1.42- 1.68- 1.77- 1.46- 1.79- 1.55- 1.34- 2.30 1.79 2.21 1.51 1.83 2.12 1.56 1.57 1.88 1.96 1.68 1.87 1.91 1.60 Mean 2.165 1.610 2.005 1.350 1.695 1.833 1.518 1.498 1.783 1.890 1.568 1.838 1.743 1.480 DP" DP" Palaeolagus sp. cf. P" DP^ P. intermedius p4 M' M" L W L AW PW L AW PW L AW PW L AW PW L AW PW L AW PW N OR Mean 3 3 1.91- 1.29- 2.20 1.42 2.067 1.340 I 1 1.98 2.62 1 2.84 1 2.22 1 4.30 — DP, P,. DP, P4 M, M., L AW PW L AW PW L AW PW L AW PW L AW PW L AW PW N OR 1 3.00 1 1.85 1 1.65 1 2.79 I 2.37 1 2.16 tooth of P. burkei. The genera! morphology agrees with that of P. burkei. The upper molars are also larger than those of P. burkei. The hypostria is shallower than inP. burkei, and on worn specimens the hypostria almost vanishes. Only a shallow groove remains on the posterolingual face of the tooth. On DP^, there are three lobes and a small accessory one. The anterior three lobes resemble those seen in DP) of P. burkei. The accessory one is half as wide transversely as the posterior lobe. The morphology of the remainder of the crown agrees with that seen in P. burkei. Discussion — Based on its larger size and the shal- low hypostria, these specimens are referred to P. intermedius . Order Carnivora Family Canidae Gray, 1821 Hesperocyon temnodon (Wortman and Matthew, 1899) (Fig. 30, Table 14) Referred specimen. — CM 21678. Locality. — Loc. 19, Badwater Creek, Wyoming. Age. — Late Oligocene. Discussion. — Only one specimen referable to this species is available. Matthew (1901:357) referred a specimen from the Leptauchenia zone in Logan County, northeastern Colorado, to this species, but he did not describe the specimen. Galbreath Table 14. — Dimension of teeth of Hesperocyon temnodon and Leptomeryx sp. near L. evansi. Sta- M‘ M" M) tistics L AW PW L AW PW L AW PW N OR Hesperocyon temnodon 1 9.88 1 4.25 1 4.16 N OR Leptomeryx sp. near L. evansi 1 5.62 1 6.29 Ill — 8.69 8.69 8.21 — — — 52 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 9 1978 SETOGUCHI— CEDAR RIDGE LOCAL EAUNA 53 (1953:76) cited Matthews but also did not describe the material. H. temnodon was originally estab- lished by Wortman and Matthew (1899:130) as Cy- nodictis temnodon based on an upper dentition. Macdonald (1963:202) listed the characteristics of Ml of Hesperocyon as: entoconid present; one or two enteroconids; deeply basined talonid; talonid closed posteriorly. The present specimen shows this morphology. In size it is smaller than H. lep- todus. Tentatively, the present specimen is referred to H. temnodon . Order Perissodactyla Eamily Equidae Gray, 1821 Miohippus sp. Referred specimen. — Astragalus: CM 21679. Locality. — Loc. 19, Badwater Creek, Wyoming. Age. — Late Oligocene. Discussion. — An astragalus is the only equid specimen available. The size of the astragalus is clearly larger than that of Mesohoppus and a little larger than that of Miohippus equiceps described and figured by Osborn (1918:324). No teeth refera- ble to this genus are available. The specific identi- fication is difficult. Eamily Hyracodontidae Cope, 1879 Hyracodontidae genus indet. Referred specimens. — Tooth and limb bone fragments: uncat- alogued. Locality. — Loc. 19, Badwater Creek, Wyoming. Age. — Late Oligocene. Discussion. — Small fragments of teeth and limb bones of hyracodont rhinoceros have been found, but nothing is generically determinable. Order Artiodactyla Eamily Hypertragulidae Cope, 1879 Leptomeryx sp. near L. evansi Leidy, 1853 (Fig. 32, Table 14) Referred specimens . — P^: CM 21694; M’: CM 21695; M^; CM 21696; M,; CM 21693. Locality. — Loc. 19, Badwater Creek, Wyoming. Age. — Late Oligocene. Discussion. — The materials at hand are too in- complete for specific identification. The size is very close to or a little larger than that of Leptomery.x evansi. The tooth morphology is almost exactly the same. CM 21696 is a second upper molar where the buccal face of the metacone is not completely flat- tened but a small ridge runs vertically on the middle of the surface. This part of the enamel is slightly thickened so that the buccal face of the metacone is rather convex buccally. On most specimens of L. evansi which I examined the buccal face of the metacone is flat or slightly concaved. Leptomeryx is a typically Oligocene genus. Scott (1940:537) stated, “It (^Leptomeryx) is one of the commoner fossils of the lower Brule (Oreodon Beds) and is much less frequently found in the up- per Brule, or the Chadron. It persisted through the John Day and into the lower Miocene (Gering stage).” Occurrences oi Leptomeryx from the lower Miocene were reported by Macdonald (1963:233) from western South Dakota and Martin (1972) from western Nebraska. The materials referrable to Lep- tomeryx reported by them are again too fragmen- tary to warrant specific identification. Hypisodus sp. near H. minimus (Cope, 1873) (Fig. 31) Referred specimens. — M‘-M^; CM 21682; M‘; CM 34111, CM 34112; M"": CM 21684, CM 21687, CM 34113; P3; CM 21688; M,; CM 21683, CM 21685, CM 21686, CM 21690, M^; CM 21689. Locality. — Loc. 19, Badwater Creek, Wyoming. Age. — Late Oligocene. Description. — The size of the present form is a little larger than that of the holotype of Hypisodus minimus. A similar form has been reported by Scott (1940;535) from “the Uppermost Brule” of eastern Nebraska and by Galbreath (his Form B, Gal- breath, 1953;91) from the Vista Member of the White River For- mation of northeastern Colorado. On specimens of the present form, the upper molars have a weak buccal style, or a median rib on the anteroexternal crescent of Scott ( 1940;525) so that the vertical valley between the style and the parastyle is very shal- low. But on a few specimens (CM 39111, M‘ and CM 21684, M'9, the upper molars have a fairly well-developed buccal style on the buccal face of the paracone so that the valley between it and the parastyle is fairly deep. The degree of the depth of that valley also depends on the orientation of the parastyle. On the specimens having a shallow valley, the parastyle projects rather anteriorly whereas the parastyle extends more anterobuccally on molars having a deeper valley. Thus, the degree of weakness of the buccal style on the buccal face of the paracone and the depth of the valley have a fairly wide range of variation. Discussion. — Hypisodus has been thought to be restricted to the Oligocene. Martin ( 1972), however. Fig. 30. — Hesperocyon temnodon. CM 21678, left M,. x4. Fig. 31. — Hypisodus sp. near H. minimus, a: CM 21682, right M‘-M^. b: CM 21685, right M,. c; CM 21689, right M.2. x4. Fig. 32. — Leptomeryx sp. near L. evansi. a; CM 21695, right M‘. b: CM 21696, left MT x4. 54 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 9 reported an occurrence of Hypisodus from the low- er Miocene Gering formation of western Nebraska. His form is larger than any other known species of Hypisodus. He stated that the labial styles are low and flattened. As stated above, based on the present samples at hand, the degree of development of styles varies from specimen to specimen. Size seems to be the only criterion for separating species of Hypisodus. SUMMARY AND CONCLUSIONS Besides a few lower vertebrates, only small mam- mals have been recovered from the upper Oligocene sediments along Badwater Creek. The preserved fauna of the Cedar Ridge local fauna is greatly biased by two major factors, which are wholly in- dependent of each other; first, due to stream action, only small teeth and bones were transported and deposited at Loc. 19 so that larger mammals, which surely lived near the site of deposition during the late Oligocene are not represented in the fauna; sec- ond, due to drier climatic conditions along Bad- water Creek in late Oligocene time, only land ani- mals, which were adapted to such kind of ecological conditions, could live there and consequently these animals are represented in the fauna, whereas the animals, which required more mesic conditions, could not have lived there and therefore they are not represented in that fauna. The drier climatic condition along Badwater Creek in the late Oligocene is indicated by the pres- ence of calcic feldspars and evaporites throughout the section. These were formed under a climatic regime with moderate temperatures. Although no botanical evidence has been recovered, the envi- ronmental situation is thought to be of a grassland, steppe to semidesert type with low precipitation and moderate temperature. Due to unfavorable ecological conditions, land micromammals, which lived there during late Oli- gocene time, were specialized in having higher crowned and more lophate teeth. This was the re- sult of adaptation to a more herbaceous diet in drier climatic conditions. Some of the rodents represent- ed had more hypsodont teeth than in their middle Oligocene counterparts, but they have never de- veloped rootless or everygrowing cheek teeth. Among rodents, many specialized eomyids are the most common constituents of the Cedar Ridge local fauna. Eomyids appear to have had their ma- jor North American radiation during the early Oli- gocene. Wilson (1949:112) pointed out that the tooth pattern in eomyids is similar to, although not identical with, the cricetodont pattern. As Black (1963;41) has stated, it is quite possible that mem- bers of the Eomyidae occupied many of the same habitats that were later filled by the cricetids. Only a few cricetids are at present known from the early Oligocene when the major radiation of the Eomyi- dae took place (Clark, et al., 1964). It seems pos- sible that many of the early Oligocene eomyids such as Yoderimys, Centimanomys, Namatomys, and Aulolithomys were replaced by the more highly spe- cialized cricetids during Chadronian time. Eomyids became abundant and varied in the late Oligocene perhaps through adaptation to drier climatic con- ditions and more open environmental situations. Eumyine cricetids are abundant in the Orellan and by the Arikareean, a number of cricetid types are known. Most of them, however, required more mesic conditions and surely lived under a more hos- pitable climate in present Oregon and Nebraska. Among cricetid rodents, Leidymys, Pacicuius, and Scottimus are known from the late Oligocene of Oregon and Nebraska, but they are not represented in the Cedar Ridge local fauna. Highly specialized eomyids and also a few highly specialized cricetids coexisted in the Badwater Creek area during the late Oligocene time. Near the Oligocene-Miocene boundary, the en- vironment returned to a more mesic condition with considerable precipitation. In relation to the cli- matic changes, the highly specialized eomyids and cricetids, which were adapted to drier conditions, could not survive and most of them became extinct by the end of the Oligocene. Subsequently, crice- tids represented by Leiydyms, Pacicuius, and Scot- timus expanded during the Miocene. An eomyine genus, Pseudotheridomys, is repre- sented in the North American early Miocene by P. hesperus (Wilson, 1960) from the Martin Canyon Quarry A fauna. This species, however, is closely related to European forms and evidently represents an early Miocene immigration into the New World. Thus, most of the North American Oligocene eomyids became extinct by the end of the Oligocene and a few forms could have survived into Miocene time. It is worth discussing whether any new forms 1978 SETOGUCHI— CEDAR RIDGE LOCAL FAUNA 55 migrated into North America from some other cen- ter of radiation during the Oligocene. Based on the study of the late Eocene Badwater local fauna. Black, (1967:63) stated, “Many groups that made their first appearances in the late Eocene may rep- resent immigrants from other areas and thus were not present in earlier North American faunas.” I agree with his conclusion. Thus, many groups rep- resent immigrants into North America from other areas in the late Eocene. Now we have to consider whether any forms represent immigrants during the Oligocene. The eomyids were probably descended from members of the Sciuravidae (Black, 1965:42), some- time during the latter half of the Eocene, but their ancestors are not known as yet. The number and diversity of eomyids now known from the early Oligocene would seem to indicate either a much greater late Eocene radiation in North America than has been recognized to date or a considerable im- migration into western North America in the latest Eocene from some other center of radiation. Only Adjidaumo and Paradjidaumo persist into the mid- dle Oligocene and, by the end of the Oligocene a number of specialized eomyids have evolved from the ancestral stocks of Adjidaumo and Panidjidau- mo. The major Old World eomyid radiation evi- dently took place during the Aquitanian and Bur- digalian (early to middle Miocene). As far as the North American eomyids are concerned, they evolved within the North American continent and they did not receive any new comers from outside North America although the possibility of an im- migration from North America to Europe can not be ruled out. After the climate returned to a more hospitable condition in the late early Miocene, some European eomyids, for example, Pseudo- theridomys (Wilson, 1960) and Eomys (Lindsay, 1974), migrated into North America. These forms are not descendents of the North American Oligo- cene eomyids. The origin of the Cricetidae is not certain. It seems likely that the cricetids along with many oth- er rodent families may be derived from late Eocene sciuravids. The North American Cricetidae are al- ready abundant in the lowermost Orellan (Martin, 1972) in Nebraska. They are very close to certain Eurasian cricetids, noiab\y Euchcetodon and P^c//- docricetodon, and it seems likely that there was an exchange of cricetids between North America and Eurasia in the Chadronian (early Oligocene). As the place of the origin of the cricetids is unknown, the direction of this exchange is not clear; however, it seems likely that it took place soon after the prob- able time of origin of the Cricetidae in the late Eocene. But, once the basal stocks of the cricetids had been established in North America sometimes during the early Oligocene, they evolved in North America and it seems likely that the cricetids re- placed eomyids in large part during the middle Oligocene. Cricelodon has been described from the Upper Oreodon Beds (middle Oligocene) of Ne- braska by Wood (1937:256). Cricetodoii is a Euro- pean genus and if the material described by Wood is referable to Cricetodon, it must represent an im- migration from Europe into North America during the middle Oligocene. Close examination, however, shows that having the mesolophid longer than the posterior arm of the protoconid on M, and M2 is seen not only in Cricetodon, as Wood believed, but in some other forms of North American cricetids as well, especially Eumys elegans. The material de- scribed by Wood thus does not represent an im- migrant from Europe, but should be regarded as a variant of the North American -complex. All the North American late Oligocene cricetids have their ancestry in the middle Oligocene of North America (Martin, 1972). The other rodents and insectivores in the Cedar Ridge local fauna have their ancestry in the middle and even in the early Oligocene of North America. Thus, the late Oligocene mammal fauna represented by the Cedar Ridge local fauna has a close relation- ship to the middle Oligocene faunas of North Amer- ica and no great faunal gap between them is rec- ognized. Because most of the "typical” North American Oligocene eomyids and some cricetids became extinct by the end of Oligocene and some forms migrated into North America during the Mio- cene, a more distinct faunal gap is recognized be- tween the late Oligocene and the early Miocene fau- nas than between the middle and the late Oligocene faunas. The rodent families Castoridae (beavers), Myla- gaulidae (mylagaulids), and Aplodontidae (aplodon- tids besides Prosciurus and Pelycomys) are almost always found in faunas of the early Miocene of North America. They are totally absent from the Cedar Ridge local fauna and this makes the faunal distinction clearer, between the late Oligocene and the early Miocene faunas. 56 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 9 LITERATURE CITED Auffenberg, W. A. 1958. A new family of Miocene salaman- ders from the Texas coastal plain. Quart. J. Florida Acad. Sci., 21:169-176. Black, C. C. 1961. New rodents from the early Miocene de- posits of Sixty-six Mountain, Wyoming. Breviora, 146:1-7. . 1963. Miocene rodents from the Thomas Farm local fau- na, Florida. Bull. Mus. Comp. Zool., 128:483-501. . 1965. Fossil mammals from Montana. Part 2. Rodents from the early Oligocene Pipestone Springs local fauna. Ann. Carnegie Mus., 38:1-48. . 1967. Middle and Late Eocene mammal communities: a major discrepancy. Science, 156:62-64. . 1968. Late Oligocene vertebrates from the northeastern Wind River Basin. Field Conf. Guidebook for the High Al- titude and Mountain Basin Deposits of Miocene Age in Wy- oming and Colorado, Pub. Univ. of Colorado Mus., pp. 50- 54. . 1969. Fossil vertebrates from the late Eocene and Oli- gocene, Badwater Creek Area, Wyoming, and some regional correlations. Twenty-first Annual Field Conference — 1969, Wyoming Geol. Assoc. Guidebook, pp. 43-47. . 1971. Paleontology and geology of the Badwater Creek area, central Wyoming. Part 7. Rodents of the family Ischy- romyidae. Ann. Carnegie Mus., 43:179-217. Black, C. C., and M. R. Dawson. 1966. Paleontology and ge- ology of the Badwater Creek area, central Wyoming. Part 1. History of field work and geological setting. Ann. Carnegie Mus., 38:197-307. Burke, J.J. 1934. New Duchesne River rodents and preliminary survey of the Adjidaumidae. Ann. Carnegie Mus., 23:391- 398. Clark, J. B., M. R. Dawson, and A. E. Wood. 1964. Fossil mammals from the lower Pliocene of Fish Lake Valley, Ne- vada. Bull. Mus. Comp. Zool., 131:29-63. Cope, E. D. 1873. Third notice of extinct Vertebrata from the Tertiary of plains. Paleont. Bull. 16:1-8. . 1884. The Vertebrata of the Tertiary Formations of the West. Book I. Report U.S. Geol. Surv. Territories, F. V. Hayden, U.S. Geologist in Charge, Washington, D. C., 1009 PP- Donohoe, j. C. 1956. New aplodontid rodent from Montana Oligocene. J. Mamm., 37:264-268. Douglass, E. 1901. Fossil Mammalia of the White River beds of Montana. Trans. Amer. Phil. Soc.. 20:237-279. . 1903. New Vertebratesfromthe MontanaTertiary. Ann. Carnegie Mus., 2:145-200. Dawson, M. R. 1958. LaterTertiary Leporidae of North Amer- ica. Univ. Kansas Paleont. Contrib., Vertebrata, 6:1-75. Emry, R. j. 1972. A new heteromyid rodent from the early Oli- gocene of Natrona County Wyoming. Proc. Biol. Soc. Wash- ington. 85: 179-190. Estes. R. 1963. Early Miocene salamanders and lizards from Elorida. Quart. J. Elorida Acad. Sci., 26:234-256. Everden, j. E., D. E. Savage, G. H. Curtis, and G. T. James. 1964. Potassium-Argon dates and the Cenozoic mammalian chronology of North America. Amer. J. Sci., 262:145-198. Gai BREATH, E. C. 1948. A new species of heteromyid rodent from the middle Oligocene of northeast Colorado with re- marks on the skull. Univ. Kansas Pub., Mus. Nat. Hist., 1:285-300. . 1953. A contribution to the Tertiary geology and pa- leontology of northeastern Colorado. Univ. Kansas Paleont. Contri., Vertebrata, 4:1-120 . 1955. A new eomyid rodent from the lower Oligocene of northeastern Colorado. Trans. Kansas Acad. Sci., 58:75-78. Hough, J., and R. Alf. 1956. A Chadron mammalian fauna from Nebraska. J. Paleont., 30:132-140. Krishtalka, L., AND C. C. Black. 1975. Paleontology and ge- ology of the Badwater Creek area, central Wyoming. Part 12. Description and review of late Eocene Multituberculata from Wyoming and Montana. Ann. Carnegie Mus., 45:287-297. Krishtalka, L., and T. Setoguchi. 1977. Paleontology and geology of the Badwater Creek area, central Wyoming. Part 13. The late Eocene Insectivora and Dermoptera. Ann. Car- negie Mus., 46:71-99. Kulp, j. L. 1961 . Geologic time scale. Science, 133: 1 105-1 1 14. Lambe, L. M. 1908. The Vertebrata of the Oligocene of the Cy- press Hills, Saskatchewan. Contrib. Canadian Paleont., 3:5- 23. Lillegraven, j. a., and M. C. McKenna. Manuscript. Phy- logenetic history and microevolutionary mosaicism of Cen- tetodon. Love, J. D. 1939. Geology along the southern margin of the Ab- saroka Range, Wyoming. Geol. Soc. Amer. Spec. Papers, 20:1-134. McGrew, P. O. 1939. Nanodelphys, an Oligocene Didelphine. Field Mus. Nat. Hist., Geol. Ser., 6:393-400. . 1941 fl. Heteromyids from the Miocene and lower Oli- gocene. Field Mus. Nat. Hist., Geol. Ser., 8:55-57. — . \9A\b. The Aplodontoidea. Field Mus. Nat. Hist., Geol. Ser., 9:3-30. — . 1959. The geology and paleontology of the Elk Mountain and Tabernacle Butte area, Wyoming. Bull. Amer. Mus. Nat. Hist., 117:117-176. McKenna, M. C. 1960. The Geolobididae, a new subfamily of early Cenozoic erinaceoid insectivores. Univ. California Publ., Geol. Sci., 37:131-164. Macdonald, J. R. 1963. The Miocene faunas from the Wound- ed Knee area of western South Dakota. Bull. Amer. Mus. Nat. His., 125:139-238. . 1970. Review of the Miocene Wounded Knee Faunas of southwestern South Dakota. Bull. Los Angeles County Mus. Nat. Hist., Sci., 8:1-82. Martin, L.D. 1972. The mammalian fauna of the lower Miocene Gering Formation of western Nebraska and the early evo- lution of the North American Cricetidae. Unpublished Ph.D. dissert., Univ. Kansas, Lawrence. Matthew, W.D. 1901. Fossil mammals of the Tertiary of north- eastern Colorado. Mem. Amer. Mus. Nat. Hist., 1:355^47. . 1903. The fauna of the Titanotherium beds at Pipestone Springs. Montana. Bull. Amer. Mus. Nat. Hist., 19: 197-226. . 1918. Contributions to the Snake Creek Fauna; with notes upon the Pleistocene of western Nebraska; American Museum Expedition of 1916. Bull. Amer. Mus. Nat. Hist., 38:138-229. Osborn, H. E. 1918. Equidae of the Oligocene, Miocene, and Pliocene of North America, Iconographic Type Revision. Mem. Amer. Mus. Nat. Hist., n. s., 2: 1-330. Patterson, B., and P. O. McGrew. 1937. A soricid and two erinaceids from the White River Oligocene. Field. Mus. Nat. Hist., Geol. Ser., 6:245-272. 1978 SETOGUCHI— CEDAR RIDGE LOCAL FAUNA 57 Reed, K. M. 1961. The Proscalopinae, a new subfamily of talpid insectivores. Bull. Mus. Comp. Zool., 125:473-494. Reeder. W. G. 1960 Reithrodontomys spp. 2.095 4 0.2 0 4 3 1 Other nocturnal species 2,095 7 0.3 0 7 0 7 Diurnal species 2,095 15 0.7 0 15 8 7 Perognathus fiavescens 1,229 1 1 0,9 0 1 1 1 1 0 Perognathus flavus 1.229 1 0.1 0 1 0 1 Dipodomys ordii 1.229 200 16.3 0 200 200 0 Dipodomys merriami 869 5 0.6 0 5 0 5 Peromyscus maniculatus 1.229 21 1.7 0 21 21 0 Onychomys leucogaster 1,229 28 2.3 0 28 28 0 The University of New Mexico provided financial assistance in the form of assistantships and an NDEA Title IV fellowship. I am grateful for that assistance, and for the equipment and facilities furnished by the University. Dr. Arthur H. Harris laid the groundwork for this study with his collection of Apache pocket mice and his investigation into their relationship with the olive-backed pocket mouse. Dr. James S. Findley suggested the original project, and gave advice and encouragement along the way. Several of my fellow students aided my studies. Kenneth Andersen. Hal Black. Gwen Britt. Michael Bogan. Jay Drueck- er, Kenneth Geluso, and Don Wilson were particularly helpful. California State College, Stanislaus paid a portion of my travel expenses incurred in examining holotypes. The Carnegie Mu- seum of Natural History furnished computer time at Carnegie- Mellon University. Teresa Bona typed the manuscript, and Nan- cy Perkins drew Figs. 3 and 6. Suzanne Braun and Hugh H. Genoways have been especially helpful in editing drafts of this report. James L. Patton and Robert F. Martin critically reviewed a draft of this paper, and offered several suggestions for im- proving it. A special note of appreciation is extended to these persons and to the many curators and curatorial assistants who made specimens available for study. RESULTS AND DISCUSSION Distribution and Habitat A summary of the relevant capture data is pre- sented in Table 2. A total of 522 individuals of Pe- rognathus was captured, representing 16.5% of the small mammals taken. A majority of traps were set in sandy areas, and no attempt was made to sample all habitats or to sample different habitats equally. The small traps undoubtedly reduced the catch of larger species, including Dipodomys. Even so, kan- garoo rats were captured from six to 16 times more frequently than were fasciatiis group pocket mice. The silky pocket mouse, P. flaviis, was captured twice as often as/*, apache, despite a trapping reg- imen that was designed to maximize the catch of P. apache. Apache pocket mice are usually limited to loose. 12 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 10 sandy soils and dunes with a sparse vegetational cover. I often captured them on sand dunes, several hundred feet from the nearest vegetation. One ex- ception was in the Unitah Basin of Utah, where 1 found P. apache to be common and widespread on a variety of substrates. In the Navajo Reservoir area of northwestern New Mexico, Harris (1963) captured two specimens on fine-textured soils. Some specimens from adjacent areas in Colorado may have come from similar habitats. Twice, in northwestern New Mexico and southeastern Utah, I took single specimens on hard-packed, fine-tex- tured soils along arroyos. In both cases, I captured several specimens in adjacent, sandy areas. The geographic distribution of P. apache is shown in Eig. 4. Apache pocket mice are most nu- merous in steppe-grassland associations between 5,000 and 7,500 ft in elevation. Common plant as- sociates are sagebrush (Artemisia), saltbush (Atri- plex), mormon tea (Ephedra), snakeweed (Gutier- rezia), juniper (Jiiniperus), rice grass (Oryzopsis), tumbleweed (Salsola), yucca (Yucca), and rabbit bush (Chrysothamnos). P. apache ranges from Lower Sonoran mesquite associations through low- er Transition pinyon-juniper associations. It is re- corded as occurring in the yellow-pine zone in the Gallina Mountains of New Mexico (Bailey, 1932). That locality is actually a stabilized dune system at the northeastern edge of the San Augustine Plains, where scattered yellow pines (Pinas ponderosa) ex- tend onto an old dune system that abuts against the mountains. I, and others, have also taken P. apache among scattered yellow pines in the vicinity of Wi- nona, east of Flagstaff, Arizona. There, mice were captured on lava sands among rabbit bush, sage- brush, and juniper, with scattered yellow pines growing mostly along the bases and slopes of cinder buttes and on rocky outcrops. In both areas yellow pines occur at lower and drier elevations than nor- mal due to local edaphic factors. P. apache is not known to occur in typical yellow-pine forests. Apache pocket mice may be prevented from spreading farther to the west by competition with P. ampins, P. longimemhris, and P. parvus. There is very marginal sympatry between P. apache and P. ampins along the western edge of the range of P. apache in northern Arizona (Fig. 4). According to Benson (1933^?), P. ampins is restricted to sand habitats in that area. In contrast, I found P. ampins common on rocky slopes and gravelly soils. Around Navajo Mountain, Utah, P. longimemhris has been taken withP. apache (Benson, 1935). It seems like- ly that Apache pocket mice are prevented from oc- cupying areas west of the Colorado River in south- ern Utah by competition with P. longimemhris and P. parvus (Fig. 4). South of Socorro Co., New Mexico, in the south- ern portion of its range (Fig. 4), P. apache appears to be confined to sandy hummocks and dunes in mesquite (Prosopis) associations. At these lower, warmer elevations, P. apache is very rarely cap- tured. Here, and farther north in New Mexico and Arizona, P. flavus is generally most numerous on both fine-textured and gravelly soils with moderate vegetational cover. In the southern portion of the range of P. apache, the desert pocket mouse, P. penicillatus, is common in sandy areas and on creo- sote flats with sparse vegetational cover. Competi- tion with P. penicillatus may be a major reason for the relative scarcity of P. apache there. The silky pocket mouse (P. flavus) does not occur much farther north than the San Juan River in southeastern Utah and southwestern Colorado. North of the range of P. flavus, P. apache has been more frequently captured on non-sand substrates. This suggests that competition with P. flavus may generally limit P. apache to sandy substrates. In this regard, I found P. flavus to be common on loose sand soils in most areas where no P. apache were captured, and I caught P. flavus in the same trap lines as P. apache at just four localities. The geographic range of P. apache was found to terminate at the White and Duchesne rivers in the Uintah Basin of Utah and Colorado (Fig. 5). I cap- tured 22 P. apache at a single locality north of the White River ( 1.5 mi E Ouray). The mice were taken on hard-packed, sand-gravel conglomerate soil. There is a bridge across the White River within a kilometer of this site, and individuals may have re- cently colonized the north bank of the White River via the bridge. About 2 km east of this collecting site (also north of the White River), I trapped for two nights in a sand dune area extending over about 100 hectares, but caught no pocket mice. The olive- backed pocket mouse, P. fasciatus, was captured at several localities north of the White River, north and east of the site where 1 captured P. apache (Fig. 5). Most P. fasciatus were taken in sandy areas, although one was captured on a rocky slope. 1 did not find them to be common at any site in the Uintah Basin. Perognathus apache and P. fasciatus occur in 1978 mLLlAMS—PEROGNATHUS SYSTEMATICS 13 I 09 Fig. 5. — Map of northeastern Utah and northwestern Colorado, showing the distribution of three species of pocket mice. Triangles = Perognalhus parvus: closed circles = P. apache: open circles = P. fasciatus. The shaded areas represent mountains and plateaus over 7,500 ft in elevation. similar habitats and are nearly the same size, and it is possible that competitive exclusion limits their ranges along a line formed by the White and Du- chesne rivers. Certainly the rivers are not barriers to these pocket mice, as they are shallow, and meander through broad floodplains near their con- fluence with the Green River. The geographic ranges of both species may be limited on the west by competition with P. parvus (Fig. 5). In areas where I caught P. parvus, they seemed to be com- mon on all types of substrates, including slopes and level areas. There appear to be no physical barriers to the spread of P. parvus to the east. All three species may be recent arrivals to the Uintah Basin, possibly within historic times. Wells (1970<7, 19706) presents evidence that both Pinus ponderosa and Juniperus scopulorum were widely distributed over the now treeless, arid Laramie Ba- sin from at least 5,600 years B.P. to 200 years B.P. He states that the hypsithermal period (from 9,000 to 2,500 years ago) was a time of higher tempera- tures and greater moisture. The drastic reduction of woodlands over the past several centuries may con- stitute the first climatically induced episode of tree- lessness in the area in post-Wisconsin time, and may have allowed the recent spread of these species into the Uintah Basin. The major habitat of Apache pocket mice extends more or less continuously from the Tavaputs Pla- teau of eastern Utah and adjacent Colorado south- ward into the Painted Desert of Arizona and the San Juan Basin of New Mexico, and southeastward into the Rio Grande Valley (Fig. 4). Most other in- habited areas are smaller, are found at higher ele- vations, and are more or less isolated by very nar- row corridors of intermittent habitat along water 14 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 10 courses. Populations appear to be isolated by wide stretches of unfavorable habitat in a few areas (for example, the Uintah Basin, San Luis Valley, San Augustine Plains, and Willcox Playa). The Uintah Basin is bounded on the south by the high, east-west oriented Tavaputs Plateau (Fig. 5). Most of the plateau is over 8,000 ft. On its southern front the plateau rises abruptly, as along the Book and Roan Cliffs, and in some places is nearly 2,000 ft higher than the land to the south. On its northern side, the plateau slopes more gradually into the Uintah Basin. Sheer cliffs and steep rocky slopes on the south provide no sandy habitat for Apache pocket mice, and only a few large drainage channels cut sharply into the plateau. In Colorado, the pla- teau does not stand out in such bold relief, and it is possible that Apache pocket mice crossed the plateau via passes, such as that north of Rifle. More likely, however, P. apache colonized the Uintah Basin via the Green River Canyon through the pla- teau (Fig. 5). This is a narrow, precipitous route, and one that is likely to be breached only rarely. The Rio Grande Valley abruptly narrows south of Socorro, New Mexico, and the river, from near Val Verde to near Las Cruces, flows through a nar- row channel along the western front of a series of low mountains (Fig. 4). A southward drop in ele- vation occurs near Socorro, and creosote (Larrea) associations replace the more northern steppe- grasslands. Near this locality, a biotic transition occurs, and the ranges of several species charac- teristic of the southern desert associations end (for example, Perognatiuis penicillatus, Geomys are- narius, and Peromyscus ere micas, Findley et al., 1975). To the east, the Jornada Del Muerto (a broad, north-south valley) extends without inter- ruption from the Deming Plains to near Val Verde. Sandy, mesquite-dominated hummocks are scat- tered throughout the Jornada, with creosote being found on hard-packed, fine-textured soils and on the gravelly slopes. Elevation increases gradually to the north, where the Jornada is partially blocked by a large lava flow. Sandy habitat continues inter- mittently northeastward in the direction of Gran Quivira, and elevation increases more rapidly. The Gran Quivira site is in a juniper association, where sands have accumulated at the base of some low hills. The San Augustine Plains are connected on the southwest by a large drainage channel with inter- mittent sandy spots along its banks (Fig. 4). On the western and northwestern perimeters mountains and low hills and stretches of rocky soils probably form a barrier to dispersal and intermingling with the Rio Grande Valley population. Only low hills form the boundaries of the San Augustine Plains to the northwest, and population interchanges in the direction of Gallup and the Painted Desert are prob- able. There are no significant physiographic barriers between the ranges of P.flavescens and P. apache, and their populations may be in contact at a few points. The plains pocket mouse {P. flavescens) may be distributed all along the Pecos River Valley, and Apache pocket mice have been collected in the upper Pecos River Valley (Fig. 4). The two popu- lations may contact each other on the plains north of Gran Quivira, although there is a stretch of hard, rocky, limestone soils between the Pecos Valley and the Gran Quivira site. However, the most likely contact zone is in the Trans-Pecos region. Neither species is known from a fairly wide area (Fig. 4), but it is likely that Apache pocket mice occur east- ward to the sands along the salt lakes west of the Guadalupe Mountains. This Trans-Pecos gap is no greater than several other gaps between known populations. That Apache pocket mice are found at nearly every sandy site within their range suggests that they have the ability to disperse through areas not suitable for supporting permanent populations. Even in areas where they are relatively abundant and widespread, such as the San Juan Basin, they are discontinuously distributed, as loose, sandy soil is a minor habitat of spotty occurrence. Intrapopulation Variation Adults averaged significantly larger than sub- adults in most characters, and larger than juveniles in all traits except the dimensions of the permanent teeth. Consequently, juveniles and subadults were excluded from interpopulation comparisons. There were no significant differences between the sexes of adults in the morphometric characters for the Albuquerque sample (39 males, 31 females) and the Painted Desert sample (35 males, 27 females). Females of the Uintah Basin sample (28 males, 33 females) averaged significantly larger in length of the interparietal and length of P4. Males of the White Sands sample (28 males, 26 females) aver- aged significantly larger than females in bullar length, width across the bullae, and in width of M3. All of the significant differences were relatively slight ( 1 to 3%) and could be due to normal sampling errors. Because of the small number of differences. 1978 WILLIAMS— PEROGNATHUS SYSTEMATICS 15 Table 3. — Premolar cusp numbers and bullae apposition in fasciatus group samples. Sample P^ cusps P4 cusps Bullae meet 1 2 4 5 Y 3 4 5 Yes No 1 . Perognaihus fasciatus fasciatus 12 12 12 2. Perognathus fasciatus olivaceogriseus 21 1 18 21 3. Perognathus fasciatus litus 19 19 18 4. Perognathus fasciatus callistus 31 31 31 5. Perognathus apache Uintah Basin 3 ^9 1 7 61 1 63 6. Perognathus apache Moab 30 1 27 1 25 7. Perognathus apache Painted Desert 66 1 2 66 6 55 8. Perognathus apache Flagstaff 36 3 34 3 30 9. Perognathus apache Gallup S Y 4 6 10. Perognathus apache San Juan Basin 30 T 1 27 1 3 32 1 1 . Perognathus apache Canyon Largo 24 5 2 23 3 27 12. Perognathus apache Estrella 12 3 1 13 14 13. Perognathus apache San Luis Valley 20 4 16 16 14. Perognathus apache Santa Fe 25 3 1 27 Y 22 Perognathus apache Rio Grande Valley 53 18 1 54 14 57 16. Perognathus apache San Augustine 19 1 14 17 17. Perognathus apache Gran Quivira 15 1 14 Y 12 18. Perognathus apache White Sands 1 1 52 2 7 16 28 3 7 40 19. Perognathus apache Deming Plains 16 2 14 4 12 20. Perognathus flavescens copei 14 ■) Y 14 5 12 Total 1 4 511 37 7 58 518 4 49 491 % 0.0 0.7 92.0 6.7 1.2 9.9 88.3 0.7 9.1 90.9 and because there was no pattern or consistency to the differences, I decided that the advantages of pooling the sexes for intersample comparisons out- weighed the possible disadvantages. There were no significant differences between the sexes in the col- or indices. Thus, all intersample univariate and multivariate comparisons were made using pooled samples of both sexes. Essentially no individual variation was noted in external morphology (other than normal meristic and color differences) in P. apache, P. fasciatus, or P. flavescens. Only a single specimen was found that lacked black-tipped guard hairs (MVZ 55716 from Ream’s Canyon, Navajo Co., Arizona). This condition must be regarded as a rare anomaly. Some authors (for example, Blair et al., 1957) have stated that the auditory bullae are in contact (apposed) ventrally, and have used this feature as a taxonomic character for E. apache. For adults, I noted if the bullae were in contact (Table 3). None of the specimens of P. fasciatus had bullae in ap- position, whereas approximately 10% of the P. apache and P. flavescens samples had apposed bul- lae. The Rio Grande Valley sample exhibited a sig- nificantly greater than expected number of individ- uals with bullae in contact, whereas the Uintah Basin sample exhibited significantly fewer individ- uals than expected. Populations with relatively large bullae had a higher proportion of bullae in apposition than those with relatively small bullae, such as the Uintah Basin sample. In any case, this character is not useful as a taxonomic trait. Several departures from the typical cusp patterns of the upper and lower fourth premolars were noted (Table 3). The normal pattern of the upper premolar consists of four major cusps, an anterior protocone (comprising the protoloph), and a three-cusped metaloph, consisting of a metacone, hypocone, and hypostyle (Fig. 6A). In about 7% of the individuals a prominent accessory cusp, representing either a paracone or a protostyle, was present on the P* (Fig. 6B, C, D, and F). In four individuals, the metaloph was compressed laterally, and the meta- loph cusps were united into a single structure, giv- ing the tooth a bicuspid appearance (Fig. 6E). The P of one individual was unicusped (Fig. 6G). The typical lower premolar has four prominent cusps on two transverse lophs. The anterior pro- tolophid consists of a protoconid and protostylid, and the posterior metalophid consists of a metaco- nid and hypoconid (Fig. 6H). The most common departure from the normal condition was the union 16 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 10 Fig. 6. — Cusp patterns and some anomalies observed in the upper and lower permanent fouth premolars of Perognathus fasciatus group pocket mice. A-E = occlusal views of right P; F and G = labial views of right P; H-L = occlusal views of left P^; M-O = labial views of left P4. A and Ft represent normal cusp patterns; B, C, D, F, I, K, M, and O depict teeth with accessory cusps; D, E, G, J, K, L, N, and O depict teeth with cusp deletions; B and F, I and M, J and N, and K and O each represent two views of similar anomalies. Hy = hypocone; Hy‘* = hypoconid; Hys = hypostyle; Me = metacone; Me"* = metaconid; Pa = paracone; Pr = proto- cone; Pr‘‘ = protoconid; Prs = protostyle; Prs"* = protostylid; X = accessory cusp of uncertain homology. of the protostylid and metaconid into a single cusp (Eig. 6J and N). Also common was the loss of one or two of the cusps. Generally, this loss involved the cusps on the metalophid (Eig. 6K, L, and O), but in a few instances cusps on the protolophid were absent (not figured). An extra cusp of uncer- tain homology was noted in a few individuals (Eig. 61, K, M, and O). Although some samples of P. apache exhibited a disproportionate number of cusp anomalies (for example, the White Sands and Rio Grande samples), and the samples of P. fascia- tus exhibited fewer than expected anomalies, no obvious geographic pattern could be discerned. Interpopulation Variation Karyology The karyotypes off*, apache, P. flavescens, and P. fasciatus are very similar (Table 4), and widely divergent from those of other members of the sub- genus Perognathus (Williams, 1978). A typical karyotype of P. apache is presented in Fig 7. The 1978 WILLIAMS— PEROG NATH US SYSTEMATICS 17 X chromosomes of this individual (Fig. 7) were dif- ferentially contracted, a condition noted frequently in the cells of females. Other karyotypes of P. fas- ciatus, P. flavescens, and P. apache are figured in Williams (1978). The chromosomes of P . flavescens copei were identical in gross structure to all P. apache samples except that from near Nueva Casas Grandes, Chihuahua. This latter population differed in having a submetacentric X, and in having a pair of small biarmed autosomes. This small autosome pair appears to be homologous to the small acro- centric pair with the secondary constriction found in the other karyotypes (the pair on the right in the bottom row in Fig. 7). A single pericentric inversion can account for this difference. The karyotype of P. fasciatus differed from the others in having ac- rocentric sex chromosomes. The nature of the karyotypic variation in theyh.v- ciatus group is not easy to interpret phyletically. According to one model of karyotypic evolution in the %ubgQnns Perognathus (Williams, 1978), the au- tosomal karyotype represented by P. fasciatus, P. flavescens, and most P. apache is primitive, and the karyotype of the Casas Grandes sample is de- rived. The differences are slight, however, and from one to three arm additions or pericentric inversions could convert any of the karyotypes into any of the others. The submetacentric X and the acrocentric Y have been regarded as primitive for Perognathus (Patton, 1969; Williams, 1978), but none of these karyotypes exhibits that combination of sex chro- mosome structure. The identical appearance of the chromosomes of P. flavescens and typical P. apache sets them apart from P. fasciatus and the Casas Grandes sample of P. apache, but the im- portance of these differences has not been estab- lished. n » HA HO Oft Oft aa *a ift *• AA ^ Fig. 7. — Representative karyotype of Perognathus apache. Fe- male P. a. gypsi from Walker Ranch, White Sands National Monument, Otero Co., New Mexico. Morphometric Variation The statistical summaries of the standard univar- iate analyses are given in Table 5. Coefficients of correlation between the morphometric traits, based upon the 21 sample means, are presented in Table 6. The number of significant correlations was high, and only width of interparietals and least interbullar distance exhibited significant negative correlations with the other characters. These two traits were highly correlated (/- = 0.91), and either expresses the degree of posterior constriction of the brain- case. Some traits (TL/HBL, IPL, IPW, RW, and LID) exhibited relatively low numbers of significant correlations. A factor analysis of the matrix of correlation demonstrated that only nine factors accounted for Table 4. — Chromosome characteristics «/' fasciatus group pocket mice. BA = biarmed: UA = uniarmed. Species S ? 2N FN Autosome BA pairs UA X Y P. fasciatus olivaceogriseus 1 1 44 48 3 18 A A P. fasciatus litus 1 1 44 48 3 18 A A P. fasciatus callisius 3 2 44 48 3 18 A A P. flavescens copei 5 2 44 48 3 18 ST ST P. apache apache 7 9 44 48 3 18 ST ST P. apache caryi 7 6 44 48 3 18 ST ST P. apache cleomophila 1 - 44 48 3 18 ST ST P. apache gypsi - 1 44 48 3 18 ST — P. apache relictus A 1 44 48 3 18 ST ST P. apache mekinotis 3 1 44 50 4 17 SM ST 18 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 10 Table 5. — Standard statistics for samples of Perognathus fasciatus, P. apache, and P. fiavescens. Trait N M SE cv Range N M SE CV Range Minimum Maximum Minimum Maximum 1. P. /. fasciatus 2. P.f. oli vaceogriseiis 1 1 1 135.7 2.269 5.54 123.0 147.0 15 133.1 1.099 3.20 126.0 142.0 ■) 11 62.6 1.410 7.46 57.0 70.0 15 62.7 0.720 4.45 59.0 68.0 3 1 1 73.1 1.504 6.82 65.0 80.0 15 70.4 0.974 5.36 65.0 81.0 4 11 17.4 0.338 6.45 16.0 19.0 15 16.7 0.157 3.63 15.7 18.0 5 1 1 7.5 0.157 6.92 7.0 8.0 15 7.1 0.134 7.37 6.0 8.0 6 1 1 0.859 0.023 8.72 0.738 0.971 15 0.894 0.016 6.92 0.738 0.985 7 8 22.60 0.208 2.59 21.65 23.50 15 22.17 0.104 1.89 21.25 22.80 8 8 22.60 0.208 2.59 21.65 23.50 15 22.14 0.101 1.77 21.25 22.75 9 10 4.88 0.063 4.08 4.55 5.20 15 4.95 0.041 3.24 4.65 5.20 10 10 22 0.041 4.01 3.05 3.40 15 3.12 0.032 3.96 2.95 3.35 1 1 8 4.41 0.063 4.04 4.05 4.70 15 4.33 0.030 2.71 4.00 4.50 12 9 7.69 0.094 3.68 7.35 8.20 15 7.60 0.086 4.39 7.00 8.05 13 8 11.92 0.082 1.96 1 1 .45 12.20 15 11.71 0.075 2.54 1 1.15 12.10 14 10 2.62 0.054 6.55 2.30 2.90 15 2.63 0.062 9.18 2.30 3.00 13 10 4.69 0.104 6.99 4.25 5.20 15 4.75 0.048 3.89 4.35 5.00 16 10 8.22 0.112 4.30 7.65 8.65 15 8.20 0.073 3.45 7.55 8.65 17 10 2.21 0.028 3.94 2.10 2.35 15 2.15 0.016 2.97 2.05 2.30 18 10 3.68 0.060 5.16 3.35 3.90 15 3.74 0.041 4.30 3.40 4.00 19 8 4.80 0.052 3.04 4.65 5.10 15 4.58 0.071 6.00 4.10 5.00 20 10 2.80 0.022 2.50 2.70 2.90 15 2.78 0.024 3.38 2.60 2.95 21 11 0.63 0.007 3.51 0.61 0.68 15 0.62 0.006 3.65 0.58 0.65 11 0.67 0.011 5.32 0.61 0.74 15 0.67 0.010 5.60 0.61 0.71 23 10 0.62 0.006 3.29 0.58 0.65 15 0.61 0.008 4.85 0.58 0.65 24 10 0.73 0.014 5.89 0.68 0.81 15 0.72 0.010 5.26 0.68 0.81 25 11 2.76 0.070 8.24 2.40 3.05 15 2.75 0.033 4.60 2.40 2.90 26 1 1 0.98 0.010 3.39 0.90 1.03 15 1.00 0.014 5.31 0.90 1.10 27 11 1.10 0.022 6.58 1.00 1.23 15 1.09 0.015 5.27 0.97 1.19 28 10 0.71 0.013 5.82 0.65 0.77 15 0.70 0.009 4.96 0.65 0.77 29 11 0.86 0.014 5.48 0.81 0.97 15 0.86 0.008 3.70 0.8! 0.90 30 11 0.98 0.008 2.69 0.94 1.00 15 0.97 0.006 2.34 0.94 1.03 90% of the total variance, and that the first five factors accounted for 85.1% of the variance (Table 7). Each of the traits except length and width of the interparietals and least interbullar distance showed high positive loading on Factor I, which is a general size factor. The posterior cranial region become more constricted with increasing size, as demon- strated by the negative loading of interparietal di- mensions and least interbullar distance on Factor I. Factor II, which accounts for 15.6% of the total variance, showed high positive values for width of interparietals, least interbullar distance, width of P4, and width of M’^. Traits with high negative val- ues were least interorbital breadth, length of M,, and length of hind foot. Factor II was most strongly influenced by traits expressing the postcranial con- striction. Factor III, accounting for 5.7% of the total vari- ance, showed a high positive value only for length of interparietal. Width of interparietals also also ex- hibited positive loading for Factor III, whereas length of tail exhibited negative loading. Samples with high positive scores for Factor III had rela- tively long interparietals and short tails. Factor IV accounted for 4.2% of the total variance and was loaded most strongly by length of articular process. Factor V, accounting for 3.5% of the total variance, had high loading on the TL/HBL ratio. Negative loading on length of head and body and positive loading on length of tail also reflect this "tail fac- tor." Samples with high factor scores for Factor V had relatively long tails. Beyond Factor V, the in- dividual factors accounted for relatively little of the variance. The factor scores of the first three factors, for each of the samples, are plotted in Fig. 8. Note that P. fasciatus samples are most distinctive in terms of Factor II, having wide interparietals and narrow 1978 WILLIAMS— PEROG NATH US SYSTEMATICS 19 Table 5. — Continued . Trait N M SE cv Range N M SE CV Range Minimum Maximum Minimum Maximum 3 .P. ,/ litus 4. P. f. caUistiis ! 27 139.8 0.824 3.06 134.0 149.0 18 134.3 1.249 3.95 123.0 146.0 2 27 66.9 0.536 4.16 61.0 75.0 18 63.3 1.105 7.41 53.0 70.0 3 27 72.9 0.53! 3.78 69.0 80.0 18 71.0 0.925 5.53 63.0 78.0 4 27 17.7 0.099 2.89 16.9 19.0 18 18.0 0.164 3.88 17.0 19.0 5 27 6.9 0.093 6.98 6.1 7.9 18 6.8 0.200 12.47 5.0 8.0 6 27 0.919 0.009 5.02 0.813 1.029 18 0.894 0.022 10.58 0.739 1.079 7 27 23.23 0.118 2.64 22.15 24.80 17 23.00 0.111 1.98 21.95 24.00 8 27 23.07 0.111 2.50 22.10 24.55 17 22.85 0.102 1.85 21.90 23.70 9 27 5.13 0.026 2.65 4.95 5.40 18 5.18 0.028 2.34 5.00 5.40 10 27 3.27 0.019 3.03 3.05 3.50 18 3.18 0.026 3.50 3.00 3.40 11 27 4.45 0.022 2.69 4.15 5.65 18 4.44 0.027 2.65 4.20 4.60 12 27 8.60 0.056 3.39 8.00 9.35 18 8.48 0.062 3.11 7.90 8.90 13 27 12.90 0.072 2.81 12.20 13.85 18 12.71 0.081 2.62 12.10 13.35 14 27 2.63 0.040 8.01 2.25 3.15 18 2.78 0.057 8.67 2.25 3.35 15 27 4.50 0.072 8.36 3.85 5.15 18 4.56 0.054 5.19 4.15 5.15 16 27 8.55 0.051 3.11 7.95 9.10 17 8.30 0.072 3.58 7.75 8.85 17 27 2.33 0.021 4.69 2.10 2.55 17 2.23 0.029 5.30 2.05 2.45 18 27 3.71 0.027 3.76 3.50 4.00 18 3.64 0.024 2.81 3.45 3.85 19 27 4.37 0.077 9.13 3.55 5.30 18 4.39 0.080 7.78 3.65 5.05 20 27 2.84 0.014 2.69 2.65 2.95 16 2.82 0.026 3.65 2.60 3.00 21 27 0.66 0.006 4.75 0.61 0.71 14 0.64 0.004 2.37 0.61 0.65 22 27 0.69 0.005 4.15 0.65 0.77 14 0.67 0.013 7.10 0.61 0.74 23 27 0.60 0.007 5.71 0.55 0.68 14 0.62 0.008 5.07 0.58 0.68 24 27 0.74 0.007 5.01 0.65 0.81 14 0.72 0.006 3.25 0.68 0.74 25 27 2.89 0.025 4.57 2.53 3.20 14 2.86 0.037 4.87 2.67 3.13 26 27 1.03 0.008 3.90 0.97 1.10 14 1.00 0.008 2.95 0.97 1 .03 27 27 1.10 0.009 4.06 1.00 1.19 14 1.10 0.012 3.95 1 .03 1.16 28 27 0.71 0.008 6.18 0.6! 0.81 14 0.73 0.012 5.94 0.68 0.81 29 27 0.91 0.010 5.92 0.81 1.00 14 0.88 0.009 3.91 0.84 0.94 30 27 0.95 0.007 3.98 0.87 1 .00 14 0.96 0.015 5.81 0.81 1 .03 interorbital regions. The P. flavescens samples are most notable in their highly negative scores for Fac- tor I and high positive scores for Factor III. They are small in size (Factor I), with relatively short tails and with broad interparietals (Factor III). Of the P. apache samples, the northern ones are, per- haps, the most unique, being large, with long tails and narrow crania (Factor I). This is, of course, a simplification of the geographic variation in the morphometric traits (Table 5), but most of the vari- ation (77.4%) is accounted for in these three factors, and the factors are most strongly influenced by suites of related characters. Tests for significant differences between samples, using the SS-STP routine, resulted in a large num- ber of superfluous comparisons between samples that are widely separated geographically. There- fore, only a summary of the comparisons between samples in geographic proximity is presented (Table 8). The comparisons were limited to samples of populations that are neighbors and, potentially, can exchange genes. The results of the SS-STP com- parisons are illustrated in Fig. 9. To simplify the picture, some of the less likely comparisons were omitted from Fig. 9, but are given in Table 8. Note that the P. fasciatus samples (1-4) exhibited high numbers of significant differences with adjacent samples of P. flavescens (21) and P. apache (5). These were not the highest numbers of significant differences, but because these populations are sym- patric, or, in the case of P. apache, in close prox- imity, reproductive isolation is suggested. Of the P. apache samples, numbers 5, 6, 10, 12, 15, 16 or 17, and 19 form a chain of populations with few signif- icant differences, extending from northern Utah to Chihuahua (Fig. 9). Most other samples peripheral to this chain are connected by only one or a few routes. For example, the Painted Desert and Flag- 20 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 10 Table 5. — Continued. 1 rail N M SE cv Range N M SE CV Range Minimum Maximum Minimum Maximum 5 . P. apache Uintah Basin 6. P. apache Moab 1 60 140.7 0.717 3.94 128.0 155.0 27 137.4 1.078 4.07 123.0 146.0 *> 60 68.3 0.489 5.55 52.0 74.0 27 66.7 0.817 6.36 59.0 73.0 3 61 72.5 0.646 6.97 61.0 87.0 27 70.7 0.636 4.68 64.0 78.0 4 61 18.2 0.095 4.08 16.7 20.0 27 18.5 0.233 6.56 15.0 21.0 5 60 7.0 0.064 7.05 6.0 8.0 24 6.7 0.107 7.79 6.0 8.0 6 60 0.949 0.012 9.67 0.667 1.147 27 0.946 0.013 7.37 0.815 1.078 7 60 23.92 0.076 2.44 22.85 25.50 23 23.53 0.133 2.71 22.30 24.60 8 59 23.74 0.075 2.42 22.55 25.30 21 23.33 0.127 2.49 22.05 24.25 9 61 5.36 0.023 3.36 5.00 5.70 25 5.26 0.030 2.85 5.00 5.60 10 61 3.30 0.015 3.61 3.05 3.55 27 3.27 0.048 7.64 2.35 3.75 1 1 61 4.58 0.017 2.83 4.35 4.90 26 4.51 0.028 3.19 4.15 4.75 12 61 8.80 0.038 3.36 8.30 9.45 26 8.51 0.051 3.08 7.85 9.05 13 61 13.08 0.042 2.50 12.45 14.00 24 12.81 0.058 2.24 12.30 13.25 14 61 3.22 0.028 6.71 2.80 3.65 26 3.14 0.037 5.95 2.80 3.50 15 61 4.35 0.034 6.07 3.75 4.95 26 4.39 0.085 9.92 3.80 5.25 16 60 8.74 0.044 3.92 8.00 9.50 25 8.54 0.075 4.38 7.90 9.25 17 60 2.23 0.014 4.90 2.00 2.50 25 2.30 0.024 5.33 2.00 2.60 18 61 3.85 0.019 3.88 3.60 4.20 26 3.79 0.043 5.79 3.45 4.30 19 61 3.94 0.027 5.27 3.50 4.35 25 4.05 0.069 8.49 3.55 4.60 20 60 2.95 0.01 1 2.88 2.70 3.10 27 2.93 0.019 3.39 2.70 3.10 21 60 0.65 0.004 4.91 0.58 0.71 27 0.63 0.007 6.26 0.52 0.68 22 60 0.66 0.004 4.23 0.61 0.71 27 0.64 0.008 6.75 0.58 0.71 23 60 0.64 0.004 4.74 0.58 0.71 27 0.64 0.006 5.28 0.58 0.71 24 60 0.75 0.005 4.61 0.68 0.84 27 0.74 0.009 6.36 0.65 0.81 25 60 2.84 0.022 6.00 2.53 3.27 27 2.96 0.038 6.65 2.67 3.40 26 61 1.05 0.005 4.00 0.94 1.16 27 1.03 0.006 3.05 1.00 1.10 27 61 1.22 0.005 3.37 1.13 1.32 27 1.20 0.009 3.98 I.IO 1.29 28 61 0.72 0.005 5.70 0.61 0.77 26 0.69 0.008 6.20 0.61 0.77 29 60 0.96 0.005 3.60 0.84 1.06 27 0.99 0.007 3.91 0.94 1.06 30 60 1.0! 0.005 4.11 0.94 1.13 27 1.01 0.010 5.28 0.87 1.13 Staff samples were closely similar, and appear to be linked with the others along an avenue to the east, in the direction of the Gallup sample (Fig. 9). The Canyon Largo sample from higher juniper and pin- yon-juniper associations (11), was very distinct from the adjacent San Juan Basin sample (10) from a lower, sage-grassland habitat. The Canyon Largo sample was more similar to the Santa Fe (14) and Estrella samples (12) from ecologically similar areas. The Rio Grande Valley sample ( 15) was very dif- ferent from the adjacent Deming Plains sample ( 19), and they appear to be linked only by an indirect route through the Gran Quivira (17) or San Augus- tine Plains sample (16). Mice typical of the Rio Grande Valley sample were found from the Rio Sa- lado, north of Albuquerque, southward to where the Valley narrows near Val Verde, New Mexico. No Apache pocket mice are known from between Val Verde and Las Cruces in the Rio Grande Val- ley. I have, however, collected a few Apache pock- et mice, typical of the Deming Plains population, as far north as Engle in the Jornada del Muerto. Only two significant differences were found be- tween the Deming Plains sample of P. apache and P. flavescens copei (Table 8, Fig. 9). Other popu- lations of the Apache pocket mouse in proximity to P. flavescens were quite distinct ivom flavescens, with from seven to 16 significant differences. These were no greater than the differences among some samples of Apache pocket mice, however. Of the potential avenues of gene exchange between Apache and plains pocket mice, the Trans-Pecos route seems most likely on the basis of the univar- iate analyses (Fig. 9). Even though the Uintah Basin appears to be iso- lated, samples of the Apache pocket mice occupy- ing the Basin were essentially the same as samples 1978 WILLIAMS— PEROGNATHUS SYSTEMATICS 21 Table 5. — Continued. Trait N M SE cv Range N M SE CV Range Minimum Maximum Minimum Maximum 7. P. apache Painted Desert 8. P. apache Elagstaff 1 52 133.1 0.987 5.26 1 19.0 156.0 32 132.2 1.098 4.69 120.0 145.0 2 52 64.1 0.595 6.64 50.0 73.0 32 63.4 0.598 5.34 58.0 73.0 3 54 69.3 0.750 7.78 61.0 89.0 32 68.8 0.724 5.95 58.0 77.0 4 54 18.5 0.119 4.97 17.0 21.0 32 19.1 0.137 4.05 17.5 20.5 5 28 6.1 0.067 8.15 6.0 7.0 7 6.4 0.190 7.86 6.0 7.0 6 52 0.929 0.012 9.49 0.714 1.076 32 0.918 0.01 1 6.73 0.824 1.086 7 49 22.60 0.090 2.90 21.30 24.35 31 22.60 0.156 3.83 21.10 24.55 8 49 22 42 0.089 2.89 21.25 23.95 32 22.50 0.149 3.73 21.00 24.50 9 51 5.14 0.022 3.07 4.75 5.55 32 5.25 0.039 4.18 4.80 5.70 10 55 3.19 0.022 5.07 2.95 3.85 32 3.19 0.028 4.99 2.85 3.60 II 53 4.37 0.021 3.58 4.05 4.65 32 4.29 0.029 3.83 3.95 4.60 12 54 8.37 0.049 4.41 7.50 9.35 32 7.97 0.071 5.02 7.25 8.60 13 49 12.58 0.059 3.29 1 1.65 13.35 31 12.25 0.076 3.49 1 1.60 13.15 14 54 2.87 0.032 8.01 2.20 3.30 32 2.88 0.041 8.02 2.35 3.40 15 54 3.94 0.040 7.26 3.30 4.65 32 3.92 0.073 10.50 2.80 4.70 16 53 7.94 0.048 4.55 7.10 8.95 32 8.15 0.070 4.87 7.40 9.25 17 51 2.26 0.016 5.30 2.05 2.55 32 2.32 0.024 5.94 2.10 2.60 18 55 3.68 0.023 4.61 3.35 4.15 32 3.69 0.028 4.24 3.40 4.00 19 53 3.90 0.034 6.06 3.45 4.45 31 3.87 0.057 8.21 3.10 4.40 20 55 2.79 0.014 2.07 2.60 3.05 32 2.76 0.021 4.13 2.50 3.05 21 55 0.58 0.006 7.33 0.48 0.65 32 0.59 0.008 7.64 0.48 0.68 22 55 0.62 0.006 6.94 0.55 0.74 32 0.60 0.007 6.43 0.55 0.68 23 55 0.61 0.005 6.17 0.55 0.68 32 0.61 0.007 6.68 0.52 0.68 24 55 0.70 0.006 6.90 0.61 0.81 32 0.70 0.008 6.25 0.61 0.81 25 54 2.80 0.002 5.80 2.33 3.20 32 2.86 0.030 5.99 2.53 3.27 26 55 0.99 0.006 4.68 0.87 1.10 32 0.97 0.007 4.36 0.87 1 .03 27 55 I.IO 0.008 5.63 0.94 1.26 32 1.10 0.01 1 5.56 0.97 1.23 28 54 0.65 0.006 7.41 0.55 0.81 32 0.66 0.008 7.33 0.55 0.81 29 55 0.93 0.007 5.35 0.81 1.06 32 0.95 0.009 5.31 0.84 1.06 30 53 0.96 0.007 5.34 0.84 1.13 32 0.94 0.009 5.41 0.84 1.03 of populations from farther south in Utah and Col- orado. On the other hand, the San Luis Valley pop- ulation of south-central Colorado appeared to be relatively isolated and well differentiated from its neighbors. Samples of the Painted Desert and Flagstaff populations were quite different from those from adjacent areas to the north and east. The lower San Juan River and Chuska Mountains ap- pear to be effective barriers to gene exchange be- tween these populations (Fig. 4). The most apparent geographic pattern to the mor- phometric variation in Apache pocket mice was a strong north-south size dine. This is best seen in the main chain of populations extending from north to south. Occipitonasal length is representative of size and well illustrates this dine (Fig. 10). A cor- relation analysis between the morphometric char- acters and the climatic and geographic variables showed the size dine to be highly significant (Table 9, Fig. 11). Latitude showed 23 significant positive correlations with the morphometric traits, of which 21 were highly significant (Table 9). The climatic severity index, growing season, and mean July min- imum temperature Were not significantly correlated with any of the morphometric traits. Mean annual temperature was negatively correlated with body size (TOTL and HBL). Latitude was, however, more highly correlated with size than the variables expressing environmental temperature. It is prob- able that these variables do not adequately express the complexities of the yearly climatic cycle or the temperature factors affecting these pocket mice. Apache pocket mice do vary in size as predicted by Bergman’s principle, and it is most likely that rel- ative heat loss is an important factor in determining size in these populations. These mice store seeds, become hypothermic at low ambient temperatures and in times of food deprivation, and are generally 22 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 10 Table 5. — Continued. Trait N M SE CV Range N M SE CV Range Minimum Maximum Minimum Maximum 9. P. apache Gallup 10. P. apache San Juan Basin 1 8 131.5 2.062 4.43 124.0 139.0 32 137.8 1.093 4.49 127.0 151.0 A 8 61.9 1.187 5.42 58.0 69.0 32 67.3 0.511 4.29 62.0 72.0 3 8 69.6 1.557 6.33 63.0 76.0 32 70.4 0.747 6.00 63.0 80.0 4 8 18.3 0.164 2.53 18.0 19.0 32 19.5 0.105 3.06 18.0 20.0 5 0 — — — — — 32 6.8 0.050 4.19 6.0 7.0 6 8 0.891 0.025 7.85 0.815 0.985 32 0.958 0.009 5.20 0.858 1.078 7 5 22.54 0.206 2.05 22.00 23.00 32 23.34 0.135 3.27 21.30 24.70 8 6 22.41 0.183 2.00 21.80 23.00 32 23.14 0.128 3.12 21.10 24.60 9 6 5.16 0.077 3.65 4.90 5.35 32 5.29 0.033 3.55 4.90 5.65 10 6 3.29 0.042 3.10 3.10 3.35 32 3.42 0.025 4.09 3.15 3,75 11 6 4.37 0.031 1.72 4.30 4.45 32 4.40 0.019 2.41 4.20 4.60 12 5 8.17 0.087 2.35 7.90 8.40 32 8.42 0.067 4.45 7.60 9.25 13 6 12.51 0.073 1.55 12.25 12.80 32 12.71 0.062 2.81 12.05 13.70 14 6 3.19 0.104 8.01 2.80 3.45 32 3.07 0.047 2.65 3.70 8.76 15 6 4.07 0.077 4.64 3.75 4.25 32 4.05 0.057 8.02 3.50 4.75 16 6 8.14 0.104 3.14 7.80 8.50 32 8.46 0.075 5.01 7.60 9.25 17 6 2.26 0.047 5.13 2.10 2.40 30 2.38 0.026 5.88 1.90 2.60 18 6 3.71 0.108 7.16 3.40 4.00 32 3.85 0.034 5.00 3.50 4.30 19 6 4.01 0.030 1.83 3.90 4.10 32 3.92 0.043 6.26 3.35 4.40 20 7 2.88 0.031 2.81 2.75 3.00 32 2.97 0.017 3.31 2.75 3.15 21 7 0.62 0.01 1 4.70 0.58 0.65 32 0.62 0.007 6.65 0.55 0.71 22 7 0.63 0.011 4.53 0.61 0.68 32 0.65 0.008 6.98 0.58 0.77 23 7 0.63 0.009 4.02 0.58 0.65 32 0.64 0.007 5.88 0.58 0.74 24 7 0.73 0.018 6.70 0.68 0.81 32 0.75 0.009 7.10 0.68 0.87 25 7 2.77 0.095 9.09 2.27 3.00 32 2.93 0.027 5.28 2.60 3.33 26 5 0.99 0.012 2.72 0.97 1.03 32 1.05 0.008 4.49 0.97 1.13 27 5 1.13 0.016 3.11 1.10 1.16 32 1.18 0.007 3.17 l.IO 1.26 28 4 0.68 0.022 6.73 0.61 0.71 32 0.69 0.007 6.01 0.61 0.77 29 7 0.99 0.010 2.45 0.97 1 .03 32 0.99 0.007 4.08 0.94 1.10 30 7 0.99 0.013 3.62 0.94 1.03 32 1.01 0.009 4.84 0.90 1.13 inactive on the surface during inclement weather (personal observations); so it would, perhaps, be naive to expect a simple relationship between size and mean annual temperature. Another factor that is possibly working in concert with temperature in selecting for size is differential resource allocation. In the southern portion of their range, Apache pocket mice are sympatric with an- other sand-dwelling species, P. penicillatus. The desert pocket mouse has about the same body size as the northern populations of P. apache (HBL = 70-75 mm), but is about 13% larger than the sym- patric population ofP. apache. Mares and Williams (1977) presented experimental evidence suggesting that the differences in body size among several het- eromyid granivores determines, in part, the sizes of the seeds gathered. They also showed that larger species were able to gather a greater size array of seeds. Thus, within limits, larger body size should be advantageous for most species. Competition be- tween P. apache and P. penicillatus could select for mice with well-differentiated body sizes, and could be partly responsible for the small size of southern populations of P. apache. North of the range of P. penicillatus, the lack of competition with that species may permit selection for larger body size in P. apache. North of the San Juan Riv- er, the absence of competition with sympatric con- geners and cooler ambient temperatures may both be important factors in the selection for even larger body size in Apache pocket mice. The size of the auditory bullae (BL, BW) and interorbital breadth exhibited significant negative correlations with mean annual precipitation (Table 9, Fig. 12). These traits were positively correlated with latitude, although latitude and mean annual 1978 WILLIAMS— PEROGNATHUS SYSTEMATICS 23 Table 5. — Conthmed . Trait N M SE cv Range N M SE CV Range Minimum Maximum Minimum Maximum II . P. apache Canyon Largo 12. P. apache Estrella 1 25 132.9 1. 171 4.48 120.0 143.0 13 137.4 1.146 3.01 131.0 144.0 1 25 63.6 0.716 5.63 55.0 71.0 13 67.6 0.738 3.94 62.0 72.0 3 26 69.0 0.860 6.35 57.0 75.0 13 69.8 0.856 4.42 64.0 75.0 4 26 17.8 0.126 3.61 16.7 19.3 13 19.3 0.166 3.11 18.5 20.5 5 26 6.6 0.062 4.76 6.1 7.2 13 6.7 0.104 5.64 6,0 7.0 6 25 0.922 0.015 8.25 0.733 1.105 13 0.970 0.015 5.74 0.837 1 .045 7 24 22.57 0.162 3.53 21.35 24.10 13 23.05 0.161 2.53 22 2^ 23.80 8 25 22.41 0.151 3.37 21.30 24.00 13 22.87 0.168 2.63 22.00 23.70 9 26 5.13 0.034 3.46 4.75 5.50 13 5.22 0.051 3.53 5.00 5.50 10 26 3.26 0.034 5.26 2.85 3.50 13 3.34 0.024 3.20 3.50 2.67 II 26 4.32 0.027 3 22 4.10 4.65 13 4.36 0.038 3.18 4.15 4.65 12 24 7.97 0.065 4.02 7.45 8.55 13 8.30 0.038 4.25 7.65 8.80 13 25 12.27 0.080 3.34 11.65 13.20 13 12.75 0.096 2,59 12.10 13.25 14 26 2.81 0.043 7.84 2.30 3.20 13 2.89 0.059 7.45 2.60 3.30 15 26 3.96 0.045 5.87 3.60 4.45 13 4.09 0.069 6,06 3.55 4.45 16 25 8.08 0.071 4.42 7.55 9.00 13 8.27 0.107 4.67 7.60 8.80 17 25 2.23 0.030 6.83 1.85 2.40 13 2.31 0.039 6.13 2.10 2.50 18 26 3.68 0.039 5.39 3.35 4.10 13 3.79 0.045 4.32 3,55 4.10 19 26 3.85 0.054 7.20 3.40 4.30 13 3.96 0.064 5.83 3.50 4.25 20 25 2.85 0.030 5.30 2.55 3.15 13 2.93 0,033 4.04 2.80 3.15 21 25 0.59 0.01 1 9.14 0.52 0.68 13 0.63 0.008 4,49 0.58 0.68 22 25 0.63 0.008 6.75 0.55 0.71 13 0.64 0.01 1 6,30 0.58 0.74 23 25 0.63 0,008 6.48 0.55 0.68 13 0.62 0.01 1 6.58 0.55 0.68 24 25 0.74 0.009 6.34 0.65 0.84 13 0.74 0.013 6.40 0.68 0.84 25 24 2.88 0.036 6.13 2.67 3.20 13 2.90 0.017 4.36 2.67 3.13 26 26 I.OI 0.01 1 5.70 0.94 1.13 13 1.02 0.013 4.78 0.97 1.10 27 26 1.13 0.010 4.58 1.00 1.19 13 1.14 0.014 4.38 1 .03 1.19 28 26 0,68 0.007 5.29 0.61 0.74 13 0.67 0.01 1 6.11 0.58 0.74 29 25 0.95 0.010 5.43 0.81 1.03 13 0.97 0.009 3.60 0.90 1.03 30 25 0.98 0.011 5.84 0.87 1.10 13 0.99 0.01 1 3.90 0.90 1.03 precipitation were not correlated (Table 10). Rela- tive bullar size appears to be related to two inde- pendent factors, the general size factor and the amount of environmental moisture. These two fac- tors were not correlated, and their associations with bullar and interorbital size were conflicting. Varia- tion in the size of the auditory bullae is in agreement with the commonly observed ecogeographic prin- ciple of animals from drier climates having larger sound sensing organs that their relatives from moist- er climates. The rostrum (NL, NW, RW, lOB) was wider and longer in samples from the more arid lo- calities, but only interorbital breadth showed a sig- nificant negative correlation with mean annual pre- cipitation. This association is as one would expect if these mice are adapted to decrease pulmonary water loss in drier climates. The strong size dine, however, partially masked these associations. Tail size did not exhibit a significant negative cor- relation with mean annual temperature, contrary to the prediciton of Allen’s principle. Rather, the TL/ HBL ratio increased with increasing latitude (r = 0.60), with increasing body size (/• with HBL = 0.68) and with head size (/■ with ONL = 0.72). This suggests that some factor or factors are selecting for increasingly longer tails with increasing body size. 1 have observed these mice in bipedal stances while foraging, and a longer tail could be necessary to counterbalance a larger body. The relationship between head size and tail length is shown in Fig. 13. The significant associations between length of in- terparietal and elevation, and between certain den- tal measurements (P*W and M,L; Table 9) and el- evation adjusted for latitude have no obvious explanations. Perhaps they are spurious correla- 24 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 10 Table 5. — Continued . Frail N M SE cv Range N M SE CV Range Minimum Maximum Minimum Maximum 13. P. apache San Luis Valley 14. P. apache Santa Ee 1 17 136.3 1.288 3.90 123.0 144.0 24 136.2 1.325 4.81 1 18.0 145.0 ■> 17 67.2 0.671 4.11 60.0 71.0 24 65.0 0.797 6.01 56.0 70.0 3 17 69.1 0.813 4.85 63.0 76.0 26 70.2 0.797 5.78 62.0 80.0 4 17 18.6 0.139 3.09 18.0 20.0 26 18.2 0.141 3.95 16.5 20.0 5 15 6.7 0.152 8.86 6.0 8.0 11 6.6 0.187 9.38 5.8 7.8 6 17 0.972 0.011 4.58 0.882 1.031 24 0.927 0.013 7.15 0.802 1 .060 7 13 22.41 0.186 3.00 21.00 23.60 22 22.46 0.136 2.85 21.25 24.20 8 15 22.39 0.162 2.80 21.00 23.55 21 22.40 0.138 2.82 21.25 23.90 9 16 5.37 0.045 3.39 5.00 5.65 24 5.10 0.034 3.27 4.80 5.40 10 17 3.12 0.030 3.97 2.90 3.35 26 3.24 0.024 3.80 3.00 3.55 1 1 16 4.21 0.029 2.78 3.95 4.35 25 4.34 0.030 3.50 4.10 4.60 12 14 7.86 0.074 3.52 7.30 8.30 25 7.73 0.066 4.27 7.00 8.45 13 15 12.14 0.080 2.58 11.60 12.50 23 12.17 0.087 3.38 11.55 13.55 14 16 2.63 0.048 7.30 2.35 3.00 24 2.81 0.059 10.25 2.35 3.55 15 16 3.83 0.065 6.81 3.35 4.50 24 4.06 0.078 9.44 3.25 4.90 16 16 7.79 0.098 5.02 7.15 8.40 25 8.07 0.064 4.02 7.35 8.65 17 15 2.25 0.027 4.70 2.05 2.35 24 2 24 0.028 6.16 1.95 2.50 18 15 3.75 0.041 4.29 3.45 3.95 25 3.74 0.038 5.13 3.35 4.10 19 15 3.99 0.052 5.01 3.65 4.40 24 4.15 0.056 6.63 3.75 4.75 20 16 2.76 0.027 3.93 2.55 2.90 27 2.81 0.017 3.24 2.65 2.95 21 16 0.57 0.010 7.27 0.48 0.65 27 0.61 0.006 5.32 0.55 0.68 22 16 0.59 0.006 4.38 0.55 0.65 27 0.64 0.006 5.05 0.58 0.71 23 16 0.61 0.007 4.71 0.55 0.65 27 0.61 0.007 5.65 0.55 0.68 24 16 0.71 0.009 5.12 0.68 0.81 27 0.73 0.007 5.19 0.65 0.81 25 16 2.78 0.021 6.36 2.40 3.07 27 2.73 0.030 5.75 2.40 3.00 26 16 0.86 0.008 3.66 0.81 0.90 26 0.97 0.011 5.93 0.89 1.06 27 16 1.02 0.010 3.77 0.97 1.10 26 1.11 0.010 4.80 1.00 1.19 28 16 0.63 0.008 ^ 2^ 0.58 0.71 26 0.65 0.008 6.19 0.55 0.74 29 16 0.95 0.008 3.26 0.90 1.00 27 0.95 0.007 3.67 0.87 1.00 30 16 0.94 0.009 3.73 0.90 1.03 27 0.97 0.008 4.34 0.90 1.06 tions, or perhaps they are somehow affected by available moisture which increases with increasing elevation. Color Variation Color was more variable geographically than were the morphometric traits, and consequently, I did not group samples as much for the analysis of color. Table 1 1 lists the means for the color indices of samples of P. apache and P. flavescens. In gen- eral, an increase in darkness was accompanied by an increase in richness (r - 0.52). Notable excep- tions were seen in samples from unusually light or dark colored soils. The Uintah Basin sample had a relatively large number of black-tipped guard hairs (darkness), but the yellowish color (richness) was quite pale. Soils in that region are very pale-tannish or grayish in color. The White Sands sample was also exceptional in that yellowish pigment was ab- sent in most adults (most young mice in juvenile pelage had a very pale yellowish tinge). Their basic color was white, overlain by a normal number of black-tipped hairs, presenting a neutral, grayish ap- pearance. The gypsum dunes upon which these mice were collected are white. Samples from red- dish sands, such as those near Caprock and Tolar were darker and more orange-colored than those from tanner soils. A correlation analysis between the color param- eters and the climatic and geographic variables is given in Table 10. The color indices were all highly correlated with mean annual precipitation, and rich- ness and relative darkness were correlated with el- evation (elevation and mean annual precipitaion were highly correlated). The relative darkness in- dex is plotted against mean annual precipitation in 1978 mLLlAMS—PEROGNATHUS SYSTEMATICS 25 Table 5. — Continued . Range Range Trait N M SE cv Minimum Maximum N M SE CV Minimum Maximum 15. P. apache Rio Grande Valley 16. P. apache San Augustine Plains 1 66 133.8 0.762 4.77 1 19.0 151.0 17 128.8 1.651 5.28 117.0 145.0 2 66 63.1 0.565 7.32 50.0 78.0 17 60.2 1.034 7.08 51.0 68.0 3 68 70.9 0.506 6.00 58.0 80.0 17 68.6 0.955 5.74 64.0 77.0 4 68 18.6 0.100 4.62 17.0 20.0 17 18.1 0.157 3.59 17.3 19.5 5 67 6.6 0.067 8.39 5.0 8.0 14 6.6 0.093 5.34 6.1 7.1 6 66 0.892 0.01 1 9.64 0.734 1.121 17 0.879 0.015 7.08 0.708 0.984 7 63 22.88 0.087 3.14 21.00 24.45 15 22.30 0.131 2 27 21.75 23.60 8 63 22.81 0.085 3.06 21.00 24.15 15 22.20 0.124 2.16 21.65 23.45 9 68 5.24 0.026 4.12 4.65 5.75 17 5.09 0.053 4.32 4.75 5.60 10 69 3.27 0.017 4.41 3.00 3.60 17 3.15 0.029 3.83 2.95 3.35 1 1 68 4.36 0.016 3.14 4.05 4.65 17 4.28 0.025 2.38 4.05 4.45 12 64 8.10 0.045 4.60 7.30 8.95 16 7.97 0.073 3.68 7.50 8.40 13 64 12.34 0.059 3.79 11.35 13.40 17 12.24 0.089 2.97 1 1 .60 12.90 14 68 2.87 0.027 7.80 2.35 3.35 17 2.81 0.060 8.79 2.40 3.35 15 68 4.07 0.039 7.93 3.45 5.00 17 4.09 0.081 8.17 3.50 4.65 16 65 8.28 0.046 4.51 7.35 9.15 15 7.95 0.072 3.49 7.60 8.55 17 66 2.28 0.020 7.02 1.95 2.80 16 2.14 0.022 4.11 2.00 2.30 18 68 3.78 0.028 6.20 3.30 4.25 17 3.71 0.047 5.31 3.50 4.15 19 68 3.99 0.040 8.19 2.65 5.10 17 4.00 0.079 8.11 3.45 4.50 20 69 2.85 0.013 3.85 2.60 3.20 17 2.74 0.025 3.81 2.55 2.90 21 69 0.58 0.005 6.91 0.52 0.71 17 0.58 0.010 6.80 0.52 0.65 22 69 0.64 0.004 6.03 0.58 0.71 17 0.60 0.009 5.81 0.55 0.68 23 69 0.61 0.004 5.88 0.52 0.71 17 0.60 0.009 6.07 0.52 0.68 24 69 0.73 0.005 5.41 0.61 0.81 17 0.71 0.01 1 6.37 0.65 0.84 25 69 2.82 0.010 6.05 2.40 3.27 17 2.73 0.016 4.95 2.40 2.93 26 69 0.98 0.006 5.01 0.84 1.10 17 0.93 0.01 1 4.65 0.87 1.03 27 69 1.19 0.006 4.43 1.03 1.23 17 1.11 0.012 4.46 0.97 1.16 28 69 0.66 0.004 5.42 0.58 0.74 17 0.65 0.008 4.79 0.61 0.7! 29 69 0.97 0.004 3.74 0.87 1.03 17 0.91 0.010 4.60 0.81 0.97 30 69 0.98 0.004 3.59 0.90 1.06 17 0.95 0.008 3.62 0.90 1.03 Fig. 14. Here too, notable exceptions to the rela- tionship between darkness and the amount of pre- cipitation are explained by unusually colored sands. For example, the White Sands sample was lighter than would be expected on the basis of precipita- tion, and the sample from near Flagstaff was darker than expected. This latter sample came from an area with a relatively high amount of precipitation, and with soils composed of black, volcanic cinders. Samples from dark, reddish sands were also darker than expected on the basis of precipitation. With two independent color parameters, individ- uals of the populations can be selected to closely match the color of the substrate. The yellowish- orange pigment varies in concentration to approxi- mate the color of the sands, which are generally some shade of tan or reddish, although both white (gypsum crystals) and black (volcanic cinders) soils occur within the range of P. apache. The number of black-tipped hairs seems most important in de- termining the overall darkness or lightness of the mice. Within limits, the actual color of the substrate appears to be less important in determining dark- ness, especially relative darkness, than is the amount of precipitation. Precipitation does not di- rectly determine soil color, although it is a well- known phenomenon that soils tend to be darker col- ored in areas of higher precipitation, due to higher humus contents. The sandy soils upon which these mice are found have very little organic matter and essentially no surface litter, and vegetation consists mostly of small, annual forbs. The perennials that occur are generally widely scattered. The amount of vegetational cover is primarily dependent upon the amount of moisture available during the growing season of small annuals. Within the range of P. 26 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 10 Table 5. — Continued . Trait N M SE CV Range N M SE CV Range Minimum Maximum Minimum Maximum 17. P. apache Gran Quivira 18. P. apache White Sands 1 13 130.5 1.651 4.56 120.0 139.0 51 128.7 0.677 3.76 120.0 140.0 A 13 62.4 1.010 5.83 57.0 69.0 51 62.2 0.456 5.23 56.0 68.0 3 15 68.6 0.979 5.53 63.0 74.0 54 66.2 0.448 4.98 60.0 73.0 4 15 17.6 0.114 2.56 16.7 18.1 54 18.3 0.103 4.13 16.0 20.0 5 10 6.5 0.139 6.72 6.1 7.5 48 6.5 0.072 7.67 6.0 7.5 6 13 0.898 0.020 8.10 0.783 1 .063 51 0.939 0.009 6.71 0.828 1.097 7 II 22.30 0.087 1.30 21.90 22.80 52 22.47 0.073 2.35 21.35 23.80 8 12 22.28 0.087 1.36 21.85 22.75 47 22.40 0.071 2.16 21.30 23.75 9 13 5.10 0.037 2.59 4.85 5.25 54 5.17 0.023 3.30 4.50 5.45 10 15 3.14 0.062 1.98 3.00 3.25 53 3.12 0.017 4.06 2.90 3.40 11 15 4.30 0.032 2.88 4.10 4.60 54 4.21 0.015 2.67 3.85 4.45 12 15 8.05 0.085 4.10 7.35 8.50 51 8.21 0.042 3.61 7.55 8.85 13 14 12.30 0.080 2.49 11.65 12.85 48 12.40 0.042 2.39 1 1 .65 13.00 14 14 2.73 0.050 6.84 2.35 3.05 52 2.91 0.029 7.13 2.55 3.50 15 14 4.06 0.088 8.09 3.40 4.65 52 3.94 0.035 6.40 3.15 4.35 16 13 7.94 0.057 2.57 7.60 8.25 54 8.00 0.045 4.15 7.40 8.80 17 13 2.20 0.029 4.67 2.00 2.35 53 2.21 0.018 5.78 1.90 2.55 18 14 3.83 0.038 3.71 3.60 4.00 54 3.83 0.020 3.93 3.45 4.15 19 14 4.00 0.064 5.96 3.60 4.50 51 3.76 0.035 6.64 3.10 4.15 20 14 2.80 0.017 2.32 2.70 2.85 49 2.78 0.015 3.72 2.55 3.10 21 15 0.60 0.009 5.51 0.58 0.68 48 0.56 0.007 8.59 0.48 0.68 22 15 0.64 0.009 5.57 0.61 0.79 49 0.60 0.006 7.05 0.52 0.68 23 14 0.61 0.006 3.86 0.58 0.65 49 0.62 0.005 5.72 0.55 0.71 24 14 0.72 0.007 3.61 0.68 0.77 49 0.72 0.005 5.18 0.65 0.81 25 15 2.77 0.028 3.94 2.53 3.00 49 2.67 0.009 4.83 2.40 2.93 26 15 0.97 0.011 4.54 0.90 1.03 49 0.94 0.011 8.29 0.55 1.06 27 15 I.IO 0.013 4.58 1.03 1.19 49 1.06 0.006 4.36 0.97 1.23 28 15 0.65 0.008 5.02 0.58 0.71 49 0.63 0.005 5.32 0.58 0.71 29 15 0.94 0.009 3.78 0.87 1.00 49 0.92 0.006 4.43 0.87 1.00 30 15 0.97 0.012 5.19 0.87 1.06 49 0.94 0.006 4.68 0.87 1.06 apache, the growth of annuals generally occurs from June through August. Sands receiving higher amounts of moisture will have a more lush vegeta- tional cover, and appear darker (due both to the plant cover, seen from above, and the shadows cast on the sand by vegetation). The highly significant correlation between mean annual precipitation and relative darkness is, I believe, attributable to this phenomenon, with predation being the ultimate fac- tor determining the color of the mice. There was no apparent geographic continuity to the observed color variation (Fig. 15), although there was a predictable pattern. Higher elevations receive more precipitation (r = 0.60) and have low- er temperatures (/• = 0.73), hence more moisture is available for plant growth. Samples of Apache pocket mice from higher areas such as Flagstaff, Coventry, Navajo Reservoir, Canyada Larga, San Luis Valley, Pecos, San Augustine Plains, Gran Quivira, and Casas Grandes were correspondingly dark (Table 11, Fig. 15). Color variation was very localized, and no broad pattern emerged from this analysis that would support the current arrange- ment of subspecies. Color variation in P. apache and P. flavescens was similar and the same pigments appeared to be involved. P. fasciatiis is colored differently and one can readily distinguish sympatric specimens of P. fasciatus and P. flavescens from the Great Plains by their color differences. P. fasciatus is darker dorsally, with an “olive” tone. The yellowish color bands of the dorsal hairs are much narrower in P. fasciatus and the dark-grayish basal bands show on the surface and contribute to the darker, olive tone. P. flavescens has a more orange lateral line. In the Uintah Basin, hoi\\ P. fasciatus andP. apache have 1978 WILLIAMS— PEROGNATHUS SYSTEMATICS 27 Table 5. — Continued . Trait N M SE cv Range N M SE CV Range Minimum Maximum Minimum Maximum 19 . P. apache Deming Plains 20. P. flavescens copei 1 23 124.8 1.320 5.07 1 13.0 134.0 17 122.1 1.257 4.24 112.0 129.0 2 23 58.2 0.868 7.16 50.0 65.0 17 56.4 0.753 6.56 50.0 61.0 3 25 66.6 0.714 5.36 57.0 73.0 17 65.7 0.817 5.13 61.0 72.0 4 26 17.6 0.193 5.61 15.0 19.0 17 16.7 0.145 3.58 15.9 17.9 5 23 6.6 0.086 6.27 6.0 7.1 17 6.6 0.129 8.07 5.8 7.6 6 23 0.874 0.014 7.85 0.725 1.000 17 0.859 0.013 6.28 0.754 0.938 7 21 21.94 0.130 2.70 20.45 22.95 16 21.23 0.179 3.37 19.65 8 •>2 21.83 0.117 2.50 20.45 22.65 17 21.19 0.160 3.11 19.65 22.10 9 24 5.12 0.032 3.07 4.90 5.40 17 5.16 0.064 5.12 4.60 5.60 10 23 3.07 0.027 4.16 2.85 3.30 17 3.04 0.028 3.73 2.85 3.20 11 23 4.25 0.027 3.07 4.00 4.50 17 4.16 0.043 4.25 3.90 4.50 12 21 7.94 0.084 4.84 7.25 8.70 16 7.48 0.089 4.80 6.85 8.00 13 22 12.10 0.074 2.89 11.60 13.00 16 11.79 0.116 3.88 11.00 12.85 14 24 2.83 0.049 8.52 2.40 3.25 17 3.06 0.077 10.37 2.35 3.40 15 24 4.06 0.069 8.27 3.15 4.65 17 4.44 0.084 7.82 3.75 5.00 16 24 7.81 0.060 3.75 7.25 8.35 16 7.70 0.094 4.89 6.85 8.15 17 23 2.24 0.020 4.32 2.00 2.35 16 2.16 0.048 8.87 1.70 2.40 18 23 3.63 0.045 6.53 3.35 4.25 17 3.70 0.038 4.29 3.40 3.95 19 23 3.95 0.078 9.52 2.85 4.85 17 4.31 0.074 7.08 3.80 4.85 20 17 2.67 0.023 3.63 2.45 2.80 15 2.62 0.028 4.13 2.50 2.95 21 17 0.56 0.009 6.75 0.52 0.65 15 0.57 0.009 6.30 0.48 0.61 TO 17 0.59 0.008 5.66 0.52 0.65 15 0.59 0.007 4.91 0.52 0.65 23 17 0.58 0.008 6.22 0.48 0.65 15 0.57 0.01 1 7.68 0.48 0.65 24 17 0.70 0.008 4.65 0.65 0.77 15 0.67 0.012 7.08 0.61 0.74 25 17 2.66 0.033 5.16 2.40 2.87 15 2.77 0.048 6.68 2.40 3.07 26 17 0.93 0.010 4.35 0.87 1.00 15 0.93 0.009 3.88 0.84 0.97 27 17 1.07 0.009 3.52 1.00 1.13 15 1.04 0.01 1 4.01 0.97 1.13 28 17 0.64 0.010 6.26 0.58 0.71 15 0.62 0.012 7.61 0.55 0.68 29 17 0.94 0.009 4.21 0.84 1.00 15 0.87 0.015 6.62 0.81 0.97 30 17 0.95 0.011 5.09 0.84 1.03 15 0.90 0.012 5.05 0.81 0.97 about the same degree of relative darkness, but P. apache has a pale yellowish-orange (Light Ochra- ceous-Buff) color, and P. fasciatus has a pale olive- yellow color (near Cream-Buff or Chamois). I have found these color differences to be reliable for dis- tinguishing these taxa. Multivariate Analyses The matrix of taxonomic distances is presented in Table 12. The least similar samples have the larg- est distance coefficients. These data are summa- rized in the phenogram of Fig. 16. Note four main clusters, a P. fasciatus cluster (samples 1-4), a P. apache cluster encompassing the southern and western samples (samples 7 through 13, in descend- ing order in Fig. 16), aP. apache cluster of northern samples (samples 5, 6, 10, and 12), and a P. flaves- cens cluster (samples 20 and 21). Two principal de- ficiencies in the phenogram are apparent. All sam- ples had an equal chance of being linked regardless of their geographic positions, and in clustering sam- ples, many of the intersample relationships were simply lost. The coefficient of cophenetic correla- tion (derived from a distance-cophenetic matrix comparison) is a measure of the amount of infor- mation lost in the phenogram. This value (0.77) falls near the lower end of the range reported by Sneath and Sokal (1973). The phenogram is weakest in ad- equately portraying relationships at the more dis- tant levels. Placing the distance values in their geo- graphic context (Fig. 17) corrects some of these deficiencies and alters the interpretations derived from the phenogram. The sample of P. f. copei was about equally similar to the Deming Plains sample and io P.f. flavescens. Samples of Apache and ol- ive-backed pocket mice from the Uintah Basin were 28 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 10 Table 5. — Continued . Trait N M SE cv Range Minimum Maximum 2 1 . />. f. fla vescens 1 10 121.7 1 .453 3.78 114.0 128.0 2 10 58.7 1.350 7.27 52.0 65.0 3 II 62.7 0.702 3.71 60.0 66.0 4 11 16.8 0.122 2.41 16.0 17.0 5 II 6.5 0.157 7.98 6.0 7.0 6 10 0.931 0.026 8.23 0.839 1.083 7 8 21.39 0.120 1.59 20.90 21.85 8 8 21.39 0.120 1.59 20.90 21.85 9 9 4.99 0.042 2.52 4.85 5.20 U) 10 3.02 0.038 3.99 2.85 3.25 II 9 4.21 0.034 2.41 4.10 4.40 12 9 6.98 0.095 4.07 6.45 7.35 13 9 1 1.46 0.102 2.76 10.95 11.90 14 9 3.01 0.044 4.43 2.90 3.35 15 9 5.01 0.074 4.42 4.75 5.40 !6 9 7.53 0.078 3.11 7.10 7.80 17 8 T ->2 0.028 3.60 2.10 2.30 18 9 3.72 0.034 2.78 3.60 3.90 19 8 4.92 0.117 6.71 4.15 5.20 20 5 2.64 0.024 2.07 2.60 2.70 21 6 0.52 0.013 6.08 0.48 0.58 22 6 0.60 0.014 5.67 0.55 0.65 23 6 0.57 0.013 5.51 0.55 0.61 24 6 0.67 0.018 6.61 0.61 0.74 25 6 2.74 0.020 1.82 2.67 2.80 26 6 0.92 0.023 6. II 0.87 1 .03 27 6 1.04 0.018 4.22 1.00 1.10 28 5 0.63 0.030 10.58 0.55 0.71 29 6 0.85 0.007 1.96 0.84 0.87 30 6 0.90 0.019 5.29 0.84 0.97 distant phenetically, as were samples of the olive- backed and plains pocket mice. The relatively small distances linking neighboring samples 5, 6, 10, 12, 15, and 17 reinforce the previous interpretation that these samples represent a more or less continuous population. Peripheral populations in Arizona (7 and 8), the San Luis Valley (13), and the San Au- gustine Plains (16) showed high similarity to only one or two other samples (see Table 12 for coeffi- cients of distance not depicted in Fig. 17). The Gran Quivira sample (17) was about equally distant to the Rio Grande Valley (15) and White Sands (18) sam- ples. Gene exchange between the White Sands and Gran Quivira populations is not too likely today, but this route may have only recently been blocked. Large lava flows, of fairly recent age, and upland rocky terrain constrict the Tularosa Valley north of the White Sands, and any movement along this route would be through non-sand habitats. A matrix of similarity, based upon the Q-mode correlation analysis, is given in Table 13, summa- rized in the phenogram of Fig. 18, and the data placed in a geographic context in Fig. 19. The higher the similarity values, the greater the similar- ity between samples. Note in the phenogram (Fig. 18) a P. fasciatus cluster (samples 1-4), a cluster including P. flavescens and a set of neighboring samples of Apache pocket mice (samples 9, 16, 19, 20, and 21), and aP. apache cluster. Those closely linked samples (that is, 1 and 2, 3 and 4, 5 and 6, 7 and 8, 10 and 12, and 20 and 21) are ones that are geographic neighbors and were shown to be closely similar in the other analyses. Otherwise, these two phonograms do not appear to be too similar (Figs. 16 and 18). A matrix comparison between the ma- trices of distance and similarity showed only an ap- proximate 50% (/• = —0.51) correspondence. They differed most in the linkages of the more dissimilar Table 6. — Interpopulation matrix of correlation between morphometric traits. The numbers of the traits correspond to those in Table I. Degrees of freedom = 19. 1978 WILLIAMS— PEROGNATHUS SYSTEM ATICS 29 r- Tf 00 00 r^. r- r-- r- 00 o fT-j ^ ri r- Tt 00 Tf r- O — -sC o r-4 r^, r-i I I I r- 00 r I I 1 I I 00 “'I o ri ri «/“, ri r^-, I O n O On ■ f f rj ~ — r«~i O O 04 n Tt 00 0 4 ON I I o — o4v“)Ov'^r- 9^-1 o- vO I o >/“j »Ot r^4 r<-i O', <0\ O- sC 04 — V“, rj- r^, r I 0-1 04o'^rt\00''OOONO~04 30 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 10 Table 7. — Factor scores for the first five factors extracted from a matrix of correlation. Sample Factors 1 II III IV V 1 . Perognathiis fasciatns fasciatus -0.26180 2.47176 -0.78360 1.84415 -1.12312 2. Perognathiis fasciatus olivaceogriseus -0.48409 2.26763 -0.08344 0.27492 0.59393 3. Perognathiis fasciatus litus 0.80838 1.5 1588 -0.74339 2.45487 -0.47299 4. Perognathiis fasciatus callistiis 0.62812 1.01588 -0.12778 1.29386 0.14666 5. Perognathiis apache Uintah Basin 1.59702 0.57906 2.17739 -1.32724 -2.18045 6. Perognathiis apache Moah 1.49471 -0.40347 1 .05894 1.69595 2.39596 7. Perognathiis apache Painted Desert -0.07471 -0.66931 0.14526 0.58163 -0.56386 8, Perognathiis apache Flagstaff -0.34763 -0.78385 0.17254 0.83553 -1.47170 9. Perognathiis apache Gallup 0.22867 0.16594 1.83041 -1.58717 -1.32679 10. Perognathiis apache San Juan Basin 1.67000 -1.20200 -0.83880 1.01375 2.61925 1 1 . Perognathiis apache Canyon Largo 0.18009 -0.16779 -0.23676 -1.55194 -0.52558 12. Perognathiis apache Estrella 0.84708 -0.03241 1.18789 -1.13915 -1.20545 13. Perognathiis apache San Luis Valley -0.56873 -1.38406 -1.44469 2.01622 -0.42079 14. Perognathiis apache Santa Fe 0.44337 -0.62562 -1.69960 0.36607 1.25001 15. Perognathiis apache Rio Grande Valley 0.34882 0.06618 0.81357 -0.12503 -0.64501 16. Perognathiis apache San Agustine Plains -0.64491 -0.50534 -0.73553 1.38297 0.32324 17. Perognathiis apache Gran Quivira -0.44968 -0.03960 -1.05570 -1.90546 -1.56102 18. Perognathiis apache White Sands -0.56426 -1.18087 0.05141 -0.42458 -0.13820 19. Perognathiis apache Deming Plains 1.15288 0.61484 0.05253 1.77366 1 .43557 20. Perognathiis flavescens copei -1.94565 -0.30397 1.28590 0.31711 -0.02803 21. Perognathiis flavescens flavescens -2.25961 0.70574 1.81326 0.86707 2.04352 Cumulative % of total variance 56.1 71.7 77.4 81.6 85.1 samples. The shortcomings noted for the distance phenogram are also apparent in the similarity phe- nogram. Namely, much information was lost (the coefficient of cophenetic correlation is 0.73), and the geographic relationships were obscured. The same overall pattern of intersample relationships are apparent in the similarity map of Fig. 19 as was shown in the distance map (Fig. 17). Samples 5, 6, 10, 12, 15, and 17 form a geographic chain, with sample 15 being the weakest link between the more northern and more southern populations. Peripheral populations exhibited much the same relationships to other samples as were shown on the distance map. Some differences stand out, such as the Can- yon Largo sample (11), which was most similar to the San Juan Basin sample (10), and was particu- larly dissimilar to the Santa Fe sample (14). The Painted Desert sample (7) showed little similarity to the San Augustine Plains sample ( 16), and was more similar to the San Juan Basin sample in this analysis than in previous ones. The Deming Plains sample ( 19) was most similar to P. f. copei (20), but copei was much closer io P . f. flavescens (21) than to the Deming Plains sample. The differences in computational procedures ac- count for much of the disparity between these dis- tance and similarity analyses. Taxonomic distance is a measure of the Euclidean distance separating the samples arrayed, in this case, in 29 dimensional hyperspace. The closer two samples are, the more similar they will be in both size and proportions of all 29 characters. Q-mode correlation analysis, on the other hand, measures the correspondence be- tween the columns (samples) in a matrix of 29 rows. If two samples differ in size, but have the same body proportions, they would have a similarity val- ue of 1.0. Samples could be significantly different in the size of all traits and have relatively large dis- tance values, yet exhibit similar proportions and have high similarity values. Conversely, samples with no significant differences in size (for example, samples 1 1 and 14), but which differ proportionately (Q - r = 0.07), appear to be quite close pheneti- cally (d = 0.59). Samples that are shown to be quite similar by both procedures are probably closely re- lated. A principal components analysis of the matrix of correlation yielded results very similar to the other factor analysis, and reinforced the conclusions of the other multivariate analyses. The first five prin- cipal components accounted for 89.6% of the total variance. The scores for the first three principal 1978 WILLIAMS— PEROGNATHUS SYSTEMATICS 31 I Fig. 8. — Three-dimensional plot of first three factors, extracted from the matrix of correlation, for samples of the Perognathus fasciatiis species group. The sample codes are defined in the text and Table 7, and are shown in Fig. 4. components for each of the samples are plotted in Fig. 20. Component I accounted for 58.3% of the total variance, and is the size component. Only width of interparietals and least interbullar distance were loaded negatively on component I. Compo- nent II, accounting for 16.8% of the variance, ex- hibited highest negative loading on traits expressing the constriction of the postcranial region (IPW, LID) and the width of P4. Traits with high positive loading were length of hind foot, skull length (ONL) and length of M,. Component III was most highly loaded by traits measuring external dimensions (TOTL, TL, HBL with positive coefficients), and bullar inflation (BW with a negative coefficient). Note, by comparing Figs. 8 and 20, that the prin- cipal components and factor analyses yielded similar results, except that the images are rotated on the horizontal axis. There are other minor differences 32 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 10 Table 8. — Summary of SS-STP analysis between geographically adjacent samples. Refer to Table 7, the text, or Figs. 4 or 9 for an explanation of sample codes. + = significant difference: — = nonsignificant difference. Traits Samples I 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 Total 1-2 ______________________________ 0 1- 21 + - + - + + - + + + + - + 14 2- 3 + + + + _ + _ + + g 2-4 2 2-20 + + + _ + 9 2- 21 +- + - + -- -- -- -- -- + -- _- + + ___4._ + __ g 3- 4 ______________________________ 0 3- 5 + 7 4 - 5 — — — — — — T + — — — — — + — — — + + + — — — — — — + — + + 9 5- 6 1 6- 7 _ g 6-10 - 2 6-11 + + + + + + 9 7- g 1 7-9 1 7- 10 - + + - + - + - + - + + - + + + + + + 14 8- 9 1 8-16 2 9- 10 _ + _ + - + + + 5 9^12 _ + 2 9u|5 1 9- 16 _ 4 10- 11 + + _ + 9 10-12 1 11-12 _ 2 11- 14 ______________________________ 0 12- 14 4 12-15 _ + _ 3 13_14 4 I4_I5 ______________________________ 0 14_I7 ______________________________ 0 14- 20 + + + + -- + + - + + -- - + -- -- + - + - + - + + - + + 16 15- 16 - I 15-17 2 15-18 _ 6 15- 19 + + - + + + - + - + _ + 9 16- 19 - 0 17- 18 I 17-19 2 17- 20 + - 7 |g_19 1 18— 20 — + — + — + + + — — — + T — "f — — — + + — — + — — — — — "t~ — 12 19^20 - 2 20-21 - 1 (the factor analysis was based upon the unstan- dardized data of a 30 characters matrix, whereas the principal components analysis utilized standard- ized data in a 29 characters matrix), but the differ- ences seem relatively trivial. Note in Fig. 20 that the Apache and plains pocket mice samples differed mostly in size (Component I). The olive-backed pocket mice differed from the others in having small bullae, short skulls, unconstricted crania, small feet, large lower premolars, and short lower first molars. The San Luis Valley sample was the most distinctive of the Apache pocket mice, having short tails and wide crania (Component III), but paralleled the White Sands sample in Components I and II. 1978 WILLIAMS— PEROGNATHUS SYSTEMATICS 33 105 Fig. 9. — Map depicting the number of significant differences between adjacent samples of the Perognathus fasciatus species group, based upon the SS-STP analysis. The lines extending northward from samples 2 and 21 represent intersample comparisons with sample 1 {P. f. fasciatus from parts of North and South Dakota). The geographic positions of the samples are only approximations. Solid lines represent most likely routes of gene exchanges, and broken lines represent unlikely routes of gene exchange, based upon the number of significant differences between samples. The lines in Fig. 20 connect samples sharing the same centroid. Seven centroids extracted from the distance matrix and the distances of each sample to its closest centroid are shown in Fig. 21. This sum- mary technique resulted in a much more satisfac- tory geographic grouping of samples than was the case with the phenograms. Only one sample (13) was erroneously placed with a nonneighhoring sam- ple. The San Luis Valley sample was closest to the Painted Desert sample, as shown in the centroid 34 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 10 Table 9. — Coefficients of correlation between morphometric traits, color indices, and climatic and geographic variables for samples of P. apache. DI = darkness; R! = richness: RDI = relative darkness; AEL = elevation adjusted for latitude; CS = climatic severity index; EL = elevation; GS = growing season: Eat = latitude; MJT = mean July minimum temperature: MR = mean annual precipitation: MT = mean annual temperature. Refer to Table I for a list of traits. The degrees of freedom for the morphometric traits and the color indices are 14 and 3J, respectively. Trait CS LAT EL GS MT MP MJT AEL 1 .40 .84** .18 -.34 -.54* -.31 -.42 -.01 2 .34 .65** .21 -.25 -.39 -.29 -.26 -.13 3 .36 .68** .27 -.33 -.53* -.11 -.44 .16 4 .35 .39 .17 -.29 -.36 -.29 -.27 .01 6 .19 .60** -.04 -.12 -.25 -.40 -.14 -.20 7 .19 .75** -.12 -.19 -.31 -.40 -.28 .18 8 .21 -.10 -.21 -.34 -.39 -.30 .14 9 .26 .60** -.19 -.21 -.34 J o -.26 -.47 10 .29 .61** .06 -.33 -.30 -.15 -.23 .38 II -.07 .78** -.23 .08 -.17 -.27 -.11 .39 12 -.08 .58* -.37 .07 -.04 -.58* -.07 .20 13 .07 .66** -.23 -.09 -.19 -.50* -.20 .26 14 -.26 .33 -.58* .14 .15 -.26 .13 .33 15 -.31 .17 -.47 .22 .19 -.09 .20 .29 16 .09 .69** -.25 -.15 -.22 -.26 -.18 .36 17 .21 .29 .04 -.18 -.16 -.13 -.07 .04 18 .15 .28 -.12 -.28 -.19 -.37 -.18 -.02 19 -.04 -.12 .16 .03 -.06 .29 .04 -.01 20 .21 yY** -.08 -.22 -.27 -.32 _ 22 .27 21 .16 yy** -.02 -.18 -.34 -.10 -.24 .39 22 .10 .67** .00 -.14 -.25 -.07 -.17 .48 23 .23 .69** -.06 -.26 -.29 -.20 -.30 .12 24 .18 .63** -.04 -.21 _ 22 -.16 -.18 .30 25 .25 .63** -.08 -.22 -.23 -.17 -.17 .13 26 -.05 .56* -.25 -.02 -.01 -.07 .01 .67** 27 -.03 .69** -.26 -.02 -.12 -.17 -.08 .54* 28 .12 yy** -.17 -.14 -.25 -.14 -.23 .39 29 .21 .60** .05 -.15 -.27 -.13 -.20 .08 30 .11 .67** -.08 -.12 -.20 -.21 -.16 .37 Dl — .13 .28 — -.36* .50** — .31 RI — -.14 .45** — -.23 .64** — .30 RDI — -.04 .43** — -.31 .66** — .34 * P « 0.05; P « 0.01. analysis, but was next most distant to the Santa Fe sample (Table 12, or Fig. 17), with which it has its only geographic affinities. Discriminant function and canonical analysis pro- vided a technique whereby individuals of each sam- ple could be tested for phenetic fidelity to their pop- ulations, and the distances between individuals could be measured. As only individuals with com- plete data could be utilized, length of ear was ex- cluded, and some of the smaller samples were either submitted for classification only or were combined with other samples. Table 14 is a classification ma- trix, based upon the squared Mahalanobius distance of individuals from the nearest group means on the discriminant functions. The samples listed in the columns of Table 14 were those utilized in comput- ing the discriminant functions and F statistics. Sam- ples in the rows without a corresponding column were submitted for classification only. Note that only one P. fasciatus was misclassified as a P. fla- vescens, and that only three Apache pocket mice were closest to P. fasciatus . Three Apache pocket mice were placed with the plains pocket mice, and two plains pocket mice were classified as Apache pocket mice. Overall, the individuals showed rela- tively high group fidelity, and misclassifications were most between neighboring groups. Samples showing relatively little misclassification were the San Luis Valley, the Uintah Basin, Moab, and the P. fasciatus samples. 1978 WILLIAMS— PEROGNATH US SYSTEMATICS 35 Table 10. — Coefficients of correlation between climatic and geographic variables. The abbreviations are defined in Table 9. Degrees of freedom = 14. CS LAT EL GS MT MP MJT AEL cs LAT .33 — EL .75** .01 — GS -.96** -.28 -.65** — MT -.89** -.52* _ -74** .83** — MP .21 -.23 .46 -.24 -.22 — MJT -.90** -.42 _ *72** .87** .96** -.26 — AEL -.25 -.07 -.10 .14 .23 .35 .17 — * P ^ 0.05; ** P =£ 0.01. Table 15 lists the group means of the first five samples of both taxa. Note that the uppermost sam- canonical variables and their cumulative propor- pie of P. apache (5) is from the Uintah Basin, and tions of the total dispersion. Twelve canonical vari- that the spatial relationships of the P. apache sam- ables accounted for 100% of the dispersion, al- pies in Fig. 22 are nearly the same as their geo- though variables beyond the third individually accounted for relatively little. The sample means of the first two canonical variables are plotted in Fig. 22. The encircled areas correspond to the distribu- tion of the 457 individual cases. There was consid- erable overlap within the samples of P. fasciatus, P. apache, ^.ndP.flavescens, and between samples oiP. apache 2tndP. flavescens, but essentially none between P. fasciatus and the other taxa. A single specimen ofP. apache from the Moab sample over- lapped the position of the P. fasciatus sample (Fig. 22). The overlap between P. apache and P. flaves- cens included numerous specimens from several graphic relationships (see, for example. Fig. 21). The P. fasciatus samples, starting on the right side of Fig. 22, are distributed from northeast to south- west. Starting on the lower left side of the P. fla- vescens samples, individuals are distributed from southwest to northeast. With only a little distortion. Fig. 22 could be placed over a map, and the geo- graphic positions of the samples would nearly cor- respond to their positions on canonical variates I and II. The space in the middle of the samples would fit over the Rocky Mountains in Colorado and New Mexico. CONCLUSIONS In summarizing the patterns of structural varia- tion, it must be emphasized that the latitudinal size dine in the Apache pocket mouse populations is the dominant trend. Apache pocket mice are larger in colder and more northern areas. With increasing size the posterior cranial region becomes progres- sively more constricted and the length of the tail increases at a rate faster than the length of the head and body. Imposed upon the general size dine are significant relationships between relative amounts of moisture and bullar and rostral sizes. Generally, these skull parts increase with increasing body size, but the rates of increase are apparently also influ- enced by the amount of available moisture. Small mice from drier areas have as large or larger bullae than larger mice from moister sites. Rostral size increases at a slower rate with increasing moisture. Color is strongly associated with a combination of climatic and local edaphic soil factors that together determine relative substrate darkness. In general, mice from higher, moister elevations or latitudes are darker than those from lower, drier dimes. This relationship is modified by unusually light or dark colored sands. Apache pocket mice are primarily limited to loose sands and this habitat type is relatively uncommon and discontinuously distributed. Therefore, most populations are probably small and well isolated from their neighbors. Under such circumstances, populations can quickly evolve according to the dic- tates of local selective forces. This undoubtedly ex- plains the great amount of local variation in color, and the relatively great degree of morphometric variation over short distances. Several peripheral populations have apparently evolved along parallel paths, due to similar selective forces. Most of these 36 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 10 6 — 1 I 2 L_ t: I 7 I 9 20 22 24 26 Fig. 10. — North-south variation in occipitonasal length for ad- jacent samples of Perognathus apache. Samples are arranged from north to south. Vertically arrayed numbers are sample codes (see Fig. 9). The scale is in mm. Horizontal lines depict the sample ranges; the vertical lines mark the sample means; the outer rectangles encompass ±1 SD from the mean; the inner rectangles encompass ±2 SE from the mean; and the vertical lines on the right side of the diagram connect nonsignificant sub- sets, based upon the SS-STP analysis. populations live in higher, moister areas, although there are some parallel developments in populations from drier sites too. Plains pocket mice, in contrast, are less variable over much greater distances. The same trends in variation are apparent, but are not nearly so dra- matic. Plains pocket mice from the drier southern and western portions of their geographic range have larger bullae and slightly more constricted crania and larger rostra than the more mesic-adapted northern and eastern populations. The relatively uniform topography and the gradually changing cli- matic patterns of the Great Plains have resulted in selection for a more uniform and gradually varying population. Populations of the olive-backed pocket mouse exhibit a similar pattern of geographic variation in relationship to changing climatic patterns. Some structural convergence with P. apache can be seen in populations that approach the range of the Apache pocket mouse (see Fig. 20). A size dine in the P . fasciatus samples runs from northeast (small mice) to southwest (larger mice). This does not defy Table 11. — Mean values for color indices of samples of Perognathus apache and P. flavescens. Sample numbers are as defined In the text and shown in Fig. 4. Within samples, localities are arranged from north to south, as shown in Fig. 15. DI = darkness; RI = richness: RDI = relative darkness ( darkness + richness). Locality Sample DI Rl RDI Uintah Basin, UT 5 3.9 2.1 6.0 Fruita-Rifle, CO 6 4.7 2.2 6.8 Green River (city), UT 6 3.8 1.2 5.0 Dewey-Castle Valley, UT 6 4.0 4.7 8.7 Moab, UT 6 4.4 4.5 8.9 Coventry, CO - 4.0 5.0 9.0 Navajo Mtn, UT-Page, AZ - 3.8 4.3 8.2 Tuba City, AZ 7 3.2 3.4 6.6 Oraibi, AZ 7 2.5 2.6 5.1 Ream’s Canyon, AZ 7 2.9 3.4 6.3 Zuni Well, AZ 7 4.0 4.0 8.0 Holbrook-Winslow, AZ 7 2.4 2.8 5.2 Flagstaff-Winona, AZ 8 4.7 5.0 9.7 Gallup, NM 9 3.7 4.0 7.7 El Morro, NM 9 3.0 5.0 8.0 Chaco Wash, NM 10 2.5 2.3 4.8 Navajo Reservoir, CO, NM 11 4.7 5.0 9.7 Canyon Largo, NM 11 3.0 2.5 5.5 Canyada Larga, NM II 4.3 4.3 8.6 Estrella, NM 12 3.2 3.0 6.2 San Luis Valley, CO 13 4.7 3.8 8.5 Espanola, NM 14 3.4 3.0 6.5 Santa Fe, NM 14 4.1 3.9 8.0 Pecos, NM 14 4.7 4.1 8.8 Albuquerque, NM 15 3.0 2.5 5.5 Socorro, NM 15 3.1 3.3 6.5 San Augustine Plains, NM 16 4.8 4.0 8.8 Gran Quivira, NM 17 4.3 4.2 8.5 White Sands, NM 18 2.6 0.2 2.8 Engle, NM 19 4.3 3.6 7.9 Las Cruces, NM 19 4.0 3.4 7.4 El Paso, TX 19 4.0 4.0 8.0 Samalayucca, CH 19 2.7 2.3 5.0 Casas Grandes, CH 19 4.1 4.5 8.6 Willcox, AZ — 3.0 3.2 6.2 Clayton, NM 20 2.0 3.0 5.0 Logan, NM 20 3.7 3.8 7.5 Tolar, NM 20 3.9 4.6 8.5 Caprock (Mescalero Sands), NM 20 4.0 4.5 8.5 Jal-Carlsbad, NM 20 3.3 3.2 6.5 Mentone, TX 20 3.0 4.0 7.0 Bergman’s principle, as elevational increases and corresponding temperature decreases occur from northeast to southwest. Along the same transect, aridity increases from northeast to southwest, and there is a corresponding increase in the relative bul- lar and rostral sizes, and color becomes progres- sively lighter. 1978 mLLlAMS—PEROGNATHUS SYSTEMATICS 37 Fig. 11. — Two dimensional plot, depicting geographic variation in total length (in mm) of samples of Perognathus apache, P < 0.001. Speciation Only two major taxonomic units are apparent from the data presented in this study. One unit, P. fasciatus, is the more northern and shows adapta- tions for life in a cooler, moister environment. It is relatively dark colored and large bodied, with a short tail, small bullae, narrow interorbital region, and an unconstricted cranium. The more southern unit, represented by Apache and plains pocket mice, exhibits adaptations for life in a warmer and drier climate. The Great Plains populations are lighter colored and smaller bodied, with relatively larger bullae than P. fasciatus. The intermountain plateau populations are much more variable, as should be expected from the great amount of to- pographic and climatic variability within their range. The variability of the mice follows a pre- dictable course, with temperature and moisture fac- tors, and, perhaps, competition from congeners de- termining to a great degree the size and shape of individuals in each population. I envision that these two units arose from a par- ent population which was widely distributed in the Great Plains and intermountain plateaus. Adapta- tion to local environmental conditions was probably similar to that exhibited by extant populations. The events leading to speciation may have been initiated as late as the last interglacial period (Sangamon), although speciation at an earlier time seems equally feasible. At the close of of the interglacial period. 38 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 10 BULLAR LENGTH Fig. 12. — Two-dimensional plot, depicting relationship between bullar size (in mm) and mean annual precipitation (in inches), of samples of Perognathus apache. P < 0.05. the southward advance of glacial conditions, in- cluding the increasingly cooler and wetter climate in the Southwest, caused a southward contraction of the range of the ancestral form. At the height of the glacial pluvial period the population was frag- mented into a Chihuahuan Plateau unit and a South- ern Plains unit. The most likely barrier separating these populations was the mountains and highlands that transect the Trans-Pecos area of Texas and eastern Chihuahua, and which are continuous with the mountain axis extending southward from the main Rocky Mountain mass. According to Wells (1970fl), much of the southern Great Plains region was an open yellow pine-sage- brush parkland during the Wisconsin pluvial period, which is the same type of habitat occupied by some P. fasciatus populations today. Northern sage- brush-grassland species, such as Lagurus curtains, have been found in late Pleistocene cave deposits of southeastern New Mexico (Harris, 1970). This supports the hypothesis that a northern plains grassland fauna occupied this region during the Wisconsin pluvial maximum. West of the highland barrier, the other population was isolated in more arid, grassland or desert conditions. Wells (1970&) stated that he found no indication that treeless grassland shifted southward into the now arid Chi- huahuan Desert during the Wisconsin glacial. Much of the slightly higher plateau regions in this area were vegetated with semiarid grasslands and pin- yon-juniper woodlands, habitats that support the denser populations of Apache pocket mice today. In isolation, these populations, which were al- ready adapted to different environments diverged even more and speciation occurred. The Great Plains isolate was adapted to conditions essentially 1978 WILLIAMS— PEROGNATHUS SYSTEMATICS 39 t/5 C dj a C/D (U > a q= CL x: o a. 03 CL CL '5' ' [jj ^ 5 H O', o o d O o ON o 00 d d ON o ON 04 o o r^, ON r^, d d — 00 o 0“| r^, 04 o NO •r*, ON o ON NO d O — r- o NO — O o I--- o r- ON •'3' 0'' o o d — “ On — o nO •o O', 04 o V'*, I"- NO — o o d O — " O '3' lo 00 O', •O’ On O', "" o ON nO o 0-, 00 — d d d — — — 04 O r^, or 'O o 0-, r^, 04 04 04 “ w 00 nO ON 04 r- ON d d d o d — — — NO 00 — O *y~, _ o r^i o I— (ON 04 ro O', oo NO o ON — o O ON 04 NO 00 d d — — — o — — ™ 00 or, o- 0" lO, r^- r 1 o •O' O r^, r^i r^. O o nO “ o r^'-, 00 J-*- '3' — r^, 00 rr •3- o — o d — “ “ — 04 OI On ■O' — ir, NO ON or 00 rr-, 3' — o •O' 00 ON ON r^< OO 04 o 00 — »0', nC 00 0- O OI o- O d O — d O O O — — — 04 o o ON o- NO 04 NO o ir, 00 On 04 o o 04 ON o 04 1'- 04 O- o 04 »/') 00 OI o 00 r- 04 r*' 00 o — d — — — — — -- 04 04 04 o lO, 04 r- ON (On nO 04 o- or '3' ON r^i ON ON tr, Of) On — '3' r^i o — nD 00 r^, nC o 00 o r*', 00 — d — d d — d d O — — — 04 o 04 0- 0- NO o- r^i _ ON f'- d NO Nc;> 04 NO r-- OO K NO r<*i or r^, or 00 o 00 r^i r- ON ON NO nO 00 00 00 O ir. 00 d d — d O d O d O O d — — — 3 r^. 00 — 00 04 _ r^. On 04 O o- NO OO; 04 r^, 04 00 »/■•, o 04 — ON ir-, r- O •O' I--- rr. NO ON ON NO NO r- NO 00 O 00 o o d — d O o d o O o d — — — o 00 o ON r^, NO •O' NO ON r- o '•O, NO x> o NO ON •rf 00 04 o- On •r, o r 1 <'^1 o ‘/“j — o- — ON r- •O' nO O tr. r- o — — — d — d — — d — — — 04 04 04 o 00 O'-! V“, O'*, 'O' NO O m o ‘O, NO O OI o 1-^ 04 oo OI nO 00 nO On NO 04 o NO nO 04 •ON •O' — — '(3' 04 ON NO ON r<‘i OO o o d -- — — d — — 04 — — — — — 04 04 rN-, ri r^, •O' o r- o- o OI On _ or nC 00 NO 00 ON 1-^ r^i 04 0-, lo*, 04 nO nO ir, r^i Tt w — ON o ON O' ON — ON ON — o •o On O — " d — d o — — d O — — — — “ 04 r^- nO ON OO r^j nO ON oc r- _ o- 00 or, ON ON o ON 00 — ri NO 'O' o o I"-- O'*, r<*. '3' On f^j o NO ri — 04 r^i 04 04 — o O'- — o V“, r^i 0- ON rr •O' o d 04 04 o V“. NO ON 'O' r- 0- o •'1' 00 •o On o ri ri 0- o NT nO r^, 04 1--' Co 04 n o rn O ON r- 04 ON — NO nC O 04 o ON •o r^, nC o O — NO •r^ ON ON ON 04 nO o •O' o- 0-- 0- 00 •o ir, ON 00 o r- r^i ON nO nO nO 1-- '0' »/“i rf', o o , 00 d 00 nO •O' 04 00 • I--- 04 04 • nC V“, 00 ON JU CL E — r 1 nO 00 ON _ 04 r^. •'(3' nO o- 00 On 04 04 C/D 40 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 10 Fig. 13. — Two-dimensional plot, showing relationship between occipitonasal length and length of tail (both in mm), of samples of Perognathus apache. P < 0.01 . the same as P. fasciatus lives under today. The Chihuahuan Desert isolate was adapted to the more arid conditions extant in much of the intermountain region. With the wane of pluvial conditions and the disappearance of woodlands from the central grass- lands, both populations began to expand north- ward. The Rocky Mountain axis of New Mexico and Colorado remained a barrier to pocket mice for some time, and probably prevented contact be- tween the two northward expanding populations. Part of the Chihuahuan isolate’s population spread northward as conditions became suitable, but part also remained in place and became progressively more adapted to warmer and more xeric conditions. In time, the Trans-Pecos barrier fell and the Chi- huahuan isolate spread onto the southern Great Plains. There, it probably contacted relictual pop- ulations of P. fasciatus. Perhaps competition with the more xeric adapted P. y/r/ve.vce//5, coupled with increasingly xeric conditions in the southern Great Plains, hastened the retreat of P. fasciatus to the north. Some populations, instead, retreated to higher elevations along the southern Rocky Moun- tain front of Colorado, such as near La Veta, Silver Cliff, and Colorado Springs, where relict popula- tions are found today. Perhaps others will be found along the Sangre de Cristo range in northern New Mexico. Apache pocket mice pushed northward through the intermountain basins, finally arriving in the Uintah Basin. Olive-backed pocket mice moved into the Uintah Basin from the northeast as forests retreated and conditions became suitable. Primary contact by individuals of these populations is though to be taking place now. Taxonomic Conclusions All evidence points to the specific status of P. fasciatus. Its different karyotype (which can be used to document interspecific hybrids), different color, and high number of morphometric differ- ences with adjacent populations of Apache and plains pocket mice suggest reproductive isolation. If, as I believe, Apache and olive-backed pocket mice are now making contact for the first time, it is possible that hybridization may occur. However, it seems unlikely that introgression of genes be- 1978 WILLIAMS— PEROGNATH US SYSTEMATICS 41 RELATIVE DARKNESS INDEX Fig. 14. — Two-dimensional plot, showing relationship between relative darkness index and mean annual precipitation (in inches), of samples of Perognathiis apache. P < 0.001. tween their populations would occur. This situation warrants further monitoring. Sympatry, without ap- parent hybridization between olive-backed and plains pocket mice, is widespread in the Great Plains, and confirms their specific integrity in that area. The evidence suggests that the Apache and plains pocket mice are conspecific. Common color pat- terns and the close phenetic similarity (including chromosome structure) between adjacent samples point to a close relationship. Greater differences are found between some Apache pocket mouse samples than between P. f. copei and the Deming Plains sample of P. apache. The approximate 200-km hia- tus between nearest collecting localities of the two populations may seem large, but is no greater than some others. The Trans-Pecos gap may be nar- rowed considerably by additional field work. Both taxa are rarely collected in the southern parts of their ranges, and in my experience, repeated trap- ping is often necessary to collect one or a few spec- imens. 42 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 10 104 Fig. 15. — Geographic variation in the relative darkness index for samples of Perognathus apache and P. flavescens. The degree of darkening of the circles represents the relative darkness of the samples. Refer to Table 1 1 for identification of the samples. Perognathus flavescens and P. apache were named by Merriam (1889), who published their de- scriptions simultaneously in the same paper. As the first reviser, I have chosen Perognathus flavescens as the species name because it has precedence of position (p. 11 versus p. 14, Merriam, 1889), and, more importantly, because the epithet flavescens will best ensure stability and universality of nomen- clature. In this regard, apache appears to be a mis- spelling of apeches (Greek: discordant, noisy, quar- relsome), probably originating through the French variant, apache (a gangster or thug of Paris). Fitting the named subspecies of Apache pocket mice into the observed pattern of geographic vari- ation is not too difficult. But, from the original sub- species descriptions, it is clear that most were based primarily on color differences. Paradoxically, the holotypes of cleomophila, caryi, relictus, and melanotis are very similar in color, all being darker and richer than the holotype of apache. To recog- 1978 WILLIAMS— PEROGNATH US SYSTEMATICS 43 o > cd cn cu 5 <6 o a o a o 0^ cd cy c ^ Oij ^ o ^ 'S o -O cd h- O o r^i o o r“. o o ri n o r- d o r^, r- Os o NO o — d d o o o ON 00 o o rr o — — O d d 1 d 1 o ON 00 r^, r-. o m Tj- r- 00 o rn ri o — ■ “ d d j d 1 d 1 O ON Os to. ON o \D o NO rn O to. — . o — — ri — — — d d d d d 1 o NO o ri ri o ri 00 ON ON o n o q ri •o. "■ d d d d d 1 d 1 O Tj- o to. 00 to. r- O — « NO r- O ri q o — o ri r<-. ri — d d 1 d d d I d 1 d 1 O lOi 00 r^t 00 ON m O r-- m 00 -=i q — — O O — o ”■ d d d [ d d d 1 d 1 d 1 o _ Os 00 o rn n o rn oo m to. NO 00 ON ON •— o n ri o ri ri q d d d d 1 d d d 1 d 1 d 1 o r- ON to. r{ 00 r- n to, — o ON ON NO to. to. NO n 00 o ro o o — — q r J to. d d 1 d d 'Y Y” d 1 d 1 d d 1 o 00 to, o nO r- r\ NO o o o r-- 00 ON 00 o NO — o •Oi — — to. ri — r^, d d d d d 1 d d Y" d 1 d 1 o rr, o fO 00 00 r-- 00 0\ ri r\ ri o »o, to, to, to. NO o ri r- ON r^. — ON o r^i r<-, q r<", O r 1 r J O q r^, O r<“, d d d 1 d 1 d d d d 1 d d d 1 d 1 o ON o 00 to, o NO r-- r- nO to. ri o r J o ON ON r 1 r- r 1 r-- •— 00 o o to — Tf o rfs ri — rj — — d d d d d d d ■y" d 1 d d Y^ Y" o NO ON o r 1 to, ON NO o rn NO o o ON nO NO o ON NO — — o r^, ri o o ro — to, ri o — — r 1 r 1 — r<-. r*. d d d d d d d 1 d d 1 d 1 d d d 1 d 1 o r-- o rr\ to. to. to, r- 00 o NO NO ON NO NO r- ri to. r- nO n r- 00 nO to-. sO q r^i n n ■rf q ri — — — O o o q n r<-j "■ d d d d d d d 1 d 1 d d 1 d 1 d d 1 d I d 1 o NO r- o tO) ON to. r^, 00 oo r- o ri NO ON 00 00 00 rf ON r-- r- r- o q ri ri q — o — r^i — — q r J •*i- — d d d 1 d d d d 1 d 1 d 1 o d d d y" d 1 d 1 O ON o NO ON r 1 00 m o ON o 00 NO o O ON 00 00 r^j oo 00 oo rj 00 On Tt O o — O ri •o, O Tf r^i o r^< to. O O d d 1 d 1 d 1 d 1 d 1 d 1 d 1 d 1 d 1 'Y d d 1 d 1 d 1 y" d 1 O n 00 00 00 NO r^j •rj' •O, o 00 00 ri to. ON 00 r- O NO NO o rj NO o ri ON r^, NO ri 00 o r- o n o ri rn q o ri — — ri — to. ri ri — d d 1 d 1 d 1 d 1 d 1 d 1 d 1 d 1 d 1 I” d 1 d 1 Y" d 1 d 1 d 1 d 1 o 00 o ri ON o rj r^i ON fO, to, 00 o o o 00 o NO ON NO NO ri ro NO Tt to. 00 NO ri ri ri o Tf o Tf NO r^i nO — to. q o — Tj- r<-. — — d d d 1 d j d 1 d 1 d [ d 1 d 1 d 1 d 1 d 1 d 1 d d d [ d 1 d d O ON nD o 00 NO o 00 r- ON o NO ri o ON o NO 00 NO r- tT 00 to, NO NO nO r^t NO NO ri — * q 00 — NO — t/"l q to, q q rj o o nO ri o — — d d d d I d d d d 1 d d d d d d d d d 1 d d E _0> 1 ’ ’ ' 1 ’ ' ' ' ' ' 1 ' cd D, — n rr^ »/~i nO r- 00 ON o ri rn to, nO r- 00 ON o —<• C/5 “ “ — ri ri 44 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 10 fasciatus I olivaceogriseus 2 / i t u s 3 c a Hi St us 4 apache 7 apache 8 apache 9 apache I I apache 14 *- apache |5 apache 16 *- apache 17 apache 18 apache I 9 apache 13 — apache 5 apache 6 apache 10 apache I 2 cope i 2 0 flavescens 2 I I I _i 1 ) 2.0 1.5 1.0 0.5 0 Fig. 16. — Phenogram, based upon taxonomic distances, of samples of the Perognathus fasciatus species group. Refer to Fig. 17 or the text for an explanation of sample codes. Grouping was by the unweighted pair-group method using arithmetic averages. The coefficient of cophenetic correlation was 0.77. nize these and all of the other unique populations (each sample is unique) would require naming a number of new taxa. A more conservative approach seems to be required. Mice from near Flagstaff differ from P. flaves- cens apache from the lower, drier Painted Desert areas only in color, and even color is quite variable among samples from those areas (Fig. 15). I can see no reason to recognize more than a single subspe- cies from northeastern Arizona, and regard P. f. cleomophila as a junior synonym of P. f. apache. The name P. f. apache applies only to populations from northern Arizona (Fig. 23) and Utah south of the San Juan River. The southern samples of Apache pocket mice (Fig. 23) comprise a recogniz- able morphologic unit that is about equally similar to P. /. copei and the more northern populations of Apache pocket mice. These samples represent pop- ulations previously known asP. a. apache (samples 16, 17, and parts of 19), P. a. melanotis (part of 19), and P. a. gypsi (18). The name P. flavescens mel- anotis is the senior unoccupied name, andF. /. gyp- si is considered to be a junior synonym. The San Luis Valley sample is divergent structurally, and the name P. flavescens relictns is retained for this population. The remaining samples (Fig. 23), ex- tending southeastward from the Uintah Basin through the northwestern quarter of New Mexico, vary clinally in size, but exhibit sufficient morpho- logic and geographic continutity to make any taxo- nomic separation highly arbitrary. The name P. fla- vescens caryi applies to these populations. 1978 WILLIAMS— PEROGNATHUS SYSTEMATICS 45 Fig. 17. — Map showing taxonomic distances between adjacent samples of the Perognathus fasciatus species group. Lines extending northward from samples 2 and 21 represent intersample comparisons with sample 1. Broken lines connect samples with distance values greater than 1.0. The geographic positions of the samples are only approximations. SYSTEMATIC ACCOUNTS Perognathus flavescens apache Meriam, 1889 1889. Perognathus apache Merriam, N. Amer. Fauna, 1:14, 25 October. 1918. Perognathus apache cleomophila Goldman, Proc. Biol. Soc. Washington, 31:23, 16 May; holotype from Winona, 6,400 ft, Coconino Co., Arizona. Holotype. — Adult male (age class 5), skin and skull, BS 4253/4984, from near Ream’s Canyon, Navajo Co., Arizona; obtained on 22 May 1888 by Jere Sullivan. Skin in good condition; skull in fair condition, bullae damaged. 46 NO. 10 L. I I I -.5 0 .5 1.0 f a sc i at u s I olivaceogriseus 2 I i t u s 3 ca 1 1 i s t u s 4 apache 9 apache 19 apache I 6 cope i 20 ftavescens 2 1 apache 5 apache 6 apache II apache 7 apache 8 apache 10 apache 12 apache 13 apache 18 apache I 4 apache I 5 apache 17 Fig. 18. — Phenogram, based upon coefficients of similarity, of samples of Ihe Perognathiis fasciatus species group. Refer to Fig. 19 or the text for an explanation of sample codes. The coefficient of cophenetic correlation was 0.73. Measurements of holotype. — Total length, 140; length of tail, 68; length of hind foot, 18.5 (all ex- ternal measurements from Merriam, 1889; measure- ments not recorded on skin tag); occipitonasal length, 24.05; interorbital breadth, 5.10; alveolar length of maxillary toothrow, 3.20; width across maxillary toothrow, 4.60; bullar length, 8.45; length of interparietal, 2.50; width of interparietals, 3.75; length of nasal, 8.70; width of nasals, 2.45; width of rostrum, 4.65; least interbullar distance, 4.00; length of mandibular toothrow, 2.90. Distribution. — Sandy areas in semiarid grass- lands and pinyon-juniper woodlands in northeastern Arizona, north and east of the Mogollon rim, west of the Chuska Mountains, and east of the Coconino Plateau, northward into southeastern Utah east of the Colorado River and south of the San Juan River (Fig. 23). Diagnosis. — See Table 5, samples 7 and 8, for measurements. Size medium, feet relatively large, ears relatively small. Skull with interparietals very narrow, bullae large, nasals short, rostrum narrow, and interbullar region constricted. Color variable (Table 11, Fig. 15), from lighter and yellower than average, as near Holbrook and Oraibi, to much darker and oranger than average, as near Flagstaff. Comparisons. — Distinguishable from P. ampins and P. longimemhris by its shorter, nonpenicillate tail and slightly stiffer pelage (length of tail averages greater than 114% of length of head and body in sympatric P. ampins and P. longimemhris, and less 1978 WILLIAMS— PEROG NATH US SYSTEMATICS 47 Fig. 19. — Map showing coefficients of similarity between adjacent Perognathus fasciatus species group samples. Lines extending northward from samples 2 and 2! represent intersample comparisons with sample 1. Broken lines connect samples with similarity values of less than 0. 1 , and samples of sympatric species. The geographic positions of the samples are only approximations. than 95% in P. f. apache). Body size about 18% larger than sympatric P. flaviis, with a relatively longer tail (length of tail averages 86% of length of head and body in P. flavus) and with smaller post- auricular spots that contrast less with dorsal color. Skull about 1 1% longer than that of P. flavus, with relatively smaller bullae (length of bullae averages 40% of occipitonasal length in P. flavus, and 37% in P. f. apache), and wider interorbital region (in- terorbital breadth averages greater than 5.1 mm in P.f. apache and less than 4.5 mm in P. flavus). Size somewhat smaller than adjacent P. f. caryi popu- 48 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 10 I Fig. 20. — Three-dimensional plot of the first three principal components, derived from the matrix of similarity, for samples of the P. fasciatus species group. Lines between samples connect samples sharing the same centroid. The sample codes are defined in the text and in Table 7. lations, with shorter skull, smaller interparietal di- mensions, and smaller teeth (Tables 5 and 8). Size larger than P. f. melanotis, with relatively longer tail and shorter nasals. Remarks. — The type locality was listed as Apache Co. by Merriam (1889), but is actually in Navajo Co. The holotype averaged larger in most dimensions than typical specimens ofP. f. apache, but was within the range of measurements for the Ream’s Canyon sample. P. f. apache populations averaged largest in measurements in the north, near the San Juan River, and smallest in the south. Spec- 1978 WILLIAMS— PEROGNATHUS SYSTEMATICS 49 Fig. 21 . — Map showing the geographic positions of the samples belonging to the seven centroids extracted from the matrix of taxonomic distances. Concentric circles depict the centroid samples. Sample 13 was placed with the Arizona samples (centroid 8). imens from southern Utah and adjacent Arizona are somewhat intermediate to P. a. caryi from north- western New Mexico, and their populations are probably continuous in the Four Corners area. Specimens from near Holbrook and Adamana ap- proach P. f. caryi of the Gallup sample in size and proportions and are somewhat similar to P. f. mel- anotis from the San Augustine Plains. Cockrum ( 1 960) assigned two specimens from 3 and 2 mi W Wupatki, Coconino Co., Arizona, to P. a. cleo- mophila. These specimens represent P. ampins cineris Benson, 1933. 50 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 10 Table 14. — Matrix of classification, based upon the discriminant functions of 29 morphometric traits. Values indicate the number of individuals classified into each group. See text for further explanation. Classification groups Samples 3 10 13 15 18 19 20 1 . Perognathus fasciatus fasciatus 2. Perognathus fasciatus oiivaceogriseus 3. Perognathus fasciatus callistus and litus 5. Perognathus apache Uintah Basin 6. Perognathus apache Moab 7. Perognathus apache Painted Desert 8. Perognathus apache Flagstaff 9. Perognathus apache Gallup 10. Perognathus apache San Juan Basin 11. Perognathus apache Canyon Largo 12. Perognathus apache Estrella 13. Perognathus apache San Luis Valley 14. Perognathus apache Santa Fe 15. Perognathus apache Rio Grande Valley 17. Perognathus apache Gran Quivira 18. Perognathus apache White Sands 19. Perognathus apache Deming and San Augustine 20. Perognathus flavescens and P.f copei 15 ____________ — 40 — — — — — — — — — I — — 53 2 1— !_____ — — 117 1— !_____ — 1 — — 28 1 2 — 2 6 1 — — — — — 123 1— 2 1 2 1 — — 1 4 2 121— I — — — — 1— 4— 3 3—5—4 1 — — — I 3 2 5 — 2 — — — 1 — — — I 3 1 2 7 — I — — — — — 3 4 6 1 36 3 4 — ________ 3 1 5_ — — — — 2 1 1 — 131 I — — — — — 1 4 — — 1 1 23 I _________ 1 I 13 Records of occurrence. — Specimens examined, 187, distrib- uted as follows: Arizona. Apache Co.: Four Corners, 1 (UlMNH); Chin Lee, 5,600 ft, 3 (BS); Zuni Well, 7.5 mi N Adamana, 5,337 ft, 1 (MVZ). Coconino Co.: 2 mi S Endische Spring, Navajo Mountain, 3 (MVZ), 1 (TCWC); 5 mi S Navajo Mountain, 1 (MVZ); Page, 1 (UIMNH); 0.5 mi NW Page, 1 (UIMNFI); 0.5 mi S Page, 7 (UIMNH); Salt River Project, Na- vajo Generating Plant Site, Page, 4,520 ft, 4 (MNA); 6 mi SE Page, I (MNA); 19 mi SW Page (hwy. 189), 1 (UIMNH); 2 mi N, I mi E Bitter Springs, 3 (UIMNH); Cedar Ridge, 6,000 ft, I (MVZ); 3 mi above mouth. Cedar Ranch Wash, 3 (BS); Tuba City, Painted Desert, 1 (BS), I (MVZ); Moa Ave, I (BS), 10 (MVZ); Moenkopi Wash, 12 mi above mouth, 4,500 ft, 3 (BS); 2.5 mi S. 2 mi E Moenkopi, 5.050 ft, 6 (UIMNH); 2.5 mi SE Moenkopi, 4,900 ft, 4 (UIMNH); 5 mi S, 2 mi E Moenkopi, 5,500 ft, 4 (UIMNH); 5 mi N Cameron, I (UIMNH); 4.5 mi N Cam- eron, 1 (UIMNH); 3 mi S Visitor Center, Wupatki National Monument, 5,000 ft, 4 (MNA), 1 (UIMNH); 4 mi W Winona, I Table 15. — Mean scores for canonical variables for samples o/ Perognathus fasciatus, P. apache, and P. flavescens. Canonical variables Samples I 11 III IV V 1. Perognathus fasciatus fasciatus -1.667 -4.449 2.163 -0.400 1.074 2. Perognathus fasciatus oiivaceogriseus -1.772 -4.215 1.878 -0.244 2.057 3. Perognathus fasciatus callistus and litus -2.802 -2.722 -1.040 0.841 -0.733 5. Perognathus apache Uintah Basin -3.107 1.489 0.207 -0.852 -0.428 6. Perognathus apache Moab -1.607 1.439 0.908 0.372 0.235 7. Perognathus apache Painted Desert 0.745 0.594 -1.648 0.329 0.115 8. Perognathus apache Flagstaff 2.094 0.160 0.659 0.854 -0.190 9. Perognathus apache Gallup 0.883 1.267 0.389 0.888 0.265 10. Perognathus apache San Juan Basin 0.283 1.668 0.601 1.960 0.373 1 1 . Perognathus apache Canyon Largo 0.740 -0.147 0.361 0.708 0.357 12. Perognathus apache Estrella 0.642 0.787 -0.057 1.396 -0.283 13. Perognathus apache San Luis Valley 3.942 -1.416 0.954 -1.057 -2.710 14. Perognathus apache Santa Fe 1.411 -0.816 1.272 0.066 -0.259 15. Perognathus apache Rio Grande Valley 1.188 -0.062 0.756 -0.184 -0.139 17. Perognathus apache Gran Quivira 0.680 -1.056 -0.205 -1.020 0.493 18. Perognathus apache White Sands 1.636 0.459 -0.953 -0.724 1.049 19. Perognathus apache Deming and San Augustine 1.284 -0.516 -0.290 -1.244 0.555 20. Perognathus flavescens copei and P.f. flavescens 0.897 -1.881 0.357 -0.614 1.461 Cumulative % of total dispersion 39.75 64.21 73.49 81.83 88.82 1978 mLLlAMS—PEROGNATHUS SYSTEM ATICS 51 Fig. 22. — Two-dimensional plot of the first two canonical variables for individuals of the P. fcisciatus species group. Circles mark the positions of the sample means, whereas the lines encompass the distribution of the individual cases for each taxon. To prevent visual confusion, the positions of the individual cases are not shown. Refer to the text or Table 7 for an explanation of the sample codes. + chromosomes (MSB); 3 mi NW Winona, 6,400 ft, 27 ( BS); Winona, 6,400 ft, 7 (MVZ); Grand Falls, Little Colorado River, 2 (UCM); 1 mi SE Grand Falls, 1 (MVZ); 30 mi NE Flagstaff, 1 (BS); 9 mi E Flagstaff, I (BS); Walnut, 5 mi from Turkey Tank, 4 (BS). Navajo Co.: 1 1 mi N Kayenta, 4 (UlMNH); 4.5 mi N, 1 mi E Kayenta, 2 (UIMNFI); 1 mi E Kayenta, 5 (UlMNH); Dogoszhi Biko Canyon, mouth of Water Lily Canyon, 1 1 (MNA); Keam’s Canyon, 7 (BS), 20 (MVZ), 8 (UlMNH); Oraibi, 6,000 ft, 7 (BS), 5 (MVZ); Holbrook, 2 (BS); 0.5 mi S, 3 mi E Holbrook, 1 (UlMNH); Winslow, Painted Desert, 5,326 ft, 1 (BS), 3 (UMMZ); 2 mi N, 2 mi E Winslow, 1 (UlMNH); 2 mi E Winslow, Little Colorado River, 1 (BS); Winslow, N side river, 1 (BS). Utah. San Juan Co.: Navajo Mountain Trading Post, 5 mi SE Navajo Mountain, I (MVZ). Additional records. — Arizona. Apache Co.: Canyon de Chel- ley (Cockrum, 1960). Coconino Co.: Tappan Spring, 4,500 ft (Cockrum, 1960). Navajo Co.: Walpi (Cockrum, 1960). Perognathus flavescens caryi Goldman, 1918 1918. Perognathus apache caryi Goldman, Proc. Biol. Soc. Washington, 31:24, 16 May. Holotype. — Adult male (age class 4), skin and skull, BS 148206, from 8 mi W Rifle, Garfield Co., Colorado; obtained on 4 October 1906 by M. Cary. Both skin and skull in good condition. Measurements of holotype. — Total length, 154; length of tail, 73; length of hind foot, 21; occipito- nasal length, 25.15; interorbital breadth, 5.60; al- veolar length of maxillary toothrow, 3.50; width across maxillary toothrows, 4.65; bullar length, 9.00; width across bullae, 13.40; length of interpa- rietal, 3.35; width of interparietals, 4.15; length of nasal, 9.30; width of nasals, 2.30; width of rostrum, 4.35; least interbullar distance, 4.20; length of man- dibular toothrow, 3.15. Distribution. — Usually in sandy areas in semiarid grasslands and pinyon-juniper associations, from near Val Verde in the Rio Grande Valley, north- ward to at least the Rio Chama; and from the upper Pecos Valley and the Rio Grande Valley westward to Gallup and the Chuska Mountains, all in New Mexico; northward from the Four Corners area through western Colorado and eastern Utah into the 52 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 10 Fig. 23. — Map showing geographic range of the intermountain races of Perognathus flavescens, and portions of the ranges of two of the Great Plains races. Circles represent localities from which specimens were examined. To prevent crowding, single circles represent two or more localities that partially overlapped on the map. A = P. f. flavescens; B = P. f. copei; C = P. f. mekinotis; D = F. /. apache; E = P. f. caryi; E = P. f. relictus. Uintah Basin, at least to the Duchesne and White rivers. Not known from west of the Colorado and Green rivers south of the city of Green River (Eig. 23). Diagnosis. — See Table 5, samples 5, 6, 9, 10, 11, 12, 14, and 15, for measurements. Size of most characters averaging from medium to large, and varying clinally, being largest in the north (sample 5) and smallest in the southeast (sample 14). Skull with relatively constricted interbullar region and with narrow interparietals. Color variable, lightest and palest in the San Juan Basin and near Green 1978 mLLlAMS—PEROGNATHUS SYSTEMATICS 53 River, Utah, and darkest and richest at higher ele- vations (Table 11, Fig. 15). Comparisons. — See account of P. f. apache for remarks on distinguishing P. flavus. Size smaller and with a relatively shorter, nonpenicillate tail than P. parvus; P. parvus is generally tannish-gray or tannish colored (not yellowish-orange), and has less contrasting postauricular spots than P. flaves- cens; occipitonasal length less than 25.5 mm in P. flavescens, but greater than 26.5 mm in P. parvus. Differs from P. fasciatus caUistus in larger size, rel- atively longer tail, and yellowish-orange, rather than olive-yellow, lateral line; interparietal of P. f. caryi longer, rostrum wider and interbullar region narrower than P. f. caUistus; mandibular toothrow longer, M‘ wider, and Mi larger in P.f. caryi. Skull larger, but interorbital region narrower and inter- parietals larger than P. f. relictus; premolars and molars larger than P.f. relictus. Size larger and tail relatively longer than P. f. melanotis; skull longer and interparietals shorter and narrower than P. f. melanotis. Remarks. — This is the most variable of the inter- mountain races of P. flavescens. The dominant trend is the north-south size dine. Specimens from Gallup are somewhat intermediate to P. f. apache and also show some similarity to P.f. melanotis from the San Augustine Plains. The Uintah Basin popu- lation is found on a variety of substrates, but most others appear to be limited to loose sands. Durrant ( 1952) thought that the Uintah Basin population rep- resented an undescribed subspecies. Size averages slightly larger in the Uintah Basin population, but it is, overall, similar to P. f. caryi from south of the Tavaputs Plateau, and subspecific recognition does not seem warranted. A specimen from San Antonio Mountains, N Tres Piedras, Rio Arriba Co., New Mexico, listed as this species by Findley et al. ( 1975), is P. flavus. Records of occurrence. — Specimens examined, 448, distrib- uted as follows: Colorado. Garfield Co.: 7 mi W Rifle, 2 (BS); 8 mi W Rifle, 2 (MVZ). La Plata Co.: 9 mi S Ignacio, 1 (UU). Mesa Co.: Sieber Ranch, Little Dolores Creek, 1 (UCM); 0.25 mi E Colorado National Monument, 1 (DCBML); 0.5 mi E Grand Junction Entrance Station, Colorado National Monument, 2 (UCM); Eruita, I (BS); Badger Wash, 8 (DCBML); State Line, 1 (MVZ). Montezuma Co.: Morfield Mesa, Mesa Verde National Park, 2 (KU). Montrose Co.: Coventry, 1 (BS); Bedrock, 3 (UCM). Rio Blanco Co.: 17 mi W Meeker, 1 (DCBML); 7 mi N, 19 mi E Rangely, 2 (DCBML). New Mexico. Bernalillo Co.: 2 mi N Albuquerque, 2 (MSB); West Mesa, W Albuquerque near Lava Elow, I (MSB); 2 mi N, 5.5 mi W Albuquerque, 2 (MSB); 2.5 mi N, 6 mi W Albuquerque, 2 (MSB); 5 mi W Albuquerque, 3 (MSB); W Albuquerque, E side Rio Puerco Valley, S U.S. 66, 2 (MSB); 18 mi W Albuquerque, Puerco Valley, 3 (MSB); 14 mi W Albuquerque, 24 (MSB), 2 -I- chromosomes (MSB); 16 mi W Albuquerque, 2 (UlMNH); 4.8 mi N, 14 mi W Albuquerque, I (MSB); 14.7 mi N, 3 mi E Suwanne, 2 (MSB); 2.5 mi S, 7.5 mi E Suwanne, 3 (MSB); 0.25 mi S, 10.2 mi W Isleta, 12 (MSB); 2.2 mi S, 10.5 mi W Isleta, 1 (MSB). McKinley Co.: Gallup, 3 (BS); Wingate, 4 (BS); 3 mi N, 2 mi W Estrella, 7 (MSB); 4 mi N, 2 mi W Estrella, 8 (MSB); 3 mi N Crownpoint, 1 (MSB), Rio Arriha Co.: Stinking Springs Lake (Burford Lake), 1 (BS); 10 mi W Lindrith along Canyada Larga, 4 (MSB); River mile 165, River Island, San Juan River, 1 (UU); River mile 166, San Juan River, 1 (UU); Rio Ojo Caliente, 1.5 mi E, I mi N Chili, 2 (MSB); Espanola, 9 (BS); 5 mi E Abiquiu, 1 (BS); 3.5 mi S junction U.S. 285 and N.M. 30, on 30, T20N, R8E, I (MSB). Sandoval Co.: 5 mi S, 3 mi E Domingo, 2 (MSB); San Eelipe Indian Reserva- tion, sec. 2, T13N, R5E, 1 (MSB); Jemez, 1 (BS); 0.25 mi S, 1 mi W San Ysidro, 1 (MSB); 6 mi S, 4.5 mi W San Ysidro, 9 (MSB); I mi SW Santa Ana Pueblo, 2 (MSB), 4 (CM), 12 + chromosomes (MSB); 4.5 mi N, 14 mi W Alameda, 6 (MSB). San Juan Co.: Chaco Wash. 6 mi E, 14 mi S Shiprock, 43 (MSB); Newcomb, 5 mi N, 6 mi E, I (MSB); Newcomb, 1 (MSB); Gal- lego Canyon, 7.5 mi S, 5 mi E Earmington, I (MSB); Gallego Canyon, 7.5 mi S, 4 mi E Farmington, 1 (MSB); 7 mi S, 6 mi W Bloomfield, sec. 4, T27N, R12W, 2 (MSB); 13 mi S, 11 mi E Farmington, 3 (MSB); 3 mi S, 3 mi E Farmington, 2 (MSB); 10 mi S, 7 mi E Farmington, 1 (MSB); 16 mi S, 1 mi W Farmington, sec. 17, T26N, R13W, 2 (MSB); upper Benito Canyon, I (UU); Lucero Place, sec. 17, T3IN, R7W, I (MSB); Pine River Road, sec. 9, T31N, R7W, I (MSB); Canyon Largo, sec. 22, T29N, R9W, 7 (MSB); 3 mi S, 3 mi E Blanco, I (MSB); Canyon Largo at Fresno Canyon, sec. 33, T28N. R8W, 27 (MSB); 0.5 mi ESE Four Corners boundary marker, 2 (MSB): 5.5 mi N, 1.5 mi W Waterflow, 1 (MSB); El Huerfano, 0.5 mi SE base, I (MSB); Chaco Canyon National Monument, 2 (MSB). San Miguel Co.: Pecos. 4 (BS); 3 mi S Pecos. 13 (BS). Santa Fe Co.: Rio Tesuque, sec. 14. TI8N. R9E. I (MSB); NW Santa Fe Airport, 2 mi W Sewage Disposal Plant. I (MSB); Santa Fe, I (BS); Galisteo Creek, I mi E U.S. 85, sec. 31, TI5N, R7E, I (MSB); I mi W Cerillos on Galisteo Creek, I (MSB); Galisteo Creek, 1 mi E Galisteo R.R. Station, sec. 26, T14N, R8E, 1 (MSB); Glorieta, 2 (BS); San Pedro, 1 (BS). Socorro Co.: I mi N Pope, 2 mi S Lava Mesa, I -r chromosomes (MSB); Lava Mesa, 2 mi SE San Marcial. I (MVZ); 0.5 mi S. 2 mi W Bernardo. 2 (MSB); 4 mi E Escondida, 2 (MSB); 2 mi N, 4.5 mi E Socorro, 3 (MSB); Lava Mesa, S of Clyde, I (MBZ). Valencia Co.: 2 mi E Valencia, 1 -I- chromosomes (MSB); 1.5 mi S, 5 mi W Los Lunas, I (MSB); 2 mi S, 8.5 mi W Los Lunas, 4 (MSB); 2 mi W Los Chavez, 2 (MSB); 7 mi W Belen, 1 (MSB); Zuni Mountains, 2.5 mi E El Moro, 1 (LACM). Utah. Duchesne Co.: S Myton Bench, 3 mi SE Myton, 3 (UU); Myton Bench, 5 mi SE Myton, 16 (UU). Emery Co.: 16 mi NW Green River, 1 (CM); Gunnison Valley, W side Green River, 7.6 mi N Green River (city), 4,200 ft, 2 (UU). Grand Co.: Castle Valley, 10 mi NE Moab, 5,000 ft, 5 (UU); Castle Valley, 8 mi NE Moab, I (UU); 1 mi E Green River (city), 4,080 ft, 8 (UU); 1 mi SE Dewey Bridge, S side Colorado River, 4,500 ft, 1 (UU); 3 mi SE Dewey, S side River. 4,810 ft, 1 (UU); 4 mi SE Dewey, 5,000 ft, 1 (UU); Big Flat, sec. 21, T26N, R19E, 6,000 ft, 2 (UCM). San Juan Co.: S end Gray's Pasture, sec. 32. T27S, RI9E, 5,960 ft, 7 (UCM); Willow Flat, sec. 6, T28S, R19E, 6,040 ft, 4 (UCM); sec. 5, T27S, R19E, 6,050 ft, I (UCM); NE Corner Gray’s Pasture, sec. 22, T26S, R19E, 6,000 ft, 1 (UCM); Chester Canyon at Beef Basin Rd., 54 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 10 sec. 7, T31S, R19E, 5,280 ft, 7 (UCM); W of Squaw Butte, sec. 25, T30S, R19E, 5,040 ft, 3 (UCM); S of Squaw Butte, sec. 30, T30S, R20E, 5,040 ft, 2 (UCM); SW Cave Spring, sec. 29-30, T30S, R20E, 5,000 ft, 4 (UCM); Canyon Lands National Park, sec. 15, T27S, R19E, 5,900 ft, I (MMNH); Dry Valley ( = Hatch Crossing, about 30 mi N Monticello), 1 (BS); 1 mi S Kern Spring, 5 + chromosomes (MSB); Highway 160, 25 mi N Monticello, 6,100 ft. 1 (UU); Bluff, 4,400 ft, 1 (MVZ); 1 mi N Bluff. 4.500 ft, 1 (UU); Noland's Ranch, San Juan River, 1 (BS); Johns Can- yon, 5,150 ft, 1 (UU); 119 mi N Lee’s Eerry, 3 (UU); 121 mi N Lee’s Eerry, 2 (UU); 142 mi N Lee’s Ferry, 2 (UU). Uintah Co.: Evacuation Creek, 2 mi S White River, 1 (MSB), 2 -f chromo- somes (MSB); S shore White River, 3 mi S Bonanza, 1 (MSB), 1 + chromosomes (MSB); Evacuation Wash, 4 mi NE Rainbow, 5,600 ft, 3 (UU); 2 mi NE Rainbow, 5,800 ft, 2 (UU); Brown’s Corral, 20 mi S Ouray, 6,250 ft, 4 (UU); Willow Creek, 25 mi S Ouray, 5,250 ft, 2 (UU); White River, 2 mi W upper White River Crossing, 14 mi N Dragon, 5,000 ft. 1 (UU); confluence of Green and White rivers, 1 mi S Ouray, 4.654 ft, 2 (UU); 2 mi S Ouray, 4,800 ft, 2 (UU); 1.5 mi S, 1 mi E Ouray, 7 (MSB); 2 -f chro- mosomes (MSB); 1.5 mi E Ouray, N White River, 20 (MSB), 2 -I- chromosomes (MSB); Pariette Bench, 4,700 ft, 6 mi SW Our- ay, W Green River, 1 (CM). Additional records. — Colorado. Mesa Co.: 0.25 mi W Red Canyon Overlook, Colorado National Monument, 6,400 ft; 0.25 mi SE East Entrance Ranger Station (Armstrong, 1972). New Mexico. Valencia Co.: near Laguna (Bailey, 1932). Utah. San Juan Co.: 0.5 mi N Bluff. 4,400 ft (Durrant, 1952); River View (Durrant, 1952). Perognathus flavescens relictus Goldman, 1938 1938. Perognathus apache relictus Goldman, J. Mamm., 19:495, 14 November. Holotype. — Adult male (age class 4), skin and skull, BS 150768, from Medano Ranch, 15 mi NE Mosca, Alamosa Co., Colorado; obtained on 2 No- vember 1907 by M. Cary. Both skin and skull in good condition. Measurements of holotype. — Total length, 137; length of tail, 68; length of hind foot, 19.0; occipi- tonasal length, 22.70; interorbital breadth, 5.45; al- veolar length of maxillary toothrow, 3.15; width across maxillary toothrows, 4.20; bullar length, 8.00; width across bullae, 12.40; length of interpa- rietal, 3.10; width of interparietals, 3.75; length of nasal, 7.90; width of nasals, 2.35; width of rostrum, 3.60; least interbullar distance, 4.00; length of man- dibular toothrow, 2.80. Distribution. — Sandy areas in arid grassland as- sociations in and around the Great Sand Dunes of the San Luis Valley, Colorado (Fig. 23). Diagnosis. — See Table 5, sample 13, for mea- surements. Size medium, with tail relatively longer than other populations. Skull with broadest inter- orbital region, short nasals, and shortest and nar- rowest interparietals; bullae relatively small, molari- form, teeth relatively narrow, expecially premolars and M^ Color dark and rich (Table 11, Fig. 15). Comparisons. — See account of P. f. apache for remarks on distinguishing P. flavus. .Skull smaller, interorbital region wider, and interparietals smaller than P. f. caryi. Size generally larger, interorbital region much wider, and interparietals shorter and narrower than P. f. melanotis. Remarks. — Goldman (1938) assigned all of the relatively dark-colored specimens of Apache pock- et mice from New Mexico, including those from Gran Quivira, Santa Fe, Pecos, Glorieta, and Bur- ford Lake to P. a. relictus. The Gran Quivira spec- imens are referred to P. f. melanotis, and the Pecos, Santa Fe, Glorieta, and Burford Lake specimens are assigned to P. f. caryi. Records of occurrence. — Specimens examined, 26, distributed as follows: Colorado. Alamosa Co.: 1.4 mi N. 9.6 mi E Mos- ca, 2 (MSB); 1.4 mi N, 1! mi E Mosca, 1 + chromosomes (MSB); Medano Ranch, 15 mi NE Mosca, 1 (BS), 2 (MVZ), 7 (UCM); Great Sand Dunes National Monument, 1.6 mi NE Headquarters Medano Springs Ranch, 11 (MVZ); 3 mi S Great Sand Dunes National Monument, 1 (MVZ). Additional records. — Colorado. Alamosa Co.: Great Sand Dunes National Monument (Armstrong, 1972). Perognathus flavescens melanotis Osgood, 1900 1900. Perognathus apache melanotis Osgood, N. Amer. Fauna, 18:27, 20 September. 1929. Perognathus gypsi Dice, Occas. Papers Mus. Zook, Univ. Michigan, 203: 1 , 19 June; holotype from White Sands, 12 mi SW Alamogordo, Otero Co., New Mexico. 1933. Perognathus apache gypsi . Benson. Univ. California Publ. Zook, 40:26, 13 June. Holotype. — Adult female (age class 5), skin and skull, BS 97416, from Casas Grandes, Chihuahua; obtained on 21 May 1899 by E. A. Goldman. Both skin and skull in good condition. Measurements of holotype. — Total length, 133; length of tail, 65; length of hind foot, 19.5; occipi- tonasal length, 22.20; interorbital breadth, 5.10; al- veolar length of maxillary toothrow, 2.85; width across maxillary toothrows, 4.20; bullar length, 7.60; width across bullae, 11.75; length of interpa- rietal, 2.60; width of interparietals, 3.95; length of nasal, 8.25; width of nasals, 2.25; width of rostrum, 3.65; least interbullar distance, 4.20; length of man- dibular toothrow, 2.70. Distribution. — Sandy areas in desert and grass- land associations from Gran Quivira and the San Augustine Plains, New Mexico, southward to the Samalayucca Sands and Casas Grandes, Chihua- 1978 WILLIAMS— PEROGNATHUS SYSTEMATICS 55 hua; and extending west from El Paso Co., Texas to Willcox Playa, Arizona (Fig. 23). Diagnosis. — See Table 5, samples 16, 17, 18, and 19 for measurements. Size small in most dimen- sions, with relatively short tail. Size varies altitu- dinally and latitudinally, being largest in the higher, northern populations and smallest in the Deming Plains and Jornada del Muerto populations. Skull short, but with relatively large bullae; interparietals not noticeably broadened. Color extremely variable geographically, from white with a grayish overwash (White Sands) to relatively dark and rich (Jornada del Muerto and Casas Grandes, Table 1 1 and Fig. 15). Comparisons. — More similar in size and propor- tions to P. flavns than the more northern popula- tions of P. flavescens. Interorbital breadth 4.9 mm or greater in P. f. melanotis, and 4.7 mm or less in P. flavns: posterior cranial region more constricted and bullae more inflated in P. flavns, least inter- bullar distance averages 2.90 mm in P. flavns and 3.95 mm in P. f. melanotis. See account of P. f. apache for additional remarks on distinguishing P. flavns. P. f. melanotis differs from P. f. copei in larger size; the skull of P. f. melanotis has a nar- rower interorbital region, the interparietals are shorter and narrower, the interbullar region is more constricted, and the articular process of the man- dible is shorter than in P. f. copei. Remarks. — The San Augustine Plains and Gran Quivira populations approach the Rio Grande Val- ley population of P. f. caryi in most characters. Members of the White Sands population have rel- atively large feet, inflated bullae, and wide rostra, and approach P. f. copei in having wide interorbital regions and long interparietals. The individuals from Willcox Playa were too few to adequately as- sess their morphologic features. However, other than a slight color difference (Table 12), that sample did not appear to be materially different from spec- imens from near Lordsburg and from Casas Grandes. Bailey (1932) allied all of the relatively dark-colored samples of Apache pocket mice from New Mexico, including those from Pecos, Santa Fe, Glorieta, Burford Lake, and Gran Quivira with the holotype of P. a. melanotis from Casas Grandes, but assigned specimens from Deming to P. a. apache. Goldman (1938) later assigned the dark-colored specimens from New Mexico to P. a. relict ns, restricted P. a. melanotis to the holotype, and retained the Deming specimens under P. a. apache. These latter specimens (BS collection) were dug from a burrow, and are too young to show any characteristics useful for distinguishing subspe- cies. Records of occurrence. — Specimens examined, 148, distrib- uted as follows: Arizona. Cochise Co.: 3 mi SE Willcox, 4,163 ft, 5 (MVZ). New Mexico. Catron Co.: 15 mi S, 15 mi W Mag- dalena, I (DCBML). Doha Ana Co.: 6 mi W La Mesa, I (ENMU); 7 mi N, 2 mi E Las Cruces, I (UA); 6 mi E Las Cruces, 1 (NMSU); 13 mi SW Las Cruces, I (NMSU). Hildago Co.: II mi N, 10 mi W Lordsburg, I (ENMU). Luna Co.: Dem- ing, 3 (BS). Otero Co.: White Sands, 10 mi SW Tularosa, 5 (MVZ); Quartz Sands, 14 mi SW Tularosa, 4,100 ft. 18 (MVZ); White Sands. 12 mi W Alamogordo. 8 (MVZ); White Sands. 18 mi W Alamogordo, 4 (AMNH), 5 ( MVZ): 15 mi SW Alamogordo, I (LACM); White Sands, 18 mi SW Alamogordo, 32 (MVZ); White Sands, I (LACM), 2 (UMMZ); Walker Ranch, White Sands National Monument, 1 -i- chromosomes (MSB); Interior of White Sands, 3 (UMMZ). Sierra Co.: 1 mi N, 4.5 mi E Engle, 1 (MSB); I mi S, 5.4 mi E Engle, 3 + chromosomes (MSB). Socorro Co.: Mesa Jumanes. southern portions, I (BS); Mesa Jumanes, Ruins of Gran Quivira, 1 (BS); Gran Quivira National Monument, TIS. R8E, 13 (MSB); San Augustine Plains, 12 mi E, 10 mi S Datil, I (MSB); San Augustine Plains, sec. 28-29, T2S, R7W, 13 (MSB); San Augustine Plains, 12 mi NW Monica Spring, 4 (BS); Gallina Mountains, 2 (BS). Texas. El Paso Co.: 7.5 mi E City Hall, El Paso. I (KU); 19.4 mi E El Paso, I -I- chromosomes (MSB); 2.5 mi N Ysleta, 10 (UIMNH); 3 mi E Ysleta, I (MALB). Chihuahua. 1 mi E Samalayucca, I (MVZ); 2.5 mi S, 2 mi W Samalayucca, I (KU); 10 mi SE Zaragosa, I (KU); Rio Casas Grandes, 9 mi N Nueva Casas Grandes, 1 (MSB); I mi E Rio Casas Grandes, 10 mi N Nueva Casas Grandes, 4 -(- chromosomes (MSB), Other Specimens Examined Perognathus fasciatus fasciatus. — Specimens examined, 33, distributed as follows: Montana. Roosevelt Co.: 9 mi SE Baine- ville, 4 (UMMZ). North Dakota. Billings Co.: I mi S, I mi W Medora, 2,300 ft, 10 (KU). Burleigh Co.: 9 mi E Bismark, 5 (UMMZ). Kidder Co.: 6 mi W Steele, 6 (UMMZ). Pembina Co.: Weeks Farm, sec. 36, T160N, R56W, I (MSB). Stutsman Co.: 7 mi N Jamestown, I (UMMZ); 14 mi W Jamestown, 4 (UMMZ). South Dakota. Todd Co.: 15 mi W Mission, I (MSB). Wal- worth Co.: Molstad Lake Park, I (KU); Swan Creek, 13 mi S Selby, 1,6(X) ft, 1 (KU). Nebraska. Cherry' Co.: Sparks, 1 (UMMZ); Ft. Niobrara Game Reserve, I (UNSM). Perognathus fasciatus olivaceogriseus. — Specimens exam- ined, 44, distributed as follows: Montana. Carter Co.: Ekalaka Hills, 4.5 mi S, I mi E Ekalaka, (MMNH). Nebraska. Banner Co.: 10 mi S, 2.5 mi E Gering, 3 (VMKSC). Dawes Co.: I mi SW Chadron. I (UNSM). Siou.x Co.: 5.5 mi W Crawford, 6 (UNSM); 6 mi W Crawford, 1 (UNSM); Glenn, I (UNSM); 3 mi N Glenn, 2 (UNSM); 3.5 mi N, 1 mi E Glenn, 3 (UNSM); 8 mi W Ft. Robinson, 1 (UNSM). South Dakota. Jackson Co.: 1 mi SW Kadoka, 1 (MMNH). Wyoming. Carbon Co.: 1 mi E Ft. Steele, 13 (MSB), 2 -I- chromosomes (MSB). Converse Co.: Van Tassel Creek, I (CM). Johnson Co.: 2 mi S, 6.5 mi W Buffalo, 5,620 ft, 3 (KU). Sheridan Co.: 5 mi NE Clearmont, 3,900 ft, 1 (KU). 56 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 10 Perognathus fusciatus infraluteus. — Specimens examined, 7, distributed as follows: Colorado. E! Paso Co.: Air Force Acad- emy. 10 mi N Colorado Springs, 4 (CAS); 6 mi N, 1 mi W Col- orado Springs, 2 (UIMNH). Huerfano Co.: 4 mi S LaVeta, 7,000 ft. 1 (KU), Perognathus fasciatus Ulus. — Specimens examined. 31. dis- tributed as follows: Wyoming. Natrona Co.: 5 mi W Indepen- dence Rock. 6.000 ft, 4 (KU). Sweetwater Co.: 25.4 mi N Table Rock. 25 (MSB). 2 + chromosomes (MSB). Perognathus fasciatus callistus. — Specimens examined. 30. distributed as follows: Colorado. Moffatt Co.: N Bank Yampa River. 5 mi NW Cross Mountain. I (CM). Rio Blanco Co.: 16 mi W Meeker. 3 mi up Scenery Gulch (N of White River), I (CM). Utah. Dagget Co.: 0.5 mi SW Clay Basin Camp, 6,300 ft, 2 (UU): Bridgeport, 1 (UU). Uintah Co.: West Rim, Dead Man Bench, opposite Leota Flats (W of Green River). 2 (CM); E Green River, 3 mi S Jensen, 1 (CM); 4.6 mi N Bonanza, 1 (MSB), 3 + chromosomes (MSB); Bonanza, 1 (UU); 1 mi S, 1.5 mi E Bonanza, 1 -I- chromosomes (MSB); 13.4 mi E Ouray, 1 -I- chromosomes (MSB). Wyoming. Sweetwater Co.: Kinney Ranch, 6,900 ft, 21 mi S Bitter Creek, 6 (KU); Kinney Ranch, 21 mi S Bitter Creek. 7.100 ft, 2(MVZ); Kinney Ranch, sec. 8. T15N, R98W. 23 mi SW Bitter Creek. 1 (MVZ); Shell Creek, 25 mi S Bitter Creek, 5 (CM); Blacks Fork, opposite mouth, 5,930 ft, 1 (UU). Perognathus ftavescens flavescens . — Specimens examined. 90. distributed as follows: Colorado. Adams Co.: Barr. 1 (UCM). El Paso Co.: Sandy Gulch, 2 mi E center Colorado Springs, 6,000 ft, I (UCM). Washington Co.: Akron, 5 (UMMZ); 8 mi W Akron, I (UMMZ). Yuma Co.: N of Wray, 2 (UCM). Nebraska. Antelope Co.: Clearwater, I (UMMZ); Neligh, I (MVZ), I (UNSM). Banner Co.: 10 mi S, 2.5 mi E Gering, 4 (VMKSC). Cherry Co.: blackberry Lake, Valentine National Wildlife Ref- uge, 2 (KU), 18 (UMMZ); 4 mi E Valentine. 1 (KU); 2 mi E Valentine, 1 (KU); Kennedy, 5 (MVZ), I (UMMZ); 4 mi S Ken- nedy, 1 (UNSM); 2 mi E Kennedy, 3 (KU); 4 mi E Kennedy, 2 (KU); 18 mi NW Kennedy, I (UNSM); Niobrara River, 10 mi S Cody, I (UNSM); 11.5 mi S, 0.5 mi W Nenzel, 3,000 ft, I (VMKSC). Custer Co.: 1 mi S, 2 mi W Broken Bow, 2 (VMKSC). Garden Co.: Crescent Lake National Wildlife Refuge Fleadquarters, sec, 29, T2IN, R44W, 3 (VMKSC); 0.75 mi E Crescent Lake National Wildlife Refuge Headquarters, sec. 29, T21N, R44W, 1 (VMKSC); 3 mi SE Crescent Lake National Wildlife Refuge Headquarters, 4 (VMKSC); 5 mi S Crescent Lake National Wildlife Refuge Headquarters, I (VMKSC). Hooker Co.: Kelso, 7 (UMMZ). Kearney Co.: 10 mi N, 1 mi E Axtell. 6 (VMKSC); 5 mi S. 2 mi E Kearney. 6 (VMKSC); Doby Town, 5 mi S, 3 mi E Kearney, 1 (VMKSC). Keith Co.: N side Kingsley Reservoir, 1 (UNSM). Lincoln Co.: 1 mi N Brady, 1 (UNSM); 2.5 mi N, 4.5 mi E North Platte, 1 (VMKSC). Sheridan Co.: 14 mi W Lakeside, 1 (MVZ). Thomas Co.: Halsey National Eorest, 1 (VMKSC). Perognathus flavescens perniger. — Specimens examined, 22, distributed as follows; Iowa. Ereemont Co.: Randolph, I (UNSM). Minnesota. Sherburne Co.: Elk River, I (UCM), 5 (MVZ); Sand Dune State Park, 3 (MMNH); 6 mi E St. Cloud, 5 (UMMZ). South Dakota. Bon Homme Co.: 0.3 mi N, 0.3 mi E Springfield, 2 (MSB). Clay Co.: 1 .5 mi N Vermillion, 3 (MSB); 1 mi W Vermillion, I (MSB); 3.5 mi N, 0.5 mi E Meckling, 1 (MSB). Perognathus flavescens cockrumi. — Specimens examined, 5, distributed as follows; Kansas. Barber Co.: 2 mi N, 2 mi W Sharon, 2 (SIUC). Geary Co.: Junction City, 1 (UCM). Harvey Co.: Section N of Harvey Co. Park, 2 (KSU). Perognathus flavescens copei. — Specimens examined, 54, dis- tributed as follows: New Mexico. Chaves Co.: 3 mi N, 9 mi W Caprock, 4 (MSB), 3 + chromosomes (MSB); 7 mi E Hagerman, 1 (MSB). Eddy Co.: 1 mi N, 26.5 mi E Carlsbad, 2 (ENMU). Lea Co.: 29 mi E Carlsbad, 1 (ENMU); 3 mi S, 29 mi E Carlsbad, 2 (ENMU); 2.5 mi S, 31 mi E Carlsbad, I (ENMU); 7 mi N, 15 mi W Jal, 2 (MSB). Quay Co.: 2 mi S, 0.5 mi E Logan, I (MSB). Roosevelt Co.: 3.3 mi S Tolar, 3 -I- chromosomes (MSB); 3.25 mi N, 1 mi E Portales, I (ENMU); 9 mi S Portales, I (ENMU); 4.5 mi S, 3 mi W Portales, 1 (ENMU). Union Co.: Perico Creek. 4 mi S Clayton, 3 (MSB). Oklahoma. Woods Co.: Waynoka, 10 (UMMZ). Texas. Andrews Co.: 14 mi S Andrews, 3 (UMMZ). Haskell Co.: 7 mi SW Rochester, 1 (MWU). Hemphill Co.: Gene Howe Refuge, 5 mi NE Canadian, I (TCWC). Loving Co.: 1 1 mi E Mentone, I -f chromosomes (MSB). Roberts Co.: 6 mi N Miami. 4 (MWU); 7 mi N Miami, I (MWU). Scurry Co.: 4 mi SW Snyder, 4 (MWU). Ward Co.: 4 mi NE Monahans, 1 (UIMNH). Wheeler Co.: 1 mi W Mobeetie, 2 (MVZ). LITERATURE CITED Armstrong, D. M. 1972. Distribution of mammals in Colorado. Monog., Mus. Nat. Hist., Univ. Kansas 3:1-415. Bailey, V. 1932. Mammals of New Mexico. N. Amer. Fauna, 53:1-412. Benson, S. B. 1933u. Concealing coloration among some desert rodents of the southwestern United States. Univ. California Publ. Zool., 40: 1-70. . 19336. Descriptions of two races of Perognathus am- pins. Proc. Biol. Soc. Washington, 46:109-112. . 1935. A biological reconnaissance of Navajo Mountain, Utah. Univ. California Publ. Zool., 40:439-456. Blair, W. F. 1954. Mammals of the Mesquite Plains biotic dis- trict in Texas and Oklahoma, and speciation in the Central Grasslands. Texas J. Sci., 6:235-264. Bi air, W. F., AND C. E. Mil l ER, Jr. 1949. The mammals of the Sierra Vieja region, southwestern Texas, with remarks on the biogeographic position of the region. Texas J. Sci., 1:67-92. Blair, W. E., A, P. Blair, P. Brodkorb, E. R. Cagle, and G. A. Moore. 1957. Vertebrates of the United States. McGraw-Hill Co., Inc., New York, ix -L 819 pp. Choate, J. R., and H. H. Genoways. 1975. Collections of Recent mammals in North America. J. Mamm., 56:452-502. CocKRUM, E. L. 1960. The Recent mammals of Arizona: their taxonomy and distribution. Univ. Arizona Press, Tucson, 276 pp. Dice, L. R. 1929. Descriptions of two new pocket mice and a new woodrat from New Mexico. Occas. Papers Mus. Zool., Univ. Michigan, 203:1-4, Dixon, W. J. (ED.). 1976. Biomedical computer programs. Univ. California Press, Berkley, 3rd ed., vii + 773 pp. 1978 WILLIAMS— PEROG NATH US SYSTEM ATICS 57 Durrant, S. D. 1952. Mammals of Utah, taxonomy and distri- bution. Univ. Kansas Publ., Mus. Nat. Hist., 6:1-549. Findley, J. S., A, H. Harris, D. E. Wilson, and C. Jones. 1975. Mammals of New Mexico. Univ. New Mexico Press, Albuquerque, xxii + 360 pp. Gabriel, K. R. 1964. A procedure for testing the homogeneity of all sets of means in analysis of variance. Biometrics, 20:459-477. Goldman, E. A. 1918. Five new mammals from Arizona and Colorado. Proc. Biol. Soc. Washington, 31:21-25. . 1938. A new pocket mouse from Colorado. J. Mamm., 19:495-496. Harris, A. H. 1963. Ecological distribution of some vertebrates in the San Juan Basin, New Mexico. Papers Anthro., Mus. New Mexico Press, 8:1-63. . 1965. The origin of the grassland amphibian, reptilian, and mammalian faunas of the San Juan-Chaco River drain- age. Unpublished Ph.D. dissertation, Univ. New Mexico, Albuquerque, 160 pp. . 1970. The Dry Cave mammalian fauna and late Pluvial conditions in southeastern New Mexico. Texas J. Sci., 22:3- 27. Mares, M. A., and D. E. Williams. 1977. Experimental sup- port for food particle size resource allocation in heteromyid rodents. Ecology. 58:1 186-1 190. Merriam, C. H. 1889. Revision of the North American pocket mice. N. Amer. Fauna, 1:1-29. Osgood, W. H. 1900. Revision of the pocket mice of the genus Perognathus. N. Amer. Fauna, 18:1-73. Patton, J, L. 1967. Chromosomes and evolutionary trends in the pocket mouse subgenus F’crog/iut/n/.v (Rodentia: Heter- omyidae). Southwestern Nat., 12:429—438. . 1969. Chromosome evolution in the pocket mouse, Pc- rognathus goUInuini Osgood. Evolution, 23:645-662. Ridgway, R. 1912. Color standards and color nomenclature. Published by the author, Washington, D.C., iii + 44 pp. + Fill plates. Rohlf, E. j. 1971. MINT user's manual. Unpublished manual. Dept, of Ecology and Evolution, State Univ. of New York, Stony Brook, New York, 45 pp. Sneaih, P. H. a., and R. R. Sokal. 1973. Numerical taxon- omy: the principles and practice of numerical classification. W. H. Freeman and Co., San Francisco, 573 pp. Wallace, J. T., and R. S, Bader. 1967. Factor analysis in morphometric traits of the house mouse. Syst. Zool., 16: 144- 148. Wells, P. V. 1970k. Postglacial vegetational history of the Great Plains. Science, 167:1575-1582. . 19706. Vegetational history of the Great Plains: a post- glacial record of coniferous woodland in souteastern Wyo- ming. Pp. 185-202, in Pleistocene and Recent environments of the central Great Plains (W. Dort, Jr., and J. K. Jones, Jr., eds.), Univ. Press Kansas, 443 pp. Williams, D. F. 1978. Karyological affinities of the species groups of silky pocket mice (Rodentia, Heteromyidae). J. Mamm., 59:599-612, Wii .SON, D. E. 1973. The systematic status of Perognutlws nier- rianii Allen. Proc. Biol. Soc, Washington, 86:175-192. I I ■ I I p r ) I It i, I I I • t' (i :(■ 1 a ?1 i Copies of the following Bulletins of Carnegie Museum of Natural History may be obtained at the prices listed from the Publications Secretary, Carnegie Museum of Natural History, 4400 Forbes Avenue, Pitts- burgh, Pennsylvania 15213. 1. Krishtalka, L. 1976. Early Tertiary Adapisoricidae and Erinaceidae (Mammalia, Insectivora) of North America. 40 pp., 13 figs $2.50 2. Guilday, J. E., P. W. Parmalee, and H. W. Hamilton. 1977. The Clark’s Cave bone deposit and the late Pleistocene paleoecology of the central Appalachian Mountains of Virginia. 88 pp., 21 figs. $12.00 3. Wetzel, R. M. 1977. The Chacoan peccary, Cu/ngonz/.v nY/g/?t'/7 (Rusconi). 36 pp., 10 figs. .. $6.00 4. Coombs, M. C. 1978. Reevaluation of early Miocene North American Moropus (Perissodactyla, Chalicotheriidae, Schizotheriinae). 62 pp., 28 figs $5.00 5. Clench, M. H., and R. C. Leberman. 1978. Weights of 151 species of Pennsylvania birds analyzed by month, age, and sex. 87 pp $5.00 6. Schlitter, D. A. (ed.). 1978. Ecology and taxonomy of African small mammals. 214 pp., 48 figs. $15.00 7. Raikow, R. J. 1978. Appendicular myology and relationships of the New World nine-primaried oscines (Aves:Passeriformes). 43 pp., 10 figs $3.50 8. Berman, D. S, and J. S. McIntosh. 1978. Skull and relationships of the Upper Jurassic sauropod Apatosaurus (Reptilia, Saurischia). 35 pp., 11 figs $3.00 9. Setoguchi, T. 1978. Paleontology and geology of the Badwater Creek area, central Wyoming. Part 16. The Cedar Ridge local fauna (Late Oligocene). 61 pp., 30 figs $4.50 .<■ ■t'. 'm 1' "'n,. :} BULLETIN of CARNEGIE MUSEUM OE NATURAL HISTORY THE BAKER BLUEE CAVE DEPOSIT, TENNESSEE, AND THE LATE PLEISTOCENE FAUNAL GRADIENT JOHN E. GUILDAY Associate Curator, Section of Vertebrate Fossils HAROLD W. HAMILTON Research Associate, Section of Vertebrate Fossils ELAINE ANDERSON 730 Magnolia Street, Denver, Colorado 80220 PAUL W. PARMALEE Department of Anthropology , University of Tennessee , Knoxville, 37916 with contributions by CHARLES H. FAULKNER Department of Anthropology , University of Tennessee, Knoxville 37916 (Archaeology) FREDERICK C. HILL Department of Biology , Bloomsburg State College, Bloomsburg, Pennsylvania 17815 (Fish) GEORGE H. VAN DAM The Museum, Michigan State University , East Lansing, Michigan 48823 (Amphibians and Reptiles) NUMBER 11 PITTSBURGH, 1978 BULLETIN OF CARNEGIE MUSEUM OF NATURAL HISTORY Number II, pages 1-67, figures I-I6, tables I-I7 Issued 27 December 1978 Price: $5.00 a copy Craig C. Black, Director Editorial Staff: Hugh H. Genoways, Editor-, Duane A. Schlitter, Associate Editor', Stephen L. Williams, Associate Editor-, Teresa M. Bona, Technical Assistant. © 1978 by the Trustees of Carnegie Institute, all rights reserved. CARNEGIE MUSEUM OF NATURAL HISTORY, 4400 FORBES AVENUE PITTSBURGH, PENNSYLVANIA 15213 CONTENTS Abstract 5 Introduction 5 The Site 5 Excavation 5 Procedure 8 Acknowledgments 8 Prehistoric Cultural Material 9 Faunal List 11 Biotic Discussions Flora 18 Fish Remains 18 Amphibians and Reptiles 19 Birds 25 Mammals 25 Faunal Summary 50 Faunal Analysis — Stratigraphic Change 51 Dating and Correlations 53 The Ridge and Valley Province Faunal Gradient 55 Family Composition of Three Ridge and Valley Mammalian Paleofaunas 63 Literature Cited 65 Paris No.4, Pa. 40°05'N. lat. Clark’s Cave,Va. 38°05' N.lat. Baker Bluff Cave, Tenn. 36°27'N.lat. A. Appalachian Plateau B. Valley & Ridge Province C Blue Ridge Province [after Hunt, 1974) Fig- 1- — Location of three late Pleistocene cave paleofaunas, eastern USA (Base map Erwin Raisz, 1954) ABSTRACT Late Pleistocene remains of 180 taxa of vertebrates and inver- tebrates are reported from a 3 m( 10 ft) column of fissure-fill, Baker Bluff Cave, Sullivan County, Tennessee, USA. Radiocarbon dates suggest deposition began about 19,100 years ago. Accom- panying faunal sequence indicates a transition from cool-temper- ate in the lower levels to boreal open woodland in the upper levels. The Pleistocene/Holocene transition was not recorded. Evidence of human occupation from Early Archaic, 8,000 to 9,000 years BP, to the Historic period was present in the top 90 cm (3 ft) of the deposit. Pollen analysis was negative. Six mammalian taxa are extinct — Dasypus bellus, Castoroides ohioensis, Felis onca cf. augusta, Sangamona fugitiva, Platy- gonus compressus, and Tapirus cf. veroensis; 16.6% of mam- malian taxa are found only north or west of the site today; 15% have retreated to higher elevations. Evidence for the contempo- raneity of two size classes of Blarina hrevicauda is presented. One bird, cf. Pica pica, no longer occurs in the eastern U.S. The fauna is composed of eastern temperate, mid western grassland, and boreal forest species, many now allopatric in distribution. No mammals of southerly distribution were present. A 500 km north-south transect of the Ridge and Valley province comparing late Pleistocene and Recent mammal faunas indicates a steeper faunal gradient undercurrent environmental conditions, but the late Pleistocene mammalian fauna was more boreal in character and richer in species, suggesting cooler but more equa- ble conditions. INTRODUCTION The Site Baker Bluff Cave, 13 km southeast of Kingsport, Sullivan County, is in northeastern Tennessee, 64 km west of the North Carolina border and 1 1 km south of the Virginia border (Fig. 1). It is perched high on a precipitous bluff overlooking the South Fork Holston River, a tributary of the Tennessee River, 5 km northwest, downstream from the junc- tion of the South Fork Holston and Watauga Rivers, latitude 36°27'30"N, longitude 82°28'W, Boone Dam quadrangle U.S.G.S. IV2' topographic map. The cave is 90 m above the west bank of the river on the eastern face of Ayers Ridge, locally known as Baker Bluff, at an altitude of 450 m (Fig. 2). The cave is little more than a single large chamber approxi- mately 10 m long and 3.6 m wide in vertically-bedded Cambrian dolomite. Prior to all excavations the cave floor was within 130 cm of the ceiling and the cave was little more than a crawlway (Figs. 3-4). The regional landscape today is one of timbered ridges and rolling farmlands. The site lies near the southern end of the Ridge and Valley physiographic province (Atwood, 1940) and is in the Carolinian biotic province (Dice, 1943). At the time of European colonization an oak/chestnut (QuercusICastanea) dominant, mixed mesophytic, closed-canopy wood- land covered the land, interspersed with river mead- ows and sometimes extensive Indian clearings and fields in the major river valleys. Annual precipitation at Knoxville, 140 km SW of the site, is 1 19 cm (46.85 in); average temperature ranges from 4.6°C (40.3°F) in January to 25.5°C (77.9°F) in July; average last frost 30 March; average first frost 2 November, with a growing season of 217 days. Rainfall is prevalent throughout the year with a slight increase during spring and summer. Snow cover is light and rarely lingers (USDA Yearbook, 1941). Excavation The top 3 ft of cave deposit was excavated by private collectors S. D. Dean, Jr., Robert Wilson, and Larry Gardner in 1968, while searching for Indian artifacts. They guided Hamilton to the site 1 September 1969, and on the basis of a caribou Rangifer tarandus pre molar from the original excavation it was decided to sample the site to a depth beyond that of their excavations. Excavation was recorded in feet and inches. Dean, Wilson, and Gardner divided the cave floor into six 3 ft squares and excavated to a depth of 3 ft, in 6-inch levels. They saved Indian artifacts, large bone fragments, and the larger land snails. We were able to examine color slides of the archaeological material. Bones and snails were donated by Dean to Carnegie Museum of Natural History. Between 15-23 August 1970, Ham- ilton and party excavated an irregular shaft 4 ft by 3 ft by 7 ft deep. It began at the base of the Dean excavation 3 ft below the original floor of the cave and extended to the limits of fossiliferous matrix approximately 10 ft below the original cave floor. A description of the surficial deposits is not now possible. A rim-fire .22 cartridge case, a scorched opossum vertebra indicat- ing at least one fire, and Indian cultural material to a depth of 3 ft suggested considerable disturbance. This disturbed material was described to us as “dark organic to dark brownish-red at the 3-foot level.” Below the level of the Dean excavation the matrix changed abruptly in character. Directly in the present entrance and slightly behind it, was a plug of dry, coarsely indurated, light yellow-tan "cave clay,” possibly the back-slope of a former en- trance cone built by slope-wash at a time when the entrance was more extended. No inclusions other than modem tree roots were noted. Directly behind this non-fossiliferous deposit and extend- ing down 10 ft from the original cave floor was a dark red-brown to yellow partially indurated matrix rich in bone fragments, snails, and hackberry seeds. Frost-spalled dolomite fragments and an occasional broken speleothem occurred at all levels but increased slightly in number with depth. There were no obvious stratigraph- ic changes in color or texture . N o evidence of water action or river 5 6 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. II Fig. 2. — Baker Bluff (Ayers Ridge), 13 km SE Kingsport, Sullivan County, Tennessee. South Fork Holston River, flowing to the right, in middle foreground. Arrow points to concealed entrance of Baker Bluff Cave. Hamilton photograph. pebbles was noted. As the excavation proceeded sterile cave fill encroached from the cave side of the deposit and at 10 ft the fos- siliferous material pinched out. Little, if any, is believed to be left in situ at the site. Preservation at all levels was chemically good but mechanically poor. Large mammals were represented by isolated teeth or heavily rodent-gnawed bone fragments. Thousands of complete or fragmentary small vertebrate bones were recovered, but no skeletons were found in articulation. Skulls of woodrats (Neot- onia) were occasionally complete. Mammal identifications are based mostly upon isolated teeth or mandible fragments contain- ing teeth. All bird remains were fragmentary. Snake vertebrae survived better than most bone because of their small size and compact structure. Gastropod remains were often complete at all levels and probably represented individuals that died in situ. The site throughout its history was a nesting area for rodents. 1978 GUILDAY ET AL.— BAKER BLUEE CAVE DEPOSIT 7 PROFILE (A-A') BAKER BLUFF CAVE SULLIVAN COUNTY, TENNESSEE SURVEYED BY R.E.WHITTEMORE, 1974 SCALE IN FEET Redrawn by James Senior Fig. 3. — Survey plan of Baker Bluff Cave, Sullivan County, Tennessee, showing location of excavations. primarily Neotoma and Peromyscus. Unlike rodent middens in the arid West (Wells, 1976; King and Van Devender, 1977) where plants remains are well preserved, most plant tissues and pollen had decayed in this humid eastern environment, leaving a lag de- posit of bones, teeth, snails, and hackberry (Celtis) seeds. Wood- rats were probably responsible for the large mammal remains — isolated teeth and heavily-gnawed bone fragments. Raptors, pri- marily owls, were responsible for the bulk of the small vertebrates in the deposit. All gastropods and some Neotoma probably died in situ, and the site may have served as a hibernaculum or shelter 8 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 11 Fig. 4. — Allen Hamilton sitting on original cave floor level, view facing NW taken from entrance of Baker Bluff Cave, Sullivan County, Tennessee. Note edge of CMNH excavation at his feet. Hamilton photograph. for snakes. Bones of a variety of Recent mammals from the upper 3 ft of the deposit were, in part, introduced by Indians as aborig- inal food items. Procedure Matrix from each 1 ft level was dry-screened through a '/* in. (5 mm) grid, and water-screened at the New Paris, Pennsylvania, field laboratory through 1 mm window screening to recover the finer fraction. It is felt that recovery of all fossil materials greater than 1 mm in diameter was total. Specimens are catalogued by stratigraphic levels in the Section of Vertebrate Fossils, Carnegie Museum of Natural History. Minimum number of individuals (MNI) was estimated for each level by counting the commonest replicable element for a given species, usually, but not always, teeth. Summing MNI by stratigraphic levels often results in a larg- er site total than if specimens from all levels were pooled before calculating MNI. This is true of large mammals represented by widely scattered identifiable elements, deer teeth, for instance. but not necessarily true of small vertebrates and invertebrates, which are either not as highly fragmented or whose MNI is based upon one specific element, such as the M,, in the case of the small rodents. Measurements less than 1 cm were taken with a Spencer Cy- cloptic stereoscopic microscope using an ocular grid at lOx. Larger measurements were taken with a dial micrometer cali- brated to 0. 1 mm. Abbreviations in this paper are: BP — before present; CM and CMNH — Carnegie Museum of Natural History; MNI — minimum number of individuals. Dental abbreviations are: I — incisor; C — canine; P — premolar; M — molar; d preceding tooth — deciduous. Tooth position indicated by super- and subscripts. Sites repeat- edly referred to throughout the text are: Natural Chimneys, Vir- ginia (Guilday, 1962); New Paris No. 4, Pennsylvania {Guilday et al., 1964); Robinson Cave, Tennessee (Guilday et al., 1969); Welsh Cave, Kentucky (Guilday et al., 1971); and Clark’s Cave, Virginia (Guilday et al., 1977). ACKNOWLEDGMENTS We thank S. D. Dean, Jr., Robert Wilson, and Larry Gardner for calling attention to the site and allowing us to examine their excavation notes and collections. For field assistance during CMNH excavations we gratefully acknowledge the help of Lee 1978 GUILDAY ET AL.— BAKER BLUFE CAVE DEPOSIT 9 Ambrose; Alan, Janet, Mary, and Melinda Bailey; Rita and Allen Hamilton; Paul, Helen and Mimi Imblum; Robert and Ann New, and Jay Smith. Site survey and mapping (Fig. 3) was done by Robert E. Whittemore. We thank Allen D. McCrady, codirector of the New Paris Field Laboratory, and the many volunteers through the years who assisted in the processing of the Baker Bluff Cave matrix. For professional assistance we thank: Charles H. Faulkner, University of Tennessee, Knoxville, archaeology; Robert Thompson and Paul S. Martin, University of Arizona, pollen; Frederick H. Utech, CMNH, botany; Juan Parodiz, CMNH, and Leslie Hubricht, Meridian, Mississippi, molluscs; George H. Van Dam and J. Alan Holman, Michigan State University, amphibians and reptiles; Frederick C. Hill, Bloomsburg State College, Penn- sylvania, fish; Mary R. Dawson, the late J. K. Doutt, Caroline A. Heppenstall, Hugh H. Genoways, Duane A. Schlitter, CMNH, and Jerry W. Nagel, East Tennessee State University, mammals. We thank Arthur E. Bogan, University of Tennessee, Knoxville, Helen McGinnis and Ronald C. Wilson, CMNH, for curatorial and technical assistance. Graphs and line drawings are by Erica Hansen, Idaho State Museum (Figs. 5, 8, and 10), and Nancy J. Perkins, James R. Senior, and Kemon N. Lardas, CMNH. Fi- nally, there are the two people most responsible for putting this paper together, Alice M. Guilday and Elizabeth A. Hill. Research was completed with funds from NSF Grant DEB 76- 07329 awarded to the senior author. For reading over preliminary drafts of this paper and for their helpful comments and assistance we thank: J. Alan Holman, Michigan State University; Mary R. Dawson, Hugh H. Geno- ways, and Allen D. McCrady, Carnegie Museum of Natural His- tory. PREHISTORIC CULTURAL MATERIAL CHARLES H. FAULKNER Eleven chipped stone artifacts were found in the top 3 ft of the cave fill by Dean, Wilson, and Gardner. Whether this represents the total lithic assemblage is not known, because no chipping debris from tool manufacturing was included in this collection. It is likely that some debitage was present but was not saved by the excavators. The excavators reported no features, pottery or charcoal in the upper fill, which would imply that the cave was occupied for extremely short periods of time. Food bone frag- ments are discussed elsewhere. One partially charred caudal vertebra of an opossum from the 0- 6 in. level suggests at least one fire. Unfortunately, the chipped stone artifacts were not available for the writer to examine, and their de- scription in this report is based on the examination of color prints and slides of fair quality. The exca- vators reported that the artifacts were originally typed by James Cambron. The writer concurs with some of Cambron’s identifications but disagrees with others. Based on the various limitations of this study, any conclusions should be considered tenta- tive. The excavated area was laid out in a 3 ft grid. At least six “squares” were excavated, presumably all to the 3 ft level. The earth was removed in arbitrary 0.5 ft cuts and the provenience of the artifacts was given by square number and level. Two artifacts came from the deepest level (2.5 ft- 3.0 ft). These are a side-notched bifurcated base pro- jectile point (Fig. 5c) and a small stemmed biface with an asymmetrical blade (Fig. 5g), the latter being tentatively identified as a knife. Both artifacts are made from a dark gray or black chert. All of the ar- tifacts except one appear to be made from this ma- terial. Although it is impossible to positively identify this chert in the photographs, it is probably the so- called “black flint” which occurs as small nodules in the shaley limestones of the eastern Tennessee Valley. This material ranges in color from black to opaque gray (Kellberg, 1963). The projectile point was identified by Cambron as aLeCroy type (Kneberg, 1956). While the bifurcated base of the specimen would certainly seem to place it into the cluster of Early Archaic bifurcated base projectile points that have been reported in the Ap- palachian uplands from Tennessee to West Virginia, the artifact does not really conform to the classic LeCroy type and if anything, might be more like the St. Albans Side Notched (Broyles, 1966). However, given the method of analysis, no specific type name should be assigned to this artifact. It can only be identified as an Early Archaic projectile point, prob- ably dating between 6,000-7,000 B.C. in the Eastern Tennessee Valley. The depth of occurrence would seem to substantiate this early placement. The 2.0 ft-2.5 ft level produced four artifacts from three squares. These include two artifacts identified as projectile points. One of these projectile points (Fig. 5d) had been classified as a Brewerton Side Notched (Ritchie, 1961). While this artifact could fall within the range of this Northeastern type, the Brew- erton series of projectile points have never been identified in the upper Tennessee Valley. The arti- fact also appears to resemble the bifurcated base type recovered in the lowest level. While the depth of recovery could indicate it is another Early Archaic artifact, the most prudent conclusion is that it is sim- 10 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 11 Fig. 5. — Flaked stone artifacts from 0-3 ft level of Baker Bluff Cave, Sullivan County, Tennessee, S. D. Dean, Jr., and R. Wilson private collection, a) Quartzite projectile point "Savannah River type”; b) "Madison” projectile point; c) "LeRoy” projectile point; d) "Brew- erton” projectile point; e) "Mt. Fork” projectile point (typology provided by Dean, see Faulkner discussion this paper); f) side-scraper; g) Archaic stemmed curved knife; h) uniface scraper; i-k) oval scrapers. Drawing by E. Hansen from photographs supplied by Dean. ply an Archaic type. The other projectile point from this level (Fig. 5e) was identified as a Mountain Fork which is considered to date from the Middle to Late Woodland periods in Alabama (Cambron and Hulse, 1964). While this spike-shaped artifact may generally conform to the Mountain Fork type, the present writer would not place it into any named type and would assign it to an Archaic horizon in the cave. The apparently retouched blade could also indicate this was a specialized tool type such as a perforator. Two additional implements were recovered in this level. These were previously identified as an “oval” scraper (Fig. 5j) and a uniface scraper (Fig. 5h). The writer will have to assume the latter was actually examined since the lack of retouch on the ventral surface of the flake is not apparent in the photograph. Because the writer did not examine these tools it is impossible to assign a definite function to them al- though their crude shape suggests they were prob- ably used for generalized scraping or cutting tasks. No projectile points were found in the 1.5 ft-2.0 ft level. Artifacts from this level have been identified as a side scraper and two oval scrapers. The former (Fig. 5f) appears to be simply a utilized flake, where- as the latter two crude implements (Fig. 5i and Fig. 5k) seem to exhibit bifacial flaking and retouched edges. Only two additional artifacts were recovered in the top 1 .5 ft of the cave floor. Both occurred above the 1 ft level. A stemless triangular projectile point which Cambron identified as a Madison type (Scully, 1951) was found in the 0.5 ft-1.0 ft level (Fig. 5b). Although the shape generally conforms to this type, the thickness and incurvate base is more reminiscent of such Early-Middle Woodland types in the upper Tennessee Valley as the Greenville and Nolichucky 1978 GUILDAY ET AL.— BAKER BLUEE CAVE DEPOSIT (Kneberg, 1957). In any case, this is certainly a Woodland artifact. The other worked piece is the distal end of a large biface made of quartzite (Eig. 5a). Although it is possible this is a large broken Late Archaic projectile point or knife such as the Appa- lachian Stemmed (Kneberg, 1957) or Savannah Riv- er Stemmed (Coe, 1964), the fragmentary nature of this artifact precludes such a precise identification. It was found in the 0.0 ft-0.5 ft level of the cave floor. The 11 artifacts described above indicate Ba- ker Bluff Cave was periodically occupied by several different prehistoric Indian groups. The presence of projectile points, scraping and cutting tools, and food bone fragments indicate they probably used it as a short-term hunting and butchering station. Be- cause the artifacts were not examined by the writer. a precise typology is impossible. It is also probable that they do not represent all of the human derived material deposited there. If this is a representative collection, however, and the tentative typology from the photographs is reasonably accurate. Baker Bluff Cave was first occupied by prehistoric Indians dur- ing the Early Archaic period, between 6,000-7,000 B.C. The cave continued to be utilized intermittently through the Late Archaic and Early Woodland pe- riods. There is no evidence that it was occupied by Indians during the late prehistoric or historic pe- riods. Perhaps the change in subsistence and settle- ment systems during these periods, or more likely the filling of the cave to within 4 ft of the ceiling, fi- nally caused the Indians to abandon it as a temporary camping site. Table 1. — Faunal list Baker Bluff Cave, Sullivan County, Tennessee . Superscript' = extinct; ^ = no longer present at site; F = no observation; * = present hut not tallied; MNI = minimum number of individuals . Depth in feel from surface Taxon 0-3 3^ 4-5 5-6 6-7 7-8 8-9 9-10 MNI Helicinidae Helicina orbiculata (Say) Hedersonia occulta (Say) Cionellidae Cionella luhrica (Muller) Valloniidae Vallonia parvula (Sterki) Pupillidae Gastrocopta armifera (Say) Gastrocopta contracta (Say) Gastrocopta corticaria (Say) Strobilopsidae Strobilops lahyrinthica (Say) Succineidae Succinea ovalis (Say) Catinella sp. Endodontidae Anguispira alternata (Say) Anguispira strongylodes (Pfeiffer) Discus patulus (L'*sh.) Discus bryanti (Harper) Discus catskillensis (Pilsbry) Helicodiscus multidens (Hubricht) Helicodiscus parallelus (Say) Helicodiscus singleyanus (Pilsbry) Helicodiscus inermis (H. B. Baker) Mollusca Class Gastropoda (snails) (identified by L. Hubricht) 9 9 1 — — — 1 5 4 4 5 22 9 4 4 5 6 9 30 ? 1 1 2 1 2 1 1 14 1 3 7 — 8 6 23 2 2 1 1 4 9 143 37 ? 2 ? ? 4 ? 3 ? 3 ? 13 ? 200 45 40 25 5 8 6 1 — 8 9 13 — 3 13 7 — 5 — 4 3 13 75 75 150 — 1 7 7 2 5 4 11 — 5 — 4 50 200 — 3 11 315 3 31 3 27 3 29 4 35 1 9 8 45 50 800 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 11 12 Table 1. — Continued. Depth in feet from surface Taxon 0-3 3^ 4-5 5-6 6-7 7-8 8-9 9-10 MNI Zonitidae Glyphyalinia wheatleyi (Bland) Glyphyalinia lewisiana (Clapp) Glyphyalinia caroliniensis (Ckll.) Glyphyalinia solida (H. B. Baker) Mesomphix capnodes (W.G.B.) Paravitrea multidentata (Binney) Paravitrea tridens Pilsbry Paravitrea blarina (Hubricht) Hawaiia minuscula (Binney) Euconulits fidvus (Muller) Gastrodonta interna fonticula (Wurtz) Ventridens pilsbryi (Hubricht) Ventridens coelaxis (Pilsbry) Ventridens detnissus (Binney) Zonitoides arboreus (Say) Haplotrematidae Haplotrema concavum (Say) Polygyridae Polygyra plicata (Say) Stenotrema spinosum (Lea) Stenotrema stenotrema (Pfeiffer) Stenotrema , undescribed species Stenotrema fraternum fasciatum (Pilsbry)^ Mesodon clausus clausus (Say) Mesodon elevatus (Say) Mesodon appressus (Say) Mesodon inflectus (Say) Mesodon rugeli (Shutt.) small Mesodon rugeli large Triodopsis tridentata (Say) Triodopsis tridentata tenneseensis (Walker) Triodopsis vulgata (Pilsbry) Triodopsis denotata (Per.) Triodopsis albolabris (Say) Allogona profunda (Say) Amnicolidae Pomatiopsis lapidaria (Say) Pomatiopsis cincinnatiensis (Lea) Pleroceridae lo fluviatilus ? — — 5 ? 16 19 50 ? 2 2 3 6 3—1 7 ? 3 10 3 ? 1—2 ? 16 14 35 ? — — 1 7 ? — — 2 ? — 1 — ? 5 13 7 ? 1—2 ? 1 — 12 ? — — 1 ? — 3 6 ? 1 — 1 ? 1—1 ? 1—1 422 26 41 33 20 1—3 ? 1—9 ? — 1 — 3 — — 4 7 111 1 — — 1 48 3 1 4 ? 3—2 ? — 1 Class Bivalvia (freshwater mussels) Unionidae, sp. 3 — — — Vertebrata Class Pisces (identified by F. Hill) Order Semionotiformes Lepisosteidae Lepisosteus sp. — gar ? — — * Order Salmoniformes Esocidae Esox sp. — pike or pickerel ? * — * 2 9 2 9 1 16 1 3 2 1 2 11 2 1 1 1 1 1 35 1 6 2 17 1 2 1 — 119 2 8 15 119 — — — 9 — 235 2 1 — 22 — —22 — — 1 18 — 1—4 — 14 25 120 — —12 — 326 4 6 5 18 — 329 — — — 2 1 2 5 35 1 — 4 - 1 3 28 -—14 - 1 2 13 1 4 3 11 -2—3 - — — 3 - — — 3 2 9 9 577 3 — 1 29 1 — 3 20 - — — 3 2 3 1 30 - — — 3 1 — — 4 - — — 3 2 — 3 63 -—18 1 7 1 10 - — — 1 1 3 * * 1978 GUILDAY ET AL.— BAKER BLUFF CAVE DEPOSIT 13 Table 1. — Continued. Depth in feet from surface Taxon 0-3 3^ 4-5 5-6 6-7 7-8 8-9 9-10 MNl Cyprinidae — Cyprinids, unidentified minnows Catostomidae Catostomus commersoni — white sucker Hypentelium nigricans — northern hog sucker Moxostoma carinatum^nver redhorse Moxostoma cf. duquesnei — black redhorse Moxostoma erythrurum — golden redhorse Moxostoma sp. — redhorse, unidentified Order Cypriniformes * * * * Order Siluriformes * * * * * Ictaluridae Ictalurus punctatus — channel catfish ? — Ictalurus sp. — catfish, unidentified ? — Noturus sp. — madtom or stonecat ? — Order Perciformes Centrarchidae Ainbloplites rupestris — rock bass Micropterus dolomieui — smallmouth bass Micropterus sp. — bass, unidentified Percidae Stizostedion canadense — sauger Stizostedion sp. Sciaenidae Aplodinotus grunniens — freshwater drum Bufonidae Bufo americanus — American toad Bufo w.fowleri — Fowler’s toad Bufo sp. — toad Hylidae Hyla sp. — tree frog Ranidae Rana sylvatica — wood frog Rana catesbeiana — bullfrog Rana sp. Ambystoroatidae Ambystoma opacum — marbled salamander Ambystoma maculatum — spotted salamander Ambystoma sp. Proteidae Necturus maculosus — mudpuppy Cryptobranchidae Cryptobranchus alleganiensis — hellbender Plethodontidae Desmognathus sp. — dusky salamander 0 * 9 * 9 * 9 * 9 + * * 4: * Class Amphibia (identified by G. Van Dam) Order Anura 12 24 10 3 3 1 4 15 4 1 5 6 8 1 — 3 2 2 Order Urodela ? — 1 1 ? — 1 1 ? 1 1 1 7 111 7 111 7 111 9 9 9 9 9 9 9 * * * * * * * * * * * * * 7 2 2 3 112 — 3 1 — 1 — 1 1 1 1 1 1 — 1 1 — I 1 1 1 — * * * * * * * * * * * * * * * 60 7 27 24 1 9 4 3 7 6 6 6 14 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. II Table 1. — Continued . Depth in feet from surface Taxon 0-3 3^ 4-5 5-6 6-7 7-8 8-9 9-10 MNI Emydidae Gniptemys geographica — map turtle Scincidae Eumeces fasciatus — five-lined skink Viperidae Crotalus horridus — timber rattlesnake Class Reptilia (identified by G. Van Dam) Order Testudines ? — 1 — Order Squamata ? 1 1 — 1111 Colubridae Heterodon platyrhinos — eastern hognose snake Diadophis punctatus — ringneck snake Carphophus amoenus — worm snake Coluber or Masticophis sp. — racer or coachwhip Lampropeltis triangulum — eastern milksnake Lampropeltis getulus — kingsnake Elaphe sp. — rat snake Nairi.x sipedon — water snake Matrix sp. Thamnophis sirtalus — garter snake Thamnophis sauritus — ribbon snake Thamnophis sp. ? — — 1 ? 1 1 1 ? 1 1 — ? — I 1 ? 1 1 1 ? 1 1 1 7 ? 1 1 1 ? 1 1 1 ? 1 1 — ? 1 1 1 Class Aves (identified by P. Parmalee) Order Podicipediformes Podicipedidae sp. — grebes ? — 1 Order Anseriformes Anatidae cf. Anas crecca — green-winged teal Anas sp. — duck Mergus sp. — merganser Lophodytes cucullatus — hooded merganser Order Falconiformes ? 1 _ _ ? 1 1 1 ? — 1 — ? 1 1 1 Accipitridae Accipiter gentilus — goshawk ? — — — Accipitridae sp. — hawk ? — 2 1 Falconidae Falco sparverius — kestrel 7 111 1 I 1 1 1 1 1 1 1 Tetraonidae cf. Bonasa umhellus — ruffed grouse Tetraonidae sp. — grouse Meleagrididae Meleagris gallopavo — turkey Phasianidae Callus gallus — chicken Order Galliformes 4 3 2 2 — ? 16—1 1 — — 1 — 1 1 1 1 1 1 1 1 1 1 1 I 1 1 1 1 3 8 3 3 2 5 6 4 1 3 1 4 3 6 1 1 3 1 4 1 3 3 11 9 2 1 1978 GUILDAY ET AL.— BAKER BLUEE CAVE DEPOSIT 15 Taxon Charadriidae cf. Arenaria interpres — ruddy turnstone Scolopacidae Philohela minor — woodcock Capella gallinago — common snipe cf. Actitis macularia — spotted sandpiper Scolopacidae spp. Laridae Lams sp. — gull Columbidae Ectopistes migratorius — passenger pigeon Strigidae Bubo virginianus — great horned owl Otus asio — screech owl Caprimulgidae cf. Chordeiles minor — common nighthawk Apodidae Chaetura pelagica — chimney swift Table 1. — Continued. Depth m feel from surface 0-3 3-4 4-5 5-6 6-7 7-8 8-9 9-10 MNI Order Charadriiformes 2 1 — — — — 3 6 6 1 — — — 13 Order Columbiformes 8 10 23 11 5 1 1 I 60 Order Strigiformes ? _ 1 Order Caprimulgiformes Order Apodiformes 7 7 7 9 7 9 Alcedinidae Megaceryle alcyon — kingfisher Picidae Colaptes auraius — common flicker Picidae sp. Tyrannidae Empidonax sp. — flycatcher Hirundinidae Petrochelidon pyrrhonota — cliff swallow Corvidae Cyanocitta cristaia — blue jay cf. Pica pica — black-billed magpie Turdidae cf. Turdus migratorius — robin Parulidae sp. — warblers Icteridae cf. Agelaius phoeniceus — red-winged blackbird Fringillidae sp. — sparrow Passeriformes sp. Order Coraciiformes ? 1 I 2 — — Order Piciformes ? 1 1 1 2 — ? 2 I — — 1 Order Passeriformes ? 1 _ — 1 _ ? — 1 — — — ? 6 4 4 3 — 4 5 4 1 2 1 1 I 1 I 21 16 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 11 Table 1. — Continued. Depth in feet from surface Taxon 0-3 3^ 4-5 5-6 6-7 7-8 8-9 9-10 MNI Class Mammalia (identified by J. Guilday and E. Anderson) Order Marsupialia Didelphidae Didelphis virginiana — Virginia opossum Order Insectivora 1 Soricidae Blarina brevicauda — short-tailed shrew ? 15 20 40 31 6 24 13 149 Cryptotis parva — least shrew ? 1 2 — — — — — 3 Microsorex hoyi^ — Pygmy shrew ? 2 5 2 — — — — 9 Sot ex arcticus'^ — arctic shrew ? 1 1 1 — — — — 3 Sorex cinereus — masked shrew ? 9 19 16 5 — 2 2 53 Sorex dispar^ — rock shrew ? 4 3 2 — — — — 9 Sorex fumeus — smoky shrew ? 2 6 6 6 1 — 1 22 Sorex fumeus or arcticus 7 4 5 3 - — — 12 Sorex cinereus or dispar 7 11 5 2 — — — 18 Talpidae Condylura cristata — star-nosed mole ? 1 2 — — — — — 3 Scalopus aquaticus — eastern mole 1 1 1 3 1 2 2 1 12 Parascalops breweri — hairy-tailed mole 1 2 Order Chiroptera 2 L 4 3 1 3 1 17 Vespertilionidae My Otis sp. — little brown bats ? 3 4 6 3 — 1 1 18 Pipistrellus subflavus — eastern pipistrelle 7 2 1 1 — 1 2 7 Eptesicus fuscus — big brown bat ? 2 3 9 2 1 1 1 19 Nycticeius humeralis — evening bat 7 — — — — 1 — 1 Plecotus sp. — big-eared bats 7 Order Edentata 1 1 Dasypodidae Dasypus bellus' — ’’beautiful” armadillo 9 * * Order Lagomorpha * * * 1 Leporidae Sylviagus or Lepus — rabbits or hares * 11 Order Rodentia 20 22 17 7 10 8 95 Sciuridae Tatnias striatus — eastern chipmunk ? 3 2 9 4 1 5 3 27 Eutamias minimus'^' — least chipmunk ? 1 2 — — — — — 3 Marmota monoj:— woodchuck 1 3 6 5 2 1 3 2 23 Spennophilus tridecemlineatus^ — thirteen- lined ground squirrel 1 4 10 8 2 1 1 1 28 Sciurus carolinensis — gray squirrel 1 — 2 4 3 1 3 3 17 Tamiasciurus hudsonicus — red squirrel ! 3 6 3 2 1 1 1 18 Glaucomys volans — southern flying squirrel 1 — 1 1 — — 2 2 7 Glaucomys sabrinus^ — northern flying squirrel ? 3 5 3 1 1 — — 13 Castoridae Castor canadensis — beaver — — 1 — 1 — 1 — 3 Castoroides ohioensis ‘ — giant beaver 1 — — — — — — — 1 Cricetidae Peromyscus maniculatus or P. leucopus — white-footed mice ? 8 26 24 18 2 8 8 94 Neotoma floridana — eastern woodrat 23 62 52 46 13 31 33 260 1978 GUILDAY ET AL.— BAKER BLUFF CAVE DEPOSIT 17 Table 1. — Continued. Depth in feet from surface Taxon 0-3 3-4 4-5 5-6 6-7 7-8 8-9 9-10 MNI Arvicolidae Clethrionomys gapped'^ — southern red-backed vole Phenacomys intermedins^ — heather vole Microtus chrotorrhinus^ — rock vole Microtus pennsylvanicus — meadow vole M. chrotorrhinus or pennsylvanicus (total figure includes identified M. chrotorrhinus and M. pennsylvanicus) Microtus xanthognathus^ — yellow-cheeked vole Microtus pinetorum and/or M. ochrogaster — woodland vole and/or prairie vole Ondatra zibethicus — muskrat Synaptomys cooped — southern bog lemming Synaptomys borealis^ — northern bog lemming Zapodidae Zapus hudsonius — meadow jumping mouse Napaeozapus insignis^ — woodland jumping mouse Erethizontidae Erethizon dorsatum^ — porcupine Canidae Vulpes vulpes — red fox Urocyon cinereoargenteus — gray fox Ursidae Ursus americanus — black bear Procyonidae Procyon lotor — raccoon Mustelidae Martes americana^ — marten Martes pennantd — fisher Mustela nivalis — least weasel Mustela frenata — long-tailed weasel Taxidea taxus^ — badger Spilogale putorius — eastern spotted skunk Mephitis mephitis — striped skunk Felidae Felis onca — jaguar Tayassuidae Platygonus compressus ' — fiat-headed peccary Cervidae. Cervus elaphus^ — elk Odocoileus virginianus — white-tailed deer cf. Sangamona fugitiva ■ — “fugitive” deer Rangifer tarandus^ — caribou Tapiridae Tapirus cf. veroensis ' — tapir 9 32 70 43 7 — 1 6 159 9 12 31 15 5 — — 1 64 ? 7 18 7 4 — 1 — 37 9 18 35 22 3 1 1 4 84 9 58 62 42 10 2 4 8 186 9 1 1 1 1 — — — 4 9 28 81 93 37 5 17 21 282 * 2 1 — 1 — — — 4 ? 27 71 43 15 3 17 15 191 9 2 3 3 3 — 1 — 12 9 - 1 1 2 ? 1 1 1 1 — — — 4 9 1 1 1 — — — — 3 Order Carnivora 1 1 1 3 * — — — — — — — 1 * 1 1 — — — 1 4 * — — 1 1 — 1 1 5 ? 2 1 1 4 9 (1) — — — — — 1 9 1 2 1 — — — — 4 9 -- 1 2 — — 2 1 6 — — depth unknown — — — 1 ? 1 — — 1 — 1 1 4 * — — — — — — — 1 9 — — — — — — 1 I Order Artiodactyla 9 _ 1 _ _ 1 2 * * * * 1 * * * * 2 10 * — 1 — — — — — I * - 1 — 1 — — — 2 Order Perissodactyla ? — 1 1 18 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 11 FLORA Botanical specimens were not saved from the up- per 0-3 ft levels of the deposit, but two fragmentary hickory nuts (Corya sp.) were present in the Dean/ Wilson collection from the top few inches of the cave floor and are undoubtedly Recent. A total of 1,071 hackberry seeds {Celtis sp.) were identified from the deposit. Many more were orig- inally present but so pulverized that they could not be counted. Thirty-two percent were puncture- gnawed by small rodents, probably Peromyscus. Seed counts by stratigraphic level, beginning at the 3^ ft level, were 31, 73, 320, 359, 73, 94, and 121 at the 9-10 ft level. A stratigraphic increase in abun- dance is suggested by the relative numbers of hack- berry seeds compared to small rodents (Arvicoli- dae). Applying the formula, Celtis seeds by Arvicolidae MNI x 100, a change in frequency with depth is noted: 3-4 ft level = .19; 4-5 ft level = .23; 5-6 ft level = 1.33; 6-7 ft level = 4.60; 7-10 ft level = 2.82. Hackberry seeds increased in relative abundance in the lower levels of the de- posit. There are two possible explanations for this. The greater relative abundance of hackberry in the lower levels may be due to the climatic shift indi- cated by the small mammals of the deposit, a shift from relatively temperate in the lower levels, fa- voring Celtis, to more boreal conditions in the up- per levels. Or the reason may simply be due to a change in the activity level of the raptors respon- sible for the small mammals in the deposit. It is possible that rodent midden activity, and its accom- panying seed caching predominated in the lower levels of the deposit while rodent activity decreased and raptor activity increased in the upper levels of the deposit, which may have been either a second- ary effect of climatic change or due simply to the configuration of the deposit. Pollen analysis was attempted, with negative re- sults, on seven samples from various stratigraphic levels, by Robert Thompson, Laboratory of Pa- leoenvironmental Studies, Department of Geosci- ences, University of Arizona, Tucson; four samples utilizing zinc bromide density separation, three samples using hydrofluoric acid. The lack of fossil pollen in this indurated, yellow-brown, basic (pH — 7. 6-7. 7) cave sediment was not unexpected and is unfortunately consistent with results of other Ap- palachian cave or talus deposits, such as the Mea- dowcroft Rockshelter, Pennsylvania, and Trout Cave, West Virginia. Apparently, unless pollen is deposited in a subaqueous or other oxygen deficient environment it is soon destroyed in the humid East. Pollen was well-preserved in the late Pleistocene fissure fill of New Paris No. 4, Pennsylvania, due to a rapid rate of deposition in a water-saturated colloidal clay matrix, which served as an imper- vious shield effectively sealing all accumulating or- ganic inclusions from the atmosphere. FISH REMAINS FREDERICK C. HILL Seventeen taxa of fishes represented by 147 iden- tifiable bones and at least 13 species were identified from the Baker Bluff Cave faunal sequence (Tables 1 and 2). Numerous pharyngeal arches of Cyprini- dae have not been studied. Fish are catalogued un- der CM 30228. The species recovered from the seven levels at Baker Bluff could be found coexisting in a moderate gradient, medium order river. Only Catostomus commersoni is characteristically found in intermit- tent streams, but it also occurs in larger streams or rivers. Most of these species, including Aplodinot us gmnniens, Moxostoma etythrurum, M. carinatuin, Micropterus dolomieui, Amhloplites rupestris, Esox sp., and Lepisosteus sp. are typically found in pools. The presence of riffles between pools is evi- denced by young Ictalurus punctatus , none of which exceed 16 cm total length (Table 3), and Mi- cropterus dolomieui. Nearly all species prefer water with a moderate to swift current although some, such as Lepisosteus sp., Aplodinotus grunniens, Moxostoma erythrurum, and possibly the Noturus sp. are better adapted to slower currents. Six of the species, including Hypentelium nigricans, Catos- tomus commersoni, Moxostoma cf. duquesnei, M. carinatum, Amhloplites rupestris, and Micropterus dolomieui are best adapted to clear streams, where- as Moxostoma erythrurum, Aplodinotus grunniens, Ictalurus punctatus, and Stizostedion canadense tolerate varying degrees of turbidity. Thus, one 1978 GUILDAY ET AL.— BAKER BLUEE CAVE DEPOSIT 19 Table 2. — Total number offish bones identified from levels 3^ ft through 9-10 ft, Baker Bluff Cave, Tennessee. ? = no observation: * = present, not tallied. Taxon Stratigraphic levels 0-3 3^ 4-5 5-6 6-7 7-8 8-9 9-10 Hypentelium nigricans (Lesueur) — — 2 — — — — Catostomus commersoni (Lacepede) ? — — 1 1 — — — Moxostoma cf. duquesnei (Lesueur) ? — — 1 — — — — Moxostoma erythrurum (Rafinesque) ? 1 2 ! — — — — Moxostoma carinatum (Cope) 7 — — — I — — — Catostomidae sp. * — — — — — — — Aplodinotus grunniens Rafinesque 7 3 5 9 10 8 30 16 Ictalurus punctatus (Rafinesque) 7 — — 5 1 — — — Noturus sp. Rafinesque 7 1 — — — — — 3 Stizostedion canadense (Smith) '? 1 — — — — — — Ambloplites rupestris (Rafinesque) 7 2 2 — — — 1 — Micropterus dolomieui Lacepede 7 — — 1 — — — — Lepisosteus sp. Lacepede 9 — — 3 — — — — Moxostoma sp. Rafinesque 4-- 9 6 3 1 — 1 — Stizostedion sp. Rafinesque 7 1 6 1 2 — 1 — Ictalurus sp. Rafinesque * — 1 — — — — — Micropterus sp. Lacepede 7 — 1 — — — — — Cyprinidae sp. 7 17 46 25 6 2 — 4 Esox sp. Linnaeus 7 1 — 1 — — — — Total 36 69 53 22 10 33 23 would expect to find these species coexisting in var- ious habitats within a small section of a stream near Baker Bluff. The estimated sizes (Table 3) of the various fishes suggests that, for the most part, small predators were responsible for their presence in the cave. Stizostedion canadense was the largest fish record- ed, 50 cm T.L. A single Ictalurus punctatus was the smallest fish, only 4 cm T.L. Moxostoma iden- tifications were based upon either dentaries or pha- ryngeal arch teeth. All of the elements from small Moxostoma were assigned only to the generic level because of the difficulty encountered in making species identifications. No fish of unusual size were recorded. The small sample size precludes the recognition of any subtle changes in this ichthyofauna. The greatest number of individuals and taxa appear in the upper 6 ft of the deposit. AMPHIBIANS AND REPTILES GEORGE HENRY VAN DAM The Baker Bluff Cave faunal sequence includes at least five species of urodeles, five species of an- urans, one species of turtle, one species of lizard, and 11 species of snakes. Class Amphibia — Amphibians Order Urodela — Salamanders Family Ambystomatidae — mole salamanders Vertebral centrum amphicoelous, weakly con- stricted ventrally, without spine produced from its posteroventral surface; neural spine obsolete and single throughout (Holman, 1962). Ambystoma opacum (Gravenhorst) — marbled salamander Material. — CM 29754-29757. 2 precaudal vertebrae (4-5 ft); 1 precaudal vertebra (5-6 ft); 5 precaudal vertebrae (6-7 ft); 1 pre-caudal vertebra (8-9 ft). Remarks. — The Ambystomatidae can be divided into major groups using vertebral ratios — (1) the length of the centrum divided by its width at the anterior end, and (2) the combined zygapophyseal width divided by the zygapophyseal length (Tihen, 1958). Tihen noted that in the A. macidatum group and in the subgenus Linguaelapsus (at least in the 20 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 11 Table 3. — Estimated live total length (in cm) of Baker Bluff fish. Total length in centimeters Taxon 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 Ictalurus punctatus 2 2 1 — — — 1 — — Ictalunis sp. — — — 1 — — — — — Moxostoma sp. — — — 2 1 — 111 Ambloplites rupestris — — — 1 — 1 1 1 1 Noturus sp. — — — — — 1 — — — Catostomus commersoni — — — — — — — 1 — Stizostedion sp. — — — — — — — — 1 Micropterus dolomieui — — — — — — — — — Micropterus sp. — — — — — — — — — Hypentelium nigricans — — — — — — — — — Moxostoma erythrurum — — — — — — — — — Stizostedion canadense — — — — — — — — — 1 1 1 1 — 2 1 posterior part of the trunk) the postzygapophyses extends as far, usually farther posteriorly, than does the neural arch. In my examination of the material, I used the length of the centrum divided by its width at the anterior end for species determination. Ambystoma maculatum (Shaw) — spotted salamander Material. — CM 29758-29760. 6 precaudal vertebrae (4-5 ft); 3 precaudal vertebrae (5-6 ft); 4 precaudal vertebrae (6-7 ft). Remarks. — Material was assigned to A. macula- tum based on criteria discussed under A. opacum. Ambystoma sp. indet. Material. — CM 29761-29767. 3 precaudal vertebrae (3-4 ft); 15 precaudal vertebrae (4-5 ft); 7 precaudal vertebrae (5-6 ft); 7 precaudal vertebrae (6-7 ft); 3 precaudal vertebrae (7-8 ft); 7 precaudal vertebrae (8-9 ft); 1 precaudal vertebra (9-10 ft). Remarks. — Material which was too fragmentary for measurement or those whose ratios fell within the ranges of both A. maculatum and A. opacum were assigned to genus only. Some specimens referred to Ambystoma species exhibited the back-swept neural arch characteristic of A. tigrinum as pointed out by Holman (1969), but these could not be assigned to species because an- terior thoracic vertebrae of A. maculatum and A. opacum have this characteristic. Conant (1975) states that Ambystoma opacum occurs in a variety of habitats, ranging from moist, sandy areas to dry hillsides. A. maculatum is occasionally found (from spring to autumn) beneath stones or boards. Family Proteidae — Mudpuppies Specimens were assigned to the Proteidae by comparison with Recent specimens and using cri- teria of Holman (1968) who noted that their verte- brae are amphicoelous and the transverse processes are undivided. Necturus maculosus (Rafinesque) — mudpuppy Material. — CM 29768-29773. 15 precaudal vertebrae (3^ ft); 22 precaudal and 2 caudal vertebrae (4-5 ft); 5 precaudal ver- tebrae (5-6 ft); 3 precaudal vertebrae (6-7 ft); 5 precaudal ver- tebrae (7-8 ft); 3 precaudal vertebrae (8-9 ft). Remarks. — The fossils were indistinguishable from Recent N. maculosus. Necturus maculosus habitats include lakes, ponds, rivers, streams, and other permanent bodies of water (Conant, 1975). Family Cryptobranchidae — Hellbenders Cryptobranchus alleganiensis (Daudin) — hellbender Material. — CM 29774-29779. 30 precaudal, 3 caudal verte- brae, and 1 right dentary (3-4 ft); 73 precaudal vertebrae and 2 vomers (4-5 ft); 33 precaudal vertebrae (5-6 ft); 9 precaudal vertebrae (6-7 ft); 1 precaudal vertebra (7-8 ft); 4 precaudal vertebrae (9-10 ft). Remarks. — Meszoely (1967) gave characteristics for the identification of Cryptobranchus — deeply amphicoelous cotyles, circular in outline; centrum relatively short in respect to the diameter of the cotyle; ventral surface of the centrum rounded without keel or processes; large lateral fossa anter- iad to the base of the transverse process; very large size. In addition, Holman (1968) notes that the transverse processes are undivided. Due to the fragmentary nature of much of the material I found it sometimes difficult to separate vertebrae of Cryptobranchus alleganiensis and Necturus maculosus. I have found the following criteria useful in differentiating these two: in C. al- leganiensis the sides of the centrum are more sculp- 1978 GUILDAY ET AL.— BAKER BLUFE CAVE DEPOSIT tured than in N. maculosus', in C. alleganiensis the upper transverse process is heavy and cylindrical in shape (A^. maculosus has a very wing-like upper transverse process); C. alleganiensis has the artic- ular facets of the transverse processes exhibiting a single opening, whereas in N. maculosus there are usually two distinct openings. C. alleganiensis is always found in rivers and larger streams where water is running and ample shelter is available in the form of large rocks, snags, or debris (Conant, 1975). Family Plethodontidae — Plethodontid Salamanders Desmognathus sp. indet. — dusky salamanders Material. — CM 29780-29785. 22 vertebrae (3—4 ft); 158 ver- tebrae (4-5 ft); 95 vertebrae (5-6 ft); 81 vertebrae (6-7 ft); 19 vertebrae (7-8 ft); 21 vertebrae (8-9 ft). Remarks. — These specimens have been assigned to Desmognathus based on descriptions by Soler (1950) who states that their vertebrae are opistho- coelous and have pointed processes arising from the dorsal surfaces of the postzygapophyses. I am un- able to carry the identification any further due to lack of Recent comparative material. Desmognathus fuscus, D. quadramaculatus, D. monticola, and D. wrighti are all present in north- eastern Tennessee today and it is not unlikely that all four are present in the collection. Desmognathus fuscus occurs in brooks, near springs, and in seep- age areas along edges of small woodland streams where stones, chunks of wood, and miscellaneous debris provide ample shelter both for the salaman- ders and for their food. D. quadramaculatus is abundant in boulder-strewn brooks and also found near waterfalls or other places where cold water drips or flows. D. monticola prefers a habitat of cool, well-shaded ravines and banks of mountain brooks. D. wrighti is today a resident chiefly of high spruce-fir forests and lives under moss and bark on rotting logs or beneath rotting wood or litter on the forest floor near seepage areas (Conant, 1975). Order Anura — Frogs and Toads Family Bufonidae — Bufonid Toads Fossil Bufo may be identified by the following characteristics (Holman, 1962): ilium with dorsal blade absent; dorsal prominence produced dorsally, well developed, grooved or irregular in shape; sa- cral vertebrae procoelus, with one anterior and two posterior condyles; sacrum free from urostyle, its diapophyses moderately expanded. Bufo americanus Holbrook — American toad Material.— CM 29786-29792. 12 right, 7 left ilia (3^ ft); 17 right, 24 left ilia (4-5 ft); 10 right, 10 left ilia (5-6 ft); 3 right, 7 left ilia (6-7 ft); 2 right ilia (7-8 ft); 2 right ilia (8-9 ft); 3 left ilia (9-10 ft). Remarks. — The ilium of Bufo woodhousei fowleri has the base of the dorsal protuberance narrower than in equal-sized B. americanus (Holman, 1967). Habitats include shallow bodies of water in which to breed (temporary pools or ditches or shallow por- tions of streams, for example), possess shelter in the form of hiding places where there is some mois- ture, and harbor an abundant food supply of insects and other invertebrates (Conant, 1975). Bufo woodhousei fowleri Girard — Fowler's toad Material. — CM 29793-29195 . 3 right ilia (3—4 ft); 2 right, 3 left ilia (4-5 ft); 1 left ilium (5-6 ft). Assignment of material to Bufo u’. fowleri was based on criteria given in the discussion of 5. americanus. Bufo w. fowleri occurs chiefly in sandy areas, along shores of lakes or in river valleys (Conant, 1975). Bufo sp. indet. Material. — CM 29796-29801. 2 left ilia, 4 sacral vertebrae (3- 4 ft); 3 right, 2 left ilia, 15 sacral vertebrae, 10 frontoparietais (4-5 ft); 4 sacral vertebrae, 6 frontoparietais (5-6 ft); 2 fronto- parietais (6-7 ft); 1 sacral vertebra, I frontoparietal (7-8 ft); 2 sacral vertebrae, 2 frontoparietais (8-9 ft). Remarks. — Ilia were assigned to Bufo sp. when the anterior or posterior portions of the prominence were missing, thus making it impossible to examine the prominence-protuberance relationship, or when the boundaries of the dorsal protuberance were not clearly defined within the prominence. Tihen ( 1 962 ) pointed out that the frontoparietal is the most reli- able single element for identification of the greatest number of New World Bufo. I could not find any distinct differences between the frontoparietais of B. americanus and B. ir. fowleri. Family Hylidae — Hylid Frogs Hyla sp. indet. — tree frog Material. — CM 29802. 1 left ilium (3—4 ft). Remarks. — Specimen is assigned to the genus Hyla based on characters given by Holman ( 1962)— ilium with dorsal blade absent; dorsal prominence produced dorsolaterally, well developed, usually round and smooth. The bone is too fragmentary for specific identi- fication, but in comparison with Recent material, it 22 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. II most closely resembles Hyla civysoscelis and H. versicolor in the shape of the dorsal prominence. Eamily Ranidae — Ranid Erogs Ilium with dorsal blade well developed and aris- ing anterior to dorsal prominence, without lateral deflection, and with a deep notch between it and dorsal acetabular expansion (Holman, 1962). Rana sylvatica LeConte — wood frog Material. — CM 29803-29807. 5 right, 4 left ilia (3—4 ft); 6 right, 6 left ilia (4-5 ft); 8 right, 3 left ilia (5-6 ft); 1 right, 3 left ilia (6- 7 ft); 2 left ilia (9-10 ft). Remarks. — Assignment to Ratio sylvatica is based on Holman (1967) who noted that Rana pa- lustris, R. pipiens, and R. sylvatica may be distin- guished from R. catesbeiana and R. clamitans in that the posterodorsal border of the ilial shaft slopes more gently into the dorsal acetabular expansion in the former group than in the latter. The prominence for the origin of the vastus externus head of the triceps femoris muscles is larger, less produced, and less roughened in R. pipiens and R. palustris than it is in R. sylvatica. Based on these criteria, the specimens are assigned to R. sylvatica. In examination of nine Recent specimens of R. sylvatica and two of R. palustris, the above char- acteristics hold in separating the two species. Rana sylvatica is usually encountered in or near moist wooded areas, but it often wanders considerable distances from water (Conant, 1975). Rana catesbeiana Shaw — bullfrog Material. — CM 29808. 1 left ilium (4-5 ft). Remarks. — This specimen is assigned to Rana catesbeiana on the basis of characters discussed above and the observation of Tihen (1954) that R. catesbeiana ilia appear to be highly sculptured. R. catesbeiana is an aquatic frog that prefers larger bodies of water than most other frogs. It is a resi- dent of lakes, ponds, bogs, and sluggish portions of streams (Conant, 1975). Rana sp. indet. Material. — CM 29809-29813. 3 sacral vertebrae (3—4 ft); 2 sa- cral vertebrae (4-5 ft); 2 sacral vertebrae (5-6 ft); 1 sacral ver- tebrae (6-7 ft); 1 sacral vertebra (8-9 ft). Remarks. — These vertebrae have the diplasio- coelous condition with cylindrical rather than ex- panded diapophyses (Holman, 1962) and more closely resemble those of R. sylvatica in the shape of the neural canal and in their small size. Class Reptilia — Reptiles Order Testudines — Turtles Eamily Emydidae — Pond Turtles Graptemys geographica (Le Sueur) — map turtle Material. — CM 29814. 1 proneural bone (4-5 ft). Remarks. — Material is assigned to G. geographi- ca based on the shape, location of shield impres- sions, and surface sculpturing compared to Recent material. Graptemys geographica occurs in large bodies of water, prefering rivers to creeks, and lakes rather than ponds (Conant, 1975). Order Squamata — Snakes and Lizards Suborder Sauria — Lizards Eamily Scincidae — Skinks Eumeces fasciatus (Linnaeus) — five-lined skink Material. — CM 29815-29817. 1 precaudal vertebra (3-4 ft); 3 precaudal vertebrae (4-5 ft); 1 precaudal vertebra (6-7 ft). Remarks. — E. fasciatus has a more backswept neural spine than E. laticeps. Eumeces fasciatus lives in rock piles and decaying debris in or near woods. The habitat is usually damp (Conant, 1975). Suborder Serpentes — Snakes Family Viperidae — Vipers Crotalus horridus Linnaeus — timber rattlesnake Material. — CM 29818-29824. 27 vertebrae (3—4 ft); 50 verte- brae (4-5 ft); 27 vertebrae (5-6 ft); 9 vertebrae (6-7 ft); 4 ver- tebrae (7-8 ft); 2 vertebrae (8-9 ft); 2 vertebrae (9-10 ft). Remarks. — Holman (1963) provides characters, which can be used to differentiate Crotalus from Agkistrodon. In Agkistrodon a distinct pit usually occurs on either side of the cotyle of the centrum. Each of these pits contains one moderately large fossa. In Crotalus the distinct pits are usually ab- sent and the one or more fossae that occur on either side of the cotyle of the centrum are minute. In addition, Crotalus horridus has a lower neural spine than either C. adamanteus or Agkistrodon pisciv- orus (Holman, 1967). The material most closely re- sembles Crotalus horridus in these characters. Cro- talus horridus lives in timbered terrain; usually it is common in second-growth timber where rodents abound (Conant, 1975). Family Colubridae — Colubrid Snakes Subfamily Xenodontinae Members of this subfamily lack hypapophyses on their lumbar vertebrae and have depressed verte- bral neural arches and wide vertebral hemal keels (Holman, 19737>). 1978 GUILDAY ET AL.— BAKER BLUEE CAVE DEPOSIT 23 Heterodon platyrhinos Latreille — eastern hognose snake Material. — CM 29825-29827. 1 precaudal vertebra (5-6 ft); 1 precaudal vertebra (6-7 ft); 1 precaudal vertebra (8-9 ft). Remarks. — The genus Heterodon Latreille may be diagnosed by the following strong characters: hypapophyses absent; vertebrae wider than long through zygapophyses; neural arch flat; neural spine longer than high, usually thickened dorsally with its anterior and posterior borders concave; prezygapophyseal processes large, pointed or trun- cated; epizygapophyseal spines absent; hemal keel very broad and indistinct on many thoracic verte- brae (Holman, 1962). In addition, Holman (1963) was able to differen- tiate between H. platyrhinos and H. nasicns in that in the former, the anterior zygapophyseal faces are more elongate, and in dorsal view, their anterior margins are much flatter than in the latter species. This material compares closest to the characteris- tics of H. platyrhinos. The eastern hognose snake is usually found in sandy areas (Conant, 1975). Subfamily Colubrinae Colubrinae never bear lumbar hypapophyses as do species in the subfamily Natricine, and they lack the combination of the depressed neural arch and the very wide hemal keel of the Xenodontinae (Hol- man, 19736). Diadophis punctatus (Linnaeus) — ringneck snake Material. — CM 29828-29830. 26 precaudal vertebrae (3—4 ft); 19 precaudal vertebrae (4-5 ft); 6 precaudal vertebrae (5-6 ft). Remarks. — Holman ( 1967) provides the following characters by which to distinguish the vertebrae of Diadophis punctatus from Carphophis amoenus: the neural spine is higher, thicker, and usually with more of a posterior overhang in the former than in the latter species. The Baker Bluff Cave fossils more closely resemble those of D. punctatus. Diadophis punctatus is a woodland snake, usu- ally most common in cutover areas that include an abundance of hiding places such as under stones, logs, bark slabs, or in rotting wood. Rocky, wooded hillsides are also favored (Conant, 1975). Carphophis amoenus (Say) — worm snake Material. — CM 29831-29832. 16 precaudal vertebrae (3^ ft); 7 precaudal vertebrae (4-5 ft). Remarks. — Eossils are assigned to C. amoenus based on criteria discussed under Diadophis punc- tatus. Carphophis amoenus is partial to moist earth and disappears deep underground in dry weather (Conant, 1975). Coluber or Masticophis Linnaeus — racer or coachwhip Material. — CM 29833-29837. 1 precaudal vertebra (4-5 ft); 3 precaudal vertebrae (5-6 ft); 7 precaudal vertebrae (6-7 ft); 1 precaudal vertebra (7-8 ft); 3 precaudal vertebrae (8-9 ft). Remarks. — Holman (1962) characterizes the lum- bar vertebrae for the genus Coluber as follows: hy- papophyses absent; vertebrae longer than wide through zygapophyses; neural arch vaulted; neural spine about as high as long, thin, and delicate, not beveled anteriorly; epizygapophyseal spines usual- ly well developed; hemal keel narrow throughout. The vertebrae of Coluber, Masticophis , and Opheodrys are elongate and the neural spine is thin and delicate. But vertebrae of the former two gen- era are larger, the neural spine is higher, and a well- developed epizygapophyseal spine is almost always present. The fossils resemble the characters of the former two genera in this respect. Based on the present geographic ranges of Masticophis and Col- uber it would appear that the fossils represent Col- uber, but I am unable to separate the two genera on vertebral remains. Lampropeltis Eitzinger Remarks. — The vertebrae of Pituophis, Elaphe, and Lampropeltis are very similar but have been separated on the basis of characters described by Holman (1965). Pituophis is distinct from the other two genera in having a higher neural spine with an indented edge. The gentra Elaphe and Lampropel- tis can be separated by the more depressed neural arch of the latter. Lampropeltis triangulum (Lacepede) — eastern milksnake Material. — CM 29838-29843. 8 precaudal vertebrae (3-4 ft); 33 precaudal vertebrae (4-5 ft); II precaudal vertebrae (5-6 ft); 5 precaudal vertebrae (6-7 ft); 1 precaudal vertebra (7-8 ft); 3 precaudal vertebrae (8-9 ft). Remarks. — Eossils are assigned to L. triangulum because the vertebrae possess lower neural spines than those of L. getulus. Also, L. getulus vertebrae are quite robust with thick neural spines and neural arches; the hemal keels and subcentral ridges are usually quite strong with the valleys between them quite deep (Holman, 1965). 24 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 11 Lampropeltis getulus (Linnaeus) — kingsnake Material. — CM 29844-29848. I precaudal vertebra (3-4 ft); 2 precaudal vertebrae (4-5 ft); 6 precaudal vertebrae (5-6 ft); 2 precaudal vertebrae (8-9 ft); 2 precaudal vertebrae (9-10 ft). Remarks. — The fossils have been assigned to L. getulus based on criteria discussed under L. trian- gulum. L. getulus occurs regionally but has not been recorded in the immediate area. It is possible that Recent L. getulus may be collected in the area in the future. Elaphe sp. indet. — rat snake Material. — CM 29849. 3 precaudal vertebrae (6-7 ft). Remarks. — These specimens are assigned to the genus Elaphe sp. indet. because they have a more vaulted neural arch than Lampropeltis but a less vaulted arch than Pituophis (Holman, 1973u). In addition, Pituophis exhibits strongly developed ep- izygapophyseal spines, which are lacking in the fos- sils (Auffenberg, 1963). The material is too frag- mentary to assign to species. Subfamily Natricine Material is assigned to the subfamily Natricine on characters given by Holman (1973/t) — hypapophy- ses on their lumbar vertebrae — and by Auffenberg (1965) — epizygapophyseal spines are usually pres- ent. Natrix sipedon (Linnaeus) — water snake Material. — CM 29850-29852. 2 precaudal vertebrae (3-4 ft); 1 precaudal vertebra (4-5 ft); 3 precaudal vertebrae (5-6 ft). Remarks. — In general, Thamnophis vertebrae are elongate when viewed from above, whereas Natrix vertebrae are almost square (Brattstrom, 1967). Natrix vertebrae tend to have higher neural spines (Holman, 1962). Natrix septemvittata and N. sipedon occur in the area today. N. septemvittata possesses a long, low neural spine and N. sipedon possesses a much higher one (Auffenberg, 1963). The fossils resemble the latter species in this respect. Natrix sp. indet. Material. — CM 29853. 1 precaudal vertebra (5-6 ft). Remarks. — The fossil is too fragmentary for spe- cific identification but the genus was determined using criteria discussed under Natrix sipedon. Thamnophis sauritus (Linnaeus) — ribbon snake Material. — CM 29854-29856. 1 precaudal vertebra (3-4 ft); 5 precaudal vertebrae (4-5 ft); 1 precaudal vertebra (8-9 ft). Remarks. — Criteria for assignment to Thamno- phis was discussed under Natrix sipedon. In T. sauritus the accessory processes are oblique to the longitudinal axis of the centrum; in T. sirtalis they are at right angles (Holman, 1962). Thamnophis sirtalis (Linnaeus) — garter snake Material. — CM 29857-29860. 11 precaudal vertebrae (3-4 ft); 6 precaudal vertebrae (4-5 ft); 2 precaudal vertebrae (5-6 ft); 1 precaudal vertebra (6-7 ft). Remarks. — The fossils were assigned to T. sir- talis based on criteria discussed under T. sauritus. Thamnophis sp. indet. Material. — CM 29861-29866. 16 precaudal vertebrae (3—4 ft); 20 precaudal vertebrae (4-5 ft); 3 precaudal vertebrae (5-6 ft); 2 precaudal vertebrae (6-7 ft); 1 precaudal vertebra (8-9 ft); 1 precaudal vertebra (9-10 ft). Remarks. — The material was too fragmentary for specific identification but could be assigned to ge- nus based on characters discussed under Natrix si- pedon. Discussion All species of reptiles and amphibians from the Baker Bluff Cave local fauna, as far as can be de- termined, live in the area today. Only Lampropeltis getulus, which occurs regionally, is not found in the immediate area. Perhaps the most striking thing about the herpe- tofauna is that there is nothing that strongly indi- cates that the climate or topography was any dif- ferent then than it is today in northeastern Tennessee. Minimum numbers of individuals from each level (Table 1) show no discernible trends in the herpetofauna to indicate that climatic or eco- logical conditions changed markedly from approx- imately 20,000 years BP to approximately 600 years BP (but see Faunal Summary). The Baker Bluff Cave herpetofauna exhibits many similarities to the late Pleistocene herpeto- fauna from Ladds Quarry, Georgia (Holman, 1967). At least 10 species, mostly snakes, from the Ladds Quarry site are also present at Baker Bluff Cave. The Baker Bluff Cave herpetofauna is indicative of four major ecological preferences; a permanent aquatic habitat based on the evidence of Rana ca- tesbeiana and Graptemys geographica\ a marsh- stream border situation indicated by the water snakes Natrix and Thamnophis and the toad Bufo. u’. fowleri', an open, sandy area indicated by Het- erodon platyrhinos\ and a moist woodland habitat 1978 GUILDAY ET AL.— BAKER BLUFF CAVE DEPOSIT 25 where the majority of the identified Colubrinae, The diversity of habitats exhibited by the herpe- Crotcdus horridus, Ainhystoina opacum, and A. tofauna suggests that the fossil remains were prob- macidatiim probably occurred. ably deposited by raptors. AVES— BIRDS Material.— CU 29725, 30176-30181, 30227, 30787-30789. MNI = 169. Remarks. — The majority of bird remains were fragmentary and represent prey items of raptorial birds. Approximately 30 species were identified (Table I), 23% of all vertebrate species from the deposit. One hundred sixty-nine individuals were represented, 7% of the combined numbers of indi- vidual birds and mammals from the deposit; this figure closely approximates the 5% at the Clark’s Cave, Virginia, local fauna, another riverside raptor roost. The riverbluff location of the site is reflected in the numerous remains of aquatic and semiaquatic species — grebe, ducks, mergansers, sandpipers, turnstone, gull, kingfisher — 30% of the recovered avian species, a figure again comparable to the 26% from the Clark’s Cave, Virginia, deposit. Despite the variety of birds, only grouse Tetraon- idae and passenger pigeon, Ectopistes migratoriiis, were present in any appreciable numbers. These two taxa accounted for 48% of all individual birds from the site, a figure identical to that of the Clark’s Cave deposit avifauna, reflecting the collecting bias of raptors. However the two sites did differ signif- icantly in the relative numbers of grouse (all species) and passenger pigeons. Grouse accounted for 34% of all birds at Clark’s Cave, but only 12% at Baker Bluff Cave. Passenger pigeon, on the other hand, accounted for only 14% of all birds at Clark’s Cave, but a high 36% at Baker Bluff, almost a direct reversal in relative numbers. Both sites are late Pleistocene in age, lie in a similar physiographic setting, and both contain boreal elements. It is probable that the higher number of passenger pi- geon remains at Baker Bluff Cave, 2° south of Clark’s Cave, is due to the site’s lower latitude and therefore greater relative freedom from boreal peri- glacial conditions. Spruce grouse {Canachites), sharp-tailed grouse (Pedioecetes), and ptarmigan {Lagopus) were recovered from Clark's Cave in ad- dition to ruffed grouse (Bonasa), the only tetraonid there today; only the ruffed grouse is definitely re- corded from Baker Bluff Cave. However, the distal end of a left tibiotarsus from the 6-7 ft level (CM 30810) may possibly be sharp-tailed grouse {Pedi- oecetes phasianedus). One bird of western affinities, the black-billed magpie (Pica pica Linnaeus), is tentatively identi- fied from the Baker Bluff Cave deposit. The deter- mination is based on a right scapula from the 6-7 ft level (CM 30181). The black-billed magpie in North America is found from southern Alaska to northern New Mexico and east to Kansas. It occurs east casually to Ontario and western Quebec in the north and to the Mississippi River in the south, with occasional strays reported as far east as South Car- olina and Florida (A.O.U. Checklist 1957:376). As- suming that the Baker Bluff Cave specimen was from a resident bird, its presence at the site would complement that of other western or midwestern forms from the deposit such as the thirteened-lined ground squirrel (Spennophdus tridecendineatus) and the badger (Tax idea taxus). The magpie has also been reported from the late Pleistocene raptor deposit at Natural Chimneys, Virginia (a complete humerus, Wetmore, 1962). All other species of birds from the site are of wide geographical distri- bution and are either common migrants or residents of the area today. Chicken (Gallus gallus) remains were found with- in the upper 2 ft of the deposit by Dean and Wilson and are Recent in origin. MAMMALIA— MAMMALS Order Marsupialia — Marsupials Family Didelphidae — American Opossums Didelphis virginiana Linnaeus — Virginia opossum Material. — Dean/Wilson collection: 6 vertebrae, 1 right jugal. Remarks. — The opossum was not a member of the Pleistocene component at Baker Bluff Cave. The six vertebrae and one skull fragment were re- covered from the top 2 ft of the deposit in an ar- chaeological context. Four of the vertebrae were 26 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. II Table 4. — Measurements (in mm) of Sorex cinereus Kerr. Anteroposterior crown length P4-M3 Locality Mean OR SD cv N Recent Pennsylvania* 3.69 3. 6-3. 9 .07 1.89 20 Late Pleistocene New Paris No. 4, Pennsylvania* 3.90 3. 7^. 3 .16 4.09 29 Clark’s Cave, Virginia* 3.98 3. 6^.4 .13 3.40 35 Robinson Cave, Tennessee** 3.90 3.6-4. 5 — — 18 Baker Bluff Cave, Tennessee 3.97 3.78^.36 .06 1.51 28 * From Guilday et al.. 1977. ** From Guilday et a!. 1969. recovered from a depth of less than 6 in. One of the latter, a caudal vertebra, was charred, the only in- dication of fire in the cave. Didelphis is common in the Pleistocene of Flori- da, from Irvingtonian and Rancholabrean sites (Webb, 1974), and has been reported from the Pleis- tocene of Georgia (Ray, 1967). It has not been found in Pleistocene sites farther north, but extended its range northward following the Wisconsinan glacial recession. Opossum remains are common in late prehistoric archaeological sites as far north as west- central West Virginia (Guilday, 1971). The spread of D. virginiano into Pennsylvania and points north was apparently associated with ecological changes brought on by European settlement of the country (Guilday, 1958). Order Insectivora — Insectivores Family Soricidae — Shrews Sorex arcticus Kerr — Arctic shrew Material. — CM 29959-29961. 3 partial left mandibles. MNl = 3. Sorex cinereus Kerr — masked shrew Material. — CM 29962-29967. 52 left, 41 right mandibles; 4 maxillae. MNI = 53. Sorex dispar Batchelder — rock shrew Material. — CM 29968-29970. 7 left, 4 right mandibles. MNI = 9, Sorex fumeus Miller — smoky shrew Material. — CM 29971-29976. 10 left, 21 right mandibles; 3 left, I right maxillae. MNI = 22. Sorex sp., large (S. arcticus or S. fumeus) Material. — CM 29977-29979. 10 left, II right mandibles. MNI = 12. Sorex sp., small (S. cinereus or S. dispar) Material. — CM 29980-29983. 13 left, 20 right mandibles. MNI = 20. Microsorex hoyi (Baird) — pygmy shrew Material. — CM 29956-29958. 6 left, 7 right mandibles; 2 partial skulls. MNI = 9. Cryptotis parva (Say) — least shrew Material. — CM 29954-29955. 3 right mandibles. MNI = 3. Blarina brevicauda (Say) — short-tailed shrew Material.— CM 29746-29753. 130 left, 130 right mandibles; partial skulls, maxillae, isolated teeth. MNI = 149. Remarks. — Seven species of shrews were iden- tified from the deposit. Three of these live in the area today. Blarina brevicauda and Cryptotis parva are common; Sorex fumeus and Sorex longirostris (the latter apparently not represented in the cave fauna) are uncommon (Smith et al., 1974). The four species of shrews from the deposit not represented in the modern fauna {Sorex cinereus, S. dispar, S. arcticus, Microsorex hoyi) are now confined either to higher latitudes or to higher alti- tudes in the Great Smoky Mountains east of the site. There is a change in the relative frequency of various soricids at successive stratigraphic levels in the deposit (Fig. 11) which suggests environmental changes during deposition. Those species requiring cooler conditions, S. arcticus, S. dispar, and Mi- crosorex hoyi, were confined to the upper 2 ft of the undisturbed sequence. Only Blarina brevicau- da, Sorex fumeus, and Sorex cinereus occurred at all levels. Sorex cinereus does not occur at the site today but is present at higher elevations in the Great Smoky Mountains at the same latitude. 1978 GUILDAY ET AL.— BAKER BLUFF CAVE DEPOSIT 27 Table 5. — Measurements (in mm) Microsorex hoyi (Baird), Baker Bluff Cave, Tennessee. Measurement Mean OR N Length of palate — 5. 0-5. 2 2 Maxillary width — 3.7-4. 1 2 p4-M* — 3.4-3. 6 2 M‘-M^ 2.4 — 1 Total length, mandible with incisor 8.0 7. 9-8. 2 3 Total length, dentary 6.2 6. 1-6.2 3 Height, ascending ramus 2.8 2.7-2. 9 7 P4-M3 3.5 3.4-3. 6 6 M.-Ms 2.9 2. 7-3.0 8 Given the fragmentary nature of the collection the southeastern shrew, S. longirostris , may be rep- resented in the referred 5. cinereus material from the deposit. The entire cave collection is identified as S. cf. cinereus, however, because the mandibles average larger than Recent Pennsylvania specimens and are comparable in size to late Pleistocene ma- terial from New Paris No. 4 and Clark’s Cave (Ta- ble 4). Mandibular measurements show no strati- graphic size shifts which suggest the absence of the smaller S. longirostris in the lower levels of the de- posit where boreal conditions were apparently less intense. The third unicuspid was larger than the fourth, a character distinguishing S. cinereus from S. longirostris, in the one example complete enough for observation (CM 29964, 6 ft level). The presence of S. cinereus in the lower more temperate levels of the deposit suggests that climatic conditions at the time of lower-level deposition, although milder than upper-level conditions that supported a larger number of boreal species, was cooler than at pres- ent. Remains of Clethrionomys gapperi and one Phenacomys intermedins from the lowest level sup- port this conclusion. The presence of Cryptotis parvo, a temperate species, in the upper levels seems anomalous in the boreal context suggested by the associated fauna. However it has also been reported from the late Wisconsinan Robinson Cave, Tennessee, and Eagle Cave, West Virginia, local faunas, both of which suggest cooler conditions, so it may have persisted in the richer boreal/temperate faunal mix of the late Wisconsinan periglacial environment. The least shrew was not present in either the New Paris No. 4, Pennsylvania, or the Clark’s Cave, Virginia, sor- icid faunas, however, and its late Wisconsinan dis- tribution is yet to be determined. The three man- dibles from the upper levels of Baker Bluff Cave may have filtered down from Recent levels through undetected oblique deposition or animal burrowing. The absence of the water shrew, Sorex palustris, a semiaquatic soricine of boreal affinities, was un- expected. Its remains have been recovered from the late Pleistocene Robinson Cave deposit, 256 km west of Baker Bluff. But 5. palustris remains are rare in eastern late Pleistocene deposits (New Paris No. 4, 0.9%; Clark’s Cave, 3%; Robinson Cave, 0.8% of all soricids) and their absence from the Ba- ker Bluff fauna may be a matter of chance. The Blarina hrevicauda collection from Baker Bluff Cave (Figs. 6 and 7) presents an interesting picture of the coexistence of two size-forms throughout a portion of the depositional time span, confirming a concept developed by Graham and Semken (1976). Blarina hrevicauda, the most com- mon shrew in temperate eastern North America to- day, is also the commonest soricid at all levels in the Baker Bluff deposit. Its remains comprise 41.6% of all shrews from the 3 ft level, increasing relatively (but not actually) to a high of 87.7% at the lowest cave level as a result of the lessening numbers of i'o/'e.v, Microsorex, and Cryptotis (Fig. 11). There are two forms of Blarina hrevicauda pres- ent in the Baker Bluff Cave deposit (Fig. 6). A small form, comparable in size to the modern mid-Ap- palachian B. h. kirtlandi, occurred at all levels; in the lower 3 ft of the deposit it was the only form represented. However, the upper level sample (3 to 7 ft) is bimodal, with modal values of P4-M3 of 5.92 mm (B. h. cf. kirtlandi) and 6.59 mm. The latter value is comparable to B. h. cf. hrevicauda from the late Pleistocene Clark’s Cave deposit (6.56 mm), somewhat larger than that of modern Min- nesota specimens of B. h. hrevicauda (6.29 mm, N = 20, Guilday et al., 1977). An additional expression of variability is the spread of the observed range expressed as the per- cent of increase from the smallest to the largest val- ues. That figure, for the upper levels of the deposit, is 21.4%, comparable to the bimodal samples from the New Paris Sinkhole No. 4, Pennsylvania, 20% and Meyer Cave, Illinois, 25%. In the lower levels the figure shrinks to 11.1%, a figure comparable to the unimodal Clark’s Cave sample. A small form of short-tailed shrew, comparable in size to modern mid- Appalachian material, oc- curred at the earliest level represented at Baker Bluff. At the 6-7 ft level, however, an influx of larger stock took place and both forms apparently 28 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 11 BLARINA BREVICAUDA (SAY) MEASURMENTS ( IN MM) P4-M3,Mi-M3, RECENT NORTHEASTERN TENNESSEE AND BAKER BLUFF CAVE , TENNESSEE. HISTOGRAMS ARRANGED BY STRATIGRAPHIC LEVELS . P4-M3 M^-M3 Fig. 6. — Histograms of mandibular measurements, Blarina hrevicuuda (Say), arranged by stratigraphic levels illustrating presence of two size groups, corresponding to RecentB. h. kirtlandi (smaller) andB. b. hrevicuuda (larger). Baker Bluff Cave, Sullivan County, Tennessee. 1978 GUILDAY ET AL.— BAKER BLUFF CAVE DEPOSIT 29 Fig. 7. — Scatter diagrams of mandibular measurements Blarina brevicauda (Say), Baker Bluff Cave, Sullivan County, Tennessee. Com- pare with Fig. 6. 30 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 11 existed in the area during the 3-7 ft depositional period. The top 3 ft of the deposit was destroyed and with it the Pleistocene/Holocene transition. The influx of the larger, presumably northern B. h. cf. hrevicauda at the 6-7 ft level coincides with an in- crease in relative numbers of boreal voles. Climatic change is inferred at this point and it would appear that both forms of Blarina were coexisting in the area during an episode that was cooler than lower level times when only the smaller form existed at the site. Graham and Semken (1976:443) speculate that "Sympatry of these phena [size-forms] during the Pleistocene suggested a more equable climate ex- isted during glacial times than at present and that sympatric phena of Blarina coexisted in partitioned niches that presently are not defined. Post-glacial continental climates subsequently divided the . . . phena into their existing parapatric distributions." They suggest that "Coexistence of these phena in the same deposit without apparent interbreeding suggest a specific rather than a subspecific relation- ship." This assumes that there were no ecological barriers. B. b. brevicauda and B. b. kirtlandi, the modern equivalents of two of these late Pleistocene phena, and so identified by Graham and Semken, are currently considered to be subspecies. Current studies underway by J. R. Choate and H. H. Gen- oways on the evolution of Blarina may clarify the situation. Family Talpidae — Moles Condylura cristata (Linnaeus) — star-nosed mole Material.— CM 30149-30150. 3 humeri. MNI = 3. Scalopus aquaticus (Linnaeus) — eastern mole Material. — CM 30151-30157. 3 humeri, 5 mandibles, skull fragments and isolated molars. MNI = 12. Parascalops brewed (Bachman) — hairy-tailed mole Material. — CM 30158-30165. 14 humeri, 15 mandibles, as- sorted limb bones, skull fragments and isolated molars. MNI = 17. Remarks. — The eastern mole, Scalopus aquati- ciis, is common in the area today. The hairy-tailed mole, Parascalops breweri, has been reported from Bristol, Sullivan County, 50 km NE of the site in the Ridge and Valley province (Smith et al., 1974), and may occur at or near the site. The star-nosed mole, Condylura cristata. does not occur in the Ridge and Valley province at this latitude but does occur east of the site in the Great Smoky Moun- tains. This species is the most demanding in its eco- logical requirements, preferring boggy or mucky areas. It is a relatively weak burrower and often semiaquatic in its habits. Remains of the star-nosed mole were confined to the upper 3-5 ft levels. Moles accounted for 1.6% of all mammals from the site. They were also scarce, 0.9% of the fauna, in the Clark’s Cave local fauna, 307 km NE of Ba- ker Bluff. Both accumulations are old raptor roosts, so the selection bias was much the same at both sites. But the relative percentages of the three species of talpids were quite different. The Con- dylnralScalopasIParascalops composition of the Baker Bluff Cave mole fauna (all levels combined) was 9%-54%-37%. At Clark’s Cave it was 50%-4%- 46%. The striking difference between the relative numbers of Condylura and Scalopus suggests that some factor other than relative availability to rap- tors was responsible for the discrepancy. The to- pography of the two sites is much the same but there is some evidence that the Cowpasture River valley, in which Clark’s Cave is located, may have been relatively wetter (number and variety of ranids and semiaquatic birds) so that the high number of Condylura at that site can be attributed to local eco- logical factors. At the latitude of Baker Bluff, Condylura and Parascalops are at the southern limits of their mod- ern ranges. The fact ihdd Parascalops was the com- monest of the three species in the deposit is in ac- cordance with the overall boreal aspect of the recovered fauna. Parascalops breweri has also been recorded from Robinson Cave, Overton County, Tennessee, 256 km west of Baker Bluff, well south of its present mid-continental range. Order Chiroptera — Bats Family Vespertilionidae — Evening Bats Myotis sp. Kaup — little brown bats Material. — CM 30143-30148. Partial mandibles and maxillae. MNI = 18. Pipistrellus subflavus (F. Cuvier) — eastern pipistrelle Material. — CM 30136. Partial mandibles; maxilla. MNI = 7. Eptesicus fuscus (Palisot de Beauvois) — big brown bat Material. — CM 30138-30142. Partial mandibles; maxilla; par- tial skull; isolated teeth. MNI = 19. 1978 GUILDAY ET AL.— BAKER BLUEE CAVE DEPOSIT 31 9 , , , . '9 centimeters Fig. 8. — Dasypiis hellus Simpson, dermal scutes: a) CM 29524; b) CM 29536; c) CM 29531, Tapiriis cf. veroensis Sellards: d) CM 29522, left P‘, occlusal view, anterior to left. Baker Bluff Cave, Sullivan County, Tennessee. cf. Nycticeius humeralis (Rafinesque) — evening bat Material. — CM 30137. Partial mandible. MNI = 1. Plecotus sp. E. Geoffrey Saint Hilaire — big-eared bat Material. — CM 30143. Partial mandible. MNI = 1. Remarks. — Baker Bluff Cave is too small and ex- posed to support a large bat colony and there is no other cave in the immediate vicinity capable of doing so. Bat remains were relatively uncommon, only 46 individuals, or 2.4%, of the total mammalian assemblage. By way of contrast bats accounted for 36% of the mammalian assemblage from Clark’s Cave, Virginia, and 74% from Robinson Cave, Ten- nessee. The high number of bats from the Clark’s Cave local fauna was due to raptor predation on a nearby cave colony; the bulk of the Robinson Cave bat remains (87% Eptesicus fuscus) resulted from natural mortality of the resident bat colony. The big brown bat (Eptesicus fuscus) was the commonest bat from the Baker Bluff fossil fauna — 19 individuals, 41% of all bats. This is a reflection of the shallowness of the cave, a condition differ- entially favoring this large hardy species. Although at least 18 little brown bats (My at is spp.) were present, only 11 mandibles were com- plete enough to measure. At least two size classes were represented. The alveolar length C-M3 of two specimens measured 5.3 mm and 5.5 mm. Nine ad- ditional mandibles measured C-Mg as follows: 6.2; 6.3; 6.4; 6.5; 6.5; 6.5; 6.5; 6.7; 6.8 mm. The smaller group lies within the modern M. leibiilaustroripar- iuslsodalisllucifugus size range, the larger series within the modern M. keeiiiilgrisescens range (Guil- day et al., 1977, Eig. 16). All species of bats recovered from the site are found in Tennessee today (Graves and Harvey, 1974). Only Plecotus rafineseptii has been reported from the state today, although a relict colony of P. townsendii, the western big-eared bat, in the central Appalachians northeast of the site (Handley, 1959), implies a former range continuum that may have included eastern Tennessee. The Baker Bluff spec- imen, a partial mandible, cannot be identified to species. Order Edentata — Edentates Eamily Dasypodidae — Armadillos Dasypus bellus Simpson — “beautiful” armadillo Material.— CM 29506, 29515, 29523, 29524, 29531, 29535, 29536. 15 fragmentary scutes (Fig. 8). MNI = 1. Remarks. — Fifteen fragmentary dermal scutes were recovered, scattered throughout the deposit, but no other skeletal remains of this extinct arma- dillo were recognized (Fig. 8). This is due to the poor condition of all large bones from the deposit and the characteristic appearance of even fragmen- tary armadillo scutes. Armadillo remains are found 32 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 11 in Appalachian late Pleistocene sites as far north as West Virginia (Guilday and McCrady, 1966). Its presence in such sites is an indication of milder winter extremes despite the presence of so many boreal forms in the deposit. Armadillo remains have also been reported from Robinson Cave, Overton County, Tennessee. Order Lagomorpha— Rabbits, Hares, and Pikas Family Leporidae — Rabbits and Hares Sylvilagus, sp. and/or Lepus, sp. — cottontail rabbit or snowshoe hare Material. — CM 30007-30013. Isolated teeth, fragmentary maxillae, fragmentary mandibles. MNI = 95. Remarks. — Leporid remains were common at all stratigraphic levels (Table 1). Preservation was so poor that identification beyond family level was not feasible and minimum number of individuals is based on counts of isolated teeth. At least three species may be represented — Syl- vilagus floridanus, the eastern cottontail, Sylvilagus transitionalis, the New England cottontail, and Lepus americanus, the snowshoe hare. The Baker Bluff leporid sample consists of remains of medium- sized animals, of 5. floridanusitransitionalis size, too small for the Recent Appalachian subspecies of snowshoe hare, L. a. virginianus. But in the light of the presence of other northern forms in the de- posit, a small late Pleistocene form of snowshoe hare may be present (see Guilday et al., 1964, for discussion of size relationships). Kellogg (1939) re- ports hearsay evidence of snowshoe hare from the Great Smokies in eastern Tennessee. S. floridanus is the only leporid now in the cave area. S. transitionalis, a species more characteristic of a northern hardwood forest situation, has been taken as far soutn as northeastern Alabama (How- ell, 1921) and its distribution prior to colonial de- forestation may have included the cave area. 5. acjuaticus, the swamp rabbit, occurs in swampy sit- uations along the Mississippi and Tennessee river valleys, farther west in the state (Kellogg, 1939), but the fossil remains are too small to be those of S. aquaticus. Minimum number of individuals was based upon upper incisors and upper second premolars per level. Crown width of 138 upper incisors produced a unimodal curve skewed to the left with a high coefficient of variation — 16.12. Based upon this measurement a single size-population with a large relative percentage of juvenile animals is most prob- able. Order Rodentia — Rodents Family Sciuridae — Squirrels Tamias striatus (Linnaeus) — eastern chipmunk Material. — CM 30083-30096. 3 left, 4 right partial maxillae, 1 1 left, 9 right partial mandibles: 25 M' or M^, 1 M^, 2 P\ 57 M, or M.2, 4 M3. MNI = 27. Remarks. — Remains of the eastern chipmunk, 19.8% of al) sciurids, were exceeded only by those of the thirteen-lined ground squirrel (Spermophilus tridecemlineatus). Measurements (Table 6) indicate that the Baker Bluff T. striatus were large-sized individuals, with a P4-M3 length averaging 10% larger than modern Pennsylvania material and some 6% larger than T. s. pipilans, the largest living subspecies. But the Baker Bluff sample averages 1 1% smaller in length of P4-M3 than the one extant mandible of the large extinct T. aristus from the Pleistocene of Georgia (USNM 23321, Ray, 1965). The largest of the Baker Bluff measurements, however, is only 2% smaller than that of T. aristus. We are dealing with small samples (seven measurements from Baker Bluff, one of T. aristus) and additional material may dem- onstrate a size-continuum. More material is needed to show whether T. aristus is a valid species or one based upon very large specimens of a large late Pleistocene form of 7. striatus. Ray (1965) discuss- es this possibility and we concur with his opinions. T. striatus increases in body size with decreasing latitude today, a condition which apparently held in late Pleistocene times as well even though all late Pleistocene T. striatus appear to have been larger than modern counterparts in the same latitude. Measurements P4-M3 of late Pleistocene material show a size increase from Pennsylvania through Virgina and Tennessee (Table 6). The eastern chipmunk, a woodland form, and the thirteen-lined ground squirrel {Spermophilus tride- cemlineatus), a prairie form, were both common in the Baker Bluff local fauna, suggesting a regional mosaic of prairie and woodland. Numbers of the thirteen-lined ground squirrel diminished with in- creasing depth in the deposit while those of the east- ern chipmunk remained relatively constant. This suggests that woodland was present throughout de- positional times, although other evidence from the site suggests that the percent of woodland to grass- land varied. Eutamias cf. minimus (Bachman) — least chipmunk Material.— CM 30114-30115. 2 left M, or Mj; CM 30116, par- tial left mandible with P4-M3. MNI = 3. 1978 GUILDAY ET AL.— BAKER BLUEE CAVE DEPOSIT 33 Table 6. — Measurements (in nun) o/Tamias striatus (Linnaeus) and Tamias aristus Ray. Sample data Mean OR SD cv N Recent Tamias striatus Pennsylvania* Alveolar length P4-M3 6.58 6.20-6.98 .24 3.64 17 New Paris No. 2, Pennsylvania 6.28 5.80-6.80 .18 2.86 114 (1,875 years B.P.) Tamias striatus pipilans** 6.16 6.30-7.40 .30 4.40 19 Late Pleistocene Tamias striatus Baker Bluff, Tennessee 7.24 6.79-7.95 7 Robinson Cave, Tennessee 7.50 7.30-7.80 — — 2 Clark’s Cave, Virginia*** 6.81 6.50-7.30 .21 3.08 28 Hartman’s Cave, Pennsylvania** 6.90 6.20-7.70 .42 6.11 9 New Paris No. 4, Pennsylvania 6.74 6.30-7.10 .20 2.96 30 Tamias aristus** 8.10 — — — 1 Recent Tamias striatus Pennsylvania* 1 1.47 Occlusal length M, 1.30-1,50 .05 3.49 16 Tamias striatus pipilans** 1.58 1.49-1.67 — — 2 Late Pleistocene Tamias striatus Baker Bluff, Tennessee 1.61 1.50-1,70 ,06 3,88 13 Ladds, Georgia** 1.54 1.49-1.60 — — 2 Tamias aristus Ladds, Georgia** 2.12 — — — 1 Recent Tamias striatus Pennsylvania* 1.61 Occlusal width M, 1.45-1.75 .09 5.33 16 Tamias striatus pipilans** 1.71 1.70-1.73 — — 2 Late Pleistocene Tamias striatus Baker Bluff, Tennessee 1.71 1.50-1.80 .07 4.48 13 Ladds, Georgia** 1.73 1.73-1.74 — — 2 Tamias aristus Ladds, Georgia** 2.29 — — — 1 Recent Tamias striatus Pennsylvania* 1 1.63 Occlusal length M2 1.45-1.75 .08 4.88 16 Tamias striatus pipilans** 1.89 1,81-1.91 — — 2 Late Pleistocene Tamias striatus Baker Bluff, Tennessee 1.68 1.45-1.84 16 Ladds, Georgia** 1.86 1.82-1.90 — — 2 Tamias aristus Ladds, Georgia** 2.42 — — — 1 Recent Tamias striatus Pennsylvania* 1.70 Occlusal width M2 1.55-1.75 .07 4.11 16 Tamias striatus pipilans** 1.89 1.82-1.96 — — 2 34 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. II Table 6. — Continued. Sample data Mean OR SD CV N Late Pleistocene Tamias striatus Baker Bluff, Tennessee Ladds, Georgia** Tamias aristas 1.86 1.64-1.94 _ _ 16 1.79 1.78-1.81 _ _ 2 Ladds, Georgia** 1 1 1 CO Late Pleistocene Occlusal length M, or Ms Tamias striatus Baker Bluff, Tennessee 1.62' 1.45-1.84 .14 8.66 45 Late Pleistocene Occlusal width M, or Mj Tamias striatus Baker Bluff, Tennessee 1.68 1.45-2.03 .13 7.70 46 Recent Alveolar length P^-M^ Tamias striatus pipilans** Late Pleistocene 6.55 6.50-6.60 _ _ 2 Tamias striatus Baker Bluff, Tennessee Tamias aristas** 6.30 — _ _ 1 7.25 — _ _ 1 Late Pleistocene Occlusal length M' Tamias striatus Baker Bluff, Tennessee 1.55 1.40-1.50 _ __ 4 Late Pleistocene Occlusal width M' Tamias striatus Baker Bluff, Tennessee 1.86 1.80-1.90 _ _ 4 Late Pleistocene Occlusal length Tamias striatus Baker Bluff, Tennessee 1.50 — — — 3 Late Pleistocene Occlusal width M'^ Tamias striatus Baker Bluff, Tennessee 1.90 1.80-2.00 — — 3 Late Pleistocene Occlusal length M‘ or Tamias striatus Baker Bluff, Tennessee 1.66 1.40-1.80 .11 6.36 24 Late Pleistocene Occlusal width M‘ or Tamias striatus Baker Bluff, Tennessee 1.89 1.60-2.10 .12 6.15 24 • CMNH Recent mammal collection: 25091, 25104, 25105, 25107, 25109-25116, 25118, 25120, 25123, 25124, 25132. «• Ray, 1965. *** Guilday et al.. 1977. 1978 GUILDAY ET AL.— BAKER BLUFF CAVE DEPOSIT 35 Remarks. — The least chipmunk occurs today in the American West and subarctic western Canada east to Lake Superior, Lake Huron, and south- western Hudson Bay, where it prefers open to brushy coniferous forest situations (Banfield, 1974). Remains have been reported from two other late Pleistocene cave deposits in the Appalachians, Back Creek Cave No. 2 and Clark's Cave, Virginia. Baker Bluff Cave marks the southernmost record for the least chipmunk in eastern North America. Rare in the deposit, and confined to the upper 3- 6 ft levels, presence of the least chipmunk in con- junction with high numbers of Spennophilus tride- cemlineatus from those levels suggests an open, predominantly coniferous parkland environment. Measurements of CM 30116 are as follows: P4-M3, 5.2 mm; P4, .97 mm x .97 mm; M,, 1.3 mm x 1.4 mm; M2, 1.4 x 1.4 mm; M3, 1.5 x 1.4 mm. Marmota monax (Linnaeus) — woodchuck Material. — CM 30166-30173. 3 skull fragments; 3 left, 1 right mandibles; 15 upper incisors; 8 dP^; 12 P'‘; 16 P^ 22 M' or M^; 1 M^; 17 M^; 14 lower incisors; 7 dP^; 8 P4; 26 M, or M2; I M,; 16 M3. MNI = 23. Remarks. — Remains of this large hibernating ground squirrel are present in all Holocene and Pleistocene Appalachian cave deposits of any size, and they are common in the Baker Bluff fauna — at least 23 individuals, based upon dentitions, 17% of all sciurids from the site. Most of the remains may represent individuals who died of natural causes as the deposit is within a few meters of the present entrance. All but ju- venile woodchucks are too large for most raptors and their diurnal habits protect them from owls. The extent to which woodchuck remains may build up in cave deposits is dramatically illustrated by the early Holocene fissure, Meyer Cave, Illinois (Par- malee, 1967). At least 597 individuals were present in that deposit, 82% of all sciurids. Woodchuck burrowing in such a small deposit as Baker Bluff Cave may have affected stratigraphic integrity, but there is no field evidence for this. Spermophilus tridecemlineatus (Mitchell) — thirteen-lined ground squirrel Material. — CM 30073-30082. 1 left, 3 right partial maxillae; 5 left, 2 right partial mandibles; 2 dP'*; 1 P'“; 24 P^; 2 M‘; 3 M^; 33 M' or M^ 10 M^; 7 P4; 2 M,; 36 M, or Mj; 22 M3. MNI = 28. Remarks. — The thirteen-lined ground squirrel, Spermophilus tridecemlineatus , was the common- est squirrel, 20% of all sciurids, from the deposit. This is a high figure for an eastern late Pleistocene cave fauna compared with 9.6% New Paris No. 4, Pennsylvania, 4.3% Clark’s Cave, Virginia, and 7.6% of the sciurids from Robinson Cave, Tennes- see. The thirteen-lined ground squirrel is a prairie species, occurring in the American Midlands from Wyoming to western Ohio and from Manitoba to Texas (Hall and Kelson, 1959). It does not now live in the forested East. It is not found south of the Ohio River except in those few instances where it has been introduced locally by man (Doutt et al., 1973). However during at least late Wisconsinan times it was distributed throughout the mid-Appa- lachian region. It has been reported from cave sites in Pennsylvania, West Virginia, Virginia, Ken- tucky, and Tennessee (Guilday et al., 1977). The presence of thirteen-lined ground squirrels at such sites, in association with boreal woodland mam- mals, suggests a mixed woodland/grassland eco- type. At Welsh Cave, Kentucky, 268 km NW of Baker Bluff Cave, '^C dated 12,950 ± 950 BP, 5. tridecemlineatus dominated the small mammal fau- na, and, other than a single red squirrel (Tamias- ciurus hudsonicus), was the only sciurid from the site. It was found with remains of Ta.xidea ta.xus, Ursus arctos, Mammuthus, and Ec/uus, as well as with boreal forest rodents and insectivores, sug- gesting a prairie parkland periglacial environment in what is now a region of deciduous forest. The relative percent of thirteen-lined ground squirrel to chipmunk (Tamias striatus) is instruc- tive. Both are diurnal, conspicuous ground squirrels of comparable size; both are susceptible to preda- tion by raptors. They occupy separate ecological niches, however. S. tridecemlineatus is a prairie form, whereas T. striatus inhabits woodland or for- est edge. Their changing relative numbers, strati- graphically or geographically, should provide a rel- ative index to the type of ground cover present at the time of deposition. 5. tridecemlineatus comprised 23% of the ground squirrels (Spermophilus, Tamias) from New Paris No. 4, Pennsylvania, 17% from Clark's Cave, Vir- ginia, 33% from Robinson Cave, Tennessee, 51% at Baker Bluff Cave, and 100% from Welsh Cave, Kentucky. It is apparent that variation in the Sper- mophilusITamias ratio does occur but not enough paleofaunas have been analyzed to recognize any geographic pattern that might reflect paleoenviron- mental trends. The data now at hand suggest that Spermophilus is relatively more common in south- 36 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 11 Table 7. — Measurements (in nun) o/' Spermophilus tridecemlineatus. Baker Bluff Cave . Tennessee. Mean OR SD cv N Mean OR SD cv N Occlusal length M' Occlusal width M 1 1.64 1.40-1.70 — — 5 2.30 2.10-2.40 — — 5 Occlusal length M [2 Occlusal width M 2 1.69 1.50-1.80 — — 7 2.30 2.00-2.40 - — 7 Occlusal length M‘ or M^ Occlusal width M' or M^ 1.67 1.40-1.80 .13 7.68 24 2.20 1.70-2.40 .16 7.30 24 Occlusal length M, Occlusal width M 1 1.55 1.40-1.60 — — 5 1.92 1.84-1.94 — — 5 Occlusal length M ^2 Occlusal width M 2 1.68 1.40-1.80 — — 3 2.16 1.90-2.30 — — 3 Occlusal length M, or Mj Occlusal width M, 01 M2 1.60 1.40-1.80 .13 8.69 20 1.98 1.50-2.10 .17 8.57 20 Alveolar length P4-! M3 8.60 — — — 1 ern (Tennessee) and western (Kentucky) late Pleis- tocene sites. There is a stratigraphic change in the ground squirrel fauna from Baker Bluff Cave itself (Eig. 12). At levels 0-7 ft 5. tridecemlineatus was the dominant species — 58.1%. In the lower levels, 7-10 ft, the relative numbers of S. tridecemlineatus and Tamias striatus shifted and T. striatus became the dominant species — 66%. This suggests a trend to- ward a denser woodland in the lower levels, an impression heightened by an accompanying shift in the relative numbers of large tree squirrels; the tem- perate deciduous forest Sciurus became common in the lower levels. In the upper levels of the deposit it was partially replaced by the red squirrel {Tam- iasciurus hudsonicus), a coniferous/northern hard- wood species that does not require as dense a forest habitat. Spermophilus (Ictomys) sp. has been reported from Haile XIV A, a Sangamonian/Wisconsinan fis- sure site in Elorida, accompanied by a predomi- nantly xeric fauna. No northern species were noted. It was suggested that either S. tridecemlineatus or S. mexicanus might be represented and that they represented a western element in this Elorida fauna (Martin, 1974). Northern species dominate all of the eastern late Wisconsinan faunas studied so far from the mid-Appalachians. The accompanying Sper- mophilus from these sites is believed to be the northern S. tridecemlineatus, which ranges north to Manitoba today, rather than S. mexicanus, which ranges only as far north as Texas and northern Mex- ico. Both species are larger than S. spilosoma (Ta- ble 7). Sciurus carolinensis Gmelin — gray squirrel Material. — CM 30106-30113. 1 right mandible with P4-M3; 1 left maxilla with P'*-P^; 1 right dP^; 4 left, 2 right P^; 19 left, 12 right M' or M'^; 9 left, 4 right M^; 3 P4; 12 left, 9 right M, or M2; 4 left, 2 right M3. MNl = 17. Remarks. — The gray squirrel and the larger fox squirrel (Sciurus niger) occur in the area at the pres- ent time. S. niger may have been present in the fossil deposit, but there is no indication of it. A right mandible from the Dean/Wilson collection (CM 30106) agrees in size with S. carolinensis, and a maxilla (CM 30107) bears the diagnostic P'^ The remainder of the collection, 81 isolated premolars and molars, are referred to 5. carolinensis on the basis of their size and the lack of any apparent bi- modality (Table 8). Gray squirrel remains in the deposit became more common relative to those of red squirrel, Tamias- ciurus hudsonicus, with increasing depth (Eig. 12). In the upper levels, 0-7 ft, the percentage of Sciurus Table 8. — Length and width of molars (in mm) of Sciurus cf. carolinensis Gmelin, Baker Bluff Cave, Sullivan Co., Tennessee. Measurements Mean OR SD cv N Crown length M’ ' or M^ 2.49 2.10-2.70 .15 6.02 26 Crown length M, 1 or M2 2.54 2.30-2.80 .20 7.87 18 Crown width M‘ or M^ 2.81 2.40-3.10 .16 5.69 26 Crown width M, or M2 2.75 2.50-2.90 .13 4.69 16 1978 GUILDAY ET AL.— BAKER BLUFF CAVE DEPOSIT 37 Table 9. — Measurements (in mm) o/ Tamiasciurus hudsonicus (Erxlehen). Age and locality Mean OR SD cv N Occlusal length P4-M3 Recent Pennsylvania* 7.39 6.80-7.80 — — 33 Natishquan River, Quebec*, and Hamilton River, Labrador* 7.66 7.30-8.00 — — 20 Hudson Bay, Quebec* 7.84 7.60-8.20 — — 15 Moorhead, Minnesota* 7.92 7.20-8.50 — — 1 1 Aklavik NWT, Seward, Alaska* 8.20 8.00-8.40 — — 9 Late Pleistocene New Paris No. 4, Pennsylvania* 8.20 8.00-8.30 — — 4 Alveolar length P4-M.1 Late Pleistocene Clark’s Cave, Virginia* 8.68 8.15-9.20 0.30 3.46 20 Baker Bluff Cave, Tennessee 8.48 8.24-8.73 — — 4 Robinson Cave, Tennessee** 8.70 8.65-8.70 — — 2 Occlusal length M, or M2 Late Pleistocene Baker Bluff Cave, Tennessee 2.01 1.84-2.32 0.12 5.85 30 Occlusal width M, or Mj Late Pleistocene Baker Bluff Cave, Tennessee 2.21 1.94-2.72 0.21 9.51 29 Occlusal length M' or M^ Late Pleistocene Baker Bluff Cave, Tennessee 1.95 1.64-2.32 0.14 7.16 31 Occlusal width M' or M^ Late Pleistocene Baker Bluff Cave, Tenn. 2.36 2.04-2.72 0.16 7.85 30 * Measurements from Guiiday el al.. 1977. ** Measurements from Guiiday et al., I%9. to Tamiasciurus was 40%, but increased to 70% in levels 7-10 ft. The gray squirrel is a more temperate species and its relative increase in numbers with depth suggests a more temperate environment dur- ing lower-level depositional times. The presence of gray squirrel throughout the stratigraphic column is a reflection of the low latitude of the cave and its distance from periglacial effects. S. Caroline nsis was scarce at Clark’s Cave, Virginia, 2° latitude father north, where the percentage of 5. carolinen- sis to T. hudsonicus was only 1.1%. Two degrees farther north of Clark’s Cave, at New Paris No. 4, Pennsylvania, only T. hudsonicus was present. Martin and Webb (1974) point out that 5. niger is absent from all but late Pleistocene/early Holo- cene deposits in Florida. They believe that S. niger dates back only 8,000 years in Florida and speculate that it may have been a northern invader. The ap- parent absence of S. niger from Baker Bluff Cave in eastern Tennessee during the late Pleistocene does not bear this out. Undated S. niger remains, two complete skulls covered with thin layers of flowstone (CM 8050-8051), were recovered from Robinson Cave in northcentral Tennessee (Guiiday et al., 1969). Tamiasciurus hudsonicus (Erxleben) — red squirrel Material.— CM 30097-30105. 2 incisors; 4 P^; 38 M' or 8 M^; 1 P4; 34 M, or M2; 10 Mj; 2 left, 4 right partial mandibles; 1 humerus. MNI = 18. Remarks. — A form of red squirrel larger than T. h. loquax, the subspecies now occupying the south- ern and central Appalachians, but similar in size to red squirrels from other Appalachian cave deposits of late Pleistocene age (New Paris No. 4, Clark’s 38 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 11 Cave) is indicated by dental and mandibular mea- surements (Table 9). The red squirrel does not occur in the area, nor in the state, outside of the Great Smokies where it occurs in spruce/fir/hemlock situations above 1,000 m (Kellogg, 1939; Smith et al., 1974). Remains of both gray squirrel (Sciurus caroUnen- sis) and red squirrel were present at all levels in the deposit, but red squirrel was relatively more com- mon in the upper levels, reinforcing the ecological shift indicated by the presence of other boreal forms in those same upper levels (Fig. 12). Only 18 individuals were indicated from the site, 13% of all sciurids. This is one-half of the relative number of red squirrel represented from the Clark's Cave local fauna, 307 km NE. But, if we include numbers of both Sciurus and Tamiasciurus from Baker Bluff Cave, the aggregate percentage of large diurnal tree squirrels is about the same at both sites. The percentage of large diurnal tree squirrels {Sciurus, Tamiasciurus) relative to small diurnal ground squirrels (Tamias, Eutamias, Spermophilus) differs from the two sites (38% arboreal. Baker Bluff Cave, 47% arboreal, Clark’s Cave, Virginia). Assuming that the squirrels from both sites were raptor prey, then raptors at Clark’s Cave captured a larger proportion of arboreal squirrels, suggesting that forest cover was denser at Clark’s Cave during depositional times than it was at Baker Bluff Cave. At Baker Bluff Cave the percentage of diurnal ground squirrels relative to that of diurnal arboreal squirrels dropped somewhat in the lower levels from 70% in levels 3-7 ft to 55% in levels 7-10 ft, suggesting a denser forest cover in lower level de- positional times. The fact that red squirrel also be- came relatively less common in the lower levels suggests a warmer deciduous woodland situation. Genus Glaucomys Thomas — flying squirrels Glaucomys sabrinus (Shaw) — northern flying squirrel Material. — CM 30117-30128. 6 left, 2 right partial mandibles; 2 right maxillae, 67 isolated molars and premolars. MNI = 13. Glaucomys volans (Linnaeus) — southern flying squirrel Material. — CM 30130, 30132-30135. 2 left, 1 right partial max- illae; 5 M‘ or M^; 5 M, or M2. MNI = 7. Glaucomys, sp. (cf. sabrinus or volans) Material. — CM 30129, 30131. 2 partial maxillae; 1 P"*; 2 M' or M2. Remarks. — Flying squirrels comprised 15% of all sciurids in the Baker Bluff deposit (19% at New Paris No. 4, Pennsylvania, 43% at Clark’s Cave, Virginia, 46% at Robinson Cave, Tennessee, 63% at Natural Chimneys, Virginia) a relatively low fig- ure for an Appalachian cave deposit, reflecting the greater relative abundance of remains of the thir- teen-lined ground squirrel {Spermophilus tridecem- lineatus). Glaucomys volans is common throughout the state today. Glaucomys sabrinus is a boreal species of the Hudsonian/Canadian Zone coniferous forests whose range extends south in the Appalachian high- lands at ever increasing elevations. It is present in the Great Smoky Mountains east and south of the site at elevations over 900 m higher than Baker Bluff in hardwood-coniferous woodlands. Glauco- mys sabrinus accounted for 65% of the Baker Bluff flying squirrels. Its presence in decreasing relative numbers in the site with depth is in accordance with other internal evidence suggesting a cooling of the environment during mid- and late-depositional times at the site. Measurements of the fossil material (Table 10) indicate a form of Glaucomys sabrinus larger than present day eastern races but comparable in size to modern northern and western material and to late Pleistocene specimens from other Appalachian cave deposits (Guilday et al., 1977). Glaucomys vo- lans also averaged larger in dental dimensions dur- ing the late Pleistocene in the Appalachians, but the Baker Bluff specimens, referred to G. volans, were insufficient for valid size inferences to be drawn. G. volans has been reported from numerous Pleisto- cene sites in Florida extending back to the Sanga- monian interglacial (Webb, 1974), but Baker Bluff Cave and Robinson Cave, Tennessee, are the most southern Pleistocene stations for Glaucomys sabri- nus. The ratio of remains of nocturnal flying squirrels {Glaucomys) to those of the diurnal chipmunk {Tamias) in a fossil bone sample give some indica- tion of the relative importance of owls in its for- mation. The percent of Glaucomys to Tamias at Baker Bluff Cave was a low 42%, suggesting that nocturnal predators such as owls played a lesser role at Baker Bluff Cave in building the fossil as- semblage than at Clark’s Cave (66%), Robinson Cave (66%), and Natural Chimneys (82%). The New Paris No. 4 deposit (38%) was a fissure pitfall where no owl activity was involved and the low percent of Glaucomys to Tamias reflects this. On 1978 GUILDAY ET AL.— BAKER BLUFF CAVE DEPOSIT 39 Table 10. — Measurements (in mm) o/Glaucomys Thomas. Age and locality Mean OR SD cv N Alveolar length P-'- M-* Pleistocene, Baker Bluff Cave, Tennessee Glaucomys cf. sahrinus 8.15 — — — 1 Glaucomys cf. volans 6.68 — — — 1 Alveolar length Pj- M,t Recent, Pennsylvania* Glaucomys sabrinus (Shaw) 7.19 6.80-7.60 .17 2.36 18 Pleistocene Glaucomys cf. sabrinus New Paris No. 4, Pennsylvania 7.70 7.60-8.10 — — 3 Natural Chimneys, Virginia 7.80 7.30-8.40 — — 14 Clark’s Cave, Virginia 8.00 7.60-8.60 .01 1.19 30 Robinson Cave, Tennessee 7.80 7.00-8.70 — — 8 Baker Bluff, Tennessee 7.90 7,70-8.40 — — 7 Anteroposterior crown length M, and Mj Recent, Pennsylvania Glaucomys volans (Linnaeus)** 1.57 1.45-1.75 .10 6.83 34 Glaucomys sabrinus (Shaw)* 1.69 1.45-1.94 .13 7.65 36 Pleistocene, Baker Bluff Cave, Tennessee 1.65 1.45-2.01 — — 29 Crown width M, and Mj Recent, Pennsylvania Glaucomys volans (Linnaeus)** 1.71 1.55-1.94 .12 6.99 34 Glaucomys sabrinus (Shaw)* 1.81 1.65-2.13 .15 8.26 36 Pleistocene, Baker Bluff Cave, Tennessee 1.97 1.69-2.42 — — 29 * CMNH Recent mammal collection: 31603, 31604, 31606, 31607, 31609-31611, 36392, 36394-36402, 37134. ** CMNH Recent mammal collection: 34318-34325, 34327-34329, 34332, 34333, 34335-34337, 34382, 34383. the face of it, this percentage suggests that owls were of relatively little importance in amassing the fossil deposit at Baker Bluff Cave. But a more likely possibility, strengthened by high numbers of thir- teen-lined ground squirrels (Spennophilus tride- cemlineatus) and the presence of badger (Taxidea taxiis), both open country forms, in the deposit is that the country was not as densely forested, a sit- uation differentially favoring the chipmunk {Tand- as). Family Castoridae — Beavers Castoroides ohioensis Foster — giant beaver Material. — CM 24680. 1 M‘ or (illustrated in Parmalee et al., 1976). MNI = 1. Remarks. — A single molar of the extinct giant beaver was recovered prior to the CMNH excava- tion by Bob Wilson and S. D. Dean, Jr., 36 to 72 in. inside the entrance, and 36 to 72 in. from the north wall at a depth of 1.5 to 2 ft, in what appears to have been the Holocene level. The molar prob- ably belongs to the underlying late Pleistocene fau- na, and the upper levels (0-3 ft) represent a chron- ological mix due to human disturbance. This tooth may have represented a woodrat hoard-object. Cas- toroides has been reported from an unnamed cave along the Clinch River, near Oak Ridge, Roane County, Tennessee (Parmalee et al., 1976). Castor canadensis Kuhl — beaver Material. — CM 30062-30064. 4 isolated molars. MNI = 3. Remarks. — The stratigraphic distribution of these isolated teeth has no significance. Beaver were probably present throughout the depositional se- quence. Lt. Henry Timberlake, in December 1761, specifically noted the abundance of beaver along the Holston River in what is now Sullivan County (Williams, 1927, in Kellogg, 1939). Smith et al. (1974) mention sporadic reports of their presence in the county today. 40 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 11 Eamily Cricetidae — New World Rats and Mice Peromyscus, sp. — white-footed mouse [either P. cf. maniculatus (Wagner) and/or P. cf. leucopus (Rafinesque)] Material. — CM 29994-30000. maxillae, mandibles, isolated molars. MNI = 94. Remarks. — White-footed mouse remains were common in all levels from 3-10 ft in the deposit (Table 1). They were undoubtedly present in the upper 0-3 ft of the deposit, but small vertebrate remains were not recovered by the original exca- vators. Peromyscus leucopus and P. maniculatus are common in the cave area at the present time. The golden mouse {Ochrotomys nut tali), close to Peromyscus in dental morphology, has also been recorded (Phillips and Richmond, 1971; Smith et al., 1974). M, morphologies, typical of P. maniculatus and P. leucopus (see Guilday et al., 1977:59, for crite- ria), were noted, but no stratigraphic percents were calculated due to the low frequency of recovered M/s. Alveolar length of Mj-Mg of 90 mandibles aver- aged 3.59 mm (OR = 3.2 to 4.2 mm). This is 0.19 mm larger than the mean of 48 mandibles of a mixed P. leucopus-P. maniculatus sample from the late Pleistocene Clark's Cave, Virginia, deposit (Guil- day et al., 1977). Some examples measured 0.5 mm larger than the largest Clark’s Cave specimen. This suggests that Ochrotomys nuttalli may be present in the Baker Bluff Cave collection, but no molars of O. nuttalli morphology were noted. A bimodal curve of Mj-Mg alveolar length with modes at 3.4 mm and 3.7 mm suggests that the smaller P. cf. maniculatus comprised about one- third of the total sample (stratigraphic levels com- bined) and the larger P. cf. leucopus about two- thirds of the population sample. Erequencies ran in 0.1 mm increments of increasing size — 2, 13, 15, 7, 13, 17, 9, 6, 5, 2, 1. Neotoma floridana (Ord) — eastern woodrat Material. — CM 24686-24692. partial crania, mandibles, iso- lated molars. MNI = 260. Remarks. — The woodrat was the commonest mammal represented at all stratigraphic levels of the deposit, with the exception of voles of the genus Micro! us. Most of the remains are, we believe, those of rats that died a “natural” death in this woodrat midden/raptor roost accumulation. This is suggested by the excellent preservation of some of the skulls compared with those of other species of small mammals, and by the high numbers of Ne- otoma remains, 18% of all rodents from the deposit. By way of contrast, woodrats accounted for only 2% of all rodents from the large, late Pleistocene raptor roost deposit at Clark’s Cave, Virginia, in a similar physiographic setting. Woodrats are common in cliff and cave habitats throughout the Ridge and Valley province but do not persist in settled areas, perhaps because they cannot compete with introduced black and/or Nor- way rats (Rattus). Smith et al. (1974) do not record Neotoma from Sullivan County, and characterize its distributional status as uncertain. They were probably widespread in suitable habitats at least until European settlement of the area. Woodrats were responsible for most of the iso- lated teeth and bones of large mammals in the de- posit. Scavenged items are a common constituent of modern woodrat middens, an expression of the animal’s well-known gathering and hoarding pro- pensities. Family Arvicolidae — Voles Clethrionomys gapperi (Vigors) — red-backed vole Material.— CM 24462-24466, 24655-24661. 132 left, 153 right mandibles and/or M,. MNI = 159. Remarks. — Red-backed vole remains were com- mon in the deposit at all levels, 17.6% of all voles. However, the relative frequency of Clethrionomys remains varied stratigraphically (Fig. 13) from over 25% of all voles in the upper levels to less than 10% in the lower levels. The species no longer occurs at the site or in the Ridge and Valley province of eastern Tennessee. It does occur as the dominant woodland vole in north- ern hardwood/coniferous woodlands at that latitude east of the Ridge and Valley province in the higher elevations of the Great Smoky Mountains. Smith et al. (1974) reported it as relatively common above 660 m in Unicoi County, Tennessee. Howell and Conaway (1952) trapped a specimen in the Cum- berland Mountains, about 232 km WSW of Baker Bluff Cave, in an “overgrown jumble of rocks in a growth of rhododendron and hemlock.” This is the first Recent record of Clethrionomys from the Cum- berland Plateau of Tennessee, although Barbour and Davis (1974) record it from the Cumberland Plateau of eastern Kentucky (Big Black Mountain) above 680 m elevation in northern hardwood situ- 1978 GUILDAY ET AL.— BAKER BLUFE CAVE DEPOSIT 41 ations. But in the Ridge and Valley province of Ten- nessee today Clethrionomys is replaced by the woodland vole, Microtus pinetonim. Clethrionomys has been reported from at least two other late Pleistocene Tennessee sites situated beyond its present range — Carrier Quarry Cave, Sullivan County, 10% of all voles (CM 30216- 30223; vole MNI = 212), and Robinson Cave, Overton County, 14% of all voles (Guilday et al., 1969; vole MNI = 227). The relative decrease in numbers of Clethriono- mys with increasing depth at Baker Bluff is in ac- cord with the decrease of all other species of boreal affinities in the lower levels of the deposit and sug- gests a change from boreal to temperate conditions with depth. Phenacomys intermedius Merriam — heather vole Material. — CM 24467-24470, 24662-24666. 45 left, 57 right Mi; 142 additional molars. MNI = 64. Remarks. — The heather vole occurs throughout the boreal forest of northern North America, but is no longer found in the Appalachians south of the St. Lawrence estuary. Although usually present in mid- Appalachian late Pleistocene vole faunas from Pennsylvania south to Tennessee, and in some mid- western and western deposits of Wisconsinan age (Guilday and Parmalee, 1972) it is usually the rarest species of vole encountered. Comparative percent- ages relative to all voles are as follows: 3.8%, New Paris No. 4, Pennsylvania; 3.6%, Natural Chim- neys, Virginia; 1.6%, Clark's Cave, Virginia; 5.6%, Carrier Quarry Cave, Tennessee; 7.1%, Baker Bluff Cave. Phenacomys remains were recovered from all levels in the deposit but diminished with depth from a high of 10% of all voles at the 2>-A ft level to a low of 1.1% at the 7-10 ft level (Fig. 13). The habitat requirements of the heather vole, within the context of boreal forest, vary from “dry, open coniferous forests . . . with an understory of heaths’’ to “borders of forest and in moist mossy meadows’’ (Banfield, 1974:193). Their present ab- sence from the closed-canopy northern hardwood/ coniferous forests of the northern Appalachians suggests more open woodland during late Wiscon- sinan times in the periglacial Appalachians. Phenacomys intermedins and Clethrionomys gapperi were present throughout the deposit. Both are woodland voles and their relative numbers in a given Appalachian cave deposit are usually posi- tively correlated. But Phenacomys appears to be unusually abundant in two Tennessee paleofaunas studied to date. Percentages of Phenacomys rela- tive to Clethrionomys at various Appalachian sites are as follows: 15%, New Paris No. 4, Pennsylva- nia; 17.5%, Natural Chimneys, Virginia; 10%, Clark’s Cave, Virginia; 28.7%, Baker Bluff Cave, Tennessee; 35.3%, Carrier Quarry Cave, Tennes- see. The comparative abundance of the heather vole in these two Tennessee sites, at what must have been at or near the southern limit of their Wisconsinan range expansion, is puzzling. Microtus chrotorrhinus (Miller) — rock vole Material.— CM 24483-24486, 24643-24747. 29 left, 25 right M^; associated molars and partial palates. MNI = 37. (Adjusted MNI = 57, see Microtus pennsylvanicus account for explana- tion). Remarks. — The rock vole no longer occurs at the site. This is the rarest Appalachian Microtus and occurs sporadically in the mountains to the east. Linzey and Linzey (1971) record the species above 800 m in the Great Smoky Mountain National Park, usually in talus or under mossy logs and rocks in high humid forest, much the same habitat as the red-backed vole Clethrionomys gapperi. The rock vole was present at all levels in the de- posit, but it was relatively more abundant in the upper levels (Fig. 13) and accounted for only 6.3% of all voles. Their numbers relative to those of the meadow vole {Microtus pennsylvanicus) were high, 30.6%, reinforcing the impression of a cool forest habitat. This percent was virtually identical with that of Clark’s Cave, 30.7%, suggesting that con- ditions for Microtus chrotorrhinus were equally fa- vorable at both sites. These figures differ markedly, however, from those of the Carrier Quarry Cave, 24 km east of Baker Bluff (CMNH collections). At that site Microtus chrotorrhinus formed 5.2% of the vole fauna but 10% of the M. chrotorrhinusIM . pennsylvanicus sample, only a third of the figure for Clark’s and Baker Bluff caves. The interpretation of such varying percentiles must await the analysis of other paleofaunas. Microtus pennsylvanicus (Ord) — meadow vole Material.— CM 24478-24482, 24636-24642. 61 left, 65 right M^; 72 left, 65 right M'*; isolated molars and fragmentary palates. MNI = 84; adjusted MNI = 129. Remarks. — The meadow vole is common in suit- able moist grassy habitat in the immediate area (Phillips and Richmond, 1971; Smith et al., 1974), but is here at the southwestern edge of its eastern 42 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. II North American range (Hall and Kelson, 1959). It is absent from west Tennessee (Severinghaus and Beasley, 1973). During late Pleistocene times its range extended much farther south, however. Webb ( 1974) records the meadow vole from at least four late Pleistocene sites in Elorida. Microtus pennsylvanicus was the commonest vole at Baker Bluff Cave from the 3—4 ft level, but decreased rapidly in relative abundance with depth, from 25% of all voles at the 3-i ft level to 9.8% at the 7-10 ft level where it ran a poor third in relative numbers of voles, exceeded by three times as many Synoptomys cooperi and four times as many M. pinetoriiin and/or M. ochrogaster (Pig. 13). The de- creasing abundance of M. pennsylvanicus with depth suggests drier conditions at the lower cave levels. This species prefers moist grasslands, al- though they may occur as forest enclaves. Meadow vole remains accounted for 14.3% of all voles from the site (Pig. 16), a relatively low figure compared with that of Clark’s Cave, Virginia, 32%, and New Paris No. 4, Pennsylvania, 30%. But this overall figure of 14.3% of all voles does not take into account the marked stratigraphic change from common in the upper levels of the site to scarce in the lower older levels. Thus it would be misleading to interpret the low relative number of M. pennsyl- vanicus from Baker Bluff Cave as due solely to the more southerly geographic location of the site. The meadow vole accounted for a high 31% of all voles from Robinson Cave, Tennessee (N = 227, Guilday et al., 1969), and an even higher 48% at Carrier Quarry Cave, Tennessee, only 24 km away (N = 212, CMNH collections) and just as far south as Baker Bluff. Both Robinson Cave and Carrier Quar- ry Cave lack some species of boreal affinity that were recovered at Baker Bluff Cave, M. xanthog- nathus for example, and probably postdate the site. It is obvious that relative numbers of M. penn- sylvanicus varied markedly from site to site, as well as stratigraphically, at Baker Bluff. But more data are necessary. We are at that unfortunate stage of too much data for simplistic explanations to suffice and too little to resolve the picture. The adjusted minimum numbers of M. pennsyl- vanicus and M. chrotoirhinus were derived by di- viding the minimum numbers of voles identified as either M. pennsylvanicus or M. chrotoirhinus on the basis of recovered M/s (the commonest ele- ments recovered) into two groups based on the per- centage of the specifically identified M®. This diag- nostic tooth is smaller than the M,, hence not as often recovered from fossil deposits due to its great- er chance of destruction. This resulted in an ad- justed increase in the minimum numbers of both species and made their numbers a truer relative ap- proximation to the minimum numbers of other species of voles recovered from the site based on M, counts. Microtus xanthognathus (Leach) — yellow-cheeked vole Material. — CM 24487-24490. 4 left, 1 right mandible; partial skull; left maxilla; left M^. MNI = 4. Remarks. — The yellow-cheeked vole is now found in the Hudsonian Life Zone of Alaska and northern Canada. Remains at Baker Bluff are con- fined to the upper levels of the deposit (Table 1, Eig. 13), suggesting at least marginal boreal condi- tions during the deposition of the upper cave levels. The former occurrence of the yellow-cheeked vole in North American periglacial sites of Wiscon- sinan age is well documented (Guilday and Bender, 1960; Hallberg et al., 1974). It was a common vole in sites of this age in the central Ridge and Valley province where its remains were encountered by the hundreds in such sites as Clark’s Cave or New Paris No. 4, where it dominated the vole faunas. At Baker Bluff Cave, however, only four individuals were recovered, 0.4% of the minimum number of all voles, and the animal was probably at or near the southern limits of its maximum Wisconsinan range extension (Fig. 16). Two other Sullivan Coun- ty sites have produced boreal voles. Carrier Quarry Cave and Guy Wilson Cave (CMNH collections), but Microtus xanthognathus was not present. These sites may not, however, have been contem- poraneous with the Baker Bluff deposit. Microtus xanthognathus has been recorded slightly farther south in the American Midlands dur- ing late Wisconsinan times, 36°N, Peccary Cave, Arkansas (Hallberg et al., 1974). Dental measure- ments are given in Table 1 1 and Fig. 9. Microtus pinetorum (Le Conte) — woodland vole and/or Microtus ochrogaster (Wagner) — prairie vole Material.— CU 24471-24477, 24648-24654. 241 left, 254 right mandibles or M,. MNI = 282. Remarks. — The woodland vole (Microtus pine- torum) is widely distributed throughout Tennessee and is the commonest vole at the site today (Phillips and Richmond, 1971). Microtus ochrogaster does not occur in East Tennessee east of the Cumberland Plateau, but is common in grassy situations in west- 43 1978 GUILDAY ET AL.— BAKER BLUEE CAVE DEPOSIT Microtus xanthognathus ( Leach) Occlusal length, lower first molar Late Pleistocene New Paris, Pa. No.4 X = 3.35 mm N = 55 Clark’s Cave, Va. X =3. 72 mm N = 60 Baker’s Bluff, Tenn. X = 3.82 mm X =3.46 N = 20 Fig. 9. — Histograms of length M, (in mm) Microtus xanthognathus (Leach), various localities. 44 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 11 Table 11. — Measurements (in nun) o/ Microtus xanthognathus (Leach), various Pleistocene localities. Locality Mean OR SD cv N CMNH catalog nos. Occlusal length M, New Paris No. 4, Pennsylvania 3.35 2.90-3.90 .22 6.34 55 6750, 6845, 6873, 6885,6930, 7150, 7171, 7179, 7185, 7192 Clark's Cave, Virginia 3.72 3.20-4.30 .215 5.78 60 24522 Baker Bluff Cave, Tennessee 3.80 3.70-3.90 — — 4 24487-24488 Occlusal width M, New Paris No. 4, Pennsylvania 1.24 0.97-1.45 .11 8.74 56 as above Clark’s Cave, Virginia 1.29 1.07-1.46 .09 7.23 60 as above Baker Bluff Cave, Tennessee 1.20 1.00-1.36 — — 5 as above central Tennessee (Phillips and Richmond, 1971; Barbour and Davis, 1974). In East Tennessee to- day, the meadow vole M. pennsylvanicus occupies the meadow, old-field niche to the exclusion of M. ochrogaster (Smith et al., 1974). We do not trust our ability to differentiate M. pinetorum from M. ochrogaster on dental charac- ters alone, the only criteria that can be used on the Baker Bluff Cave specimens. The difficulty lies in the amount of variation, individual and geographic, which cannot be satisfactorily assessed without a comprehensive study. Either or both species may be present in the deposit. We suspect both in the light of the presence of such western species as the thirteen-lined ground squirrel, least chipmunk, bad- ger, and probably the black-billed magpie in the de- posit. M. pinetorum and/or M. ochrogaster accounted for 31% of the voles from the deposit. The relative numbers varied stratigraphically, increasing from 17.5% at the ft level, to 45% at the 6-10 ft level (Eig. 13). Both are temperate species, their modern ranges stopping short of the boreal forest; M. pi- netorum at 45°N and M. ochrogaster at 53°N, in the Central Plains (Hall and Kelson, 1959). No matter which of the two species is present in the Baker Bluff Cave collection, their decreasing relative numbers in the upper cave levels agrees well with the relative decrease of other temperate species from those same levels. Relative numbers of M. pinetorum and/or M. ochrogaster, compared with those of other voles, fluctuate in late Pleistocene Ridge and Valley sites — 0.9% at New Paris No. 4; 8% at Clark’s Cave; 31% at Baker Bluff Cave (Fig. 16). The high figure from Baker Bluff Cave is due to its lower latitude or to an influx of M. ochrogaster, or both. Relative numbers of M. pinetorum to M. ochrogas- ter from eastern cave sites, and from the various stratigraphic levels at Baker Bluff Cave itself, might supply information about the relative abundance of deciduous woodland {M. pinetorum) and grassland (M. ochrogaster) during depositional times. Unfor- tunately “the state of the art’’ does not permit un- equivocal identification from the dentition alone. Ondatra zibethicus (Linnaeus) — muskrat Material. — CM 30068-30070. 2 partial palates; partial left mandible; 6 isolated molars. MNI = 4. Remarks. — Muskrat remains were scarce in the deposit, 0.4% of all arvicolids. They were also un- common at the extensive Clark’s Cave deposit, 0.2%. Both sites are primarily owl-roost deposits located beside rivers, but the low number of musk- rats is probably due to predator bias. Adults are too large for many birds of prey and are protected by their semiaquatic habits. Synaptomys cooperi Baird — southern bog lemming Material.— CM 24448-24463, 24667-24673. 185 left, 168 right partial mandibles and/or Mj; 1 partial palate. MNI = 191. Remarks. — Rare and local in the area today and indeed throughout most of its range, the southern bog lemming occurs here at the extreme southern edge of its range. Smith et al. (1974) record a spec- imen from the South Holston Dam, elevation 485 m within a few km of the site. Recent specimens are also known from Morristown, Hamblen Coun- ty, 75 km SW of Baker Bluff Cave (University of Tennessee, Knoxville, collections). The southern bog lemming is seldom taken by modern trapping methods throughout the Appalachian area. They accounted for only 5.5% of all voles (N = 1,367) collected from central Pennsylvania (Gifford and Whitebread, 1951; Roslund, 1951). These results may be biased by both trapping methods and mod- ern land usages, but at the Sheep Rockshelter, Pennsylvania, a prehistoric owl-roost deposit in the 1978 GUILDAY ET AL.— BAKER BLUFE CAVE DEPOSIT 45 Table 12. — Measurements (in mm) of Synaptomys cooper! and S. australis, occlusal length M Locality Mean OR SD cv N Synaptomys cooperi Baird Pennsylvania, Recent* 2.48 ± .038 2. 1-2.7 0.19 7.66 25 New Paris No. Pennsylvania, late Pleistocene 2.41 ± .020 2. 3-2. 5 0.09 3.73 20 Clark’s Cave, Virginia, late Pleistocene 2.40 ± .019 2. 2-2. 7 0.11 4.75 33 Natural Chimneys, Virginia, late Pleistocene 2.39 ± .013 2. 2-2. 5 0.05 2.92 14 Robinson Cave, Tennessee, late Pleistocene 2.65 ± .014 2. 3-2. 9 0.122 4.61 71 Baker Bluff Cave, Tennessee, late Pleistocene 2.53 ± .007 2. 2-2. 8 0.13 5.21 265 Carrier Quarry Cave, Tennessee, late Pleistocene 2.55 ± .018 2. 2-2. 8 0.13 5.10 48 Synaptomys australis Simpson Florida, CMNH collection 3.5 3. 3-3. 9 — — 7 * Guilday el al.. 1977:68. ** Guilday el al., 1969:57. same area (Guilday and Parmalee, 1965) S. cooperi remains comprised only 7% of the voles suggesting that it was a relatively uncommon animal even prior to colonial deforestation. Synaptomys cooperi is also uncommon in late Wisconsinan sites from Pennsylvania and Virginia (Fig. 16), where it is usually outnumbered by remains of the northern bog lemming S. borealis. Yet, surprisingly, Synaptomys cooperi was one of the commonest small mammals represented in the deposit at all stratigraphic levels, and in three other late Wisconsinan sites from Tennessee — Guy Wilson Cave and Carrier Quarry Cave, Sullivan County; Robinson Cave, Overton County. In the highest undisturbed level at Baker Bluff Cave, Syn- aptomys cooperi accounted for 17% of all vole re- mains and increased to a high of 34% in the lower 3 ft of the deposit (Fig. 13). In addition to being more common in the area during late Wisconsinan times, Synaptomys cooperi remains are larger than those of modern eastern North American specimens (Table 12). There is a significant size difference between the small late Pleistocene specimens from Pennsylvania and Vir- ginia, which average smaller in dental dimensions than Recent Pennsylvania specimens of S. c. coop- eri, and M]’s from three Tennessee Pleistocene sites. M,’s from Baker Bluff Cave, Carrier Quarry Cave, and Robinson Cave average larger than both Recent and more northerly late Pleistocene speci- mens. Synaptomys cooperi today becomes larger with decreasing latitude, as does the Pleistocene material. This suggests there may be a yet-undem- onstrated size continuum between S. cooperi and the large extinct S. australis from the Pleistocene of Florida. At present there is a size gap between the two populations, but there is also a geographic gap in Alabama and Georgia between the southern range limits of Pleistocene S. cooperi and the north- ern range limits of 5. australis from which speci- mens have yet to be collected. Synaptomys borealis (Richardson) — northern bog lemming Material.— CM 24454-24458, 24674-24676. 10 left, 9 right par- tial mandibles and/or M,. MNI = 12. Remarks. — Remains of the northern bog lemming were scarce in the deposit, 1.3% of all voles, and largely confined to the upper stratigraphic levels (Fig. 13). This species no longer occurs in the cen- tral or southern Appalachian region. Its present range includes the coniferous forest and taiga of Canada and Alaska south to Minnesota and the White Mountains of New Hampshire, some 1,200 km northeast of Baker Bluff Cave. The southeastern portion of the range of S. bo- realis overlaps the northern portion of the range of S. cooperi (Hall and Kelson, 1959). Both species are usually present in late Wisconsinan cave de- posits from Pennsylvania to Tennessee. S. borealis is never a common species relative to other voles in such deposits, however, and becomes rarer with decreasing latitude (Fig. 16). Relative numbers of S. borealis to S. cooperi also change with latitude in late Wisconsinan Appalachian sites (Table 13). Sullivan County, Tennessee, marks the southern limit of its known late Pleistocene range. Family Zapodidae — Jumping Mice Zapus hudsonius (Zimmerman) — meadow jumping mouse Material. — CM 30001-30002. 1 left, 2 right mandibles with M ,. 1 left M,. MNI = 3. Remarks. — Sparingly present in the upper levels 46 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. II Table 13. — Relative abundance of Synaptomys cooperi Baird and Synaptomys borealis (Richardson) in various late Wisconsinan Appalachian cave deposits, CMNH collections. Site Latitude MNI voles % Synaptomys cooperi % Synaptomys borealis % S. borealis to all Synaptomys New Paris No. 4, Pennsylvania 40°05'N. 1,212 1.32 5.8 82.0 Natural Chimneys, Virginia 38'’22'N. 323 4.62 1.07 19.0 Clark's Cave, Virginia 38°05'N. 2,060 1.11 3.0 73.0 Carrier Quarry Cave, Tennessee 36°29'N. 212 14.60 — — Baker Bluff Cave, Tennessee 36°27'N. 899 21.24 1.3 5.9 Robinson Cave, Tennessee 36°17'N. 227 14.80 0.4 2.0 of the deposit (Table I), remains of the meadow jumping mouse were not as common as those of the woodland jumping mouse. Both species were ab- sent from the lower levels of the deposit which, in conjunction with other stratigraphic species shifts in the deposit, suggests more mesic conditions in the upper levels. Zapiis hiidsonius has not been reported from the Ridge and Valley section of eastern Tennessee to- day, but it does occur farther east in the Great Smoky Mountains (Smith et al., 1974). It has been reported from low elevations farther west in the state, however (Severinghaus and Beasley, 1973), so it may occur near the cave at the present time. Remains of at least four individuals were recovered from the late Pleistocene deposits of Robinson Cave, Overton County, northcentral Tennessee. Length of M,-Mg, one mandible, was 3.8 mm. Length of Ml, four specimens, was 1.4, 1.4, 1,4, 1.5 mm. Comparable measurements from the Robinson Cave deposit: Mj-Mg = 3.5 mm; length M, = 1.2, 1.4, 1.55 mm. Napaeozapus insignis (Miller) — woodland jumping mouse Material. — CM 30003-30006. 3 maxillae; 2 mandibles; 2 iso- lated molars. MNI = 4. Remarks. — Remains of the woodland jumping mouse in the upper levels of the deposits, from 3- 7 ft depths, suggest a cooler more mesic environ- ment than that of the lower-level fauna or the Re- cent local mammal fauna. At the latitude of Baker Bluff Cave the woodland jumping mouse occurs in the Great Smoky Mountains to the east, at altitudes of 760 m or more, in cool Transition or Canadian Zone hardwood/coniferous forests (Smith et al., 1974). Remains of Napaeozapus insignis, consid- erably south of their modern range, have also been reported from the late Pleistocene deposits of Rob- inson Cave, Tennessee, 256 km west of Baker Bluff, in association with other small mammals. Their presence at these sites suggests a cooler en- vironment. Dental measurements (Table 14) suggest that late Pleistocene Napaeozapus insignis from Robinson and Baker Bluff caves, Tennessee, were somewhat smaller than individuals from more northerly late Pleistocene deposits such as Clark’s Cave, Virginia, and New Paris No. 4, Pennsylvania, thus paralleling the present day increase in size with increasing lat- itude (Wrigley, 1972). Family Erethizontidae — Porcupines Erethizon dorsatum (Linnaeus) — porcupine Material. — CM 30065-30067. 5 isolated molars, 3 isolated pre- molars. MNI = 3. Remarks. — Porcupine remains were confined to the upper levels of the deposit (Table 1). The animal Table 14. — Crown length (in mm) M,. Napaeozapus insignis (Miller). Locality and age Mean OR SD CV N Recent Quebec, Ontario (CMNH collections) 1.6 Pennsylvania (CMNH collections) 1.6 Late Pleistocene New Paris No. 4, Pennsylvania 1.8 Clark’s Cave, Virginia 1.7 Robinson Cave, Tennessee 1.7 Baker Bluff Cave, Tennessee 1.66 1.6-1. 6 — — 3 00 T .04 2.5 20 1. 7-2.1 .09 5.0 11 1.6-1. 9 — — 27 1.6-1. 8 .09 5.01 13 1.6-1. 8 — — 3 1978 GUILDAY ET AL.— BAKER BLUEE CAVE DEPOSIT 47 is not a member of the Recent fauna of Tennessee, or of the southern Appalachians, but its remains have been recovered from caves and archaeological sites in central Tennessee, northern Alabama, and Georgia. These sites range in age from late Pleis- tocene to an estimated 3,000 to 5,000 years old (summarized in Corgan, 1976:11). But Bogan (1976) states that no remains of porcupines have been re- covered from archaeological sites in East Tennes- see. The porcupine has a predilection for rocky sit- uations and may have survived in karst areas later than it might otherwise have during late Pleistocene climatic adjustments. Given the above fossil record its presence at Baker Bluff was expected and may not be of any climatic significance. Order Carnivora — Carnivores Eamily Canidae — Wolves and Foxes Urocyon cinereoargenteus (Schreber) — gray fox Material. — Dean/Wilson collection. Square 1, 0-6 in. Left hu- merus, immature. MNI = 1. Vulpes vulpes Linnaeus — red fox Material.— CM 29525, 29534, 29546. C,; left M^; right M,; un- assigned premolar. MNI = 3. Remarks. — Remains of the gray fox were con- fined to the superficial Recent strata. Red fox re- mains were found only in the Pleistocene levels (Table 1). Both red and gray fox are present in East Ten- nessee today. The red fox is not as common as the gray and prefers less heavily timbered areas (Smith et al., 1974). Kellogg (1939) speculates that the red fox may not have been native to the state. Evidence derived from analysis of mammal bones from pre- Columbian Indian garbage bears this out. Only gray fox remains occur in late prehistoric archaeological sites from Pennsylvania on south in eastern North America. The red fox appears to have been absent from the eastern forests south of New York state during most of Holocene times and extended its range to the south as Colonial deforestation resulted in copse- woodland, more to its liking. Red foxes have also been extensively stocked in the South- east. The presence of the red fox in the Pleistocene levels of Baker Bluff Cave does not contradict this picture. The presence of both boreal and steppe forms in the paleofauna suggest a more open, cooler environment compared to present day conditions and the red fox appears to have spread as far south as Florida, during at least the late Pleistocene, un- der the influence of expanding periglacial condi- tions. Pleistocene Vulpes remains have also been reported from Natural Chimneys, Virginia, and Vero Beach and Melbourne, Florida (Ray, 1958). With the waning of glacial conditions in post-Wis- consinan times the eastern range of the red fox con- tracted to the north, as far as archaeological evi- dence shows, to about the Pennsylvania-New York border, only to expand again within historic times under the influence of environmental changes brought on not by climatic shifts but by human land- use practices. Family Ursidae — Bears Ursus americanus Pallas — black bear Material.— CM 29500, 29504-29505, 29512-29513, 29516- 29518, 29545. 2 C‘; 1 C,; 1 PL 1 ML 1 M2; 2 M3; 1 unassigned premolar. MNI = 4. Remarks. — Black bear was represented by iso- lated teeth and a partial ulna. Teeth are comparable in size to those of Recent northeastern black bear (Table 15). These isolated elements undoubtedly represent woodrat detritus. Black bear remains are among the commonest of Appalachian cave finds. Family Procyonidae — Procyonid Carnivores Procyon lotor (Linnaeus) — raccoon Material.— CM 29526, 29532, 29537, 29544. Isolated molars and premolars. MNI = 5. Remarks. — Isolated raccoon teeth occurred at five different levels of the deposit, ranging from the surface material to the deepest 9-10 ft level. It is common in the area today and was probably so throughout the depositional history of the deposit. Family Mustelidae — Mustelid Carnivores Martes americana (Turton) — pine marten Material. — CM 29509-29510, 29519, 29542. Right mandible with P3-P4, 2 M,, 2 M2, 1 ML MNI = 4. Martes pennant! (Erxleben) — fisher Material. — CM 29514. Right mandible with P2; M,. MNI = I. Mustela nivalis Linnaeus — least weasel Material. — CM 29508, 29521, 29529. Right mandible with M,M2; left mandible with M,; left M,; 2 C; 2 PL MNI = 4. Mustela frenata Lichtenstein — long-tailed weasel Material.— CM 29507, 29520, 29527-29528, 29538-29539, 29541. Partial mandibles, isolated molars. MNI = 6. 48 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 11 Table 15. — Ursus americanus Pallas, crown length and width (in mm) Ms- Locality Age OR length OR width N East Tennessee, males Parmalee et al., in press Recent 20.5-22.7 11.9-14.1 6 West Virginia archaeological sites, 46Pu31 and 46Fa7 Recent ( 1600 A.D.) 17.7-31.1 10.7-13.3 17 Baker Bluff Cave, Tennessee, CM 29517 Rancholabrean 18.7 11.5 1 Baker Bluff Cave, Tennessee, CM 29518 Rancholabrean 17.3 12.7 1 Mephitis mephitis (Schreber) — striped skunk Material. — CM 29503. Right maxilla with P'*-P^. MNI = 1. Spilogale putorius (Linnaeus) — eastern spotted skunk Material. — CM 29511, 29533, 29540, 29543. Partial mandible, isolated molars. MNI = 3. Taxidea taxus (Schreber) — badger Material.— CU 29549. F. MNI = 1. Remarks. — Seven species of mustelids have been reported from East Tennessee within Recent times — fisher; striped skunk; spotted skunk; least weasel; long-tailed weasel; mink (Mustela vison)-, otter (Lutra canadensis) (Kellogg, 1939; Howell and Conaway, 1952; Smith et al., 1974). With the exception of the mink and the otter, both semi- aquatic, all Recent species were also present in the Baker Bluff Cave local fauna. Carnivore remains are scarce in raptor deposits, their numbers, usually in inverse proportion to their size, reflecting not their relative abundance during life but the inability of raptors to handle such aggressive animals. Iso- lated teeth or bones of the larger species may have been introduced into the deposit piecemeal by woodrats. Striped skunk remains were confined to the dis- turbed top strata and spotted skunk remains to the lower late Pleistocene levels, but this distribution may be fortuitous. The pine marten (Maries americana) is no longer found as far south nor the badger (Taxidea taxus) as far east as Tennessee. Within historic times the pine marten occurred only as far south as central Pennsylvania, 5° north of Baker Bluff Cave and its presence complements the large boreal element present in the paleofauna. Pine marten remains have also been reported from Robinson Cave, Overton County, Tennessee, and it is frequently found in mid- Appalachian late Pleistocene cave fau- nas (Guilday et al., 1977). This is the first record of badger from the Pleis- tocene of Tennessee. Unfortunately its site pro- venience is unknown. Badger remains are known from late Pleistocene sites east and south of their present prairie range at Welsh Cave, Kentucky, Bootlegger Sink, Pennsylvania, and Baker Bluff Cave, Tennessee (CMNH collections). At all three sites the thirteen-lined ground squirrel (Spermoph- ilus tridecendineatus), another western to midwest- ern grassland species, was also present, indicating an eastern spread of the modified steppe biota, at least in early post-Wisconsinan times. Pre-Wiscon- sinan, perhaps late Kansan, badgers have also been reported from two eastern cave deposits. Fort Ken- nedy Cave, Pennsylvania (Cope, 1899), and Cum- berland Cave, Maryland (Gidley and Gazin, 1938). Family Felidae — Cats Fells onca Linnaeus — jaguar Material.— CM 29547. P4. MNI = 1. Remarks. — Jaguar remains are common in Ten- nessee caves. Guilday and McGinnis (1972) list specimens from five cave sites in the state. There are no historic records and the Baker Bluff tooth from the lowest cave level is referred to F. o. au- gusta, the extinct late Pleistocene subspecies. Order Artiodactyla — Even-toed Ungulates Family Tayassuidae — Peccaries Platygonus compressus Le Conte — flat-headed peccary Material. — CM 29548, partial upper molar. CM 30229, LdP2. MNI = 2. Remarks. — These two molars, from an adult and a piglet, may have been introduced by rodents. Cor- gan (1976) records Platygonus from four Tennessee sites, two of them, Worley Cave and Guy Wilson Cave, in Sullivan County. Remains of at least 16 animals with a ^“^C date of 19,700 ± 700 yr BP (1-4163) were recovered from Guy Wilson Cave (CM 20042, 20043, 20083-20101). This extinct species is common in late Pleistocene Appalachian sites. 1978 GUILDAY ET AL.— BAKER BLUFE CAVE DEPOSIT 49 Table 16. — Anteroposterior length (in mm) teeth o/ Odocoileus virginianus (Zimmerman), Baker Bluff Cave , Tennessee, late Pleisto- cene; Eschelman Site, Pennsylvania (36 La 12), early historic; and Chota Site, Tennessee (40 Mr 2), early historic. Tooth Locality Mean OR SD cv N dP-> Baker Bluff Cave 14.0 — 1 dP^ Baker Bluff Cave 11.5 11.0-12.0 — — 2 dPg Baker Bluff Cave 7.8 7. 0-9.0 — — 5 dPg Baker Bluff Cave 9.7 9.0-11.0 — — 8 dP, Baker Bluff Cave 15.7 15.0-17.0 — — 4 p2 Baker Bluff Cave 11.4 10.0-12.0 — — 5 Eschelman Site 11.9 11.0-13.0 — — 25 p,2 Baker Bluff Cave 9.0 8.0-10.0 — — 3 Eschelman Site 10.5 9.0-12.0 — — 29 Baker Bluff Cave 13.9 12.0-15.0 — — 15 Eschelman Site 14.6 12.0-16.0 — — 80 P2 Baker Bluff Cave 8.2 7.0-9.0 — — 9 Eschelman Site 9.1 7.0-10.0 — — 20 P.3 Baker Bluff Cave 10.8 10.0-12.0 — — 5 Eschelman Site 11.4 10.0-13.0 — — 23 P4 Baker Bluff Cave 11.0 10.0-13.0 — — 10 Eschelman Site 11.7 10.0-13.0 — — 25 M| or Mg Baker Bluff Cave 13.4 12.0-16.0 — — 22 Eschelman Site 14.5 12.0-17.0 — — 53 Mg Baker Bluff Cave 19.1 18.0-23.0 — — 8 Eschelman Site 20.7 19.0-24.0 — — 26 Chota Site 19.4 17.0-22.3 1.18 6.06 49 Eamily Cervidae — Deer cf. Cervus elaphus Erxleben — elk Material. — CM 29530, unerupted right M.^; CM 30226, left dP' (0-3 ft. Dean collection). MNI = 1. Remarks. — Identification of CM 29530 is based primarily on size and is tentative because the tooth appears to be congenitally deformed. A normal cer- vid M,3 is trilobed. The posterior lobe is reduced in Rangifer but is prominent in other cervids. It may rarely be absent in Odocoileus (Guilday, 1961). The third lobe is reduced to a low cingulum in CM 29530. The external valley between the anterior and central lobe appears cramped compared with Ran- gifer or normal Cervus elaphus specimens. The ec- tostylid is absent. Size as in Cervus elaphus. Elk are not present in Tennessee today, but in pre-Colonial times were common and their remains occur in Holocene archaeofaunas as far south as North Alabama (Barkalow, 1972). Odocoileus virginianus (Zimmerman) — white-tailed deer Materia!.— cm 30051-30057. I dP\ 2 dPC 5 dp2, 8 dPj, 4 dP^, 5 P^ 1 P^ or P^ 3 P^ 2 P3 or P^ 9 M‘, 2 M' or M^ 2 M^, 2 M^ 2 M', M^ or M^ 10 Pj, 6 Pg, 9 P4, 13 M,, 11 Mg, 1 M, or Mg, 8 Mg, 3 incisors. MNI = see discussion below. Remarks. — The Baker Bluff Cave white-tailed deer, at all levels, were the same size as their Re- cent Tennessee counterparts. Molar teeth agreed in size with historic archaeological specimens from a Cherokee Indian village site (Chota Site, 40 MR 2, Monroe County, Tennessee, Table 16). This is in- teresting and somewhat surprising considering the presence of boreal small mammals in the upper levels of the deposit. Modern deer become larger with increasing latitude (Doutt, no date) and one might have expected the Baker Bluff deer to have been larger than Recent local deer as a reflection of more boreal conditions. But compared with archae- ological deer dentitions from only as far north as Pennsylvania (Eschelman Site, 36 LA 12), Guilday et al., 1962), Baker Bluff Cave deer are smaller in all dental dimensions by 4.8% in the upper molars and 7.6% in the lower molars (Table 16). This is also true of Pennsylvania versus Tennessee archae- ological material and points up the necessity for using local comparative material if valid compari- sons are to be made between Pleistocene and Re- cent Odocoileus. If the size of northern white-tailed deer today is correlated with environment, then the size of the Baker Bluff Cave specimens do not suggest boreal conditions at the site. Although the deposit extends throughout the late Pleistocene and early Holocene and environmental change during that period is ev- ident throughout the stratigraphic column there was 50 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 11 no correlation between dental dimensions of deer and depth in the deposit. Some measure of the stratigraphic dispersal of large mammal elements in the deposit may be gained from the fact that by lumping isolated teeth from all levels, a minimum of only five adult deer and five fawns could be accounted for, but if ana- lyzed by stratigraphic level with no reference to levels above or below, 14 adults and seven fawns were represented when levels were totaled. A bro- ken premolar from the 8-9 ft level fit a similar frag- ment from the 9-10 ft level. As with all large mam- mals remains in the deposit, deer bones and teeth probably accumulated by sporadic woodrat caching with a greater loss of stratigraphic integrity com- pared to the small vertebrate, owl-deposited, de- bris. Fig. 10. — Rangifer tarandus Linnaeus, CM 24588, left P4, occlu- sal and labial views, anterior to left. Baker Bluff Cave, Sullivan County, Tennessee. cf. Sangamona fugitiva Hay — “fugitive” deer Material.— CU 29501, P^; CM 30060, left P3. MNI = 1. Remarks. — The was compared with the re- ferred dentition from Frankstown Cave, Pennsyl- vania (CM 11044). Upper premolars of Sangamona are much larger relative to the molars than are those of Ochcoileus, and there is little likelihood of con- fusing them with white-tailed deer on size alone. They are, however, about the size of premolars of the large Rangifer from the deposit. As noted by Hay (1920) the weak buccal ribbing of Sangamona cheek teeth appears to be definitive. The P3, iden- tical in cusp conformation with Cervus or Odocoi- leiis, was referred to Sangamona because of its in- termediate size. It differs from a P3 of Rangifer in cusp conformation. Remains of this deer have been reported from two other Tennessee localities — the type locality, Whitesburg, Hamblen County (Hay, 1920), and Robinson Cave in Overton County (Guilday et al., 1969). Measurements of P^ are (CM 29501), length 14.0, width 15.8 mm; Pg (CM 30060), width 9.6 mm. Rangifer tarandus Linnaeus — caribou Material.— CU 24588. Left P4 (Fig 10). CM 24681, 29502, 30058, 30061. Left P^ or P^; right P-2, P4, Mj. CM 30274. Partial left humerus. MNI = 2. Remarks. — One P4, CM 24681, was found by S. D. Dean, Jr., in the upper 1 ft of the Baker Bluff Cave deposit mixed with Indian cultural refuse, but was probably displaced from the underlying Pleis- tocene levels by human activity. The occurrence of caribou from this and two other Tennessee cave deposits, Beartown Cave and Guy Wilson Cave, also in Sullivan County, is discussed and specimens figured in Guilday et al. (1975). This marks the southernmost North American extension of the ge- nus Rangifer. Order Perissodactyla — Odd-toed Ungulates Family Tapiridae — Tapirs Tapirus cf. veroensis Sellards — Vero tapir Material.— CU 29522. Left P‘ (Fig. 8). MNI = 1. Remarks. — The crown length of 18.7 mm and crown width of 15.5 mm of this isolated premolar lie within the range of T. veroensis Sellards, small- er than the mid-Pleistocene T. copei Simpson, as presented by Lundelius and Slaughter (1976). Ta- pirs are common Pleistocene fossils in eastern cave deposits from Pennsylvania south to Florida. Cor- gan (1976) records remains from nine other East Tennessee caves. FAUNAL SUMMARY The Baker Bluff Cave faunal sequence consists of at least 180 species of vertebrates and inverte- brates— 3 freshwater snails, 50 land snails, 1 mus- sel, 14 fish, 10 amphibians, 13 reptiles, 29 birds, and 60 species of mammals. About 2,600 individual mol- luscs and 2,305 vertebrates were present. Mammals accounted for 82.8% of all individual vertebrates, birds 7.2%, amphibians 6.9%, reptiles 1978 GUILDAY ET AL.— BAKER BLUFE CAVE DEPOSIT 51 2.3%, and fish .67%. Eish are underestimated be- cause of the difficulty of estimating minimum num- bers of individuals from fragmentary material. Forty-six of the 50 species of land snails still oc- cur in the area. According to Leslie Hubricht, one species of Stenotrema is undescribed and has not yet been found alive. Three other land snails do not occur at the site today — Hendersonia occulta, pres- ent in central Tennessee; Discus catskillensis, a montane form reaching its southern limits in High- land County, Virginia; and Stenotrema fraternum fasciatum, a montane form. All other invertebrates are present in the area today. All identified species of fish, amphibians, and rep- tiles still occur regionally. Two birds, cf. Pica pica and Ectopistes inigra- torius, and 10 species of mammals no longer occur in the southern Appalachian area — Sorex arcticus, Eutamias minimus, Spennophilus tridecemlinea- tus, Phenacomys intermedins, Microtus xanthog- nathus, Synaptomys borealis, Erethizon dorsatum, Martes americana, Taxidea taxus, and Rangifer tarandus. Nine species of mammals occur today only at higher elevations in the southern Appala- chians— Microsorex hoyi, Sorex cinereus, Sorex dispar, Condylura cristata, Tamiasciurus hudsoni- cus, Glaucomys sabrinus, Clethrionomys gapperi, Microtus chrotorrhinus, dct^dNapaeozapus insignis. Four species were represented by forms larger than those now inhabiting the area — Blarina brevicauda, Tamias striatus, Glaucomys volans, and Synapto- mys cooperi. Six taxa are extinct — Dasypus bellus, Castoroides ohioensis, Eelis onca cf. augusta, San- gamona fugitiva, Platygonus compressus, and Tap- irus cf. veroensis. In summary, 29 taxa of mammals, 48% of the 60 mammalian taxa from the site, differed in some fashion from the modern mammalian fauna. Ten percent are extinct; 16.6% are found only north or west of the site in either a boreal forest or a tem- perate grassland habitat today; 15% have retreated to higher elevations in the southern Appalachians; and an additional 6.6% have decreased in bodily size at least locally since depositional times. No taxa of southern distribution such as Sigmodon, Ochrotomys, Reithrodontomys , or Oryzomys were present in the deposit, although each of these four rodent genera occur at or near the site today (Smith et al., 1974). Faunal Analysis — Stratigraphic Change There is a marked change in faunal composition between the upper and lower levels of the site (Figs. 11, 12, and 13). Those species of small mammals found in cooler regions today, either at higher lati- tudes in Canadian/Hudsonian zone situations or at higher altitudes in the southern Appalachians, were relatively more abundant in the upper levels of the deposit. Stratigraphic analysis begins at the 3 ft level (3 ft from the original cave floor surface prior to the Dean excavation). All stratigraphic levels are arbi- trary. Natural stratigraphic levels were either not present or were not noticed under the awkward ex- cavating conditions. Remains of large mammals were not considered. They are present in such small numbers that statistical comparison is not feasible and their presence in the deposit is probably due to capricious woodrat scavenging. Remains of small mammals were numerous at every level. These animals were probably collected by raptors hunting the local countryside with reg- ularity and, unlike the sparse large mammal mate- rial, their presence, absence, or relative abundance up and down the stratigraphic column takes on in- terpretative significance. Minimum numbers used for stratigraphic analysis were, in the case of shrews and voles, based upon minimum numbers of individuals derived from tooth or mandible counts. In the case of the squirrels (Marmota omit- ted), numbers of M’’s and M^’s of each species were combined per level in order to increase sample sizes and lessen the effect of random fluctuations. Results of the analysis of shrew, squirrel, and vole remains tell a consistent complementary story. There were seven species of shrews recovered from the deposit. Only three, Blarina brevicauda, Sorex cinereus, and Sorex fumeus, occurred at all stratigraphic levels (Fig. 11). Two additional species of Sorex occurred sparingly in the upper 3- 5 ft levels, 5. arcticus, a boreal woodland species no longer found in the Appalachians, and S. dispar, a woodland rock-talus species found at higher ele- vations in the mountains east of the site. Two other shrews were also present sparingly in the upper levels, Microsorex hoyi, of boreal affinities and not a member of the Recent fauna of the state, and Cryptotis parva, a temperate grassland/old-field species. Long-tailed shrews, genus Sorex, were commonest in the upper stratigraphic levels, 63% of all shrews, twice as prevalent as Blarina brevi- cauda, the common temperate/woodland shrew of the Baker Bluff area today. Blarina brevicauda increased in numbers with depth while shrews of the genera Sorex, Microso- rex, and Cryptotis became relatively scarce or dis- 52 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. II Fig. 11. — Shrews, Family Soricidae. Stratigraphic distribution. Baker Bluff Cave, Sullivan County, Tennessee. 1978 GUILDAY ET AL.— BAKER BLUEE CAVE DEPOSIT 53 appeared from the stratigraphic column. An inter- esting infraspecific situation occurs in Blarina brevicauda from the deposit. The upper cave levels contained two distinct sizes, a smaller form similar to 5. b. kirtlandi, the subspecies currently found in the area, and a larger form similar to the large north- ern race, B. b. brevicauda of the Minnesota/Wis- consin area and also to the large Blarina from late Pleistocene sites farther north in the Ridge and Val- ley province. New Paris No. 4, Pennsylvania, and Clark's Cave, Virginia. The increase in numbers and variety of soricine shrews and the influx of a northern form of Blarina in the upper levels of the deposit suggest a transi- tion to cooler boreal forest conditions in the upper levels that would support shrews now confined to higher altitudes or latitudes. Paralleling the stratigraphic behavior of the voles and shrews from the deposit, numbers of boreal sci- urids, Tamiasciurus hudsonicus and Glaucomys sa- brinus, increased in the upper levels (Pig. 12), whereas relative numbers of temperate species, Sciurus caroUnensis, Glaucomys volans, and lami- as striatus increased with depth. The red squirrel (Tamiasciurus hudsonicus) and the northern flying squirrel (Glaucomys sabrinus) are characteristic of the northern hardwood/boreal coniferous forest belt of North America. The gray squirrel (Sciurus car- olinensis) and the southern flying squirrel (Glau- comys volans) are temperate deciduous-forest forms common in the Baker Bluff area today. Num- bers of the thirteen-lined ground squirrel (Sper- mophilus tridecemlineatus), a midwestern prairie form today, increased dramatically from 2.2% in the lowest levels of the deposit to over 30% in the upper levels where it was the commonest species of sci- urid. This reinforces a suggested change from closed to open-canopy forest in the upper levels of the deposit and is reflected by the correlated in- crease in numbers of the meadow vole (Microtus pennsylvanicus). The least chipmunk (Eutamias minimus), confined to the upper levels of the Baker Bluff deposit, also suggests a boreal parkland. Much the same picture of stratigraphic change from cool temperate-woodland to boreal forest con- ditions of some sort is suggested by the voles from the deposit (Pig. 13). All boreal forest species de- crease in relative numbers with depth, Clethriono- mys gapperi, Phenacomys intermedins , Microtus chrotorrhinus . Microtus xanthognathus, a boreal taiga species today, is confined to the upper levels of the deposit. Remains of Synaptomys borealis, rare in this southern site, occurred throughout the stratigraphic column but become rarer relative to numbers of the temperate 5. cooperi, with increas- ing depth. Only temperate species Synaptomys cooperi and voles of the subgenera Pitymys or Pe- domys, increased in numbers with depth, and they did so markedly. A cooling trend as one ascends the stratigraphic column is suggested. A change in the woodland to grassland ratio is also suggested by a rise in relative numbers with decreasing depth of Microtus pennsylvanicus, a moist-grassland species, an observation reinforced by the stratigraphic dis- tribution of the Sciuridae. Detailed ecological reconstruction is obscured by the complicated topography of the area and the southern location of the site. While it has undergone considerable Pleistocene and post-Pleistocene change, it has done so under relatively benign con- ditions. Water has been available throughout its Quaternary history so that climatic changes cannot be as closely monitored by accompanying biotic changes as they can in the arid (today) West where organisms exist under more confining strictures. Nevertheless a definite change from temperate to boreal small mammal species is noted at Baker Bluff Cave as one ascends the stratigraphic column. But the presence of such boreal species as Synap- tomys borealis and Phenacomys intermedins in the lowest cave levels suggests that the lower level cli- matic episode was cooler than that of the Sullivan County area today. DATING AND CORRELATIONS Carbon- 14 dates from the CMNH excavations at Baker Bluff Cave are as follows: 555 ± 185 years, BP, bone apatite, GX-3369, 4-5 ft level; 10,560 ± 220 years BP, bone apatite, GX-3370a, 6-7 ft level; 1 1 ,640 ± 250 years BP, bone collagen, GX-3370b, 6-7 ft level; 19,100 ± 850 years BP, bone apatite, GX-3495, 9-10 ft level. All dates were run on un- charred bone fragments selected from each level at 54 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. II Fig. 12, — Squirrels, Family Sciuridae (Marmota mona.x omitted). Stratigraphic distribution. Baker Bluff Cave, Sullivan County, Tennessee. 1978 GUILDAY ET AL.— BAKER BLUEE CAVE DEPOSIT 55 random. Surface contamination may be present due to modern plant roots occurring throughout the de- posit and perhaps from rodent burrowing. This is suggested by the relatively late date at the 4-5 ft level. Dates below 6 ft indicate a late Wisconsinan sequence. The basal date of 19,100 ± 850 years BP suggests that infilling began in full-glacial times. The faunal interpretation indicates a transition from cool-tem- perate to boreal as deposition proceeded. This sug- gests that deposition occurred during the recovery phase of an interstadial. Alexsis Dreimanis, Uni- versity of Western Ontario (personal communica- tion), tentatively suggests that deposition at Baker Bluff Cave may have begun during the Connersville Interstadial, from 22,000 years BP to slightly less than 21,000 years BP. During this time the Lauren- tide ice mass retreated from its maximum boundary some 40 to 100 km over a broad front in Indiana and Ohio and the climate was apparently less se- vere. “The presence of spruce wood and even ‘for- est bed’ in the Connersville interstadial silts sug- gests that open woodland probably reoccupied the area deglaciated during this interstade, and the cli- mate was not too rigorous” (Dreimanis, 1977:74). It is possible that the more temperate conditions suggested by the lower level paleofauna at Baker Bluff Cave may record this interstadial or the influ- ence of its later stages as climatic deterioration be- gan again. The depositional sequence may record the transition from interstadial to renewed glacial conditions. Pollen studies from Quicksand and Bob Black Ponds, Bartow County, Georgia, 34°19'30"N, 310 km SW of the cave in the Ridge and Valley province, do not seem to indicate any interstadial oscillations after 22,900-20,100 years BP. Rather, a Pinus-Picea full-glacial boreal pollen assemblage followed in early postglacial times by a Quercus- dominated flora superseded by a modern Pinus- dominant woodland (Watts, 1970), is indicated. It remains to be seen whether short-term inter- stadial oscillations produced relatively large-scale biotic adjustments as far south as Tennessee. The faunal change recorded in the Baker Bluff Cave de- posits suggests that they might, assuming that the accompanying '^C dates are correct. Watts' data suggest that they do not. The presence of tree roots throughout the excavation may cast doubt as to the validity of the dates. They may be too young and the deposit may represent a time interval prior to the full Wisconsinan maximum when climatic con- ditions were worsening. THE RIDGE AND VALLEY PROVINCE FAUNAL GRADIENT Three late Pleistocene cave deposits. New Paris No. 4, Pennsylvania, Clark’s Cave, Virginia, and Baker Bluff Cave, Tennessee, lie along a 550 km transit of the Ridge and Valley physiographic prov- ince, 150 km, 340 km and 370 km respectively southeast of the terminal Wisconsinan moraines (Pig. 1). Sites and morainal lines trend northeast/ southwest so that the distance differential south of the maximum continental glaciation is less than their actual distances apart. The three sites are spaced approximately two degrees of latitude from each other and extend from 40°N to 36°30'N. The maximum southern limit of the Laurentide Ice Sheet was approximately 39°N in the central low- lands of Illinois, Indiana, and western Ohio (Plint, 1971). The sites lie in the unglaciated portion of the Ridge and Valley province (= Valley and Ridge province of Hunt, 1974), a narrow belt of long even- crested NE/SW trending parallel ridges and inter- montane valleys formed of folded Paleozoic rocks 80 to 120 km wide lying between the Appalachian Plateau (Allegheny Plateau to the north, Cumber- land Plateau to the south) and the higher pre-Cam- brian Great Smoky Mountain/Blue Ridge/South Mountain range on its eastern boundary. Ridgetop elevations average 480 m, valley floors 360 m. The central portion of the Ridge and Valley province lies in the rainshadow of the Appalachian Plateau. This is especially marked in its central portion and has produced mildly xeric “shale barrens” that oc- cur along the crest of low shale hills on the valley floors from southern Pennsylvania south to south- ern Virginia (Keener, 1970) and support several plant endemics of midwestern and coastal plain af- finities. The drop in precipitation may be as much as 25-35 cm just east of the plateau at New Paris No. 4, Pennsylvania; it is also apparent at the lati- tude of Clark’s Cave, Virginia, but is hardly evident as far south as Baker Bluff, Tennessee. The topography of the Ridge and Valley province is such that a number of long parallel sheltered val- 56 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 11 Fig. 13. — Voles, Family Arvicolidae (Ondatra zibethicus omitted). Stratigraphic distribution. Baker Bluff Cave, Sullivan County, Tennessee. 1978 GUILDAY ET AL.— BAKER BLUFF CAVE DEPOSIT 57 leys lead south or southwest from the former con- tinental glacial margin directly into the southland unencumbered by any topographic obstacles for over 1,300 km. These long conduits, walled by ridges, should have furnished ideal migratory routes for northern late Pleistocene megafauna. They could also have provided relatively sheltered lanes for the northern penetration of tapir or armadillo into the central Appalachians. The three sites are all late Wisconsinan or early Holocene but may not be exactly contemporane- ous. New Paris No. 4, Pennsylvania, was a 10-m sinkhole fissure choked with colluvium, clay, and bones with a pollen profile indicative of an open PicealPinus cf. banksiana woodland and a date of 11,300 ± 1,000 years BP. Infilling was relatively rapid and the site data is interpreted as representing a transition from an open to a closed boreal wood- land. Remains of 2,430 individual mammals were recovered. Clark’s Cave, Virginia, 243 km to the south of New Paris No. 4, was an extinct raptor roost at the mouth of a large cave. Its precise age is unknown, probably late Wisconsinan. Remains of 4,343 indi- vidual mammals were recovered. Baker Bluff Cave, Tennessee, 307 km south of Clark’s Cave, 550 km south of New Paris No. 4, was a 3-m stratified fissure fill just inside the ante- chamber of a small cave. Radiocarbon dating sug- gests that deposition began at about 19,000 years BP and the fauna records a transition from cool/ temperate deciduous/coniferous woodland to boreal coniferous parkland. The top 1 m of the deposit was disturbed by both an aboriginal occupation and modern private excavations, so that evidence of the Pleistocene/Holocene transition was destroyed and late Pleistocene and Holocene material was inter- mingled. Remains of 1,909 individual mammals were recovered. On the basis of ^‘‘C dates the mid-levels of the Baker Bluff deposit, 10,560 to 11,640 years BP are contemporaneous with the New Paris No. 4 depos- it. The Clark’s Cave deposit is no younger than the late Wisconsinan/early Holocene faunal change about 10,000 years BP. All three sites contain boreal, mid-continental, and eastern temperate elements in their respective paleofaunas. They demonstrate a faunal gradient both quantitatively and qualitatively in the direction boreal to temperate, from north to south. Only small mammals, shrews (Soricidae), squirrels (Sciuridae), and voles (Arvicolidae) are analyzed because of their abundance in each of the three sites (Figs. 14, 15, 16). It is difficult to obtain a modern faunal equivalent for comparison. Most modern mammal collections are not heavily concentrated but reflect the regional ecology, agricultural lands, and multiple-growth woodlands, so that the relative numbers of small mammals do not represent the primeval climax fau- na under the present climatic regime. Data are pre- sented, however, for 1,447 soricids, 407 sciurids, and 1,367 arvicolids in the CMNH mammal collec- tions from the Ridge and Valley section of central Pennsylvania — 39°43'N to 42°N (Gifford and White- bread, 1951; Roslund, 1951). Forest cover within the area today varies from 19% to 98% with a 22 county average of 61%. Collecting was concentrat- ed in woodlands, forest edge, and meadowlands. Derived percentages of small mammals do not ap- proximate those of primeval times but do accurately represent the present picture. Small mammals at New Paris No. 4 were con- centrated by tumbling into a funnel-shaped fissure opening; those at Clark’s Cave and Baker Bluff by the hunting activities of birds of prey. The former site represents a local sample; the latter two micro- faunas were drawn from a larger collecting area. Owls are opportunistic feeders, taking small verte- brates as they are encountered, but may differ in their habitat preferences. Some, the barn owl (Tyto alba ) or the short-eared owl (Asia flam mens), hunt open country; others, such as the long-eared owl (Asia at us), prefer woodland and thicket. Despite these obvious sources of bias and the fact that the data for New Paris No. 4 and Baker Bluff Cave represent pooled strata, comparative analysis does indicate some trends. As one progresses from north to south. New Paris No. 4, Pennsylvania, to Clark’s Cave, Virginia, to Baker Bluff, Tennessee, the respective paleofaunas become progressively less boreal. Species of northern affinities either dis- appear from some faunas (collared lemming, Di- crostonyx hudsonius), or are present in diminishing numbers, whereas temperate species increase in relative numbers from north to south. Although bo- real species are found at all three sites, the most southern of the paleofaunas. Baker Bluff Cave, most closely resembles the modern Pennsylvania temperate small mammal fauna. Some few species did not follow this boreal to temperate gradient. The heather vole (Phenacomys intennedius) and the least chipmunk (Eutamias minimus) increased in relative numbers with dimin- 58 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 11 UJ s o QC 8.S LU O oc ^ ^ c CO § ^ cc O Q. P < m g C 8 S $ CO CO 0 c c 0 c oc Q Ql < ^ 3 oS I— 0 P cn !< :o _J oc 8 r- O -C- -'-i O) W 9 CL :£ 0 0 "c5 c -I _§ o3 E 8 0 0 [I. OC Q z 3 _l C/3 0 cc 5S OT Q < LU QC CQ t 1 § o cc < Q l- z LU o LU QC O CO ii lO o in o CM CO CM Fig. 14. — Shrews, Family Soricidae, late Pleistocene and Recent relative faunal composition, various sites. Ridge and Valley province, eastern USA. 1978 GUILDAY ET AL.— BAKER BLUEE CAVE DEPOSIT 59 ishing latitude. The number of Eiitamias at any one site, however, is so low that relative numbers are meaningless, but the unexpected increase in Pheii- acomys to 7.1% of the Baker Bluff Cave vole fauna appears to have been a real event. It is supported by similar large relative numbers of Phenacoinys (5.6% of all voles) from Carrier Quarry Cave, also in Sullivan County, Tennessee. One interesting lump of cave breccia from Carrier Quarry Cave (CM 30217) contains an incomplete but uncrushed skull of Phenacoinys with associated mandibles and a partial skull of either Microtus pinetornin or M. ochrogaster in direct association, indicating con- temporaneity of these now allopatric boreal and temperate voles. The relative increase of Phena- coinys at Baker Bluff Cave may be associated with a marked increase in numbers of the midcontinental thirteen-lined ground squirrel (Spennophilns tricle- cemlineatus) and the relative absence of small mammals typical of mesic environments such as soricine shrews and the meadow vole (Microtus pennsylvanicus). The voles furnish the most striking evidence of the boreal affinities of these three late Pleistocene Ridge and Valley paleofaunas relative to the fauna of the Recent Appalachian area. Eive species cur- rently live in the area (Pig. 16, central Pennsylva- nia; Ondatra omitted from analysis), but nine species were present at New Paris No. 4, and at least eight at both Clark’s Cave and Baker Bluff Cave. The minimum number of individual voles of species still extant in the area constituted 62% of the New Paris No. 4 fauna (51% of that contributed by just two species, Clethrionoinys gapperi and Microtus pennsylvanicus, both of northern affini- ties). At Clark’s Cave this figure increased to 70% of the vole fauna, and at Baker Bluff Cave 91% of the number of individual voles were of species still extant at that latitude (of which Clethrionoinys gap- peri and Microtus pennsylvanicus comprised only 34%). Thus, although all three faunas contained the same large complement of vole species (only Di- crostonyx hudsonius dropping out of the Virginia and Tennessee sites), the percentage of boreal to temperate species, in terms of individual animals, dropped an 8% from New Paris No. 4 to Clark’s Cave, and an additional 21% from Clark’s Cave to Baker Bluff. The situation in the case of the shrews is different and reflects primarily altitudinal rather than latitu- dinal range shifts. There are seven species now in the mid- and southern Appalachian area. There were seven species at both New Paris No. 4 and Clark’s Cave, with an additional eighth at Baker Bluff Cave (Fig. 14). Different species were in- volved, however. Cryptotis parva, a temperate field form, now present at all three localities, did not occur in either the New Paris No. 4 or the Clark’s Cave paleofaunas, but the arctic shrew (Sorex arc- ticus), a boreal forest form, was added to the fauna of all three sites. There was an increase from north to south of blarinine shrews (Blarina, Cryptotis). But all soricine shrews (Sorex, Microsorex), with the exception of the smoky shrew 5. fuineus, de- creased in relative numbers with decreasing lati- tude. S. fuineus is the commonest soricine in the Appalachians today, with the broadest regional en- vironmental tolerance and might be considered the most temperate of those species of Sorex identified from the three sites. Changing proportions of blar- inine to soricine shrews from the three paleofaunas are New Paris No. 4, 39%; Clark's Cave, 47%; and Baker Bluff Cave, 61%. Species of sciurids also increased slightly to the south (Fig. 15). Six species of squirrels (Mannota omitted from analysis) are present in the area today. Five species were present at New Paris No. 4 and seven at both Clark’s and Baker Bluff caves. The additional two, Eutainias tniniinus and Spennoph- ilus tridecemlineatus, are of midwestern affinities; E. minimus favoring central and western boreal woodlands and S. tridecemlineatus grasslands. Squirrels can be divided into two groups — arbo- real (Sciurus, Tamiasciurus, Glaucomys) and ter- restrial (Tamias, Eutainias, Spermophilus). Arbo- real species were commonest at Clark’s Cave, Virginia, 69% of all sciurids. At New Paris No. 4, Pennsylvania, they comprised only 53% (50% of this Tamiasciurus hudsonicus and Glaucomys sa- brinus, the only two arboreal sciurids in the North American boreal forests), suggesting a more open boreal woodland. Relative numbers of arboreal to terrestrial squirrels were lowest at Baker Bluff Cave, 47% of the sciurid fauna, again suggesting an open woodland. But, although arboreal squirrels in general declined in importance at Baker Bluff Cave, one temperate species, Sciurus carolinensis, none- theless increased from 0% at New Paris No. 4, to 3% at Clark’s Cave, to 13% of all squirrels at Baker Bluff Cave. This suggests a sequential change in forest composition in the Ridge and Valley province north to south from a predominantly open boreal forest at New Paris No. 4, Pennsylvania, to a dens- er boreal woodland cover at Clark’s Cave, Virginia, LATITUDINAL DISTRIBUTION OF SQUIRRELS, SCIURIDAE Ridge & Valley Province, Appalachian Mountains Pennsylvania through Tennessee Recent & Late Pleistocene 60 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 11 Fig. 15. — Squirrels, Family Sciuridae (Marmota monax omitted), late Pleistocene and Recent relative faunal composition, various sites. Ridge and Valley province, eastern USA. 1978 GUILDAY ET AL.— BAKER BLUEE CAVE DEPOSIT 61 to a mixed coniferous deciduous woodland at the latitude of Baker Bluff Cave, Tennessee. That all three sites were open to at least some extent is im- plied by the widespread late Pleistocene distribu- tion of the thirteen-lined ground squirrel (Sper- mophilus tridecemlineatiis) in eastern North American periglacial sites from Kentucky to east- ern Pennsylvania, south to at least Tennessee (Guil- day et al., 1977). Its numbers were high (26% of all sciurids) at Baker Bluff Cave, where it was the com- monest squirrel in the deposit, representing a dis- tinct grasslands element. Paleoenvironmental inferences based upon these few fossil faunas cannot be refined until additional sites are discovered and studied. But the analysis suggests that the effects of glacial cooling during the Wisconsinan glaciation profoundly affected mammalian distributions and, by inference, that of the entire biota in the Ridge and Valley province at least as far south as Tennessee. It also demon- strates the contemporaneity of both temperate and boreal species in these paleofaunas and the lack of any southern forms (except several extinct species. Tapir us cf. veroensis, Dasypus bellus, whose eco- logical requirements we do not know). There were no species in the Baker Bluff faunal sequence of southern affinities, which are at or near the northern edge of their modern distribution in the central Ap- palachians. So that while a more equitable climate is suggested, one that would allow presently allo- patric boreal and temperate species to coexist, the mean climate must have been cooler. The faunal gradient in late Wisconsinan times must have steepened south of Baker Bluff, Tennes- see. Of the many boreal forms from that site only Microtus pennsylvaniciis, Erethizon dorsatiim, and Vulpes vulpes penetrated as far south as Florida (Webb, 1974). Ridge and Valley karst areas extend both north and south of the sites discussed here so that the late Pleistocene faunal gradient can be ex- tended and refined by future work. Ladds Quarry paleofauna, Bartow County, Geor- gia, 34°09'N, 84°50'W, 320 km SW of Baker Bluff Cave, just east of the Ridge and Valley province and south of the Great Smoky Mountains, at an altitude of about 300 m, also presents a mixture of presently allopatric species — a northern element, wood turtle (Clemmys insculpta), spruce grouse (Canachites canadensis), masked shrew (Sorex ci- nereus), fisher {Martes pennanti). New England cottontail (Sylvilagus transitionalis), southern bog lemming {Synaptomys cooperi); a southern faunal element which does not occur at late Pleistocene sites farther north, southern toad {Bufo terrestris), opossum (Didelphis virginianus), rice rat (Oryzo- mys palustris), cotton rat (Siginodon hispidus), round-tailed muskrat {Neofiber alleni), hog-nosed skunk {Conepatus leuconotus), and ? jaguarundi (Felis (?) Herpailurus) (Holman, 1976; Ray, 1967; Wetmore, 1967). With the exception of the spruce grouse, which reaches its present southern Appalachian limits in the Adirondack Mountains of New York, 44°N, and the wood turtle, which reaches its southern limit in northern Virginia, 39°N, the other species of north- ern affinities were broadly distributed throughout the Appalachian area in Historic times near the lat- itude of Ladds Quarry. Unfortunately the age of the site is not known, or even if it is a synchronous local fauna (see discussions in Ray, 1965, and Lipps and Ray, 1967). The presence of Peromyscus cuni- berlandensis, a species otherwise known only from two Irvingtonian age sites in the Appalachians (the type locality Cumberland Cave, Maryland, Guilday and Handley, 1967; Trout Cave, West Virginia, CMNH collections; and absent to date from all Ap- palachian sites of Wisconsinan age or later), in con- junction with Platygonus compressus, a Wisconsi- nan peccary, suggests that the Ladds Quarry paleofauna may in fact be heterochronic. Remains of Neofiber alleni, the Recent round-tailed muskrat, a Wisconsinan to Recent species, now confined to Florida and southern Georgia, 31°N, 640 km south of the Ladds Quarry paleofauna, suggests a Wis- consinan date and milder winter temperatures (Ray, 1965; Frazier, 1977). But spruce (Picea) pollen, in- dicative of a cooler environment has also been re- corded from the site (Benninghoff and Stephenson, 1967). If the Ladds Quarry paleofauna proves to be con- temporaneous with the late Wisconsinan Baker Bluff Cave faunal sequence it demonstrates a steep- ening of the faunal gradient south of Baker Bluff Cave. As Ray pointed out, the problem can only be resolved by further fieldwork. A comparison of the late Wisconsinan faunal gra- dient of the Ridge and Valley province with that of today shows significant differences. Based upon Recent small mammal ecological requirements the late Pleistocene gradient, Pennsylvania to Tennes- see, was in the direction “boreal" to “temperate." Today the same gradient runs “temperate" to “aus- tral." This reflects the regional postglacial rise in mean temperature. LATITUDINAL DISTRIBUTION OF VOLES, ARVICOLIDAE Ridge & Valley Province, Appalachian Mountains Pennsylvania through Tennessee Recent & Late Pleistocene RECENT DATA FROM GIFFORD & WHITEBREAD, 1951; ROSLUND, 1951 62 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 11 Fig. 16. — Voles, Family Arvicolidae {Ondatra zibethicus omitted), late Pleistocene and Recent relative faunal composition, various sites. Ridge and Valley province, eastern USA. 1978 GUILDAY ET AL.— BAKER BLUFF CAVE DEPOSIT 63 Perhaps the most significant fact to be drawn from the present study is that the faunal gradient is more extreme today than it was under late glacial conditions although different combinations of species are involved. This is best illustrated by comparing the two gradient sets of small rodents — cricetids, arvicolids, and zapodids. In south-central Pennsylvania, at the gradient’s northern end, there are today 10 species of native small rodents (Pero- myscus ieucopus, Peromyscus manlculatus, Neot- oma floridana, Clethrionomys gapperi, Synapto- mys cooped, Microtus pennsylvaniciis, Microtus pinetorum, Ondatra zibethicus, Zapus hudsonius, and Napaeozapus insignis; Gifford and White- bread, 1951). All but two, Napaeozapus insignis and Clethrionomys gapperi, occur throughout the gradient transect in the Ridge and Valley province south to East Tennessee. At its southern terminus at Baker Bluff Cave, 12 species of small rodents have been reported (Reithrodontomys hiimidis, Oryzomys palustris, Peromyscus Ieucopus, Pero- myscus maniculatus, Ochrotomys nuttalli, Sigmo- don hispidus, Neotoma floridana, Microtus penn- sylvanicus, Microtus pinetorum. Ondatra zibethicus, Synaptomys cooperi, and Zapus hudsonius; Smith et al., 1974, Hall and Kelson, 1959). Four of these, Reithrodontomys , Ochrotomys, Sigmodon, and Oryzomys, do not range as far north as south-cen- tral Pennsylvania today. In other words, six out of a total of 14 species involved have restricted ranges within the transit (two reach their southern limits. four their northern limits); 43% of the small rodent fauna in the modern faunal gradient were involved in range termination. During the late Pleistocene 15 species of small rodents were identified from the northern end of the gradient (New Paris No. 4, Pennsylvania) and 14 from its southern limit (Baker Bluff Cave, Tennes- see). Only one species Dicrostonyx hudsonius , present at New Paris No. 4, failed to occur through- out the gradient, although there were changes in abundance of other species from north to south. There were no northern range terminations. Only 7% of the late Pleistocene small rodent fauna (one species) was involved in range terminations, con- trasted with 43% in the modern faunal gradient. A comparison of these two Ridge and Valley fau- nal gradients indicates that environmental condi- tions during late Pleistocene times were definitely boreal in character but more equable than today throughout the transect. This reinforces the concept of Pleistocene climatic equability developed pri- marily from Great Plains fossil faunas by the late C. W. Hibbard and others since the 1960s. Graham ( 1976) suggests that the increasing continental cli- mate of the postglacial (heightened seasonality) ac- centuated climatic gradients, narrowed ecotones, and established the present diversities and distri- butional patterns. The Baker Bluff Cave faunal se- quence and other Ridge and Valley late Pleistocene faunas bear out this concept. FAMILY COMPOSITION OF THREE RIDGE AND VALLEY MAMMALIAN PALEOFAUNAS Representatives of 20 mammalian families have been recovered from three large late Pleistocene cave faunas in the Ridge and Valley province — New Paris No. 4, Clark’s Cave, and Baker Bluff Cave (Table 17). Dasypodidae, Tayassuidae, and Tapiri- dae represented by extinct species, no longer occur in the Appalachian region. The opossum sole rep- resentative of the Didelphidae, has not been found in a Pleistocene context in any Ridge and Valley site, although it is common in the area today. The Baker Bluff Cave record is based on bones of a Recent animal from the surface deposits. Small mammals, marmot size or less, predomi- nate in these faunas where they form 99% of all individual mammals from New Paris No. 4, 99% from Clark’s Cave, and 96% from Baker Bluff Cave. The large numbers of small mammals reflects raptor bias at Clark’s Cave and Baker Bluff Cave, both owl-roost deposits, and at New Paris No. 4 inadvertent trapping of unwary animals tumbling into a funnel-shaped sinkhole. Bats, Vespertilionidae, fluctuated dramatically from a high of 42% MNI at New Paris No. 4, rep- resenting natural mortality of the resident bat col- ony, to 36% MNI at Clark’s Cave, due to selective predation by raptors at a large cave system en- trance, to a low of 2.4% MNI at Baker Bluff Cave, attributable to a low resident bat population in that unsuitably small cave. Small rodents of the families Cricetidae, Arvi- colidae, and Zapodidae formed the largest portion of each fauna; 50% at New Paris No. 4, 54.5%, 64 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 11 Table 17. — Mammalian composition of three Appalachian cave deposits based upon minimum number of individuals (MNI). Family Common name New Paris No. 4. Pennsylvania (MNI = 2,957) Clark’s Cave, Virginia (MNI = 4,343) Baker Bluff Cave, Tennessee (MNI = 1,910) Didelphidae opossums — — .05% Soricidae shrews 3.61% 5.30% 14.55% Talpidae moles .10% .59% 1.67% Vespertilionidae evening bats 41.76% 35.78% 2.40% Dasypodidae armadillos — — .05% Leporidae rabbits and hares 1.75% .55% 4.97% Sciuridae squirrels 2.80% 2.69% 7.12% Castoridae beavers — — .20% Cricetidae New World rats and mice 8.04% 6.26% 18.53% Arvicolidae voles 41.02% 47.43% 47.22% Zapodidae jumping mice .57% .85% .30% Erethizontidae porcupines .10% .02% .15% Canidae wolves and foxes — .02% .20% Ursidae bears — .02% .20% Procyonidae raccoons — .02% .26% Mustelidae weasels, otters, etc. .16% .36% 1.09% Felidae cats — — .05% Tayassuidae peccaries .03% — .05% Cervidae deer — .04% .78% Tapiridae tapirs — — .05% 99.94% 99.91% 99.89% Clark’s Cave, and 66% at Baker Bluff Cave. Insec- tivores, Soricidae and Talpidae, formed a low 4% MNI at New Paris No. 4, and 6% MNI at Clark’s Cave, but a much higher 16% at Baker Bluff Cave, a mathematical artifact caused primarily by the low percentage of bats from the latter site. If bats had been eliminated from consideration at New Paris No. 4 and Clark’s Cave, insectivores would have constituted a larger percentage of those faunas. The percent of small rodents to shrews is actually much the same at all three sites, about the same as that of the Baker Bluff Cave local fauna. Carnivore remains — Canidae, Ursidae, Procyon- idae, Mustelidae, Eelidae — are rare in raptor sites. Only weasels, genus M us tela, appear with some consistency; their small size makes them vulnerable to predation. The almost total absence of carnivores from the sinkhole fauna of New Paris No. 4 is due to the small size of the fissure, the alertness and agility of most carnivores, and the absence of skunks (Mephitis) from that boreal fauna. (Skunks, because of their myopic, terrestrial habits, are com- mon in Holocene sinkhole faunas in the New Paris area, Guilday and Bender, 1958.) The slightly larger percentage of large mammal remains at Baker Bluff Cave, Tennessee, all represented by isolated teeth or bone fragments, is due to woodrat scavenging. None of these sites present true cross-sections of the paleofaunas they sample. However, they may fairly approximate the relative abundance of small mammals present in the samples areas that can be profitably compared with other local faunas where the depositional and recovery factors are similar. Relative percents at these three sites were based upon recovery procedures using screens of both 5- mm and 1-mm grid size for specimen recovery. The sites had not been exposed to selective processes such as weathering, water-sorting, or differential chemical decomposition, and are all primary de- posits. As a result postdepositional disruption was minimal and specimen recovery thorough. These factors could drastically affect relative faunal com- position and should be kept in mind when compar- ing fossil faunas. 1978 GUILDAY ET AL.— BAKER BLUFE CAVE DEPOSIT 65 LITERATURE CITED American Ornithologists’ Union. 1957. Check-list of North American birds. Lord Baltimore Press, Baltimore, Mary- land, 5th edition, 691 pp. Atwood, W. W. 1940. The physiographic provinces of North America. Athenaeum Press, Boston, xvi -f 536 pp. Auffenberg, W. 1963. The fossil snakes of Florida. Tulane Stud. Zook, 10:129-216. Banfield, a. W. F. 1974. The mammals of Canada. Univ. Toronto Press, xxv -l- 438 pp. Barbour, R. W., and W. H. Davis. 1974. Mammals of Ken- tucky. Univ. Press Kentucky, 322 pp. Barkalow, F. S., Jr. 1972. Vertebrate remains from archeo- logical sites in the Tennessee Valley of Alabama. Southern Indian Studies, 24:3—41. Benninghoff, W. S., and A. L. Stephenson. 1967. Pollen analysis of cave breccia from Ladds Locality, Bartow Coun- ty, Georgia. Bull. Georgia Acad. Sci., 25: 188-191 . Bogan, A. E. 1976. A zooarchaeological analysis of vertebrate remains from Chota-Tanasi, a Historic Cherokee village in East Tennessee. Unpublished M.S. dissert., Univ. Tennes- see, Knoxville, 156 pp. Brattstrom, B. H. 1967. A succession of Pliocene and Pleis- tocene snake faunas from the High Plains of the United States. Copeia, pp. 188-202. Broyles, B. J. 1966. Preliminary report: The St. Albans Site (46 Ka 27), Kanawha County, West Virginia. The West Vir- ginia Archeologist, 19:1-43. Cambron, J. W., and D. C. Hulse. 1964. Handbook of Ala- bama archaeology: Part I, point types. Privately printed, 170 pp. Coe, j. L. 1964. The formative cultures of the Carolina Pied- mont. Trans. Amer. Phil. Soc., n.s., 54(5): 1-130. CoNANT, R. 1975. A field guide to the reptiles and amphibians of eastern and central North America. Houghton Mifflin Co., Boston, 429 pp. Cope, E. D. 1899. Vertebrate remains from Port Kennedy bone deposit. J. Acad. Nat. Sci. Philadelphia, ser. 2, 11:193-267. CoRGAN, J. X. 1976. Vertebrate fossils of Tennessee. Tennes- see Dept, of Conservation, Div. of Geol. Bulk, 77: 1-100. Dice, L. R. 1943. The biotic provinces of North America. Univ. Michigan Press, Ann Arbor, 78 pp. Doutt, j. K. n.d. Taxonomic review of white-tailed deer (Odo- coileus virginianus) in eastern North America. Unpublished manuscript on file. Section of Mammals, Carnegie Mus. Nat. Hist., 45 pp. Doutt, J. K., C. A. Heppenstall, and J. E. Guilday. 1973. Mammals of Pennsylvania. Pennsylvania Game Com- mission, Harrisburg, Pennsylvania, 3rd ed., 288 pp. Dreimanis, a. 1977. Late Wisconsin glacial retreat in the Great Lakes region. North America. Ann. New York Acad. Sci., 288:70-89. Elint, R. F. 1971. Glacial and Quaternary geology. John Wiley & Sons, Inc., 892 pp. Frazier, M. K. 1977. New records of Neofiber leonardi (Ro- dentia: Cricetidae) and the paleoecology of the genus. J. Mamm., 58:368-373. Gidley, j. W., and C. L. Gazin. 1938. The Pleistocene ver- tebrate fauna from Cumberland Cave, Maryland. Bulk U.S. Nat. Mus., 171:1-99. Gifford, C. L., and R. Whitebread. 1951. Mammal survey of southcentral Pennsylvania. Final Report Pittman-Robert- son Project 38-R, Pennsylvania Game Commission, Harris- burg, Pennsylvania, 75 pp. Graham, R. W. 1976. Late Wisconsinan mammalian faunas and environmental gradients of the eastern United States. Paleobiology, 2:343-350. Graham, R. W., and H. A. Semken. 1976. Paleoecological signficance of the short-tailed shrew (Blarina), with a sys- tematic discussion of Blarina ozarkensis. J. Mamm., 57:433- 449. Graves, F. F., Jr., and M. J. Harvey. 1974. Distribution of Chiroptera in western Tennessee. J. Tennessee Acad. Sci., 49:106-109. Guilday, J. E. 1958. The prehistoric distribution of the opos- sum. J. Mamm., 39:39^3. . 1961. Abnormal lower 3rd molar in Odocoileus. J. Mamm., 42:551-553. . 1962. The Pleistocene local fauna of the Natural Chim- neys, Augusta County, Virginia. Ann. Carnegie Mus., 36:87-122. . 1971. Biological and archeological analysis of bones from a 17th Century Indian village (46 Pu 31), Putnan Coun- ty, West Virginia. West Virginia Geol. and Econ. Surv., Rept. of Archeok Invest., no. 4, 64 pp. Guilday, J. E., and M. S. Bender. 1958. A Recent fissure deposit in Bedford County, Pennsylvania. Ann. Carnegie Mus., 35:127-138. . 1960. Late Pleistocene records of the yellow-cheeked vole, Microtus xanthognathns (Leach). Ann. Carnegie Mus., 35:315-330. Guilday, J. E., H. W. Hamilton, and A. D. McCrady. 1969. The Pleistocene vertebrate fauna of Robinson Cave, Overton County, Tennessee. Palaeovertebrata, 2:25-75. . 1971. The Welsh Cave peccaries (Pkitygonus) and as- sociated fauna, Kentucky Pleistocene. Ann. Carnegie Mus., 43:249-320. Guilday, J. E., H. W. Hamilton, and P. W. Parmalee. 1975. Caribou (Rangifer tarandus L.) from the Pleistocene of Tennessee. J. Tennessee Acad. Sci., 50: 109-1 12. Guilday, J. E., and C. O. Handley, Jr. 1967. A new Pero- myscus (Rodentia: Cricetidae) from the Pleistocene of Maryland. Ann. Carnegie Mus., 39:91-103. Guilday, J. E., P. S. Martin, and A. D. McCrady. 1964. New Paris No. 4: A Pleistocene cave deposit in Bedford County, Pennsylvania. Bulk Nat. Speleo. Soc., 26:121-194. Guilday, J. E., and A. D. McCrady. 1966. Armadillo re- mains from Tennessee and West Virginia caves. Bulk Nat. Speleo. Soc., 28:183-184. Guilday, J. E., and H. McGinnis. 1972. Jaguar (Panthera onca) remains from Big Bone Cave, Tennessee and east central North America. Bulk Nat. Speleo. Soc., 34: 1-14. Guilday, J. E., and P. W. Parmalee. 1965. Animal remains from the Sheep rock Shelter (36 Hu 1), Huntingdon County, Pennsylvania. Pennsylvania Archaeologist, 35:34—49. . 1972. Quaternary periglacial records of voles of the genus Phenacomys Merriam (Cricetidae: Rodentia). Qua- ternary Res., 2:170-175. Guilday, J. E., P. W. Parmalee, and H. W. Hamilton. 1977. The Clark's Cave bone deposit and the late Pleisto- 66 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 11 cene paleoecology of the central Appalachian Mountains of Virginia. Bull. Carnegie Mus. Nat. Hist., 2:1-87. GuiLDAY, J. E., P. W. Parmalee, and D. P. Tanner. 1962. Aboriginal butchering techniques at the Eschelman Site (36 La 12), Lancaster County, Pennsylvania. Pennsylvania Ar- chaeologist, 32:59-83. Hall, E. R., and K. R. Kelson. 1959. The mammals of North America. The Ronald Press Co., New York, Lxxx -L 1-546 -H 79: 2: viii -L 547-1083 -L 79. Hallberg, G. R., H. a. Semken, and L. C. Davis. 1974. Quaternary records of Microlus xanthognathus (Leach), the yellow-cheeked vole, from northwestern Arkansas and southwestern Iowa. J. Mamm., 55:640-645. Handley, C. O., Jr. 1959. A revision of American bats of the genera Eitderma and Plecotus. Proc. U.S. Nat. Mus., 110:95-246. Hay, O. P. 1920. Descriptions of some Pleistocene vertebrates found in the United States. Proc. U.S. Nat. Mus., 58:83- 146. Holman, J. A. 1962. A Texas Pleistocene Herpetofauna. Co- peia, pp. 255-261. . 1963. Late Pleistocene amphibians and reptiles of the Clear Creek and Ben Franklin local faunas of Texas. J. Grad. Res. Center, 31:152-167. . 1965. A Late Pleistocene herpetofauna from Missouri. Trans. Illinois Acad. Sci., 58:190-194. . 1967. A Pleistocene herpetofauna from Ladds, Georgia. Bull. Georgia Acad. Sci., 25:154-166. . 1968. Lower Oligocene amphibians from Saskatche- wan. Quart. J. Florida Acad. Sci., 31:273-289. . 1969. Herpetofauna of the Pleistocene Slaton Local Fauna of Texas. Southwestern Nat., 14:203-212. . 1973u. A new Pliocene snake, genus Elaphe, from Oklahoma. Copeia, pp. 574-580. . 19736. Reptiles of the Egelhoff Local Fauna (Upper Miocene) of Nebraska. Contrib. Mus. Paleontol., Univ. Michigan, 24:125-134. . 1976. Paleoclimatic implications of “ecologically in- compatible” herpetological species (late Pleistocene: south- eastern United States). Herpetologica, 32:290-295. Howell, A. H. 1921. A biological survey of Alabama: I, Phys- iography and life zones; II, The mammals. N. Amer. Fau- na, 45:1-88. Howell, J. C., and C. H. Conaway. 1952. Observations on the mammals of the Cumberland Mountains of Tennessee. J. Tennessee Acad. Sci., 27:153-158. Hunt, C. B. 1974. Natural regions of the United States and Canada. W. H. Freeman & Co., San Francisco, xii + 725 PP Keener, C. S. 1970. The natural history of the Mid-Appala- chian Shale Barren Flora. Pp. 215-248, in The distributional history of the biota of the southern Appalachians, Part II: Flora (Perry C. Holt, ed.), 414 pp. Kellberg, J. M. 1963. Chert and ‘Flint’ of the Tennessee area. Tennessee Archaeologist, 19:1-7. Kellogg, R. 1939. Annotated list of Tennessee mammals. Proc. U.S. Nat. Mus., 86:245-303. King, J. E., and T. R. Van Devender. 1977. Pollen analysis of fossil packrat middens from the Sonoran Desert. Quater- nary Res., 8:191-204. Kneberg, M. 1956. Some important projectile points found in the Tennessee area. Tennessee Archaeologist, 12:17-28. . 1957. Chipped stone artifacts of the Tennessee Valley area. Tennessee Archaeologist, 13:55-66. Linzey, a. V., AND D. W. Linzey. 1971. Mammals of Great Smoky Mountains National Park. Univ. Tennessee Press, Knoxville, 1 14 pp. Lipps, L., AND C. E. Ray. 1967. The Pleistocene fossiliferous deposit at Ladds, Bartow County, Georgia. Bull. Georgia Acad. Sci., 25:113-119. Lundelius, E. L., Jr., and B. H. Slaughter. 1976. Notes on American Pleistocene tapirs. Pp. 226-243, in Essays on paleontology in honour of Loris Shano Russell (C. S. Churcher, ed.). Life Sci. Misc. Publ., Royal Ontario Mus., 292 pp. Martin, R. A. 1974. Fossil vertebrates from the Haile XIV A Fauna, Alachua County. Pp. 100-113, in Pleistocene mam- mals of Florida (S. David Webb, ed.), Univ. Florida Press, Gainesville, x -L 270 pp. Martin, R. A., and S. D. Webb. 1974. Late Pleistocene mam- mals from the Devil’s Den Fauna, Levy County. Pp. 114- 145, in Pleistocene mammals of Florida (S. David Webb, ed.), Univ. Florida Press, Gainesville, x -I- 270 pp. Meszoely, C. A. 1967. A new cryptobranchid salamander from the Early Eocene of Wyoming. Copeia, pp. 346-349. Parmalee, P. W. 1967. A Recent cave bone deposit in south- western Illinois. Bull. Nat. Speleo. Soc., 29:119-147. Parmalee, P. W., A. E. Bogan, and J. E. Guilday. 1976. First records of the giant beaver (Casloroides ohioensis) from Eastern Tennessee. J. Tennessee Acad. Sci., 51:87- 88. Parmalee, P. W., P. J. Munson, and J. E. Guilday. In press. The Pleistocene mammalian fauna of Harrodsburg Crevice, Monroe County, Indiana. Bull. Nat. Speleo. Soc. Phillips, L., and M. Richmond. 1971. The distribution of vertebrates on the islands of Boone and Watauga lakes. J. Tennessee Acad. Sci., 46:124-129. Ray, C. E. 1958. Additions to the Pleistocene mammalian fauna from Melbourne, Florida. Bull. Mus. Comp. Zool., 119:421- - 449. . 1965. A new chipmunk, Tamias aristus, from the Pleis- tocene of Georgia. J. Paleontol., 39:1016-1022. . 1967. Pleistocene mammals from Ladds, Bartow Coun- ty, Georgia. Bull. Georgia Acad. Sci., 25:120-150. Ritchie, W. A. 1961. A typology and nomenclature for New York projectile points. Bull. New York State Mus. Sci. Serv. 384:1-119. Roslund, H. R. 1951. Mammal survey of northcentral Penn- sylvania. Final Report Pittman-Robertson Project 37-R, Pennsylvania Game Commission, Harrisburg, Pennsylva- nia, 55 pp. Scully, E. G. 1951. Some Central Mississippi Valley projectile point types. Mimeograph, Mus. Anthropol., Univ. Michi- gan, Ann Arbor. Severinghaus, W. D., and L. E. Beasley. 1973. A survey of the microtine and zapodid rodents of West Tennessee. J. Tennessee Acad. Sci., 48:129-133. Smith, C. R., J. Giles, M. E. Richmond, J. Nagel, and D. W. Yambert. 1974. The mammals of northeastern Ten- nessee. J. Tennessee Acad. Sci., 49:88-94. SoLER, E. I. 1950. On the status of the family Desmognathidae. Univ. Kansas Sci. Bull., 33:459^80. Tihen,J. A. 1954. A Kansas Pleistocene herpetofauna. Copeia, pp. 217-221. 1978 GUILDAY ET AL.— BAKER BLUEE CAVE DEPOSIT 67 . 1958. Comments on the osteology and phylogeny of ambystomatid salamanders. Bull. Florida State Mus., 3:1- 50. . 1962. A review of New World fossil bufonids. Amer. Midland Nat., 68:1-50. U. S. Department OF Agriculture. 1941. Climate and Man. Yearbook of Agri., 1248 pp. Watts, W. A. 1970. The full-glacial vegetation of northwestern Georgia. Ecology, 51:17-33. Webb, S. D. [ed.]. 1974. Pleistocene mammals of Florida. Univ. Florida Press, Gainesville, x -i- 270 pp. Wells, P. V. 1976. Macrofossil analysis of wood rat middens as a key to the Quaternary vegetational history of arid America. Quaternary Res., 6:223-248. Wetmore, a. 1962. Ice Age birds in Virginia. The Raven, 33(41:3. . 1967. Pleistocene Aves from Ladds, Georgia. Bull. Georgia Acad. Sci., 25:151-153. WiLi iAMS, S. C. 1927. Lieut. Henry Timberlake's memoirs, 1756-1765; with annotation, introduction and index. John- son City, Tennessee, 197 pp. Wrigley. R. E. 1972. Systematics and biology of the wood- land jumping mouse, Napaeozopus insignis. Illinois Biol. Monogr., 47: 1-1 18. ./>■ Copies of the following Bulletins of Carnegie Museum of Natural History may be obtained at the prices listed from the Publications Secretary, Carnegie Museum of Natural History, 4400 Forbes Avenue, Pitts- burgh, Pennsylvania 15213. 1. Krishtalka, L. 1976. Early Tertiary Adapisoricidae and Erinaceidae (Mammalia, Insectivora) of North America. 40 pp., 13 figs $2.50 2. Guilday, J. E., P. W. Parmalee, and H. W. Hamilton. 1977. The Clark’s Cave bone deposit and the late Pleistocene paleoecology of the central Appalachian Mountains of Virginia. 88 pp., 21 figs. $12.00 3. Wetzel, R. M. 1977. The Chacoan peccary, Catagonus wagneri (Rusconi). 36 pp., 10 figs. . . $6.00 4. Coombs, M. C. 1978. Reevaluation of early Miocene North American Moropus (Perissodactyla, Chalicotheriidae, Schizotheriinae). 62 pp., 28 figs $5.00 5. Clench, M. H., and R. C. Leberman. 1978. Weights of 151 species of Pennsylvania birds analyzed by month, age, and sex. 87 pp $5.00 6. Schlitter, D. A. (ed.). 1978. Ecology and taxonomy of African small mammals. 214 pp., 48 figs. $15.00 7. Raikow,R. J. 1978. Appendicular myology and relationships of the New World nine-primaried oscines (Aves:Passeriformes). 43 pp., 10 figs $3.50 8. Berman, D. S, and J. S. McIntosh. 1978. Skull and relationships of the Upper Jurassic sauropod Apatosaurus (Reptilia, Saurischia). 35 pp., 11 figs $3.00 9. Setoguchi, T. 1978. Paleontology and geology of the Badwater Creek area, central Wyoming. Part 16. The Cedar Ridge local fauna (Late Oligocene). 61 pp., 30 figs $4.50 10. Williams, D. F. 1978. Systematics and ecogeographic variation of the Apache pocket mouse (Roden- tia: Heteromyidae). 57 pp., 23 figs $4.00 BULLETIN of CARNEGIE MUSEUM OF NATURAL HISTORY REVISION OF THE ANTILLEAN BATS OF THE GENUS BRACHYPHYLLA (MAMMALIA: PHYLLOSTOMATIDAE) (permanent address: PIERRE SWANEPOEL Resident Museum Specialist, Section of Mammals Kaffrarian Museum, King William's Town, 5600, Republic of South Africa) HUGH H. GENOWAYS Curator, Section of Mammals NUMBER 12 PITTSBURGH, 1978 BULLETIN OL CARNEGIE MUSEUM OE NATURAL HISTORY Number 12, pages 1-53, figures I-I7, tables 1-9 Issued 27 December 1978 Price: $4.00 a copy Craig C. Black, Director Editorial Staff: Hugh H. Genoways, Editor-, Duane A. Schlitter, Associate Editor', Stephen L. Williams, Associate Editor-, Teresa M. Bona, Technical Assistant. © 1978 by the Trustees of Carnegie Institute, all rights reserved. CARNEGIE MUSEUM OE NATURAL HISTORY, 4400 EORBES AVENUE PITTSBURGH, PENNSYLVANIA 15213 CONTENTS Abstract 5 Introduction 5 Materials and Methods 7 Non-geographic Variation 8 Variation with Age 8 Secondary Sexual Variation 8 Individual Variation 17 Specific Relationships 18 Univariate Analyses 18 Multivariate Analyses 18 Variation in Color 20 Taxonomic Conclusions 26 Systematic Accounts 26 Genus Brachyphylla 26 Definition 26 Ecology 27 Brachyphylla cavernarum 27 Distribution 27 Diagnosis 27 Comparisons 27 Geographic Variation 27 Univariate Analyses 27 Multivariate Analyses 29 Taxonomic Conclusions 35 Brachyphylla cavernarum cavernarum 37 Brachyphylla cavernarum intermedia 38 Brachyphylla cavernarum minor 39 Brachyphylla nana 39 Distribution 39 Diagnosis 39 Comparisons 39 Geographic Variation 39 Univariate Analyses 39 Multivariate Analyses 40 Taxonomic Conclusions 48 Status of Fossil Specimens 49 Brachyphylla nana 49 Acknowledgments 51 Literature Cited 52 9^ • Ursi- 1*1. ‘vnr •• < V. ?if f ABSTRACT Nongeographic and geographic variation have been analyzed in the genus Brachyphylla , which belongs to the Antillean endem- ic subfamily Phyllonycterinae of the family Phyllostomatidae. Males were found to be generally larger than females; therefore, the sexes were analyzed separately for geographic variation. Ex- ternal measurements except length of forearm were found to dis- play a high degree of individual variation. They were not used in subsequent analyses. Of cranial measurements, greatest length of skull and condylobasal length showed the least individual varia- tion, whereas palatal length, postorbital breadth (in samples from west of the Mona Passage only), and rostral width at canines showed relatively high coefficients of variation. Variation in color was found not to follow any geographic pattern. Two species — Brachyphylla cavernanim and B. nana — were recognized in the genus. B. cavernanim occurs on Puerto Rico, the Virgin Islands, and the Lesser Antilles as far south as St. Vin- cent. Three subspecies are recognized. Populations of large bats occur on St. Croix in the Virgin Islands and the Lesser Antilles as far south as St. Vincent. The smallest individuals occur only on the island of Barbados. Populations of bats of intermediate size, described herein as a new subspecies, occur on Puerto Rico and most of the Virgin Islands. Brachyphylla nana is a monotypic species occurring on Cuba, Isle of Pines, Grand Cayman, Middle Caicos, and Hispaniola and as a sub-Recent fossil on Jamaica. INTRODUCTION Bats of the genus Brachyphylla belong to the subfamily Phyllonycterinae. This subfamily, which is endemic to the West Indies, belongs to the family Phyllostomatidae, the New World leaf-nosed bats. Members of the genus Brachyphylla occur through- out most of the Greater and Lesser Antilles south to St. Vincent and Barbados, and in the Bahamas on Middle Caicos Island. The genus is known on Ja- maica only from fossil material. The genus Brachyphylla was erected by Gray in 1834 to include the new speciesB. cavernarum . Gray (1838) placed the genus in the tribe Phyllostomina of the family Vespertilionidae. Gervais (1855-1856) placed the genus in the tribe Stenodermina, which subsequently was recognized as the subfamily Sten- oderminae of the family Phyllostomatidae. In 1866, Gray erected the tribe Brachyphyllina with Brach- yphylla as the sole genus. Later, Dobson (1878) in- cluded Brachyphylla in his group Stenodermata but stated that it was the most closely related of all known genera of phyllostomatids to the desmodon- tines. McDaniel ( 1976) in his study of the brain anat- omy also thought that Brachyphylla was most closely allied to the Desmodontinae or possibly the Stenoderminae. H. Allen (1898) placed Brachyphyl- la in the subfamily Glossophaginae, but separated it in a group termed Brachyphyllina along with Fhyl- lonycteris and Erophylla . Miller ( 1898) in describing Reithronycteris followed this arrangement but clear- ly allied Reithronycteris with Brachyphylla , Phyllo- nycteris , and Erophylla. Miller later changed his opinion and stated that he (Miller, 1907) could detect no indication that Brachyphylla was a phyllonycter- ine and placed it in the subfamily Stenoderminae. Here it remained until Silva-Taboada and Pine ( 1969) presented evidence based on osteology, behavioral characteristics, and host-parasite specificity for con- sidering Brachyphylla a member of the subfamily Phyllonycterinae. Slaughter (1970) reflected on the similarity between this genus and Stiirnira and thought it possible that these two genera, in addition to the glossophagines and stenodermines were re- lated to some unknown common ancestor, and con- cluded that the dentition offers no evidence that Brachyphylla is any more closely related to the sten- odermines than Sturnira. It should be pointed out, however, that Sturnira is now included in the Sten- oderminae by most authorities. In erecting the genus Brachyphylla , Gray (1834) described cavernarum from St. Vincent as the first species. Subsequently three additional species have been described, nana by Miller (1902/e 13 — St. Thomas, Virgin Is- lands; sample 14 — St. Croix, Virgin Islands; sample 15 — Saba; sample 16 — St. Eustatius; ,su/up/e 17 — Montserrat; sample 18 — Anguilla; sample 19 — St. Martin; sample 20 — Barbuda; sample 21 — Antigua; sample 22 — Guadeloupe; sample 23 — Dominica; sample 24 — Martinique; sample 25 — St. Lucia; sample 26 — St. Vincent; sample 27 — Barbados. Selected measurements were also taken from fragmented Brachyphylla Pleistocene or sub-Recent fossil material from Ja- maica. In order to compare these measurements to extant material similar measurements were also taken from adult specimens from the selected localities including both Brachyphylla cavernarum and B. nana. These were grouped into seven samples as follows: sample a — Cuba (five males, five females); sample b — Middle Caicos (five males, five females); .vw/«/?/e c — Dominican Republic (five males, five females); sample d — Jamaica (fossils); sample e — Puerto Rico (five males, five females);. vu/np/e/ — St. John (five males, two females); sample g — Norman (five males, five fe- males). The following measurements were taken from this material: palatal length — as for extant material; rostral width at canines — as for extant material; length of ma.xillary toothrow — as for extant material; interorbital breadth — least distance across interorbital region measured at right angles to the long axis of the cranium; height of coronoid process — least distance from a line connecting the angular process and ventral surface of the man- dible to the dorsalmost point of the coronoid process; width of articular process — least width across the articular process; man- dible breadth at M^ — least breadth of mandible at level of M3; length of mandibular toothrow — least distance from posterior lip of alveolus of M., to anterior lip of alveolus of canine. Dried skins examined in the study were assigned to one of the five color standards. The five specimens used for the color stan- dards and a description of their color as as follows: 1 ) TTU 2276 1 (male) — Haiti, Dept, du Sud, 1 km S, 1 km E Lebrun, on the dor- sum base of hair white, pattern blackish gray; 2) MCZ 21430 (male) — Martinique, on dorsum base of hair white, pattern black- ish brown; 3) AS 553 1 (female) — Puerto Rico, 17.7 km NE Utu- ado, on dorsum base of hair white, pattern grayish brown some- times with huffish tint; 4) TTU 20975 (female) — Guadeloupe, Grande-Terre, I km N, 1 km W St. Francois, on dorsum base of hair white, pattern dark brown with a very faint reddish tint; 5) AS 5126 (male) — Barbados, St. Thomas Parish, Cole's Cave, on dorsum base of hair white with yellowish tint, pattern dark brown with generally more of a buffy tint than color standard 3. Statistical analyses were performed on an IBM 370 computer at Texas Tech University. Univariate analyses of individual vari- ation, secondary sexual variation, and geographic variation were performed using the UNIVAR program, developed and intro- duced by Power) 1970). Standard statistics (mean, range, standard deviation, standard error, variance, and coefficient of variation) are generated by this program. In the event of two or more groups BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 12 being compared, a single-classification analysis of variance (AN- OVA) to test for significant differences between or among means is employed. Sums of Squares Simultaneous Test Procedure (SS- STP) (Gabriel, 1964) was used to determine maximally nonsig- nificant subsets, if means were found to be significantly different. See also Smith { 1972) for an overview of these statistical methods. Some of the multivariate analyses were performed using the Numerical Taxonomy System (NT-SYS) package developed by F. J. Rohlf, R. Bartcher, and J. Kishpaugh at the University of Kansas. The samples (OTUs) were grouped localities discussed above, and the values for each character were means for the mea- surements. Matrices of Pearson's product-moment correlation and phenetic distance coefficients were derived. Cluster analyses were conducted using UPGMA (unweighted pair group method using arithmetic averages) on the correlation and distance matri- ces, and phenograms were generated for both. Only distance phe- nograms were used because they gave higher coefficients of co- phenetic correlation than the correlation phenograms. These phenograms give a two-dimensional multivariate view of the data with characters unweighted. The first three principal components were then extracted from a matrix of correlation among charac- ters and three-dimensional projections of the samples onto the first three principal components were made. This provides a three-dimensional view of the data with unweighted characters. For the theory and use of these tests see Sokal and Sneath ( 1963), Schnell (1970), Atchley (1970), Choate (1970), Genoways and Jones (1971), Smith (1972), Genoways (1973), and Sneath and Sokal (1973). Other multivariate analyses performed involved use of the Sta- tistical Analysis System (SAS) package developed by Barr and Goodnight (Service, 1972). Individual specimens, and not series of means as in NT-SYS, were used in these analyses. Specimens with missing data could not be used, consequently sample sizes for SAS analyses were substantially reduced in some cases. To determine the degree of divergence among samples, a multivariate analysis of variance (MANOVA) and canonical analysis were per- formed. Canonical analysis of the data provides weighted com- binations of the characters, which maximize the distinction among groups. This analysis extracts characteristic roots and vec- tors and computes mean canonical variates for each sample. Ad- ditional orthogonal axes are constructed, which extract the next best combination of characters, emphasizing those with the least within sample and greatest among-sample variation, hence, pro- viding the next best combination of characters to discriminate among samples. Each eigenvalue and its corresponding canonical variate represents an identifiable fraction of the total variation. Sample means and individuals were plotted on those canonical variates, which account for the greatest fraction of total variation. The relative importance of each original variable (character) to a particular canonical variate was computed by multiplying the vec- tor variable coefficient by the mean value of the dependent vari- able, summing all variable values for a particular vector, and then computing the percent of relative importance of each variable per vector. These techniques have recently been used in the study of mammals by Schmidly and Hendricks ( 1976), Yates and Schmidly (1977), and Yates et al. (1978). NON-GEOGRAPHIC VARIATION Three kinds of nongeographic variation — varia- tion with age, secondary sexual variation, and indi- vidual variation — are discussed in the following sec- tion. Variation with Age One external and 12 cranial measurements of one non-adult male from Oriente Province, Cuba, and one non-adult female from Martinique are respec- tively, as follows: length of forearm, — , 56.8; great- est length of skull, 26.0, 28.6; condylobasal length, 23.3, 26.0; palatal length, 7.9, 9.9; braincase depth, 11.1, 1 L5; zygomatic breadth, 13.5, 15.1; breadth of braincase, 11.8, 11.7; mastoid breadth, 12.8, 13.3; postorbital breadth, 6.3, 5.9; length of maxillary toothrow, 8.9, 9.8; rostral width at canines, 6.5, 6.4; breadth across upper molars, 9.4, 10.2; mandibular length, — , 16.8. Comparing measurements of the subadult male from Oriente Province, Cuba, with those of adult males (1^) from Cuba (Table 1) shows that there is overlap in only four measurements (breadth of brain- case, mastoid breadth, postorbital breadth, rostral width at canines). A similar comparison between the subadult and adult females from Martinique (24) shows no overlap in measurements tested (Table 1). Only adult specimens (phalangeal epiphyses com- pletely fused) were used in the study of geographic variation. Secondary Sexual Variation External and cranial measurements of adult males from each sample were tested against those of adult females utilizing single classification ANOVA. This was done in order to establish if any significant differences in size exist between the sexes. The re- sults are shown in Table 1. In samples from west of the Mona Passage, males proved to be significantly {P < 0.05) larger than fe- males in two measurements (greatest length of skull, zygomatic breadth) in specimens from Habana Prov- ince, Cuba (sample 1); in one measurement (length of hind foot) in specimens from Las Villas Province, Cuba (sample 2), and in two measurements (length of hind foot, postorbital breadth) in specimens from the Dominican Republic (sample 8). On the other hand, females were found to be significantly larger than males in one measurement (length of ear) in specimens from Las Villas Province (sample 2). In samples from east of the Mona Passage, males 1978 SWANEPOEL AND GENOW AYS— BRACHYPHYLLA SYSTEMATICS 9 Table 1. — Geographic variation and secondary sexual variation in externa! and cranial measurements of B. nana (seven samples of males, and eight samples of females) and B. cavernarum ( 19 samples of males and 17 samples of females). Statistics given are number, mean, two standard errors, range, coefficient of variation, value. Means for males and females that are significantly different at P < 0.05 are marked with an asterisk. See text for key to sample numbers. Sample no. Male Female N X ± 2 SE Range cv N X ± 2 SE Range cv Fs Brachyphylla nana Total length 1 2 79.0 ± 2.00 78-80 1.8 1 75.0 2 2 87.5 ± 9.00 83-92 7.3 4 84.0 ± 4.90 80-90 5.8 4.581 3 1 95.0 2 93.5 ± 7.0 90-97 5.3 4 2 82.5 ± 5.0 80-85 4.3 1 81.0 7 1 80.0 1 79.0 8 26 72.0 ± 1.28 65-78 4.5 25 73.8 ± 1.74 67-84 5.9 2.68 Length of hind foot 1 2 16.5 ± 1.0 16-17 4.3 1 16.0 2 8 19.5 ± 0.38 19-20 2.7 4 18.0 ± 1.16 17-19 6.4 20.0* 3 1 21.0 2 21.5 ± 3.0 20-23 9.9 4 2 18.0 18 1 19.0 5 1 17.0 7 1 19.0 I 19.0 8 26 16.0 ± 0.43 13-18 6.8 25 15.3 ± 0.53 12-17 8.6 4.043* Length of ear I 2 20.0 20 1 19.0 2 9 17.2 ± 1.04 16-21 9.1 4 21.5 ± 1.30 20-23 6.0 33.639* 3 1 21.0 2 23.0 ± 6.00 20-26 18.4 4 2 21.0 21 1 21.0 5 1 21.0 6 7 20.7 ± 0.37 20-21 2.4 12 20.3 ± 0.05 19-21 4.3 1.674 7 1 19.0 1 20.0 8 25 19.7 ± 0.55 17-22 7.0 25 19.8 ± 0.52 17-22 6.5 0.101 Length of forearm 1 13 59.2 ± 1.34 53.0-61.4 4.1 9 58.1 ± 0.81 56.3-59.8 2.1 1.361 2 13 58.8 ± 0.80 56.8-61.3 2.5 7 58.7 ± 1.08 57.0-61.0 2.4 0.030 3 5 59.0 ± 2.02 55.2-61.0 3.8 1 60.3 ± 0.20 60.2-60.4 0.2 0.558 4 2 55.3 ± 1.00 54.8-55.8 1.3 1 57.7 5 1 60.2 6 7 56.2 ± 1.81 51.5-58.3 4.3 12 56.7 ± 0.59 54.6-58.5 1.8 0.375 7 1 58.9 4 58.7 ± 0.65 57.9-59.5 1.1 8 35 56.7 ± 0.50 53.5-59.1 2.6 29 57.2 ± 0.56 54.1-60.3 2.6 1.424 Greatest length of skull 1 12 28.7 ± 0.28 27.6-29.4 1.7 9 28.2 ± 0.42 27.1-29.0 2.2 4.547* 2 11 28.4 ± 0.30 27.5-29.2 1.7 7 28.4 ± 0.28 27.7-28.8 1.3 0.039 3 7 28.3 ± 0.29 27.5-28.6 1.3 3 28.5 ± 0.50 28.0-28.8 1.5 0.553 4 2 28.8 ± 0.50 28.5-29.0 1.2 1 27.4 5 1 28.9 6 7 28.7 ± 0.43 28.0-29.4 2.0 12 28.9 ± 0.27 28.4-29.8 1.6 0.568 7 1 28.3 4 28.6 ± 0.40 28.2-29.1 1.4 8 34 28.3 ± 0.18 27.2-29.3 1.8 33 28.2 ± 0.18 27.1-29.0 1.8 0.639 Condylobasal length 1 12 25.5 ± 0.29 24.5-26.2 2.0 7 25.3 ± 0.47 24.4-26.3 2.4 0.631 2 13 25.3 ± 0.23 24.7-26.0 1.7 7 25.4 ± 0.35 24.8-25.9 1.8 0.205 3 7 25.0 ± 0.23 24.6-25.5 1.2 3 25.3 ± 0.70 24.6-25.7 2.4 0.829 4 2 25.4 ± 0.20 25.3-25.5 0.6 1 24.4 5 1 25.6 6 6 25.4 ± 0.43 24.5-25.9 2.1 12 25.3 ± 0.23 24.7-26.0 1.6 0.280 7 1 24.9 4 25.2 ± 0.46 24.7-25.8 1.8 8 35 24.9 ± 0.16 23.7-25.7 1.9 32 24.8 ± 0.17 23.7-25.7 2.0 1.220 10 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 12 Table I. — Continued. Sample no. Male Female F, N X ± 2 SE Range cv N X ± 2 SE Range cv Palatal length 1 11 9.3 ± 0.23 8. 7-9.9 4.1 8 9.3 ± 0.26 8. 7-9.9 4.0 0.005 2 13 9.1 ± 0.12 8. 7-9.4 2.3 7 9.1 ± 0.27 8. 7-9.6 4.0 0.0 3 7 9.2 ± 0.24 9.0-9.9 3.4 3 9.5 ± 0.24 9.3-9.1 2.2 1.23 4 2 9.3 ± 0.10 9.2-9.3 0.8 1 9.0 5 1 9.4 6 7 9.8 ± 0.28 9.0-10.1 3.8 12 9.5 ± 0.19 8.9-10.0 3.5 1.685 7 1 9.4 4 9.6 ± 0.44 9.2-10.1 4.6 8 36 9.5 ± 0.13 8.7-10.4 4.0 33 9.4 ± 0.18 8.5-10.6 5.4 1.386 Depth of braincase 1 12 11.9 ± 0.15 11.5-12.2 2.2 8 11.8 ± 0.20 11.4-12.1 2.5 0.454 2 12 11.9 ± 0.17 11.4-12.3 2.5 7 12.0 ± 0.16 11.7-12.2 1.8 0.226 3 5 11.7 ± 0.35 11.3-12.1 3.3 3 11.7 ± 0.29 11.5-12.0 2.1 0.044 4 2 12.0 12.0 1 11.7 5 1 11.9 6 6 12.3 ± 0.29 11.6-12.6 2.9 12 12.2 ± 0.13 11.8-12.5 1.8 0.126 7 1 11.9 4 12.3 ± 0.22 12.1-12.6 1.8 8 32 11.9 ± 0.12 11.3-12.8 2.9 31 11.9 ± 0.12 11.3-12.6 2.9 0.261 Zygomatic breadth ! 10 15.2 ± 0.10 14.9-15.4 1.1 7 14.9 ± 0.11 14.7-15.1 1.0 17.704* 2 13 15.2 ± 0.21 14.5-16.0 2.5 7 15.3 ± 0.25 14.9-15.9 2.2 0.572 3 6 15.1 ± 0.41 14.4-15.7 3.3 4 14.9 ± 0.69 14.0-15.5 4.6 0.390 4 2 15.1 ± 0.10 15.0-15.1 0.5 1 14.8 5 1 15.1 6 7 15.1 ± 0.26 14.6-15.6 2.3 12 15.3 ± 0.17 14.7-15.7 2.0 1.094 7 1 14.7 5 15.1 ± 0.37 14.6-15.5 2.7 8 34 14.8 ± 0.12 14.2-15.5 2.3 30 14.8 ± 0.15 14.0-15.4 2.8 0.000 Breadth of braincase 1 13 11.9 ± 0.19 11.0-12.4 2.9 9 11.8 ± 0.21 11.4-12.2 2.6 0.166 2 13 11.8 ± 0.14 11.4-12.4 2.1 7 11.7 ± 0.17 11.3-12.0 1.9 0.000 3 7 11.8 ± 0.25 11.4-12.3 2.9 4 11.8 ± 0.33 11.4-12.2 2.8 0.007 4 2 12.0 ± 0.20 11.9-12.1 1.2 1 11.7 5 1 11.7 6 6 11.8 ± 0.13 11.6-12.0 1.4 12 11.9 ± 0.10 11.5-12.1 1.5 0.612 7 1 11.2 5 11.8 ± 0.20 11.5-12.1 2.0 8 37 11.8 ± 0.09 11.2-12.3 2.2 30 11.7 ± 0.09 11.2-12.2 2.2 0.182 Mastoid breadth 1 12 13.5 ± 0.19 12.9-14.0 2.5 7 13.2 ± 0.16 12.8-13.4 1.6 3.492 2 13 13.7 ± 0.16 13.1-14.1 2.1 7 13.4 ± 0.27 12.9-13.9 2.7 3.553 3 7 13.3 ± 0.33 12.7-14.0 3.2 4 13.4 ± 0.30 13.1-13.7 2.2 0.100 4 2 13.8 ± 0.40 13.6-14.0 2.0 1 13.1 5 1 13.8 6 5 13.6 ± 0.16 13.4-13.8 1.3 12 13.7 ± 0.14 13.2-14.0 1.8 0.146 7 1 13.7 5 13.1 ± 0.30 12.8-13.5 2.6 8 34 13.4 ± 0.13 12.9-14.4 2.9 31 13.3 ± 0.12 12.8-13.9 2.5 0.263 Postorbital breadth 1 12 6.2 ± 0.18 5. 7-6. 8 5.1 9 6.1 ± 0.15 5. 8-6. 5 3.7 0.283 2 13 6.1 ±0.11 5. 9-6. 6 3.4 7 6.2 ±0.11 6.0-6.4 2.4 1.028 3 7 6.0 ± 0.30 5.6-6.6 6.7 4 6.1 ± 0.27 5. 7-6.3 4.4 0.251 4 2 6.2 ± 0.10 6. 1-6.2 1.4 1 6.2 5 1 6.0 6 7 6.2 ± 0.08 6. 1-6.4 1.7 12 6.1 ± 0.12 5. 7-6.6 3.5 1.510 7 1 6.1 5 6.3 ±0.11 6. 1-6.4 2.1 8 38 6.4 ± 0.06 6. 0-7.0 3.0 33 6.2 ± 0.05 6.0-6. 5 2.5 13.688* 1978 SWANEPOEL AND GENOWA YS— CH YPH YLLA SYSTEMATICS 11 Table I.- — Continued. Sample no. Male Female N X ± 2 SE Range cv N ± 2 SE Range CV F, Length of maxillary toothrow 1 13 9.5 ± 0.11 9. 1-9.8 2.1 9 9.5 ± 0.14 9.2-9.8 2.2 0.017 2 11 9.4 ±0.11 9.2-9.8 2.0 7 9.3 ± 0.20 8. 8-9.6 2.8 1.828 3 7 9.3 ± 0.15 9.0-9.6 2.1 4 9.4 ± 0.16 9.2-9.6 1.7 1.207 4 2 9.1 9.1 1 9.0 5 1 9.4 6 7 9.5 ± 0.16 9.3-9.8 2.2 11 9.5 ± 0.12 9.2-9.9 2.2 0.488 7 1 9.4 5 9.3 ± 0.22 8. 9-9.5 2.7 8 34 9.4 ± 0.06 9.0-9.8 2.0 25 9.4 ± 0.07 9.0-9.7 1.8 0.00 Rostral width at canines 1 12 6.6 ± 0.10 6. 3-6.9 2.7 9 6.6 ± 0.20 5. 9-7.0 4.4 0.031 2 12 6.6 ± 0. 10 6.3-6.9 2.7 7 6.7 ± 0.23 6.0-6. 9 4.5 0.047 3 7 6.5 ± 0.20 6. 1-6.9 4.1 2 6.7 ± 0.10 6.6-6.7 1.1 0.365 4 2 6.7 ± 0.20 6.6-6.8 2.1 1 6.7 5 1 6.7 6 7 6.5 ± 0.15 6.2-6.S 3.1 11 6.3 ± 0.16 5. 9-6. 8 4.2 2.413 7 1 5.9 4 6.2 ± 0.20 5. 9-6.3 3.2 8 37 6.2 ± 0.08 5. 6-6.7 3.7 29 6.1 ± 0.08 5. 8-6. 7 3.5 2.755 Breadth across upper molars 1 12 10.5 ± 0.09 10.2-10.6 1.4 9 10.3 ± 0.23 9.8-10.8 3.4 1.087 2 11 10.4 ±0.11 10.1-10.6 1.7 7 10.4 ± 0.33 9.6-10.8 4.2 0.000 3 7 10.2 ± 0.24 9.8-10.6 3.0 4 10.2 ± 0.33 9.8-10.6 3.2 0.010 4 2 10.3 ± 0.20 10.2-10.4 1.4 1 10.3 5 I 10.5 6 7 10.2 ± 0.12 10.0-10.4 1.6 12 10.3 ± 0.08 10.1-10.5 1.3 2.229 7 1 9.4 4 10.1 ± 0.13 9.9-10.2 1.2 8 36 9.9 ± 0.07 9.5-10.4 2.2 26 10.0 ± 0.10 9.6-10.5 2.6 0.610 Mandibular length 1 10 17.4 ± 0.20 16.7-17.8 1.9 6 17.2 ± 0.29 16.8-17.8 2.0 0.602 2 13 17.5 ± 0.22 16.7-17.9 2.2 7 17.5 ± 0.32 16.8-17.9 2.4 0.072 3 6 17.2 ± 0.41 16.7-18.1 2.9 1 17.5 4 2 17.6 ± 0.90 17.1-18.0 3.6 1 16.9 5 1 17.4 6 7 17.3 ± 0.27 16.8-17.8 2.1 12 17.1 ± 0.24 16.4-17.7 2.4 1.054 7 1 17.9 4 17.6 ± 0.46 17.1-18.2 2.6 8 35 17.3 ± 0.13 16.3-18.2 2.2 28 17.3 ± 0.14 16.5-18.1 2.1 0.030 Brachyphylla cavernanun Total length 9 8 86.6 ± 3.33 79-92 5.4 11 88.9 ± 1.9 84-95 3.5 1.614 10 23 92.5 ± 3.4 82-118 8.9 19 96.7 ± 4.4 84-115 9.8 2.377 1! 52 94.0 ± 1.21 84-104 4.6 7 89.1 ± 3.8 82-95 5.6 7.298* 12 33 92.3 ± 1.76 88-103 5.5 19 93.3 ± 1.9 86-102 4.5 0.543 13 1 95.0 14 2 93.5 ± 7.0 90-97 5.3 2 96.5 ± 10.1 91-102 8.1 2.212 18 9 90.2 ± 2.0 85-95 3.3 19 1 90.0 2 102.0 ± 2.0 101-103 1.4 22 3 90.3 ± 2.4 88-92 2.3 2 90.5 ± 1.00 90-91 0.8 0.012 23 8 91.1 ± 2.99 87-98 4.6 4 88.3 ± 0.50 88-89 0.6 1.756 24 10 91.6 ± 1.41 89-95 2.4 19 89.6 ± 1.57 86-93 2.6 3.786 25 2 94.5 ± 1.0 94-95 0.7 26 2 84.5 ± 9.0 80-89 7.5 1 90.0 27 3 91.0 ± 1.15 90-92 1.1 8 90.6 ± 1.85 86-94 2.9 0.056 12 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 12 Table 1. — Continued. Sample no. Male Female Fs N X ± 2 SE Range cv N X ± 2 SE Range cv Length of hind foot 9 8 20.4 ± 1.60 16-23 11.1 11 21.0 ± 1.24 17-23 9.8 0.396 10 22 21.4 ± 0.41 16-23 6.5 19 21.9 ± 0.30 15-23 6.0 4.165* 11 52 22.4 ± 0.23 19-25 3.8 7 21.6 ± 0.74 20-23 4.5 5.537* 12 33 21.5 ± 0.52 18-24 6.9 19 22.1 ± 0.43 20-23 4.2 2.194 13 1 23.0 14 2 20.0 20 2 19.5 ± 1.00 19-20 3.6 3.00 18 9 22.6 ± 0.35 22-23 2.3 19 1 23.0 2 23.0 23 22 3 21.3 ± 0.67 21-22 2.7 2 21.5 ± 1.00 21-22 3.3 0.086 23 8 22.5 ± 0.53 21-23 3.4 4 22.5 ± 1.0 21-23 4.4 0.000 24 10 20.3 ± 0.85 18-22 6.6 9 19.1 ± 1.39 17-23 10.9 2.232 25 2 21.5 ± 1.00 21-22 3.3 26 2 21.5 ± 3.0 20-23 9.9 27 3 20.7 ± 1.33 20-22 5.6 8 21.1 ± 0.7 20-23 4.7 0.431 Length of ear 9 8 21.3 ± 0.73 20-23 4.9 11 20.9 ± 0.87 19-23 6.9 0.323 11 43 22.0 ± 0.26 20-26 3.8 5 23.4 ± 0.49 23-24 2.3 3.146 12 33 22.4 ± 0.38 20-24 4.9 19 22.8 ± 0.41 21-24 3.9 2.002 14 2 21.0 21 2 20.5 ± 1.00 20-21 3.4 3.000 21 1 21.0 22 3 21.0 ± 3.06 18-23 12.6 2 24.0 24 23 8 23.0 ± 0.53 22-24 3.3 4 22.3 ± 0.5 22-23 2.3 3.158 24 5 20.2 ± 1.47 19-23 8.1 6 19.7 ± 0.67 18-20 4.2 0.492 25 2 21.0 21 26 2 23.0 23 27 3 22.7 ± 0.67 22-23 2.6 8 22.4 ± 0.37 22-23 2.3 0.664 Length of forearm 9 8 64.0 ± 1.19 60.7-65.4 2.6 11 65.1 ± 1.10 60.4-67.0 2.8 1.734 10 61 65.0 ± 0.47 61.6-69.4 2.8 24 65.0 ± 0.77 60.3-68.2 2.9 0.128 11 38 63.3 ± 0.59 60.0-66.4 2.9 7 63.3 ± 1.41 60.9-65.7 2.9 0.000 12 18 62.5 ± 0.87 60.0-66.1 3.0 8 62.8 ± 1.31 60.0-65.5 3.0 0.127 13 1 64.3 14 6 64.1 ± 1.61 60.2-65.5 3.1 8 65.6 ± 0.49 64.5-66.8 1.1 4.019 15 6 65.6 ± 2.07 61.6-68.7 3.9 5 65.7 ± 2.28 62.0-68.0 3.9 0.002 16 3 65.3 ± 2.05 63.9-67.3 2.7 17 1 65.2 2 63.9 ± 0.60 63.6-64.2 0.7 18 9 65.7 ± 1.18 62.3-67.4 2.7 19 6 65.4 ± 0.10 65.3-65.6 0.2 5 65.2 ± 0.44 64.5-65.9 0.8 9.940* 20 4 65.4 ± 0.78 64.4-66.3 1.2 3 67.3 ± 2.60 65.9-69.9 3.3 2.578 21 6 66.6 ± 0.90 65.3-67.9 1.6 5 67.6 ± 0.96 65.8-68.4 1.6 2.131 22 19 65.6 ± 0.69 63.0-68.9 2.3 13 65.4 ± 0.85 63.1-68.8 2.3 0.081 23 9 63.9 ± 0.93 62.3-65.7 2.2 7 64.6 ± 2.09 60.4-67.6 4.3 0.495 24 10 65.0 ± 1.88 59.6-68.1 4.6 9 66.8 ± 1.34 64.4-71.1 3.0 2.408 25 10 65.0 ± 0.59 62.9-66.5 1.4 5 65.5 ± 1.41 63.0-66.7 2.4 0.526 26 5 64.6 ± 0.40 61.8-65.5 2.4 6 65.2 ± 0.81 64.3-66.8 1.5 0.555 27 6 61.0 ± 1.06 59.2-63.1 2.1 12 61.1 ± 0.53 59.3-62.4 1.5 0.056 Greatest length of skull 9 9 31.4 ± 0.31 30.5-32.0 1.5 11 31.3 ± 0.29 30.6-31.8 1.5 0.000 10 66 31.7 ± 0.15 30.5-33.0 1.9 27 31.4 ± 0.20 30.3-32.1 1.6 4.681* 11 48 31.4 ± 0.17 30.1-32.7 1.8 5 31.5 ± 0.71 30.6-32.7 2.5 0.098 12 26 31.6 ± 0.25 30.2-32.9 2.0 8 31.0 ± 0.49 30.2-32.2 2.3 4.309* 13 1 32.0 no. 14 15 16 17 18 19 20 21 22 23 24 25 26 27 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 SWANEPOEL AND GEhiOW AYS— BRACHYPHYLLA SYSTEM ATICS 13 Table I . — Continued. Male Female N X ± 2 SE Range cv N X ± 2 SE Range cv F. 6 32.2 ± 0.30 31.7-32.6 1.1 8 32.3 ± 0.30 31.6-32.7 1.3 0.192 6 32.1 ± 0.54 31.4-33.0 2.1 5 31.6 ± 0.43 31.0-32.3 1.5 2.036 3 32.3 ± 0.41 31.9-32.6 1.1 1 32.2 3 31.7 ± 0.81 31.0-32.4 2.2 8 32.1 ± 0.31 31.5-32.8 1.4 1 32.4 8 32.1 ± 0.41 31.3-33.0 1.8 8 31.6 ± 0.29 31.2-32.2 1.3 3.233 4 32.4 ± 0.14 32.2-32.5 0.4 7 32.0 ± 0.45 31.1-32.8 1.8 1.976 9 31.9 ± 0.26 31.2-32.5 1.2 8 31,9 ± 0.23 31.5-32.5 1.0 0.000 18 32.0 ± 0.26 30,9-32,8 1.7 13 31.6 ± 0.38 30.4-32.4 2.2 3.621 8 31.9 ± 0.40 31.2-32.8 1.8 8 31.9 ± 0.15 31.6-32.2 0.6 0.129 10 32.2 ± 0.19 31.8-32.8 0.9 9 31.7 ± 0.40 30.6-32.3 1.9 6.208* 10 31.9 ± 0.32 31.0-32.5 1.6 7 32.1 ± 0.51 30.7-32.7 2.1 0.654 5 31.9 ± 0.48 31.3-32.7 1.7 8 32.2 ± 0.36 31.7-33.3 1.6 0.505 7 30.5 ± 0.36 30.0-31.2 1.5 11 30.5 ± 0.24 29.6-30.9 1.3 0.043 Condylobasal length 8 28.0 ± 0.34 27.2-28.5 1.7 11 27.8 ± 0.29 27.2-28.4 1.7 0.580 63 28.1 ± 0.13 26.4-29.5 1.9 24 28.0 ± 0.21 27.2-29.0 1.9 0.438 49 28.2 ± 0.13 27.2-29.1 1.7 5 27.9 ± 0.65 26.8-28.7 2.6 1.402 27 28.2 ± 0.21 27.3-30.0 1.9 8 28.0 ± 0.24 27.3-28.3 1.2 1.280 1 28.4 6 28.6 ± 0.39 27.8-29.0 1.7 9 28.5 ± 0.31 28.0-29.4 1.6 0.070 6 28.6 ± 0.48 27.9-29.3 2.1 5 28.0 ± 0.39 27.6-28.6 1.6 4.073 3 28.8 ± 0.37 28.4-29.0 1.1 1 29.0 3 28.2 ± 0.58 27.7-28.7 1.8 8 28.5 ± 0.30 27.9-29.0 1.5 1 28.8 8 28.7 ± 0.53 27.7-29.8 2.6 7 28.4 ± 0.21 28.1-28.7 1.0 0.929 4 29.2 ± 0.26 28.8-29.4 0.9 4 28.2 ± 0,69 27.4-28.9 2.5 6.892* 8 28.5 ± 0.14 28.2-28.7 0.7 8 28.3 ± 0.29 27.6-28.9 1.5 1.811 19 28.4 ± 0.24 27.1-29.0 1.9 13 28.1 ± 0.34 26,8-29.0 2.2 1.479 8 28.6 ± 0.33 27.9-29.4 1.6 7 28.4 ± 0.16 28.1-28.7 0.7 0.941 9 28.5 ± 0.24 28.0-29.0 1.3 9 28.2 ± 0.25 27.6-28.6 1.3 4.330 9 28.6 ± 0.34 27.9-29.2 1.8 7 28.6 ± 0.48 27.6-29.4 2.2 0.000 4 28.6 ± 0.54 28.0-29.3 1.9 8 28.4 ± 0.24 28.0-29.0 1.2 1.032 7 27.1 ± 0.40 26.3-27.7 1.9 12 27.0 ± 0.26 26.3-27.6 1.6 0.086 Palatal length 9 11.7 ± 0.24 11.3-12.2 3.1 11 11.6 ± 0.25 10.8-12.1 3.6 0.808 67 11.7 ± 0.10 10.8-12.6 3.4 27 11.5 ± 0.18 10,8-12.6 4.1 1.445 5! 12.0 ± 0.13 11.0-12.9 3.9 6 11.3 ± 0.52 10.5-12.4 5.6 9.496* 31 12.1 ± 0.17 11.2-12.9 4.0 16 12.0 ± 0.24 11,0-12,7 4.0 1.101 1 11.8 6 12.6 ± 0.28 12.3-13.1 2.7 8 12.2 ± 0.39 11.3-13.0 4.5 3.033 6 12.2 ± 0.53 1 1.5-12.9 5.3 5 11.8 ± 0.33 11.4-12.4 3.1 1 .026 3 12.5 ± 0,58 12.0-13.0 4.0 1 11.9 3 1 1,6 ± 0.35 11.3-11.9 2.6 8 12.4 ± 0.15 12.1-12,8 1,7 1 12.2 8 12.3 ± 0.34 11.5-12.8 4.0 8 12.0 ± 0,39 11.1-13.0 4.5 0.675 4 12.4 ± 0.26 12.0-12.6 2.1 7 12.2 ± 0.40 11.6-13.0 4.4 0.499 9 12.1 ± 0.30 11,4-12,7 3.7 7 12.0 ± 0.44 11.1-12.6 4.8 0.055 19 11.9 ± 0.27 10.6-12.8 5.0 13 11,9 ± 0.26 11.2-12.7 4.0 0.013 9 11.9 ± 0.23 11.5-12.5 2.9 8 12.2 ± 0.23 11.8-12.7 2.7 2.163 9 12.0 ± 0.25 11.3-12.3 3.1 9 11.8 ± 0.23 11.3-12.4 3.0 1,712 10 12.2 ± 0.28 11.6-12.7 3.7 4 12.5 ± 0.45 12.1-13.1 3.6 0.410 5 11.9 ± 0.33 11.4-12.3 3.1 8 11.8 ± 0.12 11.5-12.0 1.4 0.958 7 11.4 ± 0.39 10.7-12.0 4.5 12 11.6 ± 0.32 10,7-12.3 4.8 0.449 no. 9 10 II 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 9 10 II 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 9 10 1 1 12 13 14 15 16 17 18 19 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 12 Table I. ontinued. Male Female N X ± 2 SE Range cv N X ± 2 SE Range cv Fs Depth of braincase 8 13.4 ± 0.17 13.1-13.7 1.8 11 13.2 ± 0.20 12.5-13.7 2.6 0.993 65 13.4 ± 0.10 12.5-13.9 3.0 25 13.1 ± 0.13 12.4-13.6 2.5 12.368* 50 13.3 ± 0.10 12.4-13.9 2.6 5 13.1 ± 0.33 12.5-13.4 2.8 0.796 29 13.3 ± 0.13 12.3-13.9 2.6 8 13.2 ± 0.26 12.7-13.6 2.8 1.103 1 13.8 6 13.6 ± 0.38 13.0-14.3 3.4 9 13.6 ± 0.20 13.3-14.1 2.2 0.048 6 13.5 ± 0.24 13.0-13.8 2.2 5 13.2 ± 0.19 13.0-13.5 1.6 2.072 3 13.7 ± 0.18 13.5-13.8 1.1 1 12.9 3 13.1 ± 0.18 12.9-13.2 1.2 8 13.4 ± 0.21 13.0-13.7 2.2 1 13.1 7 13.8 ± 0.37 13.0-14.4 3.6 8 13.4 ± 0.20 13.0-13.7 2.1 2.655 4 13.2 ± 0.35 12.8-13.6 2.7 5 13.2 ± 0.38 12.7-13.7 3.2 0.000 9 13.2 ± 0.24 12.3-13.5 2.7 8 13.2 ± 0.22 12.9-13.7 2.3 0.179 18 13.4 ± 0.19 12.4-13.9 2.9 13 13.2 ± 0.21 12.4-13.7 2.9 3.730 8 13.3 ± 0.28 12.6-13.9 3.0 7 13.3 ± 0.13 13.0-13.5 1.3 0.058 10 13.5 ± 0.15 13.1-13.9 1.7 9 13.2 ± 0.28 12.7-13.8 3.2 4.673* 8 13.4 ± 0.19 13.1-13.9 2.1 7 13.4 ± 0.22 13.0-13.9 2.1 0.028 4 13.3 ± 0.30 12.9-13.5 2.3 7 13.1 ± 0.26 12.7-13.6 2.7 0.422 6 13.1 ± 0.08 13.0-13.2 0.8 12 12.7 ± 0.18 12.2-13.3 2.4 6.921* Zygomatic breadth 7 17.0 ± 0.34 16.4-17.6 2.6 11 17.0 ± 0.21 16.5-17.7 2.0 0.026 65 17.2 ±0.11 15.8-18.1 2.6 26 17.0 ± 0.17 16.0-17.7 2.5 5.041* 47 17.2 ± 0.13 16.5-18.0 2.5 6 16.7 ± 0.42 15.9-17.2 3.1 6.857* 29 1 17.2 ± 0.12 17 1 16.7-17.8 2.0 12 17.1 ± 0.29 16.2-18.0 2.9 0.104 4 17.5 ± 0.19 17. 2-17. 6 1.1 7 17.5 ± 0.37 16.5-17.9 2.8 0.019 6 17.4 ± 0.27 16.8-17.8 1.9 5 17.1 ± 0.28 16.8-17.6 1.8 1.573 3 17.7 ± 0.37 17.5-18.1 1.8 1 17.2 3 17.0 ± 0.07 16.9-17.0 0.3 8 17.5 ± 0.24 17.0-18.0 1.9 1 17.5 7 17.5 ± 0.29 17.0-18.2 2.2 7 17.3 ± 0.28 16.8-17.9 2.1 1.136 4 17.4 ± 0.35 16.9-17.7 2.0 7 17.1 ± 0.31 16.5-17.6 2.4 1.822 9 17.5 ± 0.17 17.0-17.9 1.4 8 17.2 ± 0.22 16.5-17.4 1.8 5.436* 18 17.4 ± 0.25 16.0-18.2 3.0 11 17.3 ± 0.30 16.6-18.3 2.9 0.088 8 17.4 ± 0.21 17.0-17.9 1.7 7 17.5 ± 0.25 17.0-18.0 1.9 0.548 10 17.7 ± 0.20 17.2-18.2 1.8 8 17.0 ± 0.36 16.3-17.5 3.0 13.149* 9 17.3 ± 0.18 16.6-17.6 1.6 7 17.5 ± 0.26 17.0-17.8 2.0 1.560 3 17.1 ± 0.64 16.5-17.6 3.2 8 17.3 ± 0.29 16.8-17.8 2.3 0.376 8 16.5 ± 0.13 16.2-16.7 1.1 10 16.5 ± 0.23 16.0-17.2 2.2 0.262 Breadth of braincase 9 12.6 ± 0.10 12.4-12.9 1.2 11 12.6 ± 0.16 12.2-13.1 2.1 0.005 66 12.8 ± 0.07 12.3-13.6 2.2 28 12.5 ±0.11 11.9-12.9 2.2 19.992* 51 12.8 ± 0.07 12.3-13.2 1.8 7 12.6 ± 0.12 12.4-12.8 1.3 3.242 29 12.7 ± 0.08 12.3-13.3 1.7 II 12.6 ± 0.17 12.3-13.1 2.2 1.761 1 13.0 6 13.0 ± 0.15 12.8-13.3 1.4 9 13.0 ± 0.23 12.4-13.4 2.7 0.000 6 13.0 ± 0.15 12.8-13.3 1.4 5 12.8 ± 0.13 12.6-13.0 1.2 3.924 3 13.1 ± 0.13 13.0-13.2 0.9 1 12.6 3 12.7 ± 0.07 12.6-12.7 0.5 8 12.8 ± 0.22 12.4-13.4 2.5 1 13.0 8 12.9 ± 0.18 12.5-13.2 2.0 8 12.8 ± 0.16 12.4-13.1 1.8 0.683 4 12.9 ± 0.21 12.6-13.1 1.6 6 12.8 ± 0.15 12.6-13.0 1.5 0.151 9 12.8 ± 0.17 12.5-13.3 2.0 8 12.8 ± 0.12 12.5-13.0 1.3 0.562 19 12.7 ± 0.13 12.3-13.1 2.2 13 12.7 ± 0.16 12.4-13.3 2.3 0.332 9 12.9 ± 0.26 12.2-13.3 3.0 7 12.7 ± 0.19 12.2-12.9 2.0 1.407 1978 SWANEPOEL AND GENOW AY S—BRACHYPHYLLA SYSTEMATICS 15 Table I. — Conlinued. Sample no. Male Female Fs N X ± 2 SE Range cv N X ± 2 Sh Range tv 24 10 13.1 ± 0.14 12.7-13.4 1.7 10 12.7 ± 0.20 12.1-13.0 2.5 13.807* 25 10 12.9 ± 0.13 12.6-13.2 1.6 7 12.8 ± 0.19 12.4-13.2 2.0 0.647 26 5 12.9 ± 0.23 12.5-13.2 2.0 8 12.8 ± 0.21 12.4-13.2 2.3 0.423 27 8 12.4 ±0.11 12.2-12.7 1.3 12 12.3 ± 0.15 II. 9-12. 7 2 2 2.039 Mastoid breadth 9 8 14.8 ± 0.26 14.2-15.4 2.5 11 14.6 ± 0.21 14.2-15.4 2.4 0.995 10 65 15.0 ± 0.08 14.1-15.7 2.1 24 14.6 ±0.11 14. 1-15. 1 1.8 24.343* II 46 14.8 ± 0.10 14.0-15.3 2 2 5 14.4 ± 0.27 14.0-14.8 2.1 4.320* 12 27 14.8 ± 0.12 14.2-15.4 2.2 7 14.6 ± 0.32 14.1-15.2 2.9 1.237 13 1 14.6 14 6 15.1 ± 0.19 14.9-15.5 1.6 9 14.9 ± 0.29 14.3-15.5 2.9 0.465 15 6 15.0 ± 0.33 14.4-15.6 2.7 5 14.5 ± 0.37 14.0-14.9 2.8 3.472 16 3 15.1 ± 0.47 14.7-15.5 2.7 17 1 14.9 3 14.3 ± 0.41 14.0-14.7 2.5 18 8 14.9 ± 0.17 14.5-15.3 1.6 19 8 15.0 ± 0.25 14.5-15.6 2.4 8 14.9 ± 0.17 14.5-15.3 1.6 0.432 20 4 15.0 ± 0.44 14.6-15.6 3.0 5 14.6 ± 0.33 14.1-15.1 2.5 2.105 21 9 14.9 ±0.11 14.6-15.1 1.1 8 14.7 ± 0.25 14.2-15.3 2.4 2.319 22 18 14.9 ± 0.17 14.1-15.5 2.4 13 14.7 ± 0.22 14. 1-15.5 2.7 2.025 23 8 15.0 ± 0.16 14.7-15.4 1.5 6 14.8 ± 0.15 14.5-15.9 1.3 5.846* 24 10 15.0 ± 0.15 14.7-15.5 1.6 9 14.6 ± 0.31 14.0-15.4 3.2 6.294* 25 9 15.0 ± 0.18 14.5-15.4 1.8 7 14.8 ± 0.30 14. 1-15.4 2.7 0.448 26 3 14.7 ± 0.12 14.6-14.8 0.7 8 14.7 ± 0.18 14.4-15.0 1.7 0.000 27 7 14.4 ± 0.27 13.7-14.8 2.5 12 14.1 ± 0.17 13.7-14.6 2.0 4.007 Postorbital breadth 9 8 6.5 ±0.11 6. 3-6.7 2.3 II 6.4 ± 0.10 6. 1-6.6 2.5 0.359 10 67 6.5 ± 0.04 6.0-6. 8 2.4 28 6.5 ± 0.08 6. 1-6.8 3.2 2.127 11 53 6.4 ± 0.06 5. 8-6.9 3.2 7 6.5 ±0.11 6. 2-6.7 2.3 2.254 12 31 6.3 ± 0.07 5. 9-6.8 3.0 15 6.3 ± 0.08 6. 0-6.6 2.6 0.280 13 1 6.5 14 6 6.4 ±0.11 6. 2-6.6 2.1 9 6.4 ± 0.10 6. 2-6.6 2.4 0.078 15 6 6.4 ±0.11 6. 3-6. 6 2.1 5 6.2 ± 0.15 6. 0-6. 4 2.6 3.330 16 3 6.4 ± 0.18 6.3— 6.6 2.4 17 1 6.2 3 6.2 ± 0.07 6. 1-6.2 0.9 18 9 6.4 ± 0.12 6.2-6.7 2.7 1 6.5 19 8 6.4 ± 0.17 6. 1-6.9 3.8 8 6.3 ± 0.13 6. 0-6.5 2.9 0.211 20 4 6.1 ± 0.17 5.9-6. 3 2.8 7 6.2 ±0.11 6.0-6. 4 2.4 1.404 21 9 6.3 ± 0.13 6. 0-6.6 3.1 8 6.3 ± 0.09 6. 0-6.4 2.1 0.111 22 19 6.5 ± 0.08 6. 2-6.9 2.6 13 6.3 ± 0.09 6. 1-6.6 2.5 5.361* 23 8 6.3 ± 0.08 6. 1—6.4 1.9 8 6.3 ± 0.08 6. 2-6. 5 1.7 1.762 24 10 6.4 ± 0.07 6. 2-6. 5 1.7 10 6.4 ± 0.11 6. 1-6.6 2.8 0.102 25 1 1 6.3 ± 0.07 6. 1-6.5 1.9 6 6.3 ± 0.15 6. 1-6.6 2.8 0.133 26 5 6.4 ± 0.09 6.3-6.5 1.6 8 6.4 ± 0.12 6.2-6.7 2.6 0.202 27 8 6.3 ±0.15 6. 1-6.6 3.5 12 6.2 ± 0.10 5. 8-6. 5 2.8 2.179 Length of maxillary toothrow 9 9 10.6 ± 0.13 10.3-10.9 1.9 II 10.7 ± 0.15 10. 1-1 1.0 2.3 0.650 10 62 10.7 ± 0.05 10.1-11.1 1.9 24 10.7 ± 0.08 10.4-11.0 1.8 0.661 11 38 10.8 ± 0.06 10.3-11.2 1.9 7 10.7 ± 0.20 10.4-11.1 2.5 0.347 12 22 10.7 ± 0.10 10.3-11.1 2.2 11 10.7 ± 0.06 10.5-10.8 0.9 0.148 13 1 10.7 14 6 11.0 ± 0.10 10.8-1 1.1 i.i 8 10.9 ± 0.21 10.5-11.3 2.8 0.887 15 6 10.9 ± 0.16 10.6-11.1 1.8 5 1 1.0 ± 0.24 10.6-11.2 2.4 0.479 16 3 11.1 ± 0.18 II.0-1I.3 1.4 17 1 11.0 3 II.O ± 0.44 10.6-1 1.3 3.4 18 9 11.0 ± 0.24 10.6-11.6 3.3 1 11.5 16 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 12 Table 1. — Continued. Sample no. Male Female F, N X ± 2 SE Range CV N X ± 2 SE Range CV 19 6 10.9 ± 0.10 10.7-1 1.0 1.1 5 10.8 ± 0.15 10.7-11.0 1.5 1.129 20 4 11.3 ± 0.14 11.2-11.5 1.3 7 11.1 ± 0.20 10.6-11.4 2.4 1.900 21 9 11.0 ± 0.11 10.7-11.2 1.5 8 11.0 ± 0.14 10.5-11.1 1.8 0.877 22 19 11.0 ± 0.11 10.6-11.6 2.1 13 10.9 ± 0.13 10.6-11.4 2.1 2.659 23 9 11.0 ± 0.06 10.9-11.2 0.9 8 11.0 ± 0.12 10.8-11.2 1.6 0.000 24 10 11.1 ±0.12 10.7-11.3 1.7 9 10.9 ± 0.10 10.6-11.1 1.4 7.159* 25 11 11.0 ± 0.12 10.7-11.3 1.8 8 11.1 ± 0.13 10.7-11.2 1.7 0.026 26 4 10.9 ± 0.30 10.5-11.2 2.7 8 11.0 ± 0.15 10.8-11.4 1.9 0.355 27 8 10.6 ± 0.14 10.3-10.9 1.9 12 10.5 ± 0.13 10.0-10.8 2.1 0.173 Rostral width at canines 9 9 7.2 ±0.11 7. 1-7.6 2.3 11 7.1 ± 0.08 6.8-7. 3 1.9 1.995 10 67 7.2 ± 0.06 6.5-7. 6 3.4 28 7.1 ± 0.08 6.5-7.4 3.0 9.096* II 52 7.3 ± 0.07 6.6-7. 7 3.4 7 7.0 ± 0.21 6.6-7. 4 3.9 7.397* 12 30 7.3 ± 0.08 6.8-7. 8 2.9 14 7.2 ± 0.08 6.9-7. 5 2.2 4.375* 13 1 7.4 14 6 7.3 ± 0.12 7. 2-7. 6 2.1 8 7.2 ± 0.14 7.0-7. 6 2.8 0.962 15 6 7.2 ± 0.26 6.8-7. 7 4.4 5 7.2 ± 0.21 6.8-7.4 3.2 0.052 16 3 7.3 ± 0.18 7. 1-7.4 2.1 17 1 6.8 3 7.4 ± 0.57 6.8-7. 7 6.7 18 8 7.3 ± 0.19 7. 0-7. 8 3.6 1 7.4 19 8 7.5 ± 0.24 7.0-8. 1 4.5 8 7.2 ± 0.16 6.8-7.4 3.0 2.491 20 4 7.6 ± 0.05 7. 6-7. 7 0.6 7 7.1 ± 0.12 6.9-7. 3 2.3 37.664* 21 9 7.3 ± 0.14 7.0-7. 6 2.8 8 7.2 ± 0.18 6.8-7. 6 3.5 0.448 22 18 7.4 ± 0.13 6.7-7. 8 3.8 13 7.2 ± 0.09 6. 9-7.4 2.4 3.884 23 9 7.4 ± 0.23 6. 8-7. 9 4.6 8 7.2 ± 0.13 7.0-7. 5 2.4 2.337 24 9 7.4 ± 0.15 00 3.0 9 7.1 ± 0.15 6.7-1.5 3.1 11.422* 25 11 7.4 ± 0.12 7.0-7. 5 2.6 7 7.3 ± 0.08 7. 2-7.5 1.5 0.233 26 5 7.4 ± 0.20 7. 1-7.7 3.1 8 7.2 ± 0.12 7. 0-7.4 2.3 2.392 27 8 6.9 ± 0.15 6.6-7. 2 3.0 12 6.7 ± 0.14 6.3-7.0 3.6 2.070 Breadth across upper molars 9 9 11.5 ± 0.28 10.9-12.2 3.7 11 11.5 ± 0.10 11.2-11.7 1.5 0.244 10 66 11.5 ± 0.07 10.8-12.1 2.6 27 11.5 ± 0.11 10.9-12.1 2.4 0.558 11 50 11.6 ± 0.09 10.9-12.3 2.7 7 11.2 ± 0.22 10.8-11.7 2.6 7.009* 12 26 11.5 ± 0.11 11.0-12.1 2.5 14 11.7 ± 0.15 11.2-12.2 2.4 2.569 13 1 11.7 14 5 11.7 ± 0.29 11.2-12.0 2.7 8 11.8 ± 0.19 11.4-12.2 2.3 0.046 15 6 11.7 ± 0.27 11.2-12.2 2.8 5 11.7 ± 0.46 11.2-12.4 4.4 0.015 16 3 11.7 ± 0.18 11.6-11.9 1.3 17 1 11.1 3 11.9 ± 0.27 11.6-12.0 1.9 18 9 11.8 ± 0.21 11.4-12.3 2.7 1 12.2 19 6 11.8 ± 0.34 11.4-12.4 3.5 7 11.6 ± 0.25 11.0-12.0 2.8 0.439 20 4 12.0 ± 0.06 11.9-12.0 0.5 7 11.5 ± 0.18 11.2-11.9 2.1 11.128* 21 9 11.8 ± 0.16 11.6-12.2 2.0 8 11.8 ± 0.28 10.9-12.1 3.3 0.130 22 19 11.8 ± 0.15 11.0-12.3 2.8 13 11.7 ± 0.19 11.2-12.2 2.9 0.276 23 9 11.8 ± 0.18 11.3-12.2 2.2 8 11.8 ± 0.20 11.3-12.1 2.3 0.000 24 10 12.0 ± 0.19 11.5-12.4 2.5 10 11.6 ± 0.14 11.2-12.0 1.9 10.770* 25 10 11.7 ± 0.12 11.3-12.0 1.7 7 11.9 ± 0.11 11.8-12.2 1.2 7.171* 26 5 11.7 ± 0.19 11.6-12.1 1.8 8 11.7 ± 0.23 11.2-12.2 2.8 0.006 27 8 11.1 ±0.15 10.9-11.5 1.9 11 11.2 ± 0.19 10.8-11.9 2.9 0.576 Mandibular length 9 8 19.9 ± 0.26 19.3-20.3 1.9 11 19.9 ± 0.24 19.3-20.4 2.0 0.053 10 63 19.9 ± 0.10 19.0-20.9 2.1 26 19.9 ± 0.18 19.1-20.9 2.3 0.105 11 45 20.3 ±0.10 19.6-21.0 1.7 7 20.1 ± 0.26 19.7-20.5 1.7 2.069 12 26 20.2 ± 0.21 19.4-20.8 2.7 10 20.1 ± 0.26 19.4-20.8 2.1 0.945 13 1 20.1 1978 SWANEPOEL AND GENOW AYS— BRACHYPHYLLA SYSTEMATICS 17 Table 1. onliniied. Sample no. Male Female F, N X ± 2 SE Range CV N X ± 2 SE Riinge CV 14 4 20.6 ± 0.38 20.0-20.8 1.8 9 20.3 ± 0.19 20.0-20.7 1.4 2.328 15 6 20.6 ± 0.31 20.3-21.1 1.8 5 20.0 ± 0.54 19.4-20.6 3.0 3,938 16 3 20.5 ± 0.20 20.4-20.7 0.8 17 1 20.8 2 20.0 ± 0.60 19.7-20.3 2.1 18 9 20.6 ± 0.34 20.0-21.5 2.5 1 21.1 19 7 20.7 ± 0.39 20.0-21.3 2.5 7 20.4 ± 0.26 19.9-21.0 1.7 1.520 20 4 20.7 ± 0.49 20.1-21.3 2.4 7 20.6 ± 0.34 19.8-21.0 2.2 0.184 21 7 20.4 ± 0.29 19,8-20.9 1.9 8 20.4 ± 0.21 19.9-20.8 1.5 0.000 22 19 20.5 ± 0.16 19.7-21.0 1.7 12 20.4 ± 0,28 19.5-21.0 2.4 0.809 23 8 20.5 ± 0.19 20.1-20.9 1.3 7 20.4 ± 0.17 19.9-20.5 1.1 0.875 24 9 20.8 ± 0.13 20.5-21.1 0.9 7 20.3 ± 0.26 19.8-20.8 1.7 14.000’*’ 25 9 20.6 ± 0.27 19.8-21.1 1.9 6 20.5 ± 0.26 20.0-20.9 1.5 0.429 26 4 20.7 ± 0.47 20.2-21.3 2.3 8 20.7 ± 0.34 19.9-21.3 2.3 0.000 27 7 19.5 ± 0.30 18.9-20.0 2.0 1 1 19.7 ± 0.20 19.1-20,3 1.7 1.781 proved to be significantly larger than females in the following measurements from localities shown in parentheses: total length (St. John, 11); length of hind foot (St. John, 1 1); length of forearm (St. Mar- tin, 19); greatest length of skull (eastern Puerto Rico, 10; Norman Island, 12; Martinique, 24); condylo- basal length (Barbuda, 20); palatal length (St. John, 11); braincase depth (eastern Puerto Rico, 10; Mar- tinique, 24; Barbados, 27); zygomatic breadth (east- ern Puerto Rico, 10; St. John, 11; Dominica, 23; Martinique, 24); breadth of braincase (eastern Puerto Rico, 10; Martinique, 24); mastoid breadth (eastern Puerto Rico, 10; St. John, 11; Dominica, 23; Martinique, 24); postorbital breadth (Guadeloupe, 22); length of maxillary toothrow (Martinique, 24); rostral width at canines (eastern Puerto Rico, 10; St. John Island, 11; Norman Island, 12; Barbuda, 20; Martinique, 24); breadth across upper molars (St. John Island, 11; Barbuda, 20; Martinique, 24); man- dibular length (Martinique, 24). Although males exceeded females significantly in size in all 16 measurements except length of ear from one or more localities, females proved to be signifi- cantly larger than males in length of hind foot in the sample from eastern Puerto Rico ( 10), and in breadth across upper molars in specimens from St. Lucia (25). Samples showing males to be significantly larger than females in more than one character include eastern Puerto Rico, St. John Island, Norman Is- land, Barbuda, and Martinique. With the exception of the sample from Barbuda, all these correspond to fairly large samples. However, Guadeloupe, also represented by a large (males 19, females 13) sample. showed significant differences in males over females only in postorbital breadth. Forearm measurements, which because of loading in pregnant females, might be expected to be greater in females than males, average longer in females than males in 1 1 of 15 samples, but never significantly. In two samples the sexes have the same average length of forearm. In specimens from St. Martin, length of forearm in males was significantly longer than that of females. Conclusions . — In general, males are larger than females in the genus Brachyphylla . Therefore, in all subsequent analyses, where size was involved, males and females were treated separately. Individual Variaiion In samples from west of the Mona Passage, exter- nal measurements, excluding length of forearm, were found to vary much more (CV, 1 .8 to 18.4) than forearm and cranial measurements (CV, 0.2 to 6.7) (Table 1). Of forearm and cranial measurements, palatal length (CV, 0.8 to 5.4), rostral width at canines (CV, 1.1 to 4.5), and postorbital breadth (CV, 1.4 to 6.7) showed the highest individual variation, whereas greatest length of skull (CV, 1.2 to 2.2) and condy- lobasal length (CV, 0.6 to 2.4) showed the least. In samples from east of the Mona Passage, vari- ation in external measurements (excluding length of forearm) was again found to be higher (CV, 0.6 to 12.6) than in forearm and cranial measurements (CV, 0.2 to 6.7). Of the latter, palatal length showed the most variation (CV, 1.4 to 5.6) and greatest length 18 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 12 of skull (CV, 0.4 to 2.5) and condylobasal length (CV, 0.7 to 2.6) the least. Rostral width at canines also showed relatively high coefficients of variation (CV, 0.6 to 6.7). Coiiclusions . — Erom both east and west of the Mona Passage, external measurements taken from the skin tags proved to be highly variable. As pointed out by Sumner (1927), external measurements can be expected to vary more because of the fact that these were usually taken by various collectors under different circumstances. Because of missing data and high individual variation, total length, length of hind foot, and length of ear were excluded from sub- sequent analyses. SPECIFIC RELATIONSHIPS Because of the discordance in the literature (see Introduction) concerning the specific relationships within the genus, both univariate and multivariate analyses were employed to compare the geographic samples. Standard statistics for samples of males and females from geographic samples are given in Table 1. Univariate Analyses The SS-STP analyses revealed geographic sam- ples west of the Mona Passage (samples 1 to 8) grouped in one subset, differing significantly from all other samples in the following cranial measure- ments: greatest length of skull (females); condylo- basal length (males and females); palatal length (males); zygomatic width (males and females); length of maxillary toothrow (females); breadth across upper molars (females); mandibular length (males). The results of these analyses for condylo- basal length and mastoid breadth are shown in Table 2. This division corresponds to the specific division in the genus as recently suggested by Silva-Taboada (1976) in which he recognized two species, fi. mina from west of the Mona Passage and B. cavernanun from the remainder of the geographic distribution of the genus. Characters that showed wide overlap of subsets were depth of braincase (males) and postorbital breadth (males and females). The remainder of the characters all tend to show basically a break across the Mona Passage, with varying numbers of over- lapping subsets. Multivariate Analyses Distance phenograms for both males and females generated with the NT-SYS program package are il- lustrated in Pig. 2. In addition, a map (Pig. 3), in- cluding values for both sexes, presents appropriate distance coefficients between the connected sam- ples; in most cases, distance coefficients have been given only for contiguous samples. The first three principal components extracted from the principal component analyses are shown for males and fe- males (Pig. 4). The distanee phenograms for both male (cophe- netic correlation value, 0.975) and female (cophe- netic correlation value, 0.965) Brachyphylla clearly show two major groups. In both cases the upper clus- ter corresponds to samples west of the Mona Pas- sage (Cuba, 1 to 4; Grand Cayman, 5; Middle Caicos, 6; and Hispaniola, 7 and 8), whereas the lower clus- ter corresponds to samples east of the passage (Puer- to Rico, 9 and 10; Virgin Islands, 1 1 to 14; and the Lesser Antilles, 15 to 27). Distance coefficients on the map also clearly show this break across the Mona Passage with values of 1.96 for males and 2.04 for females. On the other hand, these values between contiguous samples west of the passage, and be- tween similar samples to the east of it are less than 1 .00, except between St. Lucia and Barbados where it is 1.03 in the females. The amount of phenetic variation explained by the first three principal components, for males and fe- males, respectively, was 90.6% and 91 .3%, 5.1% and 4.6%, and 2.1% and 1.7% (total, males, 97.8%; fe- males, 97.6%). Results of factor analyses showing characters influencing the first three components for both males and females are given in Table 3. The high percentage of variation explained by the first com- ponent in both males and females reveals that size is the major factor separating the two groups in the principal component analyses. From the factor anal- ysis it can be seen that on the first component, post- orbital width is not weighted heavily (males 0.643 and females 0.677) in separating the groups, whereas all the other characters contribute heavily (above 0.900). Postorbital breadth (Component II) and ros- tral width at canines (Component III) influence the other components most heavily. Examination of three-dimensional plots reveals basically the same pattern as the distance pheno- grams for both sexes. Samples on the left of the plot are the same samples that were found in the upper cluster of the phenograms, which are the samples 1978 SWANEPOEL AND GENOW AYS— BRACHYPHYLLA SYSTEMATICS 19 Table 2. — Results of two SS-STP analyses (coinlylohasal length and mastoid breadth) of geographic variation in Brachyphylla nana and B. cavernarum. Vertical lines to the right of each set of means connect maximally nonsignificant subsets at the 0.05 level. See text for key to sample numbers. Males Females Sam- Sam- pie p!e num- Results num- her Means SS-STP her Means Results SS-STP Condylobasal length 20 29.2 16 28.8 19 28.7 26 28.6 14 28.6 15 28.6 25 28.6 23 28.6 24 28.5 21 28.5 18 28.5 22 28.4 11 28.2 12 28.2 10 28.1 9 28.0 27 27.1 1 25.5 4 25.4 6 25.4 2 25.3 3 25.0 8 24.9 16 15.1 14 15.1 24 15.0 23 15.0 19 15.0 15 15.0 10 15.0 25 15.0 20 15.0 18 14.9 21 14.9 22 14.9 9 14.8 12 14.8 It 14.8 26 14.7 27 14.4 4 13.8 2 13.7 6 13.6 1 13.5 8 13,4 3 13.3 25 28,6 14 28.5 19 28.4 23 28.4 26 28.4 21 28.3 17 28.2 24 28.2 20 28.2 22 28.1 U) 28.0 15 28.0 12 28.0 II 27.9 9 27.8 27 27.0 2 25.4 3 25.3 6 25.3 I 25.3 7 25.2 8 24.8 Vfastoid breadth 14 14.9 19 14.9 25 14.8 23 14,8 21 14.7 26 14.7 22 14.7 10 14.6 24 14.6 12 14.6 9 14.6 20 14.6 15 14.5 11 14.4 17 14.3 27 14.1 6 13.7 2 13.4 3 13.4 8 13.3 1 13.2 7 13.1 from west of the Mona Passage. Samples on the right of the plot correspond to all samples east of the pas- sage. Sample 27 (Barbados) is somewhat separated from the cluster of samples on the right, and corre- sponds to the presently recognized subspecies/?, c. minor . In both male and fema\e Brachyphylla. multivari- ate analysis of variance (MANOVA) showed that there were significant (P < 0.0001) morphological differences among samples for all characters in the following statistical tests ( Hotelling- Lawley's Trace, Pillai's Trace, Wilks' Criterion, and Roy's Maxi- mum Root Criterion). Two-dimensional plots of the samples onto the first two canonical variates based on a matrix of vari- ance-covariance among one external and 12 cranial characters are presented for 26 male samples in Fig. 5 and for 24 female samples in Fig. 6. The amount (percentage) of phenetic variation represented in the first three canonical variates for male and female Brachyphylla, respectively, was 87.1 and 76.9 for variate I, 4.2 and 7.5 for variate II, and 3.2 and 4.5 for variate III. Combined the first three canonical variates express 94.5% in males and 88.9% in fe- males. In both males and females it took ail 13 ca- nonical variates to explain all the variation. The rel- ative contributions of each character to the first three canonical variates in males and females are given in Table 4. Examination of the two-dimensional plots of the samples of both males and females reveals two dis- tinct groups well separated on the first variate. Sam- ples of the population east of the Mona Passage are grouped in the cluster at the top and those from west of the passage in the cluster at the bottom. In both males and females, length of maxillary toothrow (males 23.5, females 15.7) and mandibular length (males 15.4, females 20.2) contributed the heaviest toward separating the two groups on the first variate. Other characters that contributed more than 10% on the first variate include breadth across upper molars in males, and condylobasal length in females. The following characters in males contributed more than 10% on the second variate, condylobasal length, pal- atal length, depth of braincase, postorbital breadth, and rostral width at canines, and on the third variate, forearm length, greatest length of skull, postorbital breadth, and mandibular length; and in females on the second variate, greatest length of skull, condy- lobasal length, and rostral width at canines, and on the third variate, greatest length of skull and man- dibular length. 20 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 12 2.16 1.86 1.56 1.26 0.96 0.66 0.36 0.06 7 8 9 24 26 10 11 12 17 15 20 19 22 23 25 21 14 18 27 I I 1 1 I i 1 I 2.13 1.83 1.53 1.23 0.93 0.63 0.33 0.03 Fig. 2. — Phenograms of numbered samples (see Fig. 1 and text) of Brachyphylla (males left, females right) computed from distance matrices based on standardized characters and clustered by unweighted pair-group method using arithmetic averages (UPGMA). The cophenetic correlation coefficient for males is 0.975 and for females 0.965. The SAS canonical variate analyses, therefore, closely correspond to the NT-SYS cluster analysis and the principal component analysis in separating the two groups. Variation in Color Color in the genus Brachyphylla does not exhibit a great deal of variation. Typically the hair is white to yellowish white at the base with the tips darker in some areas on the dorsum. These darker areas, which vary in size, occur as a distinct patch on top of the head and neck and a V-shaped mantle starting approximately at the shoulders and meeting posteriorly in the middle of the dorsum. The flanks are ususally lighter colored. The darker areas may be blackish gray, blackish brown, grayish brown, or dark brown in color. In 38 skins from Cuba, 47% correspond to color standard 5, whereas nearly an equal proportion (37%) are comparable to color standard 3 (see Ma- 1978 SWANEPOEL AND GENOVA AYS—BRACHYPHYLLA SYSTEMATICS 21 Fig. 3. — Map showing distance coefficients (from distance matrices) between samples of BnichyphylUi that were analyzed in the study of geographic variation. The upper coefficients are for males and the lower for females. See Fig. 1 and text for key to samples. BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 12 Fig. 4. — Three-dimensional projections of samples of Brachyphylla (males above, females below) onto the first three principal com- ponents based on matrices of correlation among one external and 12 cranial measurements. Components I and II are indicated in the figure and component III is represented by height. See Fig. I and text for key to samples. terials and Methods). Therefore, the majority have the base of the hair white to yellowish white with the tips of the hair in the dorsal V-pattern varying from grayish brown to dark brown with varying shades of buff. The dark brown specimens having a yellowish tint, all from the Albert Schwartz Col- lection, have a more washed-out appearance than the color standard 5. Other specimens (16%) from Cuba were blackish brown (color standard 2). Of 56 skins examined from Hispaniola, 63% have hair white at the base with blackish gray tips (color standard I). However, there is also a large per- centage (35%) that are grayish brown colored. sometimes tinted huffish (color standard 3), which corresponds in color to all specimens examined from Middle Caicos (19) and Grand Cayman (1). Erom Puerto Rico, 57 skins were examined. Of these, 42% were blackish brown (color standard 2) in color; however, nearly an equal number (35%) were grayish brown, some with a buffy tint (color standard 3). The remainder consisted of 18% black- ish gray specimens (color standard 1), and 5% yel- lowish dark brown specimens (color standard 5). The latter specimens are mostly from the Albert Schwartz Collection. The majority (54%) of the 41 bats from St. John Island are blackish brown in col- 1978 SWANEPOEL AND GENOW AY S~B RAC HYP HYLLA SYSTEMATICS 23 Fig. 5. — Two-dimensional projection of male samples (mean and one standard deviation) of BrachypItylUi onto the first two canonical variates based on a matrix of variance-covariance among one external and 12 cranial measurements. See Fig. I and text for key to samples. or (color standard 2). The remainder varied from grayish brown (34%) (color standard 3) to dark brown (12%), tinted buff or reddish (color standard 4). Over 30 specimens from Norman Island were found to be molting and were excluded from color analysis. Of the 26 remaining skins that were stud- ied, 46% were found to be grayish brown (some with a huffish tint) (color standard 3), 35% blackish brown (color standard 2), 15% dark brown with a reddish tint (color standard 4), and 4% blackish gray (color standard 1). All specimens from St. Thomas (1), St. Croix (4), Anguilla (9), St. Martin (16), and Antigua (1) were blackish brown in color (color standard 2). Of the seven specimens examined from Guade- loupe, three were blackish gray (color standard 1), two dark brown (color standard 4), one grayish brown (color standard 3), and one dark yellowish brown (color standard 5). Ten of 12 bats from Do- minica were blackish brown colored (color standard 2); the remaining two were grayish brown (color standard 3). Of nine specimens from Martinique, 24 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 12 Fig, 6. — Two-dimensional projection of female samples (mean and one standard deviation) of Brachyphylla onto the first two canonical variates based on a matrix of variance-covariance among one external and 12 cranial measurements. See Fig. 1 and text for key to samples. six were blackish brown (color standard 2) and three yellowish dark brown (color standard 5). All three specimens from St. Vincent were blackish gray (color standard 1). Coat color in most (nine of 14) specimens from Barbados have the base of the hair yellowish white with the tips of the hair dark brown and tinted buffy (color standard 5). All (nine) of these specimens are from the Albert Schwartz 1978 SWANEPOEL AND GEJ^OW AYS— BRACHYPHYLLA SYSTEM ATICS 25 Table 3. — Factor matrix from correlation among 13 characters of Brachyphylla studied, showing characters influencing the first three components. Characters Males Females Component I Component 11 Component III Component 1 Component II Component III Length of forearm 0.940 0.125 -0.163 0.947 0.084 -0.020 Greatest length of skull 0.993 0.058 -0.049 0.991 0.017 0.069 Condylobasal length 0.987 0.108 -0.040 0.996 0.030 0.002 Palatal length 0.983 0.034 -0.089 0.981 0.052 0.129 Depth of braincase 0.967 -0.162 -0.002 0.969 -0.051 0.156 Zygomatic breadth 0.994 0.055 -0.004 0.993 0.019 0.031 Breadth of braincase 0.972 -0.025 0.173 0.988 -0.015 0.037 Mastoid breadth 0.978 0.050 -0.097 0.975 0.041 0.036 Postorbital breadth 0.643 -0.760 -0.063 0.677 -0.728 -0.107 Length of maxillary toothrow 0.979 0.113 -0.100 0.985 0.049 0.046 Rostral width at canines 0.933 0.027 0.343 0.905 0.184 -0.377 Breadth across upper molars 0.972 0.026 0.210 0.979 0.099 -0.108 Mandibular length 0.980 0.092 -0.128 0.990 -0.020 0.042 Collection. Other material from Barbados have the There is little variation in color in bats of this base of the hair white with blackish gray tips (color genus. All have the same basic pattern of color. The standard 1) in two specimens, and grayish brown variation that is present is in color of the tips, which with a huffish tint (color standard 3) in three others. varies from grayish brown to blackish gray, and in Table 4. — Eigenvalues of canonical variates showing the percentage influence among 13 characters of Brachyphylla. Eigenvalues shown represent the normalized vector coefficient of each character. Characters Vector I Vector II Vector III Eigenvalue Percent influence Eigenvalue Percent influence Eigenvalue Percent influence Males Length of forearm 0.0072 4.8 -0.0088 6.3 -0.0264 15.8 Greatest length of skull -0.0204 6.8 -0.0032 1.0 -0.0411 12 2 Condylobasal length -0.0199 6.0 -0.0479 14.5 -0.0307 8.2 Palatal length 0.0408 5.0 0.1000 12.4 0.0613 6.7 Depth of braincase 0.0319 4.6 0.0884 12.6 0.0289 3.7 Zygomatic breadth 0.0095 1.9 -0.0386 7.1 -0.0435 7.1 Breadth of braincase 0.0665 9.1 0.0380 5.3 -0.0050 0.6 Mastoid breadth -0.0148 2.4 0.01.30 2.1 -0.0262 3.7 Postorbital breadth -0.0995 6.9 0.2331 16.4 -0.2212 13.6 Length of maxillary toothrow 0.2049 23.5 -0.0121 1.4 -0.0051 0.5 Rostral width at canines -0.0376 3.0 -0.1400 10.9 0.0440 3.0 Breadth across upper molars 0.0883 10.8 -0.0482 5.9 -0.0089 1.0 Mandibular length 0.0716 15.4 0.0192 4.1 0.1275 24.1 Females Length of forearm 0.0151 9.2 -0.0325 8.4 0.0088 2.6 Greatest length of skull 0.0197 6.0 0.1546 19.2 -0.1887 27.0 Condylobasal length -0.0415 11.1 -0.2381 26.2 -0.0696 8.9 Palatal length 0.0486 5.3 0.0965 4.3 0.0566 3.0 Depth of braincase -0.0233 2.9 0.0737 3.9 0.1341 7.4 Zygomatic breadth 0.0467 7.6 0.0990 6.6 -0.0830 6.4 Breadth of braincase -0.0278 3.4 0.0006 0.1 -0.0909 5.3 Mastoid breadth -0.0361 5.0 0.0257 1.5 -0.0385 2.6 Postorbital breadth 0.0662 4.1 -0.2989 7.7 0.1967 5.8 Length of maxillary toothrow 0.1551 15.7 0.2197 9.3 -0.0719 3.5 Rostral width at canines -0.0189 1.3 -0.3608 10.1 -0.0120 0.4 Breadth across upper molars 0.0757 8.3 -0.0546 2.5 0.0280 1.5 Mandibular length 0.1067 20.2 0.0051 0.4 0.2836 25.6 26 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 12 the bases of the hair, which vary from whitish to reddish and yellowish white. Some authors (Good- win, 1933; Sanborn, 1941; Buden, 1977) believed that variation in color in Bnichyphylla followed a geographic pattern. Basically, they felt that the un- derfur of specimens from Hispaniola was more dis- tinctly white than in specimens from Cuba, Puerto Rico, and Lesser Antilles. They also stated that the tips of the hair were more conspicuously pale brown with reddish or yellowish tones compared to specimens from the remainder of the geographic range of the genus. We have not been able to detect these differences in the material that we have studied. Specimens on Cuba were mostly grayish brown to dark brown with a buffy or reddish tint but some specimens lacked this tint. The same was true for the underfur, which had a reddish or buffy tint in most individuals but in some it was white. Most of the specimens from Hispaniola corresponded to color standard 1 but 35% matched with color standard 3 as did 37% from Cuba. On Puerto Rico, Norman Island, and Guade- loupe, specimens matched four of the five color standards, indicating that color variation on these islands nearly spans that found in the entire genus. Specimens from St. John Island and Barbados, rec- ognized as a distinct subspecies, corresponded to three of the color standards. We have not been able to detect any geographic trends in this variation in color. There appears to be little variation in color and what variation is pres- ent can nearly be spanned by individuals from a single island. Taxonomic Conclusions We interpret the univariate and multivariate anal- yses as revealing that the genus Brachyphylla rep- resents two species, Brachyphylla nana from Cuba, Grand Cayman, Middle Caicos, and Hispaniola, and B. cavernamm from Puerto Rico, Virgin Is- lands, and the Lesser Antilles as far south as St. Vincent and Barbados. The latter species is clearly the larger of the two; the range of some measure- ments of B. cavernarum not overlapping those of B. nana in some characters. It is also worthy of note that no species of para- sites are known to be common to both B. caver- narian and B. nana. However, within nana, Cuba and the Dominican Republic share one species of the genus Trichohiiis and within cavernarum Gua- deloupe and Martinique share a species of Ornitho- iloros (Webb and Loomis, 1977). B. cavernarum and/?, nana do share the streblid genus Trichobius, but host different species. Buden (1977), considering these two species to be conspecific, argued that the size differences be- tween the two allopatric taxa are nearly matched by those found among Middle American populations of Artiheus jamaicensis , which were treated as sub- species by Davis (1970). However, these differ- ences are in fact more comparable to size differ- ences seen between A. Jamaicensis and A. iituratus in Central America. A further argument presented by Buden (1977) for recognizing only one species is that there are no differences in the standard karyotypes of the two taxa. However, when considering the fact that, for example, species included in Artiheus, Sturnira, Vampyrops, and Myotis show no intrageneric vari- ation in chromosomal complements (Baker, 1973; Bickham, 1976), this argument is of little value. It should also be pointed out that ErophyUa bomhi- frons and Phyllonycteris poeyi, both endemic West Indian phyllonycterines, have identical karyotypes (Baker and Lopez, 1970; Nagorsen and Peterson, 1975) to Brachyphylla, but no one has considered even placing them in the same genus. Throughout the remainder of this study, we have considered the genus Brachyphylla to be composed of two species — B. cavernarum and B. nana. SYSTEMATIC ACCOUNTS Genus Brachyphylla 1834. Brachyphylla Gray, Proc. Zool. Soc. London, pp. 122- 123, 12 March. Type specie,'). — Brachyphylla cavernarum Gray. Definition Resembles the other phyllonycterines externally in all respects except for having a more stocky build with a shorter snout; lower lip with median groove 1978 SWANEPOEL AND GENOVA AYS— BRACHYPHYLLA SYSTEMATICS 27 ridged by papillae; nodular ridges on chiropata- gium; calcar absent; five lumbar vertebrae, fifth lacking neural spine; skull relatively long, narrow; upper incisors markedly different in size and shape, inner one large, higher than long, recurved, outer one rounded, minute, flat-crowned; anterior upper premolar minute; posterior upper premolar high and short; crowns of upper and lower molars heavily wrinkled; first lower molar with distinct posterioin- ternal cusp, differing markedly from last premolar; interpterygoid space not extending forward as a pal- atal emargination; nasal region without emargina- tion; ears small, separate; nose-leaf rudimentary; tail very short if present and wholly enclosed by interfemoral membrane. Dentition, 1,2/2; C,l/I; P,2/2; M,3/3 = 32, karyotype 2N = 32, FN = 60. Ecology Brachyphylla occupies most of the islands in the Greater and Lesser Antilles. A notable exception is Jamaica from where it is known only from Pleis- tocene or sub-Recent fossil material. These bats are primarily cave dwelling but have been recorded from an old sugar factory by Bond and Seaman ( 1958), from an underground unused sugar house by Koopman (1975), and from a large well by Nellis and Ehle (1977). For the observations on roosting sites of Brachyphylla, see Allen (1911), Barbour (1945), Goodwin (1933), Gundlach (1877), Miller (19026, 1913), and Nellis and Ehle (1977). The mi- croclimate in the caves inhabited by this bat varies from relatively hot, humid, and stable on Cuba (Sil- va-Taboada and Pine, 1969) to relatively cool, not too humid, and less stable on Middle Caicos (Bu- den, 1977). The diet of B. cavernarum is pollen, fruit, and insects (Bond and Seaman, 1958; Nellis, 1971; Gardner, 1977; Nellis and Ehle, 1977) and that of B. nana is fruit, pollen, nectar, and insects (Silva- Taboada and Pine, 1969; Gardner, 1977). Indica- tions are that B. cavernarum is a good thermoreg- ulator (McManus and Nellis, 1972). Nellis and Ehle (1977), however, noted that the body temper- ature of the young, in contrast to adults, seemed to be lowered during sleep. Only ectoparasites have been reported from the genus Brachyphylla (Silva-Taboada and Pine, 1969; Ubelaker et al., 1977; Webb and Loomis, 1977). Webb and Loomis (1977) summarized the ectopar- asites known to be found on Brachyphylla nana (six species of five genera) and B. cavernarum (six species of five genera). No species of parasites are common between nana and cavernarum . However, two genera, Ornithodoros (Argasidae) and Tricho- hius (Steblidae), have been found on both. Two species of Ornithodoros have been found on nana from Cuba and one on cavernarum from Guade- loupe and Martinique. One species of Trichohius has been found on each nana and cavernarum . The same species of Trichohius known from Cuba was found also on these bats from the Dominican Re- public. Brachyphylla cavernarum DlSl RIBUTION This species occurs on Puerto Rico, the Virgin Islands, and down the Lesser Antillean chain as far as St. Vincent and Barbados. Diagnosis Distinguished by large external and cranial size. Various other cranial and dental characteristics sug- gested in the literature to separate the two species appear to be attributable to individual, age, and sec- ondary sexual variation. Comparisons The two species, which occur allopatrically, can be readily distinguished. Brachyphylla cavernarum is larger than Brachyphylla nana, especially in cra- nial measurements (Table 1). In length of maxillary toothrow and mandibular length, there is no overlap in measurements between the two species. No overlap in measurements between males of the two species is present in palatal length, breadth across upper molars, greatest length of skull, and condy- lobasal length. In the latter two characters, overlap of measurements in females occurs only between the sample of B. cavernarum from Barbados in the southern Lesser Antilles and samples of B. nana in the Greater Antilles. Geographic Variai ion Standard statistics for males and females from geographic samples (9 to 27, Fig. 1) are given in Table 1. Univariate Analyses External measurements . — Because of missing data and consequent small or non-existing samples, external measurements, with the exception of fore- arm length, were not subjected to SS-STP analysis. Variation in length of forearm for Brachyphylla cavernarum shows the population from Barbados 28 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 12 (27) to have the shortest forearm of all samples for both sexes, and those from St. John (11) and Nor- man (12) islands to be the next smallest-sized. The range of forearm length in males from Barbados does overlap, to a certain extent, with most other populations, except St. Eustatius (16), St. Martin (19), Barbuda (20), and Antigua (21). This was not the case in females where overlap was found only with samples from Puerto Rico (9, 10), St. John (11) , Norman (12), Saba (15), and Dominica (23). Males and females from Antigua (21) had on the average the longest forearms for the species. No clinal variation in forearm length was apparent. Cranial measurements . — The 12 cranial measure- ments analyzed are discussed below in three groups — 1) five measurements dealing with length of the skull (greatest length of skull, condylobasal length, palatal length, length of maxillary toothrow, and mandibular length); 2) six measurements deal- ing with breadth of the skull (zygomatic breadth, breadth of braincase, mastoid breadth, postorbital breadth, rostral width at canines, breadth across upper molars); 3) one measurement dealing with depth of the skull (depth of braincase). Geographic variation in greatest length of skull for Brachyphylla cavernarum also shows, as for forearm length, the population from Barbados (27) to be the smallest in size. The range of this mea- surement in the Barbados population was clearly lower than that found in samples from the remain- der of the geographic range of the species. The male Barbados sample showed range overlap in greatest length of skull only with samples from St. Lucia (25), Dominica (23), Guadeloupe (22), Antigua (21), Puerto Rico (9, 10), St. John (11), and Norman (12) and females showed overlap only with samples from St. Lucia (25), Martinique (24), Guadeloupe (22), Puerto Rico (9, 10), St. John (11), and Norman ( 12) . In both sexes there was no overlap in this mea- surement between the Barbados sample and the nearest population, St. Vincent (26). However, in both sexes overlap was found between measure- ments of specimens from Barbados and the next to the nearest population, St. Lucia (25). In both sexes, the two samples from Puerto Rico (9, 10) are grouped with those from St. John (1 1) and Norman (12), being the next four smallest-sized samples. These four areas are, however, at the opposite end of the geographic range of the species from Bar- bados. The one specimen examined from St. Thom- as has a greater skull length than the means ob- served for the four samples discussed above (9 to 12), but it falls within the range of observed mea- surement in these samples and because of its geo- graphic position, it is thought to be grouped best with the samples from Puerto Rico, St. John, and Norman. The one male specimen examined from Montserrat (17) corresponds in greatest length of skull to surrounding localities. The sample of males from Barbuda (20) has the largest mean for this character. The one female specimen examined from Anguilla ( 18) was larger in greatest length of skull than the means of all other samples and above the upper range of this measurement in some samples. The population of females from St. Croix (14) had the longest skull. As in forearm length, no geo- graphic dine in this measurement was apparent. In both sexes, samples from Barbados, Puerto Rico, and the Virgin Islands, although overlapping, tend to be grouped in subsets showing a break with the others. Variation in condylobasal length of Brachyphylla cavernarum follows basically the pattern of varia- tion found in greatest length of skull. Palatal length displays a pattern of variation somewhat different from the two previous measure- ments of length. In males the sample from Barbados (27) is again the smallest with the next smallest two being the samples from Puerto Rico (9, 10). How- ever, the palate in the samples from St. John (11) and Norman (12) is relatively much longer. In fe- males this is only true for the sample from Norman (12). The one from St. John (11) is in fact the small- est in size of all samples. The only other measure- ment in which the population from Barbados (27) was not the smallest is in postorbital breadth for males. The mean palatal length for females from Saba (15) falls between those of Puerto Rico and St. John on the one hand and Norman on the other. Eairly broad overlap in palatal length was found between the different samples of the species. This is also evident from the SS-STP analyses where four broadly overlapping subsets in males and three in females are evident. Variation in length of maxillary toothrow is es- sentially the same as for greatest length of skull. However, a somewhat broader overlap of subsets occurs. The pattern of variation displayed in mandibular length is essentially the same as for greatest length of skull. However, the four subsets in which the female sample means fall overlap much more ex- tensively than in greatest length of skull. The means of the female samples from Saba (15) and Montser- 1978 SWANEPOEL AND GENOW AYS— BRACHYPHYLLA SYSTEMATICS 29 rat (17) fall among the means of the populations from Puerto Rico (9, 10), St. John (11), and Norman (12). The pattern of variation displayed in zygomatic breadth of Brachyphylla cavernarinn is essentially the same as for greatest length of skull. However, in the males the population from St. Vincent (26) falls within the grouping of populations from Puerto Rico (9, 10), St. John (11), and Norman (12), where- as in greatest length of skull it was just slightly lon- ger than the means of these populations. In females, samples from Martinique (24) and Montserrat (17) displayed a relatively narrow zygomatic breadth, falling within the range of means exhibited by the populations from Puerto Rico, St. John, and Nor- man. Because of broadly overlapping subsets in fe- males, this could be due to random variation. In males, there is less overlap and an indication of a break between the Virgin Islands and the Lesser Antilles is evident as it was for both sexes in great- est length of skull. The samples from Barbados (27) again averaged the smallest in size for the species. Variation in breadth of braincase is essentially as in greatest length of skull, with somewhat wider overlap of subsets. It also differs in that the male sample from Guadeloupe (22) displays a relatively narrower breadth of braincase. Variation in mastoid breadth, judged by the broadly overlapping subsets displayed in SS-STP analysis, could perhaps be explained mainly by ran- dom variation. However, the population from Bar- bados (27) still had the narrowest braincase, and the populations from Puerto Rico, St. John, and Nor- man still tend to group together exhibiting relatively narrow braincases. In males, the one sample from Puerto Rico (10) exhibited a relatively wide brain- case. Variation in postorbital breadth reveals that the populations from Puerto Rico (9, 10), St. John (11), and Norman (12) have a relatively broad postorbital region, falling among the samples with the largest means. The male sample from Barbuda (20) dis- plays the narrowest postorbital breadth of all sam- ples. The female Barbuda (20) sample also averaged relatively narrow for the species but the Barbados (27) population averaged the narrowest. Eairly widely overlapping subsets in both sexes indicate that little variation is present. The pattern of variation displayed by rostral width at canines shows very much the same pattern observed in most of the characters studied. Speci- mens from Barbados (27) have the narrowest ros- trum with those from Puerto Rico (9, 10), St. John (11), and Norman (12) being relatively narrow as well. The males from Barbuda (20) have the broad- est rostrum, whereas in the females from Barbuda (20) it is relatively much narrower, grouping with the smallest-sized samples. Variation in width across upper molars follows that of rostral width at canines. Pour broadly over- lapping subsets are exhibited in both sexes. Variation in depth of braincase shows little geo- graphic variation, exhibiting only two broadly over- lapping subsets in both sexes. The samples of both sexes from Barbados (27) still have the shallowest braincase but the Barbuda (20) samples of both males and females also have a relatively shallow braincase in contrast to the situation in most other characters where this sample averaged relatively large-sized. Multivariate Analys es Distance phenograms for both males and females generated with the NT-SYS program package are illustrated in Pig. 7. In addition a map (Pig. 8), in- cluding values for both sexes, shows the appropri- ate distance coefficients between the connected samples; in most cases distance coefficients have been given only for contiguous samples. The first three principal components extracted from the prin- cipal component analysis are shown for both males and females in Fig. 9. A factor matrix from corre- lation among one external and 12 cranial measure- ments for both sexes is given in Table 5. Two- dimensional plots of the first two variates in a canonical variate analysis generated with the Sta- tistical Analysis System (SAS) package are illus- trated for males in Fig. 10 and females in Fig. II. The relative contribution of each original variable to a particular canonical variable is shown in Table 6. The distance phenogram (cophenetic correlation coefficient, 0.910) for male Brachyphylla caverna- riun shows the samples falling into five major groups. The first cluster contains samples from Puerto Rico (9, 10), St. John (11), Norman ( 12), and St. Thomas (13). Specimens from samples in this cluster are of medium size. The second group in- cludes samples from St. Croix (14), Saba (15), St. Eustatius (16), Anguilla (18), St. Martin (19), Anti- gua (21), Guadeloupe (22), Dominica (23), Marti- nique (24), St. Lucia (25), and St. Vincent (26). Al- though this cluster could be divided into two subclusters, the groupings would not be logical on 30 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 12 2.73 2.38 2.03 1.68 1.33 0.98 0.63 0.28 2.46 216 1.86 1.56 1.26 0.96 0.66 0.36 Fig. 7. — Phenograms of numbered samples (see Fig. 1 and text) of Bnichyphylla cavenuinun (males left, females right) computed from distance matrices based on standardized characters and clustered by unweighted pair-group method using arithmetic averages (UPGMA). The cophenetic correlation coefficient for males is 0.910 and for females 0.864. geographical grounds. Groups 3, 4, and 5 in- clude one sample each — Barbuda (20), Montser- rat (17), and Barbados (27). The sample of four specimens from Barbuda is large-sized with a rel- atively narrow postorbital region and shallow brain- case. The one specimen from Montserrat (17) is characterized by a long skull that is relatively nar- row and shallow. The sample from Barbados con- sistently averaged among the smallest in size for the species. The distance phenogram (cophenetic correlation coefficient 0.864) for female B. cavenuirum reveals the samples falling into five groups. The first cluster consists of samples from Puerto Rico (9, 10) and St. John (11). The second cluster contains samples from Norman (12), Saba (15), and Montserrat (17). The third cluster consists of the following samples: St. Croix (14), St. Martin (19), Barbuda (20), Anti- gua (21), Guadeloupe (22), Dominica (23), Marti- nique (24), St. Lucia (25), St. Vincent (26). This cluster could be divided into two subclusters but again this would not be logical on geographic grounds. The fourth and fifth clusters each consist of only one sample each, Anguilla (18) and Barba- dos (27). The sample from Anguilla consists of only one specimen, which is characterized by a large skull with a relatively shallow braincase. The sam- ple from Barbados, as in the males, is the smallest- sized population within the species. In both sexes samples from Puerto Rico (9, 10), and St. John (11), group in the one cluster. How- ever, in the case of the males, the sample from Nor- man ( 12) is also contained in this cluster, whereas in the females it groups with another cluster, which has no counterpart in the males. This might be in- dicative of some past gene flow among populations 1978 SWANEPOEL AND GENOVA AYS— BRACHYPHYLLA SYSTEMATICS 31 Fig. 8. — Map showing distance coefficients (from distance matrices) between samples of Bnichyphyliti ciiveniarum that were analyzed in the study of geographic variation. The upper coefficients are for males and the lower for females. See Fig. 1 and text for key to samples. 32 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 12 Table 5. — Factor matrix front correlation among 13 characters of B . cavernarum studied, showing characters influencing the first three components. Characters Males Females Component I Component Component 111 Component I Component 11 Component III Length of forearm 0.772 -0.063 -0.341 0.679 -0.069 0.123 Greatest length of skull 0.938 -0.095 -0.178 0.915 -0.136 -0.110 Condylobasal length 0.880 -0.321 -0.064 0.970 -0.116 -0.025 Palatal length 0.826 -0.062 0.005 0.762 0.407 0.247 Depth of braincase 0.409 0.836 -0.068 0.702 -0.078 0.655 Zygomatic breadth 0.930 0.070 -0.150 0.932 0.058 0.186 Breadth of braincase 0.812 0.374 0.049 0.931 -0.009 0.011 Mastoid breadth 0.823 -0.067 -0.276 0.880 -0.248 0.242 Postorbital breadth -0.121 0.855 -0.405 0.181 -0.949 -0.134 Length of maxillary toothrow 0.839 -0.416 0.156 0.854 0.154 -0.454 Rostral width at canines 0.652 0.414 0.575 0.826 0.106 -0.185 Breadth across upper molars 0.800 0.332 0.398 0.845 0.230 -0.257 Mandibular length 0.854 -0.313 -0.043 0.855 -0.047 -0.258 from Puerto Rico through the Virgin Islands to the remainder of the Lesser Antilles. The population from St. Croix (14), geographically intermediate be- tween the two areas but fairly well isolated from the remainder of the Virgin Islands by a deep chan- nel, do not seem to be instrumental in the relation- ship. The distinct cluster formed by four male spec- imens from Barbuda is not matched in the sample of seven females. The amount of phenetic variation represented in the first three principal components for male and female Brachyphylla cavernarum, respectively, was 60.1 and 67.0 for component I, 17.3 and 9.7 for component II, and 7.2 and 7.7 for component III. Erom the factor analysis it can be seen that in males the first and most important component is heavily influenced by general size; however, depth of brain- case showed a relatively low positive value and postorbital breadth a low negative value. This neg- ative influence of postorbital breadth corresponds to what we have seen in the univariate analysis, where this measurement tended to become narrow- er when others became larger. Component II is in- fluenced by depth of braincase and postorbital breadth. Component III is negatively influenced by length of forearm and postorbital breadth and pos- itively by rostral width at canines and breadth across upper molars. In females, component I is heavily influenced by all characters except postor- bital breadth, although not negatively so as in males. Component II is negatively influenced by postorbital breadth. Component III is positively in- fluenced by depth of braincase and negatively by length of maxillary toothrow. Examination of the three-dimensional plot of the male samples reveals a pattern similar to that of the distance phenogram, whereas in the plot of the fe- male samples the two analyses differ in some ways. The sample of females from Norman (12) clustering in the distance phenogram with samples from Saba (15) and Montserrat (17), appears in the three-di- mensional plot to be closer to samples from Puerto Rico (9, 10). Samples from St. Croix (14) and St. Lucia (25) form a distinct cluster in the three-di- mensional plot but this is not evident in the distance phenogram. In both the distance phenogram and principal component analysis the samples from An- guilla (18) and Barbados (27) form their own clus- ters. In both male and female Brachyphylla caverna- rum, a MANOVA showed that there were signifi- cant (P < 0.0001) morphological differences among samples in all four statistical tests (Hotelling-Law- ley’s Trace, Pillai’s Trace, Wilks’ Criterion, and Roy’s Maximum Root Criterion) utilized. Among individual measurements only depth of braincase in males showed no significant differences among samples. In the univariate analysis of depth of braincase, two broadly overlapping subsets resulted from the SS-STP analysis in both males and fe- males, also suggesting little variation in this mea- surement between different samples of Brachy- phylla cavernarum . The amount (percentage) of phenetic variation represented in the first three canonical variates for male and female Brachyphylla cavernarum, respec- tively, was 53.7 and 33.0 for variate I, 15. 1 and 23. 1 for variate II, and 8.3 and 15.0 for variate III. Com- 1978 SWANEPOEL AND GENOV/ AYS— BRACHYPHYLLA SYSTEMATICS 33 Fig. 9. — Three-dimensional projections of samples of Brai hyphylla cavenuirmn (males above, females below) onto the first three principal components based on matrices of correlation among one external and 12 cranial measurements. Components 1 and II are indicated in the figure and component III is represented by height. See Fig. 1 and text for key to samples. bined these variates express 77.1% in males and 71.1% in females. In both males and females it took all 13 canonical variates to explain all the variation. The relative contributions of each character to the first three canonical variates in males and females are given in Table 6. Separation on the first variate in males is heavily (10%) influenced by greatest length of skull, post- orbital breadth, length of maxillary toothrow, and mandibular length, and in females by condylobasal length and mandibular length. The second variate in males is heavily (10%) influenced by length of forearm, greatest length of skull, length of maxillary toothrow, and mandibular length, and in females by condylobasal length, length of maxillary toothrow, and rostral width at canines. The third variate in males was most heavily influenced (10%) by con- dylobasal length, breadth of braincase, and mandib- ular length. In females length of forearm, condylo- basal length, zygomatic breadth, and length of maxillary toothrow contributed more than 10% to the separation of the samples on the third variate. Examination of the two-dimensional canonical variate plot of the 19 male samples generally reveals a pattern of variation similar to that found in the distance phenogram and principal component anal- ysis. On the first variate, three groups are evident. The one at the top consists of only one sample (Barbuda, 20), one at the bottom consists of the Puer- to Rican samples (9, 10), and the main group in the middle includes all other samples, including the one specimen from Montserrat (17), which in both the 34 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 12 Table 6. — Eigenvalues of canonical variates showing the percentage influence among 13 characters of B. cavernarum. Eigenvalues shown represent the normalized vector coefficient of each character. Vector I Vector 11 Vector ill Character Eigenvalue Percent influence Eigenvalue Percent influence Eigenvalue Percent influence Length of forearm -0.0066 Males 2.3 0.0285 11.2 0.0136 4.3 Greatest length of skull -0.0642 10.3 0.0929 17.7 0.0360 5.9 Condylobasal length -0.0351 5.0 -0.0293 4.9 -0.2016 29.3 Palatal length 0.0117 0.7 -0.0897 6.5 0.0126 0.8 Depth of braincase 0.0303 2.0 -0.1117 9.0 0.0535 3.6 Zygomatic breadth -0.0218 2.0 0.0594 6.1 0.0151 1.3 Breadth of braincase 0.0926 6.0 0.0140 1.1 -0.1604 10.5 Mastoid breadth -0.0897 6.9 0.0038 0.4 -0.0586 4.5 Postorbital breadth -0.4782 15.5 0.1109 4.2 0.1826 6.0 Length of maxillary toothrow 0.3126 29.9 0.2143 24.1 -0.0534 3.1 Rostral width at canines 0.0187 0.7 0.0359 1.6 -0.0762 2.9 Breadth across upper molars 0.0421 2.5 0.0289 2.0 0.1257 7,5 Mandibular length 0.1587 16.3 -0.0933 11.3 0.1751 18.2 Length of forearm 0.0362 Females 8.6 0.0264 4.8 0.0363 11.5 Greatest length of skull -0.0152 1.7 0.0998 9.0 0.0348 5.5 Condylobasal length 0.2787 28.9 -0.3084 24.5 -0.1641 22.7 Palatal length -0.0976 4.3 0.1219 4.1 -0.1068 6.3 Depth of braincase -0.1637 8.0 0.0329 1.3 -0.0750 4.1 Zygomatic breadth -0.0618 3.9 0.0987 4.8 -0.1677 14.2 Breadth of braincase 0.1207 5.7 0.2673 9.6 0.0933 5.8 Mastoid breadth 0.0732 4.0 -0.1189 4.9 0.0205 1.5 Postorbital breadth 0.2399 5.7 -0.4875 8.8 0.0208 0.6 Length of maxillary toothrow -0.0988 4.0 0.3853 11.8 0.3602 19.2 Rostral width at canines 0.3576 9.4 -0.0586 11.9 -0.0147 0.5 Breadth across upper molars -0,0425 1.8 0.0085 0.3 -0.0165 0.9 Mandibular length -0.1880 14.0 -0.0728 4.2 -0.0625 6.2 distance phenogram and principal component anal- ysis is clearly separated from the other samples. On the second variate the population from Barbados (27) is well separated, showing one standard devia- tion overlap only with the samples from St. John (11) and Norman (12). The sample from St. John (11) is somewhat removed on the hrst variate from the middle cluster and shows some overlap with the western Puerto Rican sample (9) at the bottom. The Norman population ( 12) falls between the Barbados sample (27), and the main cluster of samples. At the right of the plot, the sample from Antigua (21) shows some separation from the main cluster on the second variate. Examination of the two-dimensional canonical variate plot of 16 female samples onto the first two variates reveals a pattern of variation generally sim- ilar to that found in the distance phenogram and the principal component analyses. Two main groups of samples are evident on the first variate. The one at the bottom consists of only one sample (Barbados, 27) and the group at the top contains the remainder of the samples. The eastern Puerto Rican (10), St. John (11), and Norman (12) populations are clearly separated from the main cluster on the second vari- ate. None of these three sample means are included within a one standard deviation range of any of the other samples nor do their ranges (1 SD) include means of any other samples. The western Puerto Rican sample (9) overlaps extensively with the main cluster, whereas a clear separation from the main cluster and a grouping with eastern Puerto Rico (10), St. John (11), and Norman (12) is illustrated in the distance phenogram and principal component analyses. The sample of two specimens from Mont- serrat (17) forms a subgroup somewhat removed from the main group on the first variate to the top of the plot and overlaps with the one standard de- viation range of the samples from Martinique (24) and St. Lucia (25). At the right of the plot, a sub- 1978 SWANEPOEL AND GENOW AYS— BRACHYPHYLLA SYSTEMATICS 35 Fig. 10. — Two-dimensional projection of male samples (mean and one standard deviation) of Brachyphylla (. avcnuirum onto the first two canonical variates based on a matrix of variance-covariance among one external and 12 cranial measurements. See Fig. 1 and text for key to samples. group separated on the first variate, with no counter- part in the other multivariate analyses, is formed by samples 15 (Saba) and 20 (Barbuda). The means of these two samples fall outside the one standard deviation range of all other samples. Taxonomic Conclusions Based upon our assessment of geographic varia- tion in Brachyphylla cavernarnin , we believe there are three identifiable populations. The smallest in- dividuals in the species, and phenetically the most 36 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 12 Fig. 11. — Two-dimensional projection of female samples (mean and one standard deviation) of Brachyphylla cavernarum onto the first two canonical variates based on a matrix of variance-covariance among one external and 12 cranial measurements. See Fig. 1 and text for key to samples. distinct, occur on Barbados and the name Brachy- phytla cavernarum minor Miller, 1913, applies to them. The nominate subspecies, Brachyphylla cav- ernarum cavernarum, representing the largest in- dividuals of the species, occurs on St. Croix in the Virgin Islands and Anguilla southward through the Lesser Antilles to St. Vincent. A third subspecies, which is characterized by intermediate size and is 1978 SWANEPOEL AND GENOWAYS—BRACHYPHYLLA SYSTEMATICS 37 described herein as new, occurs on Puerto Rico and most of the Virgin Islands (St. John, Norman, and St. Thomas excluding St. Croix). This subspecies is not distinguished by any one single character but its overall size as measured in multivariate analyses indicates that 80% to 90% of the individuals in this population are distinguishable from Lesser Antil- lean populations. The population from Barbuda may represent a phenetically identifiable population and, therefore, may represent a separately evolving lineage. However, because our data are inconclu- sive, we have thought it best not to recognize this population for the time being. Brachyphylla cavernarum cavernarum Gray, 1834 1834. Brachyphylla cavernarum Gray, Proc. Zool. Soc. London, p. 123, 12 March. Leciotype. — Adult male, in alcohol with skull not removed, BMNH 77.2746, from St. Vincent, Lesser Antilles, obtained by L. Guilding. Measurements of leetotype . — Length of forearm, 65.5. Distribution . — Known from St. Croix in the Vir- gin Islands and Anguilla southward through the Lesser Antilles to St. Vincent. Comparisons . — The nominate subspecies can be distinguished from minor and intermedia by its larg- er overall size (see also Comparisons under B. c. intermedia). Remarks . — Brachyphylla cavernarum caverna- rum is a large-sized subspecies potentially in con- tact with the medium-sized B. c. intermedia in the north on the Virgin Islands and Puerto Rico, and to the southeast with the small-sized B. c. minor from Barbados. The only indication of possible past con- tact between cavernarum and intermedia was the grouping of samples of females from Norman Is- land, Saba, and Montserrat in the female distance phenogram. There is no evidence for intergradation between these two subspecies through the popula- tion on St. Croix. This population is clearly related to B. c. cavernarum . If there is intergradation between cavernarum and minor, it is probably through the population to the northwest of Barbados on St. Lucia rather than the population to the west on St. Vincent. In great- est length of skull (both sexes), condylobasal length (females), breadth across upper molars (males), and mandibular length (males) there was no overlap in the range of measurements between populations on Barbados and St. Vincent; however, there was overlap in both sexes between Barbados and St. Lucia populations. In the original description Gray listed two co- types, a male and a female, from St. Vincent. Gray (1838) again stated that this species is known only from St. Vincent. In listing the mammalian speci- mens present in the collection of the British Mu- seum, Gray (1843) indicated that at that time an additional specimen from Cuba, presented by W. S. MacLeay, was in the collection. From the above it is clear to us that this female specimen from Cuba was not available to Gray when he described B. cavernarum . Therefore, Dobson (1878) incorrectly listed this specimen from Cuba as the holotype. Dobson does list a male from St. Vincent and a female from the “West Indies," which may repre- sent the cotypes. The female paralectotype mentioned by Gray (1834) could not be located in the British Museum (Natural History) collection. The specimen presum- ably has been destroyed or was exchanged with another institution sometime in the past. However, according to John Edwards Hill (in litt 16 Novem- ber 1977) "There are in the collections male and female specimens of B. cavernarum BM(NH) 7.1.1. 701-702, that came here from the collection of R. F. Tomes. The documentation indicates that Tomes obtained these from the Zoological Society of Lon- don and there is every probability that these, too, are from the original series. Both are in good con- dition: the male is BM(NH) 7.1.1. 701, the female BM(NH) 7.1.1. 702." Specimens examined (206). — Si. Croix: Sion Hill, 11 (AMNH); no specific locality, 6 (4 AMNH, 2 AS), Saba: Bat hole near Land Point, 2 (RMNH); Ladderberg, 6 (RMNH); Windwardside, 1 (AMNH); no specific locality, 2 (RMNH). St. Eustatius: rim of The Quill, 2 (AMNH); no specific locality, I (MCZ). Montserrat: no specific locality, 5 (USNM). St. Martin: Lowlands, 16 (AMNH). Barbuda: no specific locality, 12 (USNM). Anguii la: Island Harbor, Fountain Cave, 7 (AMNH); Valley, 3 (AMNH). Aniigua: 1 mi E English Harbor, 1 (KU); St. Paul Parish. 2 (FMNH); no specific locality, 17 (3 BMNH, 14 USNM). Guadeloupe: 2 km S, 2 km E Baie-Ma- hault, Basse-Terre, 1 (TTU); 2 km N Baillif, Basse-Terre, 1 (TTU); I km S Basse-Terre, Basse-Terre, I (TTU); I km S, 4 km W Vernou, Basse-Terre, 1 (TTU); 1 km W Vernou, Basse- Terre, 1 (TTU); I km N, 1 km W St. Francois, Grand-Terre, 27 (TTU); no specific locality, 1 (MCZ). Dominica: Clarke Hall Estate, 100 ft, St. Joseph Parish, 10 (KU); 6 mi NE Roseau, St. Paul Parish, 2 (AS); no specific locality, 6 (1 AS, 5 USNM). Martinique: Bellefontaine, 2 (AMNH); Case Pilote, 5 (AMNH); 6 km E La Trinite, 4 (AMNH); no specific locality, 9 (1 AMNH, 8 MCZ). Sr. Lucia: no specific locality, 20 (USNM). St. Vincent: Clifton Hill, 400 ft, St. George Parish, 2 (KU); Kingstown, 150 ft, St. George Parish, 1 (KU), no spe- cific locality, 18 (3 BMNH, 15 USNM). 38 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 12 Brachyphylla cavernarum intermedia, new subspecies Holotype. — Adult female, skin, skull, and skele- ton, CM 44707; from 1 mi W Corozal, Puerto Rico; obtained by R. J. Baker on 22 July 1969, original no. 1375. Skin, skull, and body skeleton in good condition. Paratypes . — Two adult males and one adult fe- male, skin, skull, and skeleton, TTU 9819, CM 44708, and TTU 9820; from 1 mi W Corozal, Puerto Rico; obtained by R. J. Baker on 21 July 1969, original nos. S. L. Williams 319, 321, and 320, respectively. Skins, skulls, and body skeletons in good condition. Measurements . — External and cranial measure- ments of the holotype and paratypes, respectively, were as follows; total length, 86, 93, 87, 91; length of hind foot, 18, 17, 16, 15; length of ear, 20, 22, 21, 21 ; length of forearm, 66.5, 66.6, 66.5, 68.0; greatest length of skull, 32.1, 32.1, 32.7, 32.0; condylobasal length, 28.9, 28.4, 28.9, 28.6; palatal length, 12.0, 12.4, 11.8, 11.7; depth ofbraincase, 13.7, 13.6, 13.8, 13.6; zygomatic breadth, 17.6, 17.5, 17.4, 17.5; breadth ofbraincase, 12.6, 12.6, 12.9, 12.7; mastoid breadth, 15.0 15.0, 14.9, 14.9; postorbital breadth, 6.5, 6.4, 6.6, 6.6; length of maxillary toothrow, 10.8, 10.8, 10.7, 10.7; rostral width at canines, 7.1, 7.2, 7.5, 7.0; width across upper molars, 12.1, 11.6, 11.8, 11.5; mandibular length, 20.5, 20.5, — , 20.3. Distribution. — Puerto Rico and Virgin Islands (excluding St. Croix). Comparisons . — Brachyphylla cavernarum inter- media is distinguished from Brachyphylla caver- narum cavernarum by its smaller cranial size. From B. c. minor, with which it is not potentially in contact, B. c. intermedia differs in being larger, both externally and cranially (see Tables 1 and 2). Specimens herein referred to B. c. intermedia pre- viously have been reported as B. c. cavernarum . No overlap was found in sample means of either sex among intermedia, cavernarum, and minor in one measurement (range of means in intermedia, cavernarum , and minor, respectively) — greatest length of skull (males, 31.4-31.7, 31.9-32.4, 30.5; females, 31.0-31.5, 31.6-32.3, 30.5). In length of maxillary toothrow (males, 10.6-10.8, 10.9-11.3, 10.6; females 10.7, 10.8-11.1, 10.5) overlap was ob- served in sample means of males of minor and in- termedia only. Overlap in sample means of only one of the sexes among the three subspecies is pres- ent in condylobasal length (males, 28.0-28.2, 28.4- 29.2, 27.1; females, 27.8-28.0, 28.0-28.6, 27.0), breadth of braincase (males, 12.6-12.8, 12.7-13.1, 12.4; females, 12.5-12.6, 12.7-13.0, 12.3), breadth across upper molars (males, 11.5-11.6, 11.7-12.0, 11.1; females, 11.2-11.7, 11.5-11.9, 11.2), and man- dibular length (males, 19.9-20.3, 20.4-20.8, 19.5; females, 19.9-20.1, 20.0-20.7, 19.2). Remarks. — In our opinion, there are populations of three distinct sizes in Brachyphylla cavernarum . The populations on Puerto Rico and most of the Virgin Islands are intermediate in size between the large B. c. cavernarum of St. Croix and the Lesser Antilles as far south as St. Vincent and the small- sized population of B. c. nunor, which is restricted to Barbados. This new taxon, B. c. intermedia, is potentially in contact with B. nana on the west and B. c. cavernarum on the east. Although B. c. intermedia is smaller than B. c. cavernarum, it is still distinctly larger than .6. nana (range of greatest length of skull, male, 30.5-33.0, female, 30.3-32.1 in Puerto Rican samples as com- pared with 27.2-29.3 and 27.1-29.1 in Hispaniolan samples, see also Table 1). We have seen no evi- dence to indicate intergradation or hybridization between these taxa. See account of B. c. caverna- rum for possible intergradation with that taxon and the status of the population on St. Croix. Coloration in intermedia is generally blackish brown, or grayish brown tinted buff, whereas cav- ernarum is mostly blackish brown, with a few gray- ish brown individuals being found. Choate and Birney (1968) reported on sub-Recent fossil material from Puerto Rico. The only mea- surements they took that are comparable to ours in the way they were taken are zygomatic breadth, breadth of braincase, and height of coronoid. In both zygomatic breadth and breadth of braincase, ranges of measurements of Recent material encom- pass those of the sub-Recent material and the means are very close. However, in the sub-Recent material, height of coronoid process ranged lower in addition to averaging smaller. Anthony (1925) after comparing and measuring fossil and Recent Brachyphylla from Puerto Rico could find “no dif- ferences worthy of mention." We consider the sub- Recent as belonging to the new subspecies. Specimens e.xitmined (233). — Puerto Rico: 1.5 km N, 13.5 km E Adjuntas, 1 (LSU); Iglesia de la Mora Comerio, 11 (USNM); 1 mi Corozal, 48 (2 CM, 46 TTU); El Verde Eield Station, 2 (TTU); 5 km E Guanica, 1 (LSU); 7.5 km E Guanica, 12 (AS); Pueblo Viejo, 13 (9 AMNH, 4 USNM); Cueva de Fari, San Juan, 7 (UMMZ); Trujillo Alto, 4 (AMNH); La Cueva de Mollfulleda, Trujillo Alto, 13 (USNM); 17.7 km NE Utuado, 7 1978 SWANEPOEL AND GENOW AYS— BRACHYPHYLLA SYSTEMATICS 39 (AS). St. John: Cruz Bay, 4 (AMNH); Lameshur, 14 (AMNH); V^ mi S, % mi W Lameshur, 42 (40 KU, 2 TCWC). Norman: west end, 53 (15 AMNH, 36 KU, 2 TCWC). Sr. Thomas: Bot- any Bay, 1 (AMNH). Brachyphylla cavernarum minor Miller, 1913 1913. Brachyphylla minor Miller. Proc. Biol. Soc. Washington, 26:32, 8 February. 1968. Brachyphylla cavernarum minor. Koopman, Amer. Mus. Novit., 2333:5, 19 July. Holotype. — Adult female in alcohol with skull re- moved, USNM 101,528, from Cole's Cave, St. Thomas Parish, Barbados, Lesser Antilles, ob- tained by P. McDonough on 14 June 1899. Measurements of holotype .—ToisA length, 78; length of forearm, 61.5; condylobasal length, 26.3; palatal length, 10.8; depth of braincase, 12.6; zy- gomatic breadth, 15.8; breadth of braincase, 12.0; mastoid breadth, 13.8; postorbital breadth, 6.1; length of maxillary toothrow, 10.3; rostral width at canines, 6.4; width across upper molars, 11.0. Distribution . — This subspecies is restricted to Barbados, Lesser Antilles. Comparisons . — Size small for the species crani- ally; averaging the smallest-sized sample of B. eav- ernarum in all characters except palatal length for females and postorbital breadth in males. Remarks. — Brachyphylla cavernarum minor is well differentiated and is potentially in contact only with B. c. cavernarum and can be distinguished from it by its generally shorter forearm and smaller- sized cranium (see also Comparisons under B. c. intermedia). Brachyphylla c. minor from Barbados shows no overlap in measurements with both its nearest neighbors, St. Vincent (26) and St. Lucia (25) in condylobasal length (males) and forearm length (females), and no overlap with St. Vincent (26) only, in the following characters: greatest length of skull (males and females); condylobasal length (females); breadth across upper molars (males); and mandibular length (males) (see Table 1). This taxon was considered to be a distinct species until Koopman (1968) reviewed its status. He pre- sented evidence, and our study supports his find- ings, that this taxon is distinct but only at the sub- specific level. The isolation of the island of Barbados to the east of the main chain of the Lesser Antilles undoubtedly has provided the isolation necessary for the genetic differentiation of this pop- ulation to occur. Most of the bats from Barbados have hair yellow- ish white at the base with dark buffy tinted tips. All these specimens are Albert Schwartz Eield Series material and as in the case of the Cuban material from this collection might have been exposed to some bleaching. Other material from Barbados have base of hair white with blackish gray tips, or grayish brown with a huffish tint. Specimens examined (24). — Barbados: Brighton, 250 ft, St. George Parish, 3 (KU); Cole's Cave, St. Thomas Parish, 6 (5 AMNH, ! USNM); St. Thomas Parish, I (USNM); no specific locality, 14 (II AS, 1 BMNH, 2 FMNH). Brachyphylla nana Distribution This species occurs on Cuba, Isle of Pines (Va- rona, 1974), Grand Cayman, Hispaniola, Middle Caicos, and as a Pleistocene or sub-Recent fossil on Jamaica. Diagnosis See account for Brachyphylla cavernarum . Comparisons See account for Brachyphylla cavernarum . Geographic Variation Univariate Analyses Standard statistics for geographic samples of Brachyphylla nana (samples 1-8, Fig. 1) are given in Table 1 . External measurements . — As in Brachyphylla cavernarum. because of missing data and conse- quent small or nonexisting samples, external mea- surements except length of forearm, were not sub- jected to ANOVA and SS-STP analyses. However, in spite of small sample sizes, it is apparent that the sample from the Dominican Republic (8) is relative- ly smaller sized than the others at least in total length. Length of forearm of the samples from Middle Caicos (6) and the Dominican Republic (8) is rela- tively short for the species in both males and fe- males. The small sample size available from the Haitian (7) population makes meaningful conclu- sions difficult concerning the relationship between the Haitian and Dominican Republic populations. In males, the SS-STP analysis shows that the three samples from Cuba (1, 2, 3) fall in one subset, dif- fering significantly from the second subset, which includes samples from the Dominican Republic (8), Middle Caicos (6), and Cuba (Camagiiey, 4). Sam- ple 4 consists of only two specimens and their fore- 40 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 12 arm measurements appear to fall within the normal variation of most Cuban samples. Females do not exhibit such a clearcut break in the SS-STP analy- sis. Although no clinal trend exists in males from Cuba (1-3), there appears to be an increase in size from the small-sized specimens in Habana Province (1) eastward to Oriente Province (3) in females. Cranial measurements . — The 12 cranial measure- ments analyzed are discussed below in three groups — 1) five measurements dealing with length of the skull (greatest length of skull, condylobasal length, palatal length, length of maxillary toothrow, mandibular length); 2) six measurements dealing with breadth of the skull (zygomatic breadth, breadth of braincase, mastoid breadth, postorbital breadth, rostral width at canines, breadth across upper molars); 3) one measurement dealing with depth of skull (depth of braincase). Geographic variation in greatest length of skull for Brachyphylla nana males shows no significant differences among the seven samples tested, as re- vealed by an ANOVA. In females the values for greatest length of skull fall into two broadly over- lapping subsets. In both sexes the population from the Dominican Republic (8) has a relatively short skull and the one from Middle Caicos (6) a relatively long one. Female samples from Cuba (1-3) show a clinal trend similar to length of forearm, but male samples follow a reverse trend. The two males from Camagiiey Province (4) average large for the species. The one female available from this locality is among the smallest for females in the species. Variation in condylobasal length in Brachyphylla nana follows the pattern of variation for greatest length of skull. In this character, however, the fe- males from Cuba (1, 2, 3) do not show any clinal variation, whereas the males do. Palatal length displays a pattern of variation dif- fering from the previous two cranial measurements. Male samples from the Dominican Republic (8) and Middle Caicos (6) have on the average the longest palate for the species. Although no significant dif- ferences were detected among the samples of fe- males with an analysis of variance test, the Domin- ican Republic (8) and Middle Caicos (6) samples have on the average relatively long palates for the species. The clinal trend among samples from Cuba (1, 2, 3) is not observed in this character. There is no significant variation in length of max- illary toothrow in both males and females. No significant variation in mandibular length is displayed. Although differences among samples could be ascribed to random variation. Middle Cai- cos (6) and Dominican (8) populations tend to have relatively short mandibles. Variation in zygomatic breadth essentially fol- lows the pattern of variation for greatest length of skull and condylobasal length. However, no clinal variation is present. No significant differences in breadth of braincase were detected among samples of both males and females of B. nana with analysis of variance tests. There were only slight differences (range of 0.2) among samples in both sexes. No significant differences in mastoid breadth were detected among samples of males, whereas in females two overlapping subsets were present. In both sexes, it was found that the Dominican Re- public (8) population clearly averages narrower than did the Middle Caicos population. Variation in postorbital breadth for males, falling into two overlapping subsets, shows the Dominican Republic (8) population characterized by a relative- ly broad postorbital region and those from Cuba (1, 2, 3) and Middle Caicos (6) by relatively narrow postorbital regions. No significant differences were detected among the samples of females. Rostral width at canines displays a pattern of variation in which the population from the Domin- ican Republic (8) averages the narrowest and those from Cuba (1, 2, 3) relatively wide, whereas the Middle Caicos (6) population is of intermediate size. In males the means fall into two slightly overlapping subsets. The Dominican Republic (8) and Middle Caicos (6) samples are in one subset and all other samples in the second subset. Overlap between the two occurs only in the Caicos sample. Females fall into four overlapping subsets. The pattern of variation present in breadth across upper molars for males is essentially the same as that found in both sexes for rostral width at canines. The pattern of variation in breadth across upper molars in females differs in that the mean for fe- males from Middle Caicos is closer in size to those of the Cuban samples than to the Dominican Re- public one. The clinal variation found in some measurments for Cuban males is also found in breadth across upper molars. Variation in depth of braincase in both sexes shows the Middle Caicos (6) population character- ized by a relatively deep braincase, and the Cuban and Dominican Republic populations by a relatively shallow braincase. The four female specimens from Haiti have relatively deep braincases. 1978 SWANEPOEL AND GENOW AYS— BRACHYPHYLLA SYSTEMATICS 41 1.65 1.50 1.35 1.20 1.05 0.90 0.75 0.60 1.75 1.60 1.45 1.30 1.15 1.00 0.85 0.70 Fig. 12. — Phenograms of numbered samples (see Fig. I and text) of Bnu hyphylla naiui (males left, females right) computed from distance matrices based on standardized characters and clustered by unweighted pair-group method using arithmetic averages (UPGMA). The cophenetic correlation coefficient for males is 0.808 and for females 0.831. Multivariate Analyses Distance phenograms for both males and females, generated with the NT-SYS program package, are given in Fig. 12. In addition a map (Fig. 13), in- cluding values for both sexes, shows the appropri- ate distance coefficients between the connected samples; in most cases distance coefficients have been given only for contiguous samples. The first three principal components extracted from the prin- cipal component analysis are shown three-dimen- sionally for both males and females in Fig. 14. A factor matrix from correlation among one external and 12 cranial measurements in both males and fe- males are given in Table 7. Two-dimensional plots of the first two variates in a canonical analysis gen- erated with the Statistical Analysis System (SAS) package are illustrated for males in Fig. 15 and for females in Fig. 16. The relative contribution of each character to the first three canonical variates is shown in Table 8. The distance phenogram (cophenetic correlation coefficient, 0.808) for male Brachyphvlla nana shows the samples falling into three major groups. The first cluster contains three samples from Cuba (1,2, 3). The second cluster contains the samples from Middle Caicos (6), and Camagiiey Province, Cuba (4). The two samples in the latter cluster are phenetically quite distinct. The Camagiiey sample (4) consists of only two specimens and, as seen in the univariate analysis, they are medium to large sized except in length of forearm and length of max- illary toothrow where they averaged the smallest. The third cluster consists of two phenetically quite Table 7. — Factor matrix from correlation among 13 characters Brachyphylla nana studied, show ing characters influencing the first three components . Characters Males Females Component I Component II Component III Component I Component 11 Component III Length of forearm 0.316 -0.752 0.047 -0.562 -0.395 0.697 Greatest length of skull -0.813 0.375 0.277 0.983 0.125 -0.012 Condylobasal length -0.909 0.160 0.163 0.937 -0.221 0.136 Palatal length 0.073 0.832 -0.426 0.651 0.596 0.156 Depth of braincase -0.346 0.858 -0.048 0.479 -0.659 -0.168 Zygomatic breadth -0.889 -0.349 0.093 0.694 -0.058 -0.271 Breadth of braincase -0.921 -0.113 -0.214 0.485 -0.421 -0.477 Mastoid breadth -0.181 0.461 0.829 0.730 -0.337 -0.319 Postorbital breadth 0.010 0.617 -0.446 -0.490 0.639 0.268 Length of maxillary toothrow 0.172 0.085 -0.503 0.753 0.215 -0.114 Rostral width at canines -0.968 -0.209 0.002 -0.013 -0.944 0.107 Breadth across upper molars -0.930 -0.291 -0.156 0.358 -0.807 -0.251 Mandibular length 0.421 0.237 0.843 0.556 0.200 0.771 42 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 12 1978 SWANEPOEL AND GENOW AYS— BRACHYPHYLLA SYSTEM ATICS 43 Fig. 14. — Three-dimensional projections of samples of Brachyphylla nana (males above, females below) onto the first three principal components based on matrices of correlation among one external and 12 cranial measurements. Components 1 and II are indicated in the figure and component III is represented by height. See Fig. 1 and text for key to samples. 44 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 12 distinct samples, Haiti (7) and the Dominican Re- public (8). The sample from Haiti (7) consists of only one specimen, which varies from relatively small to large in the different measurements taken. The distance phenogram (cophenetic correlation coefficient, 0.831) for female Brachyphylla luina shows the samples falling into four major groups. The first cluster contains samples from Cuba (1,2, 3) and Grand Cayman (5). The second cluster con- sists of only one sample. Middle Caicos (6). The third cluster contains the samples from Hispaniola (7, 8). The fourth cluster consists of the sample from Camagiiey Province, Cuba (4), and represents a single specimen. This specimen (4) is character- ized by a relatively small skull, but its values fall within the range of at least one other Cuban sample. The distance phenograms of both males and fe- males essentially show the same picture. Samples from Cuba tend to cluster together with the excep- tion of the sample from Camagiiey Province (4). Although distantly, the Haitian sample clusters with that from the Dominican Republic. Both male and female distance phenograms show the Middle Cai- cos samples clustering closer to the Cuban than the Hispaniolan samples, although only distantly so. The amount (percentage) of phenetic variation represented in the first three principal components for male and female Brachyphylla nana, respective- ly, was 41.5 and 40.8 for component I, 23.7 and 25.7 for component II, and 17. 1 and 13.3 for component III. Combined these first three components express 82.3% in males and 79.8% in females. Prom the factor analysis, it can be seen that characters influ- encing the different components differ between sexes. In males the first component is heavily neg- atively influenced by the following characters: greatest length of skull, condylobasal length, zy- gomatic breadth, breadth of braincase, rostral width at canines, and breadth across upper molars. In fe- males the first component is most heavily influ- enced positively by greatest length of skull and con- dylobasal length. In males the second component is heavily positively weighted for palatal length, depth of braincase, and postorbital breadth and neg- atively for length of forearm. In females a heavy negative weighting was found on component II for rostral width at canines and breadth across upper molars. The third component in males is heavily positively influenced by mastoid breadth and man- dibular length. In females the third component is weighted (positive) for length of forearm and man- dibular length. Examination of the three-dimensional plot of the male samples reveals a pattern more or less similar to that of the distance phenogram. The Dominican Republic (8) and the Haitian (7) samples grouped in the lower cluster of the phenogram are shown on the right in the three-dimensional plot, differing from each other on the first and third components. The Middle Caicos (6) and Cuban samples (1, 2, 4) are arranged on the right of the plot with the Oriente Province, Cuba (3) sample falling nearly midway between the samples from the Dominican Republic (8) and Las Villas Province, Cuba (2). This seems to correspond to the conclusion reached in the uni- variate analysis, where the Cuban samples (1, 2, 3) displayed clinal variation in some measurements, becoming progressively smaller from west to east, with the population from Oriente Province (3) gen- erally approaching the Dominican Republic sample (8) in size. Examination of the three-dimensional plot of the female samples reveals a pattern with some basic differences from the distance phenogram. The Ca- magiiey Province sample (Cuba, 4) on the left in the three-dimensional plot corresponds to the lower cluster in the distance phenogram. The Dominican Republic sample (8), well removed from sample 4 (Camagiiey Province) to the left and the other Cu- ban (1, 2, 3) and Haitian (7) samples to the right is however, grouped with the Haitian sample (7) in the distance phenogram. The Haitian sample is sepa- rated from samples 1, 2, and 3 (Cuba) only on the second component. Therefore, it differs mostly in shape rather than size from the Cuban material. The one specimen from Grand Cayman (5) is grouped with the Cuban (1, 2, 3) populations in the pheno- gram. It is, however, well separated on the first component in the principal component analysis from these populations. The Grand Cayman speci- men is close to the Middle Caicos population on the first component but well separated on the second and third components, suggesting a difference in shape rather than size between the two. In both male and female Brachyphylla nana, multivariate analysis of variance (MANOVA) showed that there were significant (P < 0.00001) morphological differences among samples in all four statistical tests (Hotelling-Lawley’s Trace, Pillai’s Trace, Wilks’ Criterion, and Roy’s Maximum Root Criterion) utilized. In males the following individual measurements, however, failed to show significant differences among samples: greatest length of skull, breadth of braincase, mastoid breadth, and mandib- 1978 SWANEPOEL AND GENOWAYS—BRACHYPHYLLA SYSTEMATICS 45 Table 8. — Eigenvalues of canonical variates showing the percentage influence among 13 characters of Brachyphylla nana. Character Vector I Vector 11 Vector III Eigenvalue Percent influence Eigenvalue Percent influence Eigenvalue Percent influence Males Length of forearm -0.0354 7.5 0.0767 12.8 -0.0803 14.5 Greatest length of skull 0.0274 2.6 -0.0738 6.1 0.0983 8.4 Condylobasal length -0.2065 17.7 -0.0125 0.9 0.1004 7.6 Palatal length 0.2340 7.4 -0.1406 3.8 0,0260 0.7 Depth of braincase -0.0240 1.0 -0.1344 4.6 0.1664 6.0 Zygomatic breadth -0.0500 2.6 0.2355 10.2 -0.0987 4.5 Breadth of braincase -0.0825 3.3 0.1709 5.8 -0.2042 7.2 Mastoid breadth 0.0612 2.8 -0.2066 8.1 0.1980 8.1 Postorbital breadth 0.5387 11.5 0.2337 4.2 -0.1181 2.2 Length of maxillary toothrow 0,4703 15.1 0.7122 19.3 0.6221 17.7 Rostral width at canines -0.2343 5.1 -0.6273 11,6 -0.2832 5.5 Breadth across upper molars -0.5052 17,5 0.1961 5.7 0.5698 17.5 Mandibular length -0.0983 6.9 -0,1384 6.9 -0.2081 10.9 Females Length of forearm -0,1055 II. 8 0.0530 6.3 -0,0009 0.2 Greatest length of skull 0.4319 24.0 -0.0344 2.0 0.1146 8.7 Condylobasal length -0.0382 1.9 -0.5990 30.8 0.0328 2.2 Palatal length 0.0723 1.3 0.1991 3.8 -0.1424 3.6 Depth of braincase 0.1318 3.1 0.1387 3.4 0.4349 13.9 Zygomatic breadth -0.0100 0.3 -0.4200 13.0 0.3237 13.0 Breadth of braincase 0.2782 6.4 0.1611 3.9 -0.1780 5.6 Mastoid breadth 0,3047 8.0 0.4157 11.4 -0.6531 23.3 Postorbital breadth -0.9303 11.3 -0.0250 0.3 0.4387 7.2 Length of maxillary toothrow 0.1762 3.2 0.3551 6.8 0.1089 2.7 Rostral width at canines -0.7437 9.2 0.1798 2.3 -0.3533 5.9 Breadth across upper molars -0.2416 4.8 -0.2189 4.6 -0.3170 8.5 Mandibular length -0.4332 14.7 0.3225 11.4 0.1131 5.2 ular length. These measurements as well as length of maxillary toothrow revealed no significant dif- ferences among samples in the univariate analysis. In females, condylobasal length, palatal length, breadth of braincase, mastoid breadth, postorbital breadth, length of maxillary toothrow, and mandib- ular length showed no significant differences among samples in the MANOVA. In the univariate anal- ysis palatal length, breadth of braincase, postorbital breadth, length of maxillary toothrow, and mandib- ular length also showed no significant difference among samples. The amount (percentage) of phenetic variation represented in the first three canonical variates for male and female Brachyphylla nana, respectively, was 68.5 and 66.4 for variate I, and 12.7 and 20.0 for variate II, and 12.1 and 7.1 for variate III. Com- bined these three canonical variates express 93.3% in males and 93.7% in females. In males all the vari- ation was explained by the first five canonical vari- ates, whereas in females it was expressed in the first four canonical variates. In males the following characters contribute more than 10% to variate 1 in distinguishing among sam- ples; condylobasal length, postorbital breadth, length of maxillary toothrow, and breadth across upper molars; more than 10% to variate II: length of forearm, zygomatic breadth, length of maxillary toothrow, and rostral width at canines; and more than 10% to variate III: length of forearm, length of maxillary toothrow, width across upper molars, and mandibular length. In females, characters contrib- uting more than 10% to variate I are length of fore- arm, greatest length of skull, postorbital breadth, mandibular length, in variate II, condylobasal length, zygomatic breadth, mastoid breadth, and mandibular length, and in variate III, depth of braincase, zygomatic breadth, and mastoid breadth. Examination of the two-dimensional canonical variate plot of the male samples reveals the follow- ing pattern of variation. Samples from Middle Cai- cos (6) and Hispaniola (7, 8) are grouped together and are clearly separated from the Cuban samples (1, 2, 3, 4) on the first variate. These two major 46 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 12 Fig. 15. — Two-dimensional projection of male samples (mean and one standard deviation) of Brachyphylla nana onto the first two canonical variates based on a matrix of variance-covariance among one external and 12 crania! measurements. See Fig. 1 and text for key to samples. 1978 SWANEPOEL AND GENOW AYS— BRACHYPHYLLA SYSTEMATICS 47 Fig. 16. — Two-dimensional projection of female samples (mean and one standard deviation) of BmchyphylUi nana onto the first two canonical variates based on a matrix of variance-covariance among one external and 12 cranial measurements. See Fig. 1 and text for key to samples. groups show no one standard deviation overlap on the first variate. The Cuban group, however, shows overlap between samples 1, 2, and 3, but these are clearly separated from sample 4 on the second vari- ate. The canonical variate analysis shows some ba- sic differences when compared to the principal component analysis. In the principal component analysis the Hispaniolan (7, 8) samples are also sep- arated from the Cuban samples on the first com- ponent. However, the Middle Caicos (6) sample is grouped with Cuban material, although differing from these samples on the second and third com- ponent. Therefore, in the case of the distance phe- nogram and principal component analysis the sam- 48 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 12 Table 9. — Geographic variation in eight cranial measurements of six samples of Recent , and one of Pleistocene or suh-Recent fossil material of Brachyphylla. See text for key to samples. Sample N Mean ± 2 SE Range cv Palatal length a 10 9.3 ± 0.23 8. 9-9.9 3.8 b 10 9.5 ± 0.25 8.9-10.0 4.2 c 10 9.5 ± 0.26 9.0-10.2 4.4 d 1 10.1 e 10 11.8 ± 0.24 10.9-12.2 3.3 f 7 11.8 ± 0.42 11.1-12.8 4.7 g 10 11.9 ± 0.36 11.0-12.8 4.8 Rostral width at canines a 10 6.6 ±0.11 6.3-6. 8 2.6 b 10 6.4 ± 0.17 5.9-6. 8 4.3 c 10 6.2 ± 0.15 5. 8-6. 5 3.8 d 1 6.6 e 10 7.2 ± 0.13 6.8-7. 6 2.9 f 7 7.2 ± 0.21 6. 8-7. 5 3.9 g 10 7.2 ± 0.09 7. 1-7.5 2.0 Length of maxillary toothrow a 10 9.4 ± 0.16 9.0-9.8 2.7 b 10 9.5 ± 0.15 9.2-9.9 2.6 c 10 9.5 ± 0.07 9.3-9.6 1.1 d 1 9.6 e 10 10.7 ± 0.18 10.1-11.0 2.7 f 7 10.7 ± 0.12 10.4-10.9 1.5 g 10 10.7 ± 0.10 10.5-11.0 1.5 Interorbital breadth a 10 7.8 ± 0.15 7.4-8. 1 3.0 b 10 8.4 ±0.11 8. 2-8. 7 2.0 c 10 8.1 ± 0.15 7. 7-8. 6 2.9 d 3 7.7 ± 2.0 7.4-7. 8 2.2 e 10 9.0 ± 0.10 8. 9-9.4 1.7 f 7 8.6 ± 0.13 8.4-8. 9 2.0 g 10 8.5 ±0.11 8. 3-8. 9 2.1 Height of coronoid process a 10 7.3 ± 0.14 7. 0-7. 8 3.1 b 10 7.4 ± 0.12 7. 1-7.7 2.5 c 10 7.3 ± 0.13 7. 0-7. 7 2.9 d 3 7.6 ± 0.14 7. 5-7. 7 1.5 e 10 9.0 ± 0.15 8. 5-9.4 2.6 f 7 9.1 ± 0.18 8. 8-9.5 2.6 g 10 9.0 ± 0.18 8. 5-9.5 3.1 Width of articular process a 10 2.6 ± 0.12 2. 3-2. 9 7.5 b 10 2.9 ± 0.06 2. 7-3.0 3.3 c 10 2.5 ± 0.10 2. 3-2. 8 5.9 d 4 2.4 ± 0.18 2. 3-2. 7 7.9 e 10 3.3 ± 0.12 2. 9-3. 5 5.6 f 7 3.2 ± 0.12 3.0-3. 5 5.0 g 10 3.2 ± 0.10 2. 8-3. 3 5.2 Breadth of mandible at M3 a 10 1.2 ± 0.05 1. 1-1.3 6.7 b 10 1.2 ± 0.03 1.2-1. 3 3.9 c 10 1.1 ± 0.04 1.0-1. 2 5.6 d 7 1.3 ± 0.06 1.2-1. 4 6.9 Table 9. — Continued. Sample N Mean ± 2 SE Range cv e 10 1.4 ± 0.04 1.3-1. 5 4.9 f 7 1.5 ± 0.04 1.4-1. 6 3.8 g 10 1.5 ± 0.04 1.4-1.6 4.8 Length of mandibular toothrow a 10 9.8 ± 0.10 9. 5-9.9 1.7 b 10 10.0 ± 0.16 9.6-10.4 2.5 c 10 10.0 ± 0.09 9.7-10.1 1.4 d 2 10.2 ± 0.30 10.0-10.3 2.1 e 10 11.0 ± 0.17 10.5-11.4 2.5 f 7 10.9 ± 0.13 10.7-11.2 1.6 g 10 10.9 ± 0.08 10.7-11.1 1.2 pie from Middle Caicos is placed closer to the Cuban populations, whereas in the canonical anal- ysis it is grouped with the Hispaniolan populations. In females the two-dimensional canonical variate plot of the samples onto the first two variates shows the Middle Caicos (6) population to be well sepa- rated on the first variate and to some extent on the second, from both the Cuban and Hispaniolan pop- ulations. Cuban and Hispaniolan samples are closer to each other than either is to the Middle Caicos sample. Therefore, all multivariate analyses of fe- male samples show the Middle Caicos sample to be well separated from the others. In the canonical analysis the Hispaniolan material is grouped with the Cuban material, whereas in both the cluster and principal component analyses they are separated. Taxonomic Conclusions Based upon our study of geographic variation in Brachyphylla nana, we have chosen to consider it a monotypic species. In five measurements for males and seven measurements for females, either the ANOVA or MANOVA was non-significant. In four of the 13 measurements for the samples ofB. nana either the ANOVA or MANOVA was non- significant for both sexes, whereas a total of eight were non-significant for at least one sex. The results of the multivariate analyses were inconsistent. There appears to be very little morphometric variation among our samples of B. nana. The range of this variation is, in many cases, encompassed by the four samples from Cuba. Other cranial features used to distinguish B. nana and B. piiniila prove to be inconsistent when large samples are examined. Therefore, we believe the best course of action to follow is to consider Brachyphylla nana as being a monotypic species. 1978 SWANEPOEL AND GENOW AYS— BRACHYPHYLLA SYSTEMATICS 49 Status of Fossil Specimens The genus BrachyphyUa is known only as a Pleis- tocene or sub-Recent fossil from the island of Ja- maica. This material was assigned to B. pumila by Koopman and Williams (1951). We have taken the opportunity to re-examine this material and to com- pare it with the two species that we have recog- nized. Standard statistics from geographic samples listed in Materials and Methods are given in Table 9. All characters of Pleistocene or sub-Recent fossil material studied with the exception of interorbital breadth showed basically the same pattern of geo- graphic variation. In all cases the fossils grouped with populations that we consider to be B. nana. The populations of BrachyphyUa from Puerto Rico (sample e), St. John (f), and Norman (g) were usu- ally grouped into a subset or subsets significantly different from those populations from Cuba (a), Middle Caicos (b), and Dominican Republic (c). Of the eight measurements, four (rostral width at ca- nines, interorbital breadth, width of articular pro- cess, and width of mandible at Mg) showed overlap between the two main areas. The Pleistocene or sub-Recent fossil material generally averaged larger than the Recent material from Cuba, Middle Cai- cos, and Dominican Republic, but falls within the range of variation displayed by the Recent material. In only two measurements (width of articular pro- cess and interorbital breadth) did the Jamaican ma- terial average less than the Recent material from Cuba, Middle Caicos, and Dominican Republic. Only in breadth of mandible at Mg did the Jamaican material show any overlap with the ranges of mea- surements obtained from specimens from Puerto Rico, St. John, and Norman. Interorbital width in BrachyphyUa displayed a great deal of geographic variation. Individual variation as indicated by coef- ficients of variation show width of articular process and breadth of mandible at Mg to be the most vari- able. The cluster, principal components, and canonical variate analyses of these samples reveal the same basic picture. We have illustrated the principal components analysis as being typical. The first two principal components extracted from the principal component analysis for three B. nana, one fossil, and three fi. cavernarum samples are shown two-dimensionally in Fig. 17. The amount of phenetic variation represented in the first three components was 90.2 for component I, 0.08 for component II, and 0.02 for component III. From the factor analysis (not tabled) it was obvious that the first component is heavily influenced by all char- acters. Both the second and third components are not notably influenced by any character. Examination of the two-dimensional plot of the first two principal components reveals two groups of samples. The cluster on the right consists of sam- ples from Puerto Rico, St. John, and Norman; the one on the left contains samples from Cuba, Middle Caicos, Dominican Republic, and Jamaica. The lat- ter group contains the smaller specimens as clearly revealed by the univariate analysis. Although the Jamaican fossil material tends to be somewhat larger than the Recent material from Cuba, Middle Caicos, and Dominican Republic, it clearly has its relationship to these populations. Decision on whether the bats in the sub-Recent population were actually somewhat larger than in the Recent population or not, must await the dis- covery of further fossil material. However, we do not believe that the differences noted in the current material warrant taxonomic recognition. Therefore, we assign the Jamaican Pleistocene or sub-Recent fossils to BrachyphyUa nana . BrachyphyUa nana Miller, 1902 1902. BrachyphyUa nana Miller, Proc. Acad. Nat. Sci. Phila- delphia, 54:509, 12 September. 1918. BrachyphyUa piiniila Miller, Proc. Biol. Soc. Washington, 31:39, 16 May, holotype from Pont de Baisc, Haiti. 1974. BrachyphyUa cavernarum nana. Varona, Acad. Cien. Cuba, p, 27. 1974. BrachyphyUa cavernarum pumila, Varona, Acad. Cien. Cuba, p. 27. 1976. BrachyphyUa nana nana, Jones and Carter, Spec. Publ. Mus., Texas Tech Univ., 10:30, 25 June. 1976. BrachyphyUa nana pumila, Jones and Carter, Spec. Pubi. Mus., Texas Tech Univ., 10:30, 25 June. Holotype. — Skull of an unsexed adult recovered from owl pellets, USNM 103,828 from El Guama, Cuba, obtained by William Palmer and J. H. Riley on 10 March 1900; original no. 108. Measurements of holotype . — Condylobasal length, 24.9; palatal length, 8.7; zygomatic breadth, 14.6; braincase breadth, 11.3; postorbital breadth, 5.9; rostral width at canines, 6.4. Distribution . — This species is known from Cuba, Isle of Pines (Varona, 1974), Grand Cayman, Mid- dle Caicos, Hispaniola, and as a Pleistocene or sub- Recent fossil from Jamaica. Comparisons . — See Specific Relationships. Remarks. — Populations described as pumila and nana were long considered distinct species and most recent authors have considered them to be 50 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 12 Fig. 17. — Two-dimensional projection of seven samples (six Recent and one Pleistocene or sub-Recent) of Brachyphylla onto the first two principal components. See text for key to samples. distinct at least at the subspecific level (see Silva- Taboada, 1976; Jones and Carter, 1976). However, based upon our analyses and studies, we cannot support this distinction. The populations do not dif- fer much in size and Cuban populations encompass most of the range of variation observed. Various dental and cranial characters, such as difference in size and morphology of M' (Miller, 1918), a broader rostrum and palate and larger molars in B. pumila (Miller, 1929), shape of interpterygoid fossa (Good- win, 1933), and depth of pit between orbit and ant- orbital foramen (Koopman and Williams, 1951), have been used to distinguish these taxa. We have examined these characters in the large series avail- able to us. These characters were found to be in- dividually variable or nonexistent. Buden (1977) found nana to have a deeper and more robust zy- gomatic arch than pumila \ however, we are unable to appreciate this character in our material. Dorsal pelage coloration does not appear to sep- arate taxa either. Individuals corresponding to color standard 3 were found in relatively high numbers on all islands — Cuba, 37%; Hispaniola, 35%; Mid- dle Caicos, 100%; Grand Cayman, 100%. The ma- jority of the specimens (63%) from Hispaniola are slightly darker than the majority of material (47%) from Cuba being a blackish gray (standard 1) as compared to dark brown (standard 5). However, in view of the only slight differences in color found throughout this genus and fairly broad overlap be- tween all island populations ofB. nana, we see no reason based upon color to consider this taxon to be polytypic. Two recent authors, Varona (1974) and Buden (1977), have recognized nana dind pumila as distinct subspecies but placed them in B. cavernarum and considered the genus to be monotypic. Buden ( 1977) claimed that “differences in size among these 1978 SWANEPOEL AND GENOV/ AYS— BRACHYPHYLLA SYSTEMATICS 51 allopatric populations is nearly matched by those found among Middle American populations of Ar- tibeus jamaicensis that are treated as subspecies by Davis (1970).” We disagree with this conclusion based upon our studies. Bnichyphylla cavernarum and B. nana differ considerably in size; there is no overlap between these two species in six of 12 cra- nial measurements taken. In our opinion, these dif- ferences more nearly resemble those found between sympatric populations of the Middle American species Artibeiis jamaicensis and A. lituratiis. We, therefore, believe that the differences observed be- tween these allopatric populations of Brachyphylla are best represented by considering them to be dis- tinct species. Brachyphylla nana is known on the island of Ja- maica only as a Pleistocene or sub-Recent fossil (Koopman and Williams, 1951). Based upon the re- construction of the fossil bat faunas by Williams (1952), B. nana occurred in about the middle of the known record for bats on the island but no time frame is possible. It was contemporary with mem- bers of the genera Ariteus, Monnoops, Phyllonyc- teris, Erophylla, Monophyllus, and Macrotiis, but had disappeared before Artibeiis appeared in the fossil record. Although it is tempting to theorize some sort of competition to account for the extinc- tion of Brachyphylla on Jamaica, the reasons must be far more complex because almost identical fau- nas occur today on Cuba and Hispaniola, but Brachyphylla has survived there (Baker and Gen- oways, 1978). Specimens examined (185). — Cuba: 12 mi E Moron, Cama- giiey Province, 3 (AS); Cueva de los Indios, Habana Province, 6 (1 AS, 5 MCZ); Cueva de! Indio, 3 mi E Tapaste, Habana Province, 12 (AMNH); Cueva de Costilla San Jose de las Lajas, Habana Province, 3 (TCWC); 4 mi S San Jose de las Lajas, Habana Province, 2 (AMNH); 9 km SW San Jose de las Lajas, Habana Province, 8 (AS); Cantabria Cave, Hormiguero, Las Villas Province, 11(1 KU, 10 UMMZ); Cantabria Cave, 14 km NE Cienfuegos, Las Villas Province, 4 (ROM); Einca de Mo- rales, 8 mi NW Trinidad, Las Villas Province, 5 (AS); Guatana- ma, Oriente Province, 3 (USNM); Los Angeles, Oriente Prov- ince, 1 (MCZ); Santiago, Oriente Province, 3 (EMNH); Santiago de Cuba, Oriente Province, 7 (3 AMNH, 4 EMNH); Cueva de la Cantera, Siboney, 14 km SE Santiago de Cuba, Oriente Prov- ince, 2 (ROM); El Guama, Pinar del Rio Province, 1 (USNM). Grand Cayman: Old Man Bay, 1 (LSU). Dominican Repub- lic: Cueva no. 2 Los Patos, Barahona Province, 47 (1 AMNH, I EMNH, 43 PSNH, I TCWC, 1 USNM); Upper Los Patos Cave, Barahona Province, 8 (4 AMNH, 4 PSNH); Los Patos, Barahona Province, I (ROM); Cueva Wunker, 19.3 km W La Romana, La Romana Province, 6 (PSNH); Sosiia, Puerta Plata Province, 7 (AS); Cueva el Limon, Samana, Samana Province, 3 (PSNH); Cueva de Sierra de Agua San Cristobal, Samana Province, 2 (ROM). Caicos Islands: Conch Bar, Middle Cai- cos, 19 (LSU). Halil Daiquini [ = Diquini|, 3 (2 BMNH, I EMNH); 1 km S, 1 km E Lebrun, Department du Sud, 4 (TTU); Port de Paix, 1 (USNM). Jamaica: Dairy Cave, Dry Harbor | = Discovery Bay], St. Ann Parish, 12 (AMNH). ACKNOWLEDGMENTS We wish to thank Rina Swanepoel for assisting this study in numerous ways. Teresa M. Bona typed the final copy of the manuscript and Margaret Popovich aided with proofreading. Some field work in the Antilles was supported by National Sci- ence Eoundation grant GB-41105 to R. J. Baker and H. H. Gen- oways. Various phases of the laboratory studies were aided by funds from the Institute of Museum Research, Texas Tech Uni- versity. We are grateful to the following curators and their institutions for allowing us to examine material housed in their collections (abbreviations used to identify specimens in text): Karl E. Koop- man, American Museum of Natural History (AMNH); Albert Schwartz, private collection (AS); John Edwards Hill, British Museum (Natural History) (BMNH); Carnegie Museum of Nat- ural History (CM); Luis de la Torre, Eield Museum of Natural History (EMNH); Robert S. Hoffmann, Museum of Natural His- tory, University of Kansas (KU); George H. Lowery, Jr., Mu- seum of Natural Science, Louisiana State University (LSU); Barbara Lawrence, Museum of Comparative Zoology, Harvard University (MCZ); Murray L. Johnson, Puget Sound Museum of Natural History, University of Puget Sound (PSNH); A. M. Husson, Rijksmuseum of Natural History, Leiden (RMNH); Randolph L. Peterson, Royal Ontario Museum (ROM); David J. Schmidly, Texas Cooperative Wildlife Collection, Texas A & M University (TCWC); Robert J. Baker, The Museum, Texas Tech University (TTU); Emmet T. Hooper, Museum of Zoolo- gy, University of Michigan (UMMZ); Don E. Wilson, National Museum of Natural History (USNM). We particularly wish to thank Karl E. Koopman and David Klingener for reviewing an earlier draft of this manuscript. Terry L. Yates assisted with some of the statistical analyses on the IBM 370 computer at the Computation Center, Texas Tech Uni- versity. The Department of Nature and Environmental Conservation of the Cape Provincial Administration and the administration of the Kaffrarian Museum, Republic of South Africa, are gratefully acknowledged for allowing the senior author to pursue studies in the United States. 52 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 12 LITERATURE CITED Allen, G. M. 1911. Mammals of the West Indies. Bull. Mus. Comp. Zool., 54:175-263. Allen, H. 1898. On the Glossophaginae. Trans. Amer. Phil. Soc., 19:237-266. Anthony, H. E. 1925. Mammals of Porto Rico, living and extinct — Chiroptera and Insectivora. New York Acad. Sci. Scientific Survey of Porto Rico and Virgin Islands, 9:1-96. Atchley, W. R. 1970. A biosystematic study of the subgenus Seljiu oi CulicoiJes (Diptera; Ceratopogonidae). Univ. Kan- sas Sci. Bull., 49:181-336. Baker, R. J. 1973. Comparative cytogenetics of the New World leaf-nosed bats (Phyllostomatidae). Periodicum Biologo- rum, 75:37^5. Baker, R. J., and G. Lopez. 1970. Karyotypic studies of the insular populations of bats on Puerto Rico. Caryologia, 23:465^72. Baker, R. J., and H. H. Genoways. 1978. Zoogeography of Antillean bats. Spec. Publ., Acad. Nat. Sci. Philadelphia, 13:53-97. Barbour, T. 1945. A naturalist in Cuba. Little, Brown and Co., Boston, X -f 317 pp. Bickham, j. W. 1976. Chromosomal banding and phylogenetic relationships of vespertilionid bats. Unpublished Ph.D. dis- sertation, Texas Tech Univ., Lubbock, v -I- 72 pp. Bond, R. M., and G. A. Seaman. 1958. Notes on a colony of Bnichyphylla cavermimm . J. Mamm., 39:150-151. Buden, D. W. 1977. First records of bats of the genus Bnich- yphylla from the Caicos Islands, with notes on geographic variation. J. Mamm., 58:221-225. Choate, J. R. 1970. Systematics and zoogeography of Middle American shrews of the genus Cryptotis. Univ. Kansas Publ., Mus. Nat. Hist., 19:195-317. Choate, J. R., AND E. C. Birney. 1968. Sub-Recent Insectiv- ora and Chiroptera from Puerto Rico, with the description of a new bat of the genus Stenoderma . J. Mamm., 49:400- 412. Davis, W. B. 1970. The large fruit bats (genus Artibeus) of Middle America, with a review of the Artibeus jamaicensis complex. J. Mamm., 51:105-122. Dobson, G. E. 1878. Catalogue of the Chiroptera in the collec- tion of the British Museum. British Museum (Nat. Hist.), London, xlii -L 567 pp. Gabriel, K. R. 1964. A procedure for testing the homogeneity of all sets of means in analysis of variance. Biometrics, 20:459^77. Gardner, A. L. 1977. Feeding habits. Pp. 293-350, in Biology of bats of the New World family Phyllostomatidae, Part II (R. J. Baker, J. K. Jones, Jr., and D. C. Carter, eds.). Spec. Ffiibl. Mus., Texas Tech Univ., 13:1-364. Genoways, H. H, 1973. Systematics and evolutionary rela- tionships of spiny pocket mice, genus Liomys. Spec. Publ. Mus., Texas Tech Univ., 5:1-368. Genoways, H. H., and J. K. Jones, Jr. 1971. Systematics of southern banner-tailed kangaroo rats of the Dipodomys phil- lipsii group. J. Mamm., 52:265-287. Gervais. P. 1855-1856. Documents zoologiques pour servir a la monographie des cheiropteres sud-americains. In Mam- miferes. In F. de Castelanu. Animaux nouveaux ou rares recueillis pendant I'expedition dans les parties centrales de Amerique du Sud. Paris Part 7 (Zoologie). Vol. for 1855: 25-88. Goodwin, G. G. 1933. The external characters of R/-«(7;yp/iv//a piintila Miller. J. Mamm., 14:154-155. Gray, J. E. 1834. Characters of a new genus of bats {Brachy- phylla), obtained by the Society from the collection of the late Rev. Lansdown Guilding. Proc. Zool. Soc. London, pp. 122-123. . 1838. A revision of bats (Vespertilionidae), and the de- scription of some new genera and species. Mag. Zool. Bot., 2(121:484-505. . 1843. List of the specimens of Mammalia in the collec- tion of the British Museum. Trustees of British Museum, London, xxvii -1-216 pp. . 1866. Revision of the genera of Phyllostomidae, or leaf- nosed bats. Proc. Zool. Soc. London, pp. 111-118. Gundlach, j. 1877. Contribucion a la mamalogia Cubana. Im- pienta G. Montiel, La Habana, 53 pp. Jones, J. K., Jr., and D. C. Carter. 1976. Annotated check- list, with keys to subfamilies and genera. Pp. 7-38, in Bi- ology of bats of the New World family Phyllostomatidae, Part I (R. J. Baker, J. K. Jones, Jr., and D. C. Carter, eds.). Spec. Publ. Mus., Texas Tech Univ., 10:1-218. Koopman, K. F. 1968. Taxonomic and distributional notes on Lesser Antillean bats. Amer. Mus. Novit., 2333:1-13. . 1975. Bats of the Virgin Islands in relation to those of the Greater and Lesser Antilles. Amer. Mus. Novit., 2581:1-7. Koopman, K. F., and E. E. Williams. 1951. Fossil Chirop- tera collected by H. F. Anthony in Jamaica, 1919-1920. Amer. Mus. Novit., 1519:1-29. McDaniel, V. R. 1976. Brain anatomy. Pp. 147-200, in Biol- ogy of bats of the New World family Phyllostomatidae, Part 1 (R. J. Baker, J. K. Jones, Jr., and D. C. Carter, eds.). Spec. Publ. Mus., Texas Tech Univ., 10:1-218. McManus, J. J., and D. W. Nellis. 1972. Temperature reg- ulation in three species of tropical bats. J. Mamm., 53:226- 227, Miller, G. S., Jr. 1898. Descriptions of five new phyllostome bats. Proc. Acad. Nat. Sci. Philadelphia, 50:326-337. . 1902lj. Twenty new American bats. Proc. Acad. Nat. Sci. Philadelphia, 54:389^12. . \902h. The external characters of Brachyphylla nana Miller. Proc. Biol. Soc. Washington, 15:249. . 1907. The families and genera of bats. Bull. U. S. Nat. Mus., 57:xvii + 1-282. . 1913. Five new mammals from tropical America. Proc. Biol. Soc. Washington, 26:31-33. . 1918. Three new bats from Haiti and Santo Domingo. Proc. Biol. Soc. Washington, 31:39-40. . 1929. A second collection of mammals from caves near St. Michel, Haiti. Smithsonian Misc. Coll., 81(9): 1-30. Nagorsen, D. W., and R. H. Peterson. 1975. Karyotypes of six species of bats (Chiroptera) from the Dominican Repub- lic. Life Sci. Occas. Papers, Royal Ontario Mus., 28:1-8. Nellis, D. W. 1971. Additions to the natural history of Rrac/i- yphylla (Chiroptera). Caribbean J. Sci., 11:91. 1978 SWANEPOEL AND GENOW AYS— BRACHYPHYLLA SYSTEMATICS 53 Nellis, D. W., AND C. P. Ehle. 1977. Observations on the be- havior of Brachyphylla cavernamm (Chiroptera) in Virgin Islands. Mammalia, 41:403-^09. Power, D. M. 1970. Geographic variation of red-winged black- birds in central North America. Univ. Kansas Publ., Mus. Nat. Hist., 19:1-83. Sanborn, C. C. 1941. Descriptions and records of Neotropical bats. Field Mus. Nat. Hist., Zool. Ser., 27:371-387. ScHMiDLY, D. J., AND F. S. Hendricks. 1976. Systematics of the southern races of Ord’s kangaroo rat, Dipodomys ordii. Bull. Southern California Acad. Sci., 75:225-231 . ScHNELL, G. D. 1970. A phenetic study of the suborder Lari (Aves). 1. Methods and results of principal component anal- yses. Syst. Zool., 19:35-57. Service, J. 1972. A user's guide to the Statistical Analysis System. Student Supplies Stores, North Carolina State Univ., Raleigh, 260 pp. Silva-Taboada, G. 1976. Historia y actualizacion taxonomica de algunas especies Antillanas de murcielagos de los gen- eros Pteronotus, Brachyphylla. Lasiurus, y Anirozous. (Mammalia: Chiroptera). Poeyana, 153:1-24. Silva-Taboada, G., AND R. H. Pine. 1969. Morphological and behavioral evidence for the relationship between the bat genus Brachyphylla and the phyllonycterinae. Biotropica, 1:10-19. Slaughter, B. H. 1970. Evolutionary trends of chiropteran dentitions. Pp. 51-83, in About bats (B. H. Slaughter and D. W. Walton, eds.). Southern Methodist Univ. Press, Dal- las, Texas, vii -I- 339 pp. Smith, J. D. 1972. Systematics of the chiropteran family Mor- moopidae. Misc. Publ. Mus. Nat. Hist., Univ. Kansas, 56:1-132. Sneaih, P. H. a., and R. R. Sokal. 1973. Numerical tax- onomy: the principles and practices of numerical classifi- cation. W. H. Freeman and Co., San Francisco, xv -L 573 pp. Sokal, R. R., and P. H. A. Sneath. 1963. Principles of nu- merical taxonomy. W. H. Freeman and Co., San Francisco, xvi + 359 pp. Sumner, F. B. 1927. Linear and colorimetric measurements of small mammals. J. Mamm., 8:177-206. Ubelaker, j. E., R. D. Specian, and D. W. Duszynski. 1977. Endoparasites. Pp. 7-56, in Biology of bats of the New World family Phyllostomatidae, Part II (R. J. Baker, J. K. Jones, Jr., and D. C. Carter, eds.). Spec. Publ. Mus., Texas Tech Univ., 13:1-364. Varona, L. S. 1974. Catalogo de los mamiferos vivientes y extinguidos de las Antillas. Acad. Sci. Cuba, 139 pp. Webb, J. P., Jr., and R. B. Loomis. 1977. Ectoparasites. Pp. 57-1 19, in Biology of bats of the New World family Phyl- lostomatidae, Part II (R. J. Baker, J. K. Jones, Jr., and D. C. Carter, eds.). Spec. Publ. Mus., Texas Tech Univ., 13:1- 364. Williams, E. E. 1952. Additional notes on fossil and subfossil bats from Jamaica. J. Mamm., 33:171-179. Yates, T. L., H. H. Genoways, and J. K. Jones, Jr. 1978. Rabbits of Nicaragua. Mammalia, in press. Yates, T. L., AND D. J. ScHMiDLY. 1977. Systematics of 5’r«/- opus aquaticus (Linnaeus) in Texas and adjacent states. Occas. Papers Mus., Texas Tech Univ., 45:1-36. I I I V Copies of the following Bulletins of Carnegie Museum of Natural History may be obtained at the prices listed from the Publications Secretary, Carnegie Museum of Natural History, 4400 Forbes Avenue, Pitts- burgh, Pennsylvania 15213. 1. Krishtalka, L. 1976. Early Tertiary Adapisoricidae and Erinaceidae (Mammalia, Insectivora) of North America. 40 pp., 13 figs $2.50 2. Guilday, J. E., P. W. Parmalee, and H. W. Hamilton. 1977. The Clark’s Cave bone deposit and the late Pleistocene paleoecology of the central Appalachian Mountains of Virginia. 88 pp., 21 figs. $12.00 3. Wetzel, R. M. 1977. The Chacoan peccary, Curugouz/5 vraguen (Rusconi). 36 pp., 10 figs. .. $6.00 4. Coombs, M. C. 1978. Reevaluation of early Miocene North American Moropus (Perissodactyla, Chalicotheriidae, Schizotheriinae). 62 pp., 28 figs $5.00 5. Clench, M. H., and R. C. Leberman. 1978. Weights of 151 species of Pennsylvania birds analyzed by month, age, and sex. 87 pp $5.00 6. Schlitter, D. A. (ed.). 1978. Ecology and taxonomy of African small mammals. 214 pp., 48 figs. $15.00 7. Raikow,R. J. 1978. Appendicular myology and relationships of the New World nine-primaried oscines (Aves:Passeriformes). 43 pp., 10 figs $3.50 8. Berman, D. S, and J. S. McIntosh. 1978. Skull and relationships of the Upper Jurassic sauropod Apatosaurus (Reptilia, Saurischia). 35 pp., 11 figs $3.00 9. Setoguchi, T. 1978. Paleontology and geology of the Badwater Creek area, central Wyoming. Part 16. The Cedar Ridge local fauna (Late Oligocene). 61 pp., 30 figs $4.50 10. Williams, D. E. 1978. Systematics and ecogeographic variation of the Apache pocket mouse (Roden- tia: Heteromyidae). 57 pp., 23 figs $4.00 11. Guilday, J. E., H. W. Hamilton, E. Anderson, and P. W. Parmalee. 1978. The Baker Bluff Cave deposit, Tennessee, and the Late Pleistocene faunal gradient. 67 pp., 16 figs $5.00 B U L t B T I N 0/ CARNEGIE MUSEUM OF NATURAL HISTORY QH I awb ■Str MODELS AND METHODOLOGIES IN EVOLUTIONARY THEORY Edited by JEFFREY H. SCHWARTZ and HAROLD B. ROLLINS NUMBER 13 PITTSBURGH, 1979 BULLETIN of CARNEGIE MUSEUM OF NATURAL HISTORY MODELS AND METHODOLOGIES IN EVOLUTIONARY THEORY Edited hy JEFFREY H. SCHWARTZ Research Associate, Section of Vertebrate Fossils, and Department of Anthropology, University of Pittsburgh, Pennsylvania 15260 HAROLD B. ROLLINS Department of Earth and Planetary Sciences, University of Pittsburgh, Pittsburgh, Pennsylvania 15260 NUMBER 13 PITTSBURGH, 1979 BULLETIN OF CARNEGIE MUSEUM OF NATURAL HISTORY Number 13, pages 1-105, 37 figures, 7 tables Issued 12 April 1979 Price: $6.00 a copy Craig C. Black, Director Editorial Staff: Hugh H. Genoways, Editor, Duane A. Schlitter, Associate Editor, Stephen L. Williams, Associate Editor-, Barbara Farkas, Technical Assistant. @ 1979 by the Trustees of Carnegie Institute, all rights reserved. CARNEGIE MUSEUM OF NATURAL HISTORY, 4400 FORBES AVENUE PITTSBURGH, PENNSYLVANIA 15213 CONTENTS Preface — Jeffrey H. Schwartz and Harold B. Rollins 5 Alternative approaches to evolutionary theory — Niles Eldredge 7 The synthetic explanation of macroevolutionary change — a reductionistic approach — Walter J. Bock . . 20 Canalization model of chromosomal evolution — John W. Bickham and Robert J. Baker 70 Biases in the fossil record of species and genera — David M. Raup 85 Tetrapod monophyly: a phylogenetic analysis — Eugene S. Gaffney 92 *}< I ' s © '■ i‘ _ ^ *ii ii v " ". . ■■■ ■ ' :i A ( J PREFACE During the academic years 1976-1977, the Uni- versity of Pittsburgh and the Carnegie Museum of Natural History jointly sponsored two colloquia se- ries dealing with aspects of evolution, “Models and Methodologies in Evolutionary Theory.” The first year’s participants were: Niles Eldredge, The American Museum of Nat- ural History; Steven Jay Gould, The Museum of Compara- tive Zoology, Harvard University; David M. Raup, University of Rochester; Thomas J. M. Schopf, University of Chicago; Eugene S. Gaffney, The American Museum of Natural History. Those of the second series were: Albert A. Bartlett, University of Colorado; Robert J. Baker, Texas Tech University; Walter J. Bock, Columbia University; Steven M. Stanley, Johns-Hopkins University; Leonard Radinsky, University of Chicago. Each participant presented two formal talks and five also submitted original manuscripts for publi- cation. The latter include three papers on ap- proaches to evolutionary theory (Eldredge, Bock, and Baker); one discussing the evolutionary conse- quences of preservational biases of the fossil record (Raup); and another dealing with philosophy and methodology in phylogeny reconstruction, in this case the origin of Tetrapoda (Gaffney). We are indebted to Dr. Craig C. Black, Director, Carnegie Museum of Natural History, and the Pro- vost’s Office, University of Pittsburgh, for gener- ously funding these colloquia and providing the nec- essary facilities and equipment. Among the many of both institutions who helped to make these col- loquia successful, we wish to especially thank Dr. R. Raikow and Mr. J. Harper (University of Pitts- burgh) and Drs. M. Dawson, H. Genoways and L. Krishtalka (Carnegie Museum of Natural History). In addition, we wish to acknowledge the editorial assistance of Ms. G. LoAlbo. Jeffrey H. Schwartz Harofd B. Roffins 5 ALTERNATIVE APPROACHES TO EVOLUTIONARY THEORY Niles Eldredge Department of Invertebrates, The American Museum of Natural History, Central Park West at 79th Street, New York, New York 10024 INTRODUCTION To most North Americans who have ever thought about evolution at all, there is only one "evolution- ary theory” — ”neo-darwinism.” We look askance upon other sets of explanations of evolution, such as the various versions of saltationism, much in the way that Judaeo-Christian tradition views devotion to any but the One God. My purpose here is not to espouse pagan alternatives to orthodox darwinism, but rather to develop the theme that neo-darwinian theory is not at all as monolithic as we might sup- pose. Rather, all evolutionary thinking, darwinian and non-darwinian, pre- and post- 1859, has been beset by a curious duality, which has effectively hindered a truly integrated theory of any guise from emerging. Evolution can be viewed in two basic ways. If evolution is “descent with modification” (Darwin, 1859:171) or “change in the genotype of a popula- tion” (Dobzhansky, 1951:21), then we must ask how that change occurs. What are the mechanisms? Whether we are discussing genes, their phenotypic expression, physiology, or behavior, variation and factors of heritability lie at the core of the problem. Thus, any standard text properly devotes much of its space to genetics, and mutation, recombination, natural selection and related concepts weigh heavi- ly in our theories of evolutionary mechanics. Darwin (1859) also gave us our other basic way of looking at evolution — the “origin of species.” We see around us today perhaps as many as two million discrete species. If we assume, as seems reasonable, that species are real entities in nature, their origins must be explained, and because species are aggregates of populations, which are themselves aggregates of individuals, the mechan- ics of speciation do not seem immediately reducible to the principles of genetics (see for example, Avise and Ayala, 1975). Thus, standard texts also devote a good deal of space to speciation. No one would argue that we can have an evolu- tionary theory without either one of these two com- ponents. A theory is simply incomplete if it lacks either a set of statements concerning (1) changes in genes and their expression or (2) the origins of new species. The problem with neo-darwinian theory to- day is not that it lacks one or the other of these components, but that they have as yet to be suc- cessfully fully integrated. This problem comes into focus more sharply if we first generalize these two aspects of the evolu- tionary process. The first aspect can be labelled the “transformational approach”; under this approach, the central question in evolution is: how do genes and their products become modified in the evolu- tionary process? How are karyotypes, behavior, pelage colors, hormones, and so on, modified by the evolutionary process? In darwinian theory, natural selection is the key concept of the transformational approach, and evo- lutionary adaptation seems to be the prime focus of most of the research conducted from this point of view. Trends, adaptive radiations, rates of morpho- logic change, and even the origin of taxa of higher categorical rank (which superficially sounds as though it belongs under the “taxic approach”) are examples of the kinds of evolutionary topics typi- cally addressed by paleontologists and others. These problems are almost entirely defined in terms of the central issue of the transformational ap- proach: how are these particular biochemical, cy- tological, anatomical, physiological, and behavioral traits changed in evolution? An adaptive radiation, for example, is viewed as a problem of divergent anatomical specializations among a series of related organisms, rather than as a spectrum of discrete species occupying a diverse array of ecological niches. The emphasis in this approach on morpho- logic (sensu lata) change inevitably results in the entire disregarding of taxa in the analysis. And thus we have, particularly in the literatures of genetics and paleontology, a large corpus of statements per- taining to the nature of the evolutionary process which almost completely ignores the existence of taxa and the problem of their origins. Concern for species and other taxa under the transformational approach is limited to use of Lin- 7 8 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 13 naean names to refer to complexes of anatomical (or genetic, physiologic, and others) traits. Indeed, most adherents of this approach implicitly or ex- plicitly deny the "reality” of species as discrete entities in nature. Eor if species are not considered as real, discrete entities, the problem of their origin may of course be safely ignored, and we need not be concerned with integrating speciation theory with the concerns of genetic and morphologic trans- formation. This attitude is especially prevalent in paleontology and takes several guises. The more obvious, transparent version of the "non-reality” of species is well developed in Shaw (1969) whose main point concerning species is that they are nei- ther operationally nor objectively recognizable, hence can be considered in this sense not to exist. Having thus done away with species as something to be concerned with, Shaw is then at liberty to develop a strategy for analyzing phylogenetic lin- eages (for the express purpose of correlating rock sequences) which pays no attention to component taxa, but instead focuses on correlation of particu- lar character states. For a far more subtle rejection of species as ob- jects of central interest in evolutionary theory, con- sider the work of Simpson (for example, 1944, 1951, 1953). Simpson (for example, 1951) does not reject outright the reality of species'. But he does deny their discreteness, as in his rejection of the "genet- ical definition” of species in favor of a concept of species in evolutionary time: "An evolutionary species is a lineage (an ancestral-descendant se- quence of populations) evolving separately from others and with its own unitary evolutionary role and tendencies.” (Simpson, 1961:153, modified slightly from Simpson, 1951:289). Thus, though any lineage may, obviously, have a definite beginning ' The argument about the ‘Teality” of taxa of different categorical rank is an old and interesting one. As may be anticipated, all possible views have been taken from time to time. Thus Simpson ( 1953:350) has rejected the idea expressed by Willis ( 1940) and others who claim that kingdoms appear first, followed by phyla, classes, and so forth; in terms of actual evolutionary mechanics, to nearly everyone’s evident satis- faction. the evolutionary process goes on at the within- and among-population level. We inductively assume this to have been the case since life began. In this sense, genera and taxa of even higher categorical rank simply do not exist in the same sense as do species and populations. Ghiselin (1974) has recently stated this most cogently, insisting that species are individuals, not classes, and have a specifiable economic role in nature not possessed by taxa of higher categorical rank (which he calls “classes”). Inasmuch as it is populations, rather than species, which are economically integrated into ecosystems, it is perhaps preferable to bestow the quality of “reality” on them rather than species; others would prefer individuals. In any case, a counter- argument has recently appeared among the phylogenetic systematists (for example, Bonde, 1974:567). In this view, all monophyletic groups, of whatever rank, are equally real, presumably in the sense that they are all defined and recognized in the same fashion. Though there is an appealing quality to this argument, considerations of ecological integration cause me to favor the view that populations, and perhaps species, are endowed with an aspect of “reality” not shared by taxa of higher cate- gorical rank. and a definite end (that is, "extinct without issue” as opposed to "extinction by transformation” — see Simpson, 1974: 14 for an example of the use of these concepts), the door is purposely left open for any- thing from negligible to vast amounts of morpho- logical or genetical transformation within the indef- inite time segment of this lineage. This is a rather elusive concept of the species, though it may be the best we can have. It is apparent that Simpson’s sin- cere attempt to formulate a more "evolutionary” concept of "species” stemmed from the historical consensus — by no means restricted to paleontolo- gy— which views evolution solely as change through time of genes and their products. Such change is construed as an almost inevitable conse- quence of the mere passage of time. And, of course, this is a reflection of the transformational approach. Simpson’s definition, however different in intent from Shaw’s ( 1969) explicit rejection of species, has the identical effect of removing the necessity of considering the origin of species when we confront the tempos and modes of evolution. In one guise or another, the transformational approach to evolution ignores species and hence is under no obligation to integrate speciation theory into the paradigm. If species are not real, or are at best arbitrarily delin- eated segments of lineages, speciation can safely be viewed as a useful adjunct to evolutionary theory, which merely tells us how new lineages get going in the first place. But the nearly total neglect of speciation by paleontologists and others wedded to the transformational approach can only imply that the truly interesting and meaningful evolutionary phenomena take place subsequent to the origin of a lineage (sensu Simpson, as cited above). The second, or "taxic” approach, looks at evo- lution the other way around. Under this view, the central issue is: how do new taxa (usually species) originate? A fundamental assumption under this ap- proach, of course, is that, at some level, aggregates of individuals (populations, perhaps species) are actual and discrete entities in nature, with their own roles in the economy of nature. Speciation is the basic evolutionary model under the taxic approach. There are many particular models of speciation in the neo-darwinian literature (see Bush, 1975, for a useful review, as well as a stimulating advocacy of sympatric speciation). If fault can be found with the taxic approach, it is simply that its adherents have tended to limit them- selves to the topic of speciation. But those who work primarily under the taxic approach have, from 1979 ELDREDGE— ALTERNATIVE EVOLUTIONARY THEORY 9 time to time, addressed the same variety of topics studied under the transformational approach, and have shown, with varying degrees of success, how the same problems may be addressed through the taxic approach. The crucial point here is that, under the taxic approach, the key concepts of adaptation and se- lection, as well as the remainder of the principles of population genetics, have not been lost sight of in speciation theory. Rather, no matter how unsa- tisfyingly and incompletely, they have been inte- grated with speciation theory. Thus, all the mech- anisms invoked under the transformational approach (except some of the fanciful, nonexistent ones, such as “orthogenesis”) are very much a part of both the problems addressed and the mechanisms in- voked under the taxic approach. The reverse is sim- ply not true — for the transformational approach to operate, at least to date, a prime requisite has been either the outright denial of the existence (“reali- ty”) of species or denial of their discreteness, to the extent that the problem of their origin can safely be viewed as a special case, or at most as a small part of the general problem. Because both specia- tion theory and genetic mechanisms are required in a complete theory, we are justified in rejecting a priori any segment of evolutionary theory as at best incomplete, and at worst irrelevant, if it does not contain elements of both. The taxic approach, thus far, at least, offers the only promise of complete integration — under the taxic approach, we are free to investigate mode and degree of genetic and mor- phologic change in the context of speciation. Under the transformational approach we must ignore the origin of species and view them as a posteriori PALEONTOLOGICAL APPROACH By the very nature of its materials, paleontology can have nothing direct to say about evolutionary mechanisms. Using fossils, we cannot study muta- tion, recombination, and selection with any degree of practicality, and certainly these and related con- cepts would never have come from paleontology. Likewise our modern concepts of speciation, while perhaps less difficult to apply to paleontogical data, do not emerge as self-evident precepts when one examines data on the fossil record. Both these as- pects of evolutionary theory are quite difficult to relate hypothetico-deductively to the fossil record. Thus, the relationship between paleontology and evolutionary theory has been mostly a matter of products of the mind of the systematist who uses the results of evolution (genetic and morphologic change) to recognize, name, and pigeon-hole taxa. It must still be shown, however, that the same ques- tions raised under the transformational approach can effectively be attacked under the taxic ap- proach. It is possible to examine the various subdisci- plines of evolutionary biology (genetics, ecology, systematics, and paleontology — this list is perhaps not all-inclusive; certainly there is a large amount of overlap among these four semiarbitrarily delin- eated fields) and assess the relative degrees to which each of the two approaches has contributed to past and present theoretical work. Certainly ele- ments of both appear in each of these disciplines. For true integration we must take genetics and see how it looks in the context of speciation (see Ayala, 1975, for a valuable review), look at ecological con- trols of speciation and of the resultant patterns of species diversity, and then turn to systematics and paleontology to look at patterns of taxic evolution in time as well as in space. Only then shall we be effectively translating the genuine problems posed under the transformational approach into the real- istic terms of the taxic approach. The remainder of this paper treats just one of the disciplines — pa- leontology— and restates in taxic terms a single ex- ample of an evolutionary problem traditionally for- mulated and addressed in terms of the transformational approach. But the basic view ex- pressed can be applied to any branch of evolution- ary biology and to any problem usually expressed in terms of the transformational approach. TO EVOLUTIONARY THEORY application of neontological concepts to paleonto- logical data and it is indeed difficult to identify any coherent segment of evolutionary theory, no matter how broadly construed, as having originated in pa- leontology. In an evolutionary context, paleontologists have always had the option of looking at the fossil rec- ord, in either or both of two ways — ( 1) distributions in space and time of discrete taxa, which differ among themselves to a greater or lesser extent, and (2) distributions in space and time of different states of morphological characters assumed to be evolv- ing. The post- 1859 history of paleontology reveals a nearly total dedication to the transformational ap- 10 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 13 proach as opposed to the taxic approach to evolu- tionary problems. Waagen’s (1869:185 ff) study of the evolution of "'Ammonites suhradiatus in which he coined the expression “mutation,” is an early exception to this generalization. As pointed out by Simpson (1942:48; 1953:81) Waagen’s “mu- tations” were actually discrete taxa, an early ex- ample of the taxic approach to evolutionary studies in paleontology. But the preponderance of subse- quent studies in evolutionary theory fall clearly into the transformational camp — to cite but a few out- standing examples, consider Rowe (1899) on Cre- taceous echinoids, Carruthers (1910) on Carbonif- erous corals, Trueman (1922) on Jurassic oysters, and Brinkmann (1929) on Jurassic ammonites. This work culminated in the general syntheses of Simp- son (1944, 1953), who integrated paleontological data and concepts with many of the evolutionary concepts of the rest of the “modern synthesis” — with the notable exception of speciation theory. We might well wonder why paleontology re- mained on the whole oblivious to the taxic view of the evolutionary process when other areas of biol- ogy— particularly ecology and systematics — early on become aware of the problem of diversity posed by the existence of discrete taxa. The works of Wagner (1869) and Romanes (for example, 1886) (cited by Mayr, 1963) were early forerunners of the works of Mayr (for example, 1942, 1963), which remain the most complete and thorough statements of the taxic view of the evolutionary process in the English language^. But only in relatively recent years has the basic thrust of paleontological evo- lutionary thought begun to switch over to the taxic approach. This change is symbolized by the titles and contents of the two most recent book-length treatments of evolution by paleontologists — Val- entine’s (1973) “Evolutionary Paleoecology of the Marine Biosphere” and Boucot’s (1975) “Evolution and Extinction Rate Controls.” To both of these authors, evolution consists essentially of the origin, maintenance, and degradation of diversity — a pure- ly taxic approach to evolution. There is indeed little in common between these two books on the one hand, and those of Simpson (especially 1944 and 1953) on the other. Many of the issues Simpson raised and discussed so ably under the transfor- mational approach are given short shrift by Boucot (1975) and Valentine (1973) — including origin of taxa of higher categorical rank, trends, adaptive ra- diations, morphologic evolutionary rates (usually considered “macroevolutionary” topics — Simp- son’s “major features”) as well as microevolution- ary problems centering around species. Inasmuch as evolution seems to be more than merely the or- igin, maintenance, and degradation of diversity, we should continue to look for a fuller integration of the issues that Simpson and other paleontologists have addressed under the transformational ap- proach, with the basically taxic/ecological approach advocated by Valentine, Boucot, and a host of oth- er contributors to the recent paleontological litera- ture. AN EXAMPLE: MORPHOLOGIC RATES OF EVOLUTION AND THE TAXIC APPROACH Simpson (1944:3) defined “rate of evolution . . . as amount of morphological change relative to a 2 Lesch ( 1975) has reviewed the work of Romanes and other evolutionary biologists of the latter half of the nineteenth century who stressed the importance of isolation in the speciation process. That a distinction between the transformational and taxic approaches was evident to Romanes is well illustrated by Lesch, who quotes Romanes (1886:347): ' it (i.e,, natural selection) is not, strictly speaking, a theory of the ongm of species: it is a theory of the origin — or rather of the cumulative develop- ment— of adaptations, whether these be morphological, physiological, or psycholog- ical, and whether they occur in species only, or likewise in genera, families, orders and classes.” Lesch (1975:487) then elucidates: ”In other words, while the theory of natural selection has succeeded in accounting for the great central fact of adaptation m nature, it had failed to give a complete explanation of the special phenomena associated with species formation, in particular the splitting up of a single species into two or more distinct species.” This issue is an old one in evolutionary biology! There are, of course, exceptions to the generalization that paleontologists have remained oblivious to the taxic approach. For example, J. M. Clarke (1913:17 ff.). citing the works of Wallace, Wagner, and Jordan, was clearly thinking in "taxic” terms in his analysis of the role of geographic isolation in the development of the Devonian faunas of the southern hemisphere. But such examples are few and far between, and constitute only a muted counterpoint to the main theme of "transfor- mationism” in paleontological research over the past 100 years. standard,” and paid relatively less attention to taxonomic rates. In 1953, Simpson made more ex- plicit the distinction between morphologic and taxo- nomic evolutionary rates and expanded his discus- sion of the latter. Boucot (1975) in contrast, used the expression “evolutionary rate” solely to mean taxonomic rate (specifically, rate of origin of new brachiopod genera) and this corresponds only to the taxonomic frequency rate subdivision of Simpson’s (1953:10) category of taxonomic rates. In the vo- luminous literature on evolutionary rates, it is usu- ally unclear precisely what kind of evolutionary rate is intended — “genomic” (see Schopf et al., 1975, for a discussion of this type of evolutionary rate and its relationship to paleontology), morphologic, or taxonomic. Although Simpson (1944, 1953) was 1979 ELDREDGE— ALTERNATIVE EVOLUTIONARY THEORY 11 correct in separating these various kinds of rates and asserting the unique interest attached to each, the tendency not to distinguish among them implies that the fundamental problems underlying such in- vestigations are interrelated, if not identical. Degree, hence rate, of genetic change has long been known not to be invariably closely correlated with morphologic change. The existence of sibling species (for example. Drosophila persimilis and D. pseudohscura — see Dobzhansky, 1951:267) when compared with the rampant variation within a single species (for example. Homo sapiens) suffices to dramatize this long-known point, as does the recent demonstration (King and Wilson, 1975) of the close genetic similarity between Pan troglodytes and Homo sapiens, two species which appear to us to be so different. Similarly this point is directly re- lated to degree, hence rate, of morphological change versus speciation rates; a highly speciose group (such as most drosophilid faunas, perhaps excepting Hawaii — see Hardy, 1970:451) may re- main morphologically rather uniform, or produce a broad spectrum of morphologically distinctive species (as in African cichlid fish of the genus Hap- lochromis — see Eryer and lies, 1969). Finally, de- gree, hence rate, of genetic change is not clearly related to speciation rate (Avise and Ayala, 1975). These various sorts of rates are “decoupled” (sen- sn Stanley, 1975) to some extent, and it is legitimate to investigate each separately. When we turn to a specific problem involving rates and examine the hypotheses proposed to ex- plain it, the distinction between the transformation- al approach and the taxic approach becomes rele- vant. For example, bradytelic lines — defined by Simpson (1953:113) as “low rates” or “arrested evolution” — imply, above all, extremely low rates of morphological evolution over a considerable span of time — usually at least 100 million years. Most studies of bradytelic lines have focused on the question: what are the factors governing this ob- served slow rate of morphological transformation? The difficulty in applying a purely hypothetico-de- ductive approach to evolutionary paleontology is nicely exemplified in the set of explanations devised by paleontologists (and others) to explain bradyte- ly. Simpson (1953:319 ff., see also p. 303 ff) sum- marizes many of them ably, himself believing that bradytelic lines are qualitatively (as well as quan- titatively) different from moderate and fast rates, and concluding (1953:334) “bradytelic lines are merely the residuum of a process that regularly re- duces the percentage of unchanged groups but that stops short of reduction to zero . . . .” There seem to be three general “transformation” sets of explanations of bradytely. First, lack of suf- ficient genetic variability (for a variety of reasons) has been cited (see Selander et al., 1970, and ref- erences cited therein) as a major factor. If genetic varibility be the raw stuff of evolution, lack of such variability might be construed as an impediment to change. But Selander et al. (1970) have shown that Li mains polyphemns, a surviving member of a clas- sically bradytelic lineage discussed more thorough- ly below, is about as polymorphic at a sample of its loci as most other organisms. A second set of ex- planations involves selection; for example, Simp- son (1953:331) asserts that bradytelic lines are not subject to intense (directional) selection pressures, but rather are always under intense stabilizing (cen- tripetal) selection pressure. The final general kind of explanation of bradytely involves adaptation per se — organisms have persisted essentially un- changed because their adaptations were successful, their habitats persisted, their niches remained re- cognizably the same, hence so did they. This ar- gument is appealing, but amounts to a tautological restatement of the problem- — it essentially says that organisms have persisted because they have per- sisted. There seems to be few avenues to test most of these hypotheses directly. Fig. 1, taken from Westoll (1949: Fig. 11), is a typical bradytelic curve for rate of character loss within Dipnoi (following an early tachytelic period), compared with the actual number of dipnoan genera given by Romer (1966). The curves are similar:^ where there is a high rate of morphological change, there tend to be more taxa recognized by system- atists. This poses an important conundrum: do lin- eages undergoing a relatively high rate of morpho- logical transformation appear to be more taxically diverse simply because their greater morphological variety allows us to recognize more taxa within them? Or is it the other way around — are lineages, which exhibit relatively greater amounts of mor- phologic change, more taxically diverse because they have a higher rate of speciation ? If the former is true there is no compelling reason to consider ^ Raup and Gould (1974) have illustrated the pitfalls of concocting evolutionary explanations for spindle diagrams of diversity and the lesson is directly applicable to the interpretation of these taxonomic frequency and survivorship curves. Eldredge (1978: Fig. 10) has recently presented a curve of trilobite family diversity which is geometrically very similar to that for Dipnoi (Fig. lb) and Limulicina (Fig. 2) presented here. Trilobites, as a group, were by no stretch of the imagination bradytelic. Thus, low diversity, on the species or at most the generic level, must be an added element to the very definition of bradytely — see Eldredge, 1975. for a further discussion. 12 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 13 Fig. 1. — Comparison of Westoll’s “graph showing rate of loss of characters [for entire body] of the ancestral type during evolution of Dipnoi” (dashed line with scale on right; redrawn from Westoll, 1949: Fig. 1 1) with a tabulation of dipnoan generic diversity through time tabulated from Romer (1966; solid line with numbers of genera on left). Both diversity and rate of morphological evolution were initially high (“tachytelic”). The bradytelic phase, comprising some 87% of the total history of the group (345 of 395 million years) is accompanied by a greatly reduced but apparently stabilized generic diversity. Compare with Fig. 2. arguments other than the sort listed above to ex- plain differential rates of morphological change. In that case, it would appear that the problem of brad- ytely has been exhausted and we are left with a number of narrative, essentially inductive explan- atory generalizations (not all mutually exclusive) with which we must be content. However, if we examine the alternative hypothesis, that morpho- logical evolutionary rates reflect, however indirect- ly, speciation rates, the problem of bradytely takes on an entirely different guise and perhaps admits to a more compelling narrative explanation, and pos- sibly to a more rigorously hypothetico-deductively based investigation. Fisher's ( 1975/7) recent detailed study of the evo- lution and functional morphology of xiphosurans provides a comprehensive review of the diversity and morphological evolution of the infraorder Li- mulicina from the Middle Devonian through the Recent. Fig. 2, based on data from Fisher (1975/7) graphs the approximate number of valid species ex- tant during some part of each third of the various geologic periods. Fisher (1975u, 1975/7) has dem- onstrated that the classic view of virtually total sta- sis in limuline evolution has been overstated. For example, Fisher (1975/7) convincingly shows that the Carboniferous Euproopacea attained full fusion of the opisthosomal segments independently from the true Limulacea. This interpretation is contrary to conventional interpretations (or assumptions) most recently advocated by Eldredge (1974u) who cited opisthosomal fusion as a synapomorphy link- ing the two groups. Fisher (1975«) also established actual differences in shape (that is, not just size) of various prosomal and opisthosomal features which bespeak rather more profound differences in behav- ior among genera of Limulacea. In spite of the enu- meration of a large number of morphological differ- ences among the 23 genera and approximately 45 species of living and fossil Limulicina that Fisher has documented, this infraorder remains a typical example of a bradytelic lineage. Fisher (1975/7) ac- cepts four living species of Limulicina as valid — Limulus polyphemus , Carcinoscorpio rotiindatus, Tachypleiis gigas, and Tachypleus tridentatus. In- spection of Fig. 2 reveals that this low diversity has been characteristic of the infraorder from the Low- er Permian through the Recent. The fact that Re- cent diversity is so similar to the known diversity during this interval suggests that the fossil record is a fairly accurate reflection of the actual diversity, though of course we must expect it to be an under- estimation. Thus, this bradytelic lineage is charac- terized by relatively little morphologic diversity, low rate of morphologic change, and low taxic (species) diversity for all but the earliest phase of the evolution of the Limulicina. Bradytelic lineages thus exist and their existence remains to be explained. What follows, first, is an attempt to supplant the inductive narratives, which 1979 ELDREDGE— ALTERNATIVE EVOLUTIONARY THEORY 13 Fig. 2. — Species diversity of Limulicina through time, based on analysis and compilation by Fisher (I975f>). approach the problem from a purely transforma- tional standpoint by framing a narrative in the con- text of taxic evolution. This inductive narrative is then presented as an hypothesis, the components and testability of which will be considered imme- diately following its presentation. Assume that very low rates of morphological change are generally correlated with low taxic di- versity and that very high rates are correspondingly correlated with high taxic diversity. Counter-ex- amples to these generalizations are easily found; for example, Cambrian echinoderms exhibit substantial morphologic diversity but little taxic diversity (see Sprinkle, 1976 and references therein), whereas North American minnows are speciose and mor- phologically uniform (Avise and Ayala, 1975, 1976). Nevertheless, insofar as bradytelic lineages are concerned, low rate of morphologic change is cor- related with low taxic diversity — compare, for in- stance, rhynchocephalians versus other diapsid “reptiles,” coelacanths and lungfish versus Actin- opterygii, Monoplacophora versus Gastropoda, as well as Xiphosura versus both Crustacea and Tri- lobita. Inasmuch as the problem of bradytely is being recast in terms of the taxic approach, we consider the possibility that low speciation rates determine the low observed rate of morphologic change, rath- er than asserting that the low rate of morphologic change results in such little morphologic diversity that systematists do not, or cannot, recognize many discrete taxa. There is the third possibility — that the correlation may be fortuitous and thus low rates of morphologic change may in no way be dependent upon low rates of speciation. But since we are stating a problem in taxic evo- lution, we might inquire if there might be a causal connection between speciation rate and degree of morphological change in the evolutionary process. Under orthodox darwinian notions of the signifi- cance of evolutionary morphological change, we assume that such change is generally adaptive — morphology is related to ecological niche occupa- tion and exploitation. Similarly, species diversity is related to ecological niche occupation and exploi- tation, an area of intense investigation among con- temporary ecologists, and a theme developed prom- inently in the recent paleontological literature, notably by Valentine (1973 and elsewhere) and by Bretsky and colleagues (for example, 1970). In view of these relationships, niche theory holds promise for shedding light on aspects (including rates) of the 14 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 13 relationship between the evolution of morphology and the evolution of diversity. How might this work? Organisms can be char- acterized on a sliding scale of eurytopy-stenotopy as a means of describing specific aspects of an or- ganism's utilization or tolerance of a given habitat parameter. 1 use these terms explicitly to refer to relative degrees or breadth of such utilization or tolerance, and not as a measure, for example, of the range of specifiable habitats a species may be found in. The concept of eurytopy-stenotopy is largely qualitative and evaluations are generally subjective, though many physiological parameters are easily quantifiable (for example, Bradshaw, 1961, on rel- ative temperature and salinity tolerances of six species of benthonic foraminiferans). There is a fur- ther important caveat in such considerations — a species may be stenotopic in one respect, and eu- rytopic in another; furthermore a species may be stenotopic in one part of its range, life cycle, or season (for example, rely on a single food resource) while quite eurytopic under other conditions. Yet it appears to be valid to distinguish between “ba- sically” eurytopic species (Limulus polyphemus being an excellent example) and stenotopes (for ex- ample species of Haplocliromis cited above), in terms of a comparison of feeding and locomotory behavior and ranges of physiological tolerance. To be useful, such comparisons should be made be- tween closely related organisms in comparable en- vironments. Stenotopic organisms, almost by definition, are specialists narrowly adapted when compared to their closest analogs. Although there would seem to be a limit to resource subdivision (but see Elessa and Levinton, 1975), stenotopes are those species which exploit only a small portion of the resource space theoretically available to species of their gen- eral kind. Eurytopes, in contrast, are broadly adapt- ed generalists. Most paleontologists (for example, Bretsky and Lorenz, 1970) have emphasized the im- plication of these alternative adaptive strategies in terms of ability to survive extinction events. But it might be asked how these different strategies came about in the first place and came to be maintained in the second place. There are three major possible outcomes to the onset of sympatry of two closely related species (see Mayr, 1963:81, for a discussion): competitive exclusion, hybridization, and accommodation. Ac- commodation between two species with initially overlapping niches is typically accomplished through resource subdivision, which frequently results from intensification of specializations already begun prior to sympatry — an accentuation of differences already latent in the allopatric state (Eldredge, 1974fi). Accommodation amounts to a de facto gen- eralization of the typical response of stenotopes to interspecific competition — that is, there frequently ensues a further division of resources. Most closely related sympatric species are rather narrowly adapted organisms. Therefore we may hypothesize that stenotopy increases the probability of survival of new species budded off essentially by accidents of changing geography. On the average, taxic di- versity has to be higher among stenotopes. Fur- thermore, ecological specialization also implies, on the average, that behavioral differences among species will ultimately show up as anatomical spe- cializations. Therefore morphological change itself will be more common and more apparent among stenotopes. Because successful speciation will be more common, morphological change should ac- cumulate more rapidly within stenotopic lineages. Again, African cichlids of the genus Haplocliromis seem to fit this scenario well, and much of the nar- rative model corresponds very closely to specific conclusions reached by Fryer and lies (1969) in their analysis of these fish. The opposite should be true of eurytopes. Eury- topes are generalized in behavior and physiology, and therefore have relatively unspecialized somatic organizations. Eurytopes, furthermore, tend to be far-flung (Jackson, 1974) and tend not to occur sym- patrically with closely related species (for example, the largely allopatric distribution of the relatively few species of the eurytopic cichlid Tilapia com- pared with the great amount of sympatry among Haplocliromis species; Fryer and lies, 1969) — pre- sumptive evidence that eurytopes competitively ex- clude, far more often than they accommodate, close- ly related species. The data pertaining to limuline evolution also fit these generalizations quite closely. It would thus appear that speciation rates within eu- rytopic lineages are automatically dampened by their ecological strategy, and, as a corollary, mor- phological change will be retarded. Remembering that rate of morphological change is not directly or fully correlated with speciation rates, it is further relevant that when successful speciation events do take place within eurytopic lineages, the new species themselves tend to be eurytopic, hence re- main morphologically unspecialized. Jackson (1974) reaches much the same conclusions for somewhat different reasons. We may summarize the above inductive narrative 1979 ELDREDGE— ALTERNATIVE EVOLUTIONARY THEORY 15 as a single hypothesis with several component sub- hypotheses, all of which were contained in the nar- rative as assumptions. Thus the single hypothesis is rapid rates of morphologic change are correlated with rapid speciation rates; slow rates of morpho- logic change are correlated with low speciation rates^. Correlation is statistical, and a value of R = 1 is not expected; nor, of course, is the narrative held to be true if this initial hypothesis is rejected. But it may be falsified, though not by pointing to a single counter example (for example, Cambrian echinoderms as cited above), as inviolate laws seem impossible to find in evolutionary biology. Boucot ( 1975) has recently amassed an enormous amount of data on mid-Paleozoic brachiopods, con- cluding that eurytopic genera® have relatively long stratigraphic ranges, whereas stenotopes (endem- ics) have relatively short stratigraphic ranges. The former evolves more slowly than the latter. Bou- cot’s data are taxa, that is, aggregates of individuals that Boucot and many others had previously de- fined on the basis of observed morphological simi- larities and differences. Thus, in an important way, Boucot has thoroughly confounded morphological and taxic evolutionary rates — the issue he confronts is rate of morphological change, but the data he uses are of generic diversity, and thus we have in a real sense an independent test of the hypothesis above. Based on Boucot’s presentation of the bra- chiopod data, rapid rates of morphological change are indeed highly correlated with more speciose groups, whereas slower rates of morphological change are highly correlated with less speciose groups. Because the interval of time is the same for all groups examined, “more speciose” means “spe- ciating at a higher rate.” Boucot (1975 and manu- script) does not seek a purely taxic (speciational) explanation for the correlation he reports, but his data do serve to corroborate this first hypothesis — a further warning that a large if not infinite number of narratives can be concocted for any given set of observations. * Stanley (1975:647) has recently graphed the same data on Dipnoi presented here, and has briefly discussed bradytelic rates. He argues that “groups of taxa that have survived at consistently low diversities over long periods of time should exhibit very little evolutionary change,” and that in fact living fossils provide a test of the general validity of the “rectangular model” of evolution. Thus our hypotheses on bradytely are identical, apparently because a preference for the taxic approach immediately leads to such a notion. I stop short of citing this coincidence (Eldredge, 1975; Stanley, 1975) as corroboration of the hypothesis itself. ® Boucot’s criterion for determining eurytopy-stenotopy is basically the relative breadth of geographic distribution of a genus. This criterion was explicitly rejected earlier in this paper, as specialists may be widely distributed and generalists decidedly less so, all according to the distribution of their habitats. Yet the correlation Boucot relies upon seems real enough and, in any case, no other criterion forjudging eury- topy-stenotopy with fossils comes readily to mind. Insofar as the specifics of the competition model are concerned, the above inductive argument is only as good as prior theoretical analysis of empir- ical data already performed by others. Bock (1970, 1972) has utilized competition theory extensively in his analysis of the adaptive radiation of the Hawaiian honeycreepers and other groups. Bock specifi- cally invokes (divergent) character displacement, which results from sympatric interactions among closely related (hence initially similarly adapted) species, to provide the selection force for the di- rection of morphological change (and specializa- tion) that his prior analysis of relationships re- vealed. Studies of competitive interaction involving niche subdivision and divergent character displace- ment are legion, though their interpretation remains arguable (see Grant, 1972, for a recent review of the literature and a skeptical view of the success of the concept; Eldredge, 1974/7, summarizes the mea- ger paleontological literature on this subject). The studies by Bock and others tend to corroborate the notion that significant amounts of morphological change take place in conjunction with species in- teractions. But they tell us nothing about the taxic and morphologic stasis hypothesized to take place when competition results in mutual exclusion. Stanley (1973) in a lucid paper has utilized com- petition theory to explain the disparity in taxic rate of evolution of mammals vis d vis marine bivales. He concluded that, on the whole, sessile bivales are less competitive inter se than are mammals®. The argument is appealing. To test the hypothesis ad- vocated here, that eurytopic/stenotopic ecological strategies are the prime determinant of a given species’ reaction to sympatric competition with a close relative, it would be more germane to com- pare relatively faster and more slowly evolving lin- eages within both the mammals and clams. But for the most part, the model developed herein is closely similar to Stanley’s model linking differential rates of taxic evolution with interspecific competition. How, in fact, do we test the hypothesis — eury- topes react to competition by exclusion, whereas stenotopes tend to subdivide niches (hence interact and undergo divergent character displacement as per above hypothesis)? There are several ways — the first being the familiar demonstration of corre- lation. Eryer and lies (1969) reach precisely these conclusions upon comparison of the distributions, morphology, and behavior of the cichlid taxa they ^ Van Valen (1976) has taken issue with Stanley’s evidence, arguments, and con- clusion that there is relatively little interspecific competition among clams. 16 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 13 studied. In a way the hypothesis merely generalizes an empirical observation long known to ecologists and systematists. A further test of the hypothesis that eurytopes tend to mutual exclusion upon the onset of sympatric competition could be gained from mathematical modelling, particularly via sim- ulation. Studies of this kind specifically addressed to this issue have yet to be performed. Implicit in the inductive model linking morpho- logical r ate of evolution with speciation r ates, is the notion that morphological change in evolution is largely effected through speciation. This may be stated in another way — we may test the hypothesis that, once established, species tend not to exhibit much morphological change cumulatively (over time) though they would of cour se be expected to display geographic variation as well as the effects of sympatric inter action with closely r elated species should this ever occur. Avise and Ayala (1975) re- cently tested the rival hypotheses that genetic change within a lineage is correlated with (1) amount of speciation, and (2) time (duration of the lineage). Their r'esults were ambiguous, but if any- thing favored (2). Vuilleumier (1969) has docu- mented allopatric diver'gence among closely r'elated species of Andean birds, indicating that phyletic evolution is a real possibility. Given time, selection as well as r andom sampling effects fr om generation to generation ar e expected and observed to lead to cumulative changes in both genotype and pheno- type within species’’. But it is still of interest to investigate the r elative importance of phyletic morphological change within a species versus the impor tance of speciation in ef- fecting that change (via a non-random sampling of the parent species genotype/phenotype and post- speciational interspecific interactions more than any “genetic revolution” that might be imagined to take place at the “moment” of speciation). How do we test the r'elative contr ibution of these two factor's in effecting morphological change? Close analysis of low-level (ca. species) taxa in the fossil record may be used to arbitrate the dis- cussion. But the situation is not at all straightfor- ward. Gingerich (1974, 1976) concludes that phy- letic change in morphological features is the rule, not the exception within lineages of ancestor-de- ^ It should be noted that the relatively speciose lineage (North American minnows) investigated by Avise and Ayala (1975, 1976) is morphologically uniform. We may still hypothesize that relatively greater amounts of genetic change, when observed, are strongly correlated with relatively high rales of speciation, in common with the hypothesis above linking morphological rates with speciation rates. scendant population samples. His procedur'e is straightforward — select a ser ies of stratigraphically closely spaced samples of specimens which, by pr ior analysis, are thought to belong to a monophy- letic assemblage. Select one or more morphological features and simply document what happens to them up the stratigraphic section. Some characters exhibit change, others do not. Gingerich (1974, 1976) has found that the surface ar ea of M, of sev- eral differ ent Eocene mammalian taxa exhibit grad- ual, progressive change (see Gould and Eldredge, 1977, for a specific critique of Gingerich’s meth- odology). Gr aphing the changes against stratigraph- ic position, Gingerich then simply encircles seg- ments of lineages most clearly linked continuously, and names taxa on the basis of these cluster s. Mor - phology evolves inexorably, and we chop it up to name taxa. Evolution is phyletic. Eldredge and Gould ( 1972) and Eldr edge and Tat- tersall (1975) have argued that the first step in pa- leontological systematics is to recognize basic taxa. Citing examples from trilobites, gastropods, and hominids, they contend that it is possible to rec- ognize and diagnose species at any one point in time, to distinguish such taxa from others living sympatrically (including synchronically, of course) as well as allopatrically and/or allochronically. It is possible to specify in what respects those species differ among each other and, in most cases, those species-specific differentia are found in other sam- ples, be they older, younger, or elsewhere. In other words, speeies are real entities with both geograph- ic and stratigr'aphic distributions (few species ar'e known from but a single bedding plane). Tempo- rally, ther e is no significant change in these species- specific differentia — after all, it is this very conti- nuity, which allows us to recognize a species in more than one place at more than one time,. Other mor phological features within the str atigraphie dis- tribution of a species — features which wer e not cit- ed as differentia — often do show sequential change within the history of that species. They conclude that species are distinguished both spatially and tempor ally by virtue of a number of specifiable at- tributes, which tend not to change inexorably as time goes by. Most morphological change is found to be among-species, hence assoeiated with specia- tion. Stanley (1975) has called this the “rectangu- lar” model of evolution. Gingerich’s entire methodology, hence conclu- sion, springs from the transformational approach, whereas those of Eldredge and Gould (1972) and 1979 ELDREDGE— ALTERNATIVE EVOLUTIONARY THEORY 17 Eldredge and Tattersall ( 1975) derive from the taxic approach. The disagreement among these authors is ultimately derived from starkly different ways of looking at the evolutionary process. The problem lies more in the fundamental assumptions rather than in faulty logic on either part within the frame- work of the chosen approach. Thus I would con- clude that the patterns of morphological change both within and among species in the fossil record provides a crucial test of species stability through time, and thus tends strongly to corroborate the in- ductive narrative above on evolutionary rates. In the framework of the taxic approach, this may be so. But as long as we retain these two alternative approaches to evolutionary theory, we shall contin- ue to argue issues from different premises, and we may confidently expect to get nowhere. SUMMARY AND CONCLUSIONS 1) A complete evolutionary theory requires the presence of two distinct components — (a) a theory of mechanics to explain genetic, morphologic, and behavioral change, and (b) a theory pertaining to the origin of species. 2) Many of the classic areas of investigation, es- pecially in paleontology and genetics, emphasize the aspect of evolutionary mechanics (the “trans- formational” approach) to the point of near exclu- sion of consideration of the origin of taxa (the “tax- ic” approach). 3) Integration of the two approaches is best ef- fected by considering the issues of the transforma- tional approach as a subset of those of the taxic approach. This amounts to saying that taxa evolve, not individual organisms or parts thereof. 4) The problem of bradytely (“arrested evolu- tion”) is chosen as an example of a topic in evo- lutionary biology nearly always stated and investi- gated purely in terms of the “transformational” approach. Previous hypotheses (actually inductive narratives) proposed to explain bradytely include (a) lack of available variability, (b) lack of direc- tional selection and/or presence of strong “centrip- etal” or “stabilizing” selection, and (c) retention of original adaptation in conjunction with habitat per- sistence. 5) Restated in taxic terms, a narrative explanation for bradytely is proposed: a) Bradytelic lineages are characterized by low diversity for the greatest part of their temporal per- sistence. Therefore, there is a problem in deciding whether diversity is low simply as a result of there being little morphological diversity for a systematist to use to recognize and name taxa, or whether mor- phological diversity is low because there has been relatively little speciation within the lineage. The narrative proposed under the taxic approach poses the latter possibility. b) Taxonomic, morphologic, and genetic evolu- tionary rates are “decoupled” to a significant ex- tent. c) However, there is evidence that lineages ex- hibiting truly low rates of morphologic change have low taxic diversity, and that lineages characterized by high rates of morphologic change have high taxic diversity. d) Eurytopic organisms (ecologic generalists) react to competition with close relatives largely by mutual exclusion, whereas stenotopes tend toward accommodation by specialization and partitioning of resource space. e) Thus stenotopes have a higher probability of successful speciation and thus behavioral special- izations ultimately are expressed in terms of ana- tomical modifications. Speciation rates of eurytopes are dampened by their ecological strategy and thus their relatively generalized behavior and physiology leads to retention of relatively primitive pheno- types. 6) Several ways of testing the component hy- potheses of the narrative are suggested. The prob- lem remains difficult to attack in an hypothetico- deductive manner. Correlation between different parameters (for example, between low rate of mor- phologic change and low taxic diversity; between degree of sympatric occurrence among congeneric stenotopes versus congeneric eurytopes, and oth- ers) offers a means of rejection, though not a par- ticularly strong source for corroboration. 7) Many of the controversies in contemporary evolutionary biology stem not so much from the relative merits of the logic or data used in arguing a particular point of view, but rather from funda- 18 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 13 mentally different approaches taken by the investi- gators. The current controversy in paleontology over the relative importance of phyletic evolution versus speciation is an excellent case in point. Most adherents of the view that evolution is essentially phyletic see evolution primarily as the cumulative change of gene content, frequency, and expression. The fundamental assumption of the opponents of this view appears to be that evolution is quintes- sentially the origin of new taxa (species). Thus the two opposing views (that is, on the relative impor- tance of phyletic evolution versus speciation) are not sufficiently comparable to allow rejection of one in favor of the other. The argument is, in the end, over which approach to evolutionary theory is the more appropriate. The taxic approach seems the superior of the two because of its capacity for sub- suming the transformational approach. The con- verse does not appear to be true. ACKNOWLEDGMENTS I have benefited from a great deal of discussion with students and colleagues resulting from presentation of some of the ideas in this paper in seminars at Queens College and Wake Forest University. 1 am particularly grateful to Drs. H. B. Rollins and J. H. Schwartz, both of the University of Pittsburgh, who or- ganized the Colloquim and prevailed upon me to organize my thoughts for formal presentation. Drs. Rollins and Schwartz, together with a number of their students and colleagues, made very valuable suggestions which helped improve the final ver- sion. I thank Drs. J. Cracraft, E. S. Gaffney, S. J. Gould, E. Mayr, M. C. McKenna, B. Schaeffer, and E Tattersall for read- ing an early draft of this paper. I acknowledge with thanks the fine artwork of Ms. Marjorie Shepatin. I am particularly grateful to Dr. Tattersall for spotting the worst errors in logic and gram- mar in the final version. None of the above are to be held re- sponsible for the heresies and errors that remain. LITERATURE CITED Avise, j. C., and F. j. Ayala. 1975. Genetic change and rates of cladogenesis. Genetics, 81:757-773. . 1976. Genetic differentiation in speciose versus depau- perate phylads: evidence from the California minnows. Evo- lution. 30:46-58. Ayala, F. J. 1975. Genetic differentiation during the speciation process. Pp. 1-78, in Evolutionary biology (T. Dobzhansky et al., eds.) Plenum, New York, 8:1-396. Bock, W. J. 1970. Microevolutionary sequences as a funda- mental concept in macroevolutionary models. Evolution, 24:704-722. . 1972. Species interactions and macroevolution. Pp. I- 24, in Evolutionary biology (T. Dobzhansky et al., eds.) Appleton-Century-Crofts, New York, 5:1-317. Bonde, N. 1974. Review of "Interrelationships of Fishes.” Syst. Zool., 23:562-569. Boucot, a. j. 1975. Evolution and extinction rate controls. Elsevier, Amsterdam, 427 pp. Bradshaw, J. S. 1961 . Laboratory experiments on the ecology of Eoraminifera. Contrib. Cushman Eound. Foram. Res., 22:87-106. Bretsky, P. W., and D. M. Lorenz. 1970. Adaptive response to environmental stability: a unifying concept in paleoecol- ogy. North Amer. Paleont. Conv., Chicago, 1969, Proc. E:522-550. Brinkmann, R, 1929. Statistisch-biostratigraphische Unter- suchungen an mittel jurassischen Ammoniten iiber Artbe- griff Lind Stammesentwicklung. Abh. Ges. Wiss. Gottingen, 8(3). Bush, G. L. 1975. Modes of animal speciation. Ann. Rev. Ecol. Syst., 6:339-364. Carruthers, R. G. 1910. On the evolution of Zaplirentis de- lanouei in Lower Carboniferous times. Quart. J. Geol. Soc., 66:523-538. Clarke, J. M. 1913. Fosseis devonianos do Parana. Monog. Serv. Geol. Mineral, do Brasil., 1:1-353. Darwin, C. 1859. On the origin of species. Atheneum, New York, 1967 facsimile edition, 502 pp. Dobzhansky, T. 1951. Genetics and the origin of species. Co- lumbia Univ., New York, 3rd ed., 364 pp. Eldredge, N. I974<;. Revision of the suborder Synziphosurina (Chelicerata, Merostomata), with remarks on merostome phylogeny. Amer. Mus. Novitates, 2543:1-41. . 19746. Character displacement in evolutionary time. Amer. Zool., 14:1083-1097. . 1975. Survivors from the good old, old, old days. Nat. Hist., 84:60-69. . 1978. Trilobites and evolutionary patterns. Pp. 305- 332, in Patterns of evolution (A. Hallam, ed,), Elsevier, New York, 591 pp. Eldredge, N., AND S. J. Gould. 1972. Punctuated equilibria: an alternative to phyletic gradualism. Pp. 82-1 15 in Models in paleobiology (T. J. M. Schopf, ed.) Freeman, Cooper and Co., 250 pp. Eldredge. N., and I. Tattersall. 1975. Evolutionary models, phylogenetic reconstruction, and another look at hominid phylogeny. Pp. 218-242, in Approaches to primate paleobiology (F. Szalay, ed.). Karger, Basel, 326 pp. Fisher, D. C. 1975u. Swimming and burrowing in Limiiliis and Mesolimtilus. Fossils and Strata, 4:281-290. . 19756. Evolution and functional morphology of the Xiphosurida. Unpublished Ph.D. dissert.. Harvard Univ., 393 pp. 1979 ELDREDGE— ALTERNATIVE EVOLUTIONARY THEORY 19 Flessa, K. W., and J. S. Levinton. 1975. Phanerozoic di- versity patterns: tests for randomness. J. Geol., 83:239-248. Fryer, G., and T. D. Iles. 1969. Alternative routes to evo- lutionary success as exhibited by African cichlid fishes of the genus Tikipia and the species flocks of the great lakes. Evolution, 23:359-369. Ghiseein, M. 1974. A radical solution to the species problem. Syst. Zool., 23:536-544. Gingerich, P. D. 1974. Stratigraphic record of Early Eocene Hyopsodus and the geometry of mammalian phylogeny. Nature, 248: 107-109. . 1976. Paleontology and phylogeny: patterns of evolu- tion at the species level in early Tertiary mammals. Amer. J. Sci., 276:1-28. Gould, S. J., and N. Ei dredge. 1977. Punctuated equilibria: the tempo and mode of evolution reconsidered. Paleobiol- ogy, 3: 1 15-151. Grant, P. R. 1972. Convergent and divergent character dis- placement. Biol. J. Linn. Soc., 4:39-68. Hardy, D. E. 1970. Drosophilid fauna. Pp. 450-469, in Evo- lutionary biology (M. K. Hecht and W. C. Steere, eds,), Appleton-Century-Crofts, New York, suppl., 594 pp. Jackson, J. B. C. 1974. Biogeographic consequences of eury- topy and stenotopy among marine bivalves and their evo- lutionary significance. Amer. Nat., 108:541-560. King, M. C., AND A. C. Wilson. 1975. Evolution at two levels in humans and chimpanzees. Science, 188: 107-1 16. Lesch, j. E. 1975, The role of isolation in evolution: George J. Romanes and John T. Gulick. Isis, 66:483-503. Mayr, E. 1942. Systematics and the origin of species. Columbia Univ, Press, New York, 334 pp. . 1963. Animal species and evolution. Harvard Univ. Press, Cambridge, 797 pp. Raup, D. N., and S. j. Gould. 1974. Stochastic simulation and evolution of morphology — towards a nomothetic pa- leontology. Syst. Zool., 23:305-322. Romanes, G. J. 1886. Physiological selection: an additional suggestion on the origin of species. J. Linn. Soc. Zool , , 19:337-411. Romer, a. S. 1966. Vertebrate paleontology. Univ. Chicago. Chicago, 3rd ed., 468 pp. Rowe, A. W. 1899. An analysis of the genus Micraster. as determined by rigid zonal collecting from the zone of Rhyn- clioncdla Cuvieri to that of Micraster coranguinum. Quart. J. Geol. Soc. London, 55.494-547. ScHOPF, T. J. M., D. M. Raup, S. J. Gould, and D. S. Sim- BERLOFF. 1975. Genomic versus morphologic rates of evo- lution: influence of morphologic complexity. Paleobiology, 1:63-70. Selander, R. K., et ai . 1970. Genetic variation in the horse- shoe crab (Liniuhis polyphemiis), a phylogenetic “relic.” Evolution, 24:402^14. Shaw, A. B. 1969. Adam and Eve, paleontology and the non- objective arts. J, Paleont., 43:1085-1098. Simpson, G.G. 1944. Tempo and mode in evolution. Columbia Univ., New York, 237 pp. . 1951. The species concept. Evolution, 5:285-298. . 1953. The major features of evolution. Columbia Univ., New York, 434 pp. . 1961. Principles of animal taxonomy. Columbia Univ., New York, 247 pp. . 1974. Notes on Didelphidae (Mammalia, Marsupialia) from the Huayquerian (Pliocene) of Argentina. Amer. Mus. Novitates, 2559:1-15. Sprinkle, J. 1976. Classification and phylogeny of “pelmato- zoan” echinoderms. Syst. Zool., 25:83-91. Stanley, S. M. 1973. Effects of competition on rates of evo- lution, with special reference to bivalve mollusks and mam- mals. Syst. Zool., 22:486-506, . 1975. A theory of evolution above the species level. Proc. Nat. Acad. Sci,, 72:646-650. Trueman, A. E. 1922. The use of Gryphaea in the correlation of the Lower Lias. Geol. Mag., 59:256-268. Valentine, J. W. 1973. Evolutionary paleoecology of the ma- rine biosphere. Prentice-Hall, Inc., Englewood Cliffs, New Jersey, 5 1 1 pp. Van Valen, L. 1976. Energy and evolution. Evol. Theory, 1:179-229. Vuii leumier, F. 1969. Pleistocene speciation in birds living in the high Andes. Nature, 223:1179-1180. Waagen, W. 1896. Die Formenreihe des Ammonites suhradia- tiis. Geogn.-Palaeont. Beit., 2:181-256. Wagner, M. 1869. Die Entstehung der Arten durch raumliche Sonderung. Benno Schwalbe. Basel. (Not seen.) Westoi L, T. S. 1949. On the evolution of the Dipnoi. Pp. 121- 184, in Genetics, paleontology and evolution (G. L. Jepsen et al., eds.), Princeton Univ. Press, xiv-l-474 pp. Willis, J. C. 1940. The course of evolution. Cambridge Univ. Press. Cambridge, 207 pp. THE SYNTHETIC EXPLANATION OE MACROEVOLUTIONARY CHANGE— A REDUCTIONISTIC APPROACH Walter J. Bock Department of Biological Sciences, Columbia University, New York, New York 10027, and Department of Ornithology, American Museum of Natural History, New York, New York 10024 INTRODUCTION Formulation of the synthetic theory of evolution began in the early 1930’s with the high hopes of explaining all evolutionary phenomena with a single unified theory. Many of the hopes of these evolu- tionary biologists were realized, some more than they had dared to believe possible. Other facets of evolutionary theory, once considered incontestably established, were subsequently questioned. Yet the single aspect of evolutionary biology that eluded attempts for successful and convincing explanation is macroevolution — the appearance and subsequent specialization of distinctive new features and taxa. Advocates of the synthetic theory assumed, largely as untested conviction, that major evolutionary changes were simply the consequence of many small modifications and that these large changes would be understood if the microevolutionary events were fully comprehended. Although 1 be- lieve that this assumption is correct and will advo- cate it in this paper, one of the major failures of the synthetic theory has been to provide a detailed and coherent explanation of macroevolution based on the known principles of microevolution. In spite of the beliefs of the advocates of the synthetic theory, macroevolution has not been reduced successfully to microevolution. Because of this failure of the synthetic theory, two major explanations have been advocated for major evolutionary modifications. One is the reduc- tionistic theory as mentioned above and which will be advocated herein. The other stems from the pe- riod of idealistic morphology of the second half of the nineteenth century with strong pre-Darwinian roots, and can be termed as the quantum or salta- tion theory. Differences between these two conflict- ing theories, as well as the confusing complex of ideas associated with macroevolutionary explana- tion, are demonstrated by two recent textbooks of evolution published by the same house. Dobzhan- sky et al. (1977) advocate strongly the synthetic view with a firm statement that macroevolutionary change must be adaptive throughout. Grant (1977) leans strongly toward the quantum theory of major evolutionary change although he rejects extreme versions of saltation. Yet both texts rely heavily on the ideas of Simpson (1944, 1953) and both cite my analysis (Bock, 1970) of the evolution of the Hawai- ian honeycreepers (Drepanididae) as an example of adaptive radiation. I would like, in this paper, to provide a detailed reductionistic explanation of macroevolution within the tradition of the synthetic theory. An important part of this analysis will be an examination of the role of the species and of speciation in this expla- nation. At this point, I would like to acknowledge my debt to my teacher and mentor Ernst Mayr whose name is usually associated with the species concept and speciation, but who has done more to establish the foundations of an explanation of macroevolutionary phenomena than any other liv- ing evolutionist. LIMITS OF MACROEVOLUTION Before a causal theory of macroevolution can be postulated, it is necessary to establish the limits of this type of evolutionary change and to distinguish it from microevolutionary phenomena. Unfortu- nately no definite boundary can be drawn between macro- and microevolutionary changes; I believe that the two grade smoothly into one another. Yet even when no sharp demarkation can be established between two concepts or phenomena, such as night and day, frequently the two may be easily recog- nized. Hence, by microevolutionary change, I mean those modifications of the level studied by popula- tion geneticists and by animal and plant breeders. These would be differences of the degree that dis- tinguish populations and subspecies of the same species or that distinguish congeneric species. 20 1979 BOCK— SYNTHETIC EXPLANATION OF MACROEVOLUTION Microevolutionary events are ones amenable to ex- perimentation and direct observation. Macroevo- lutionary changes are those modifications of the level studied by comparative anatomists and pale- ontologists. They would be differences of the de- gree that separate families, orders, and groups of higher categorical rank. Differences between gen- era are usually considered to be at the lower end of the scale of major evolutionary change. The origin of a distinctive new feature, such as the vertebrate eye or the mammalian jaw articulation, or the rad- ical modification of an existing feature, such as the evolution of the tetrapod limb from a crossoptery- gian fin or the avian wing from a reptilian forelimb, would all be major changes. Macroevolutionary 21 changes are usually not open to experimentation or direct observation, although some of the results of animal and plant breeding, for example, the fancy breeds of goldfish and the diverse breeds of dogs, surely must be regarded as major modifications. I do not want to accept any level as the demar- cation between micro- and macroevolution in a hard and fast way. Successful development of the reductionistic explanation of major evolutionary change is not dependent upon any particular limit between these two degrees of evolutionary change. More important is the continuum between these facets of evolutionary change and the fact that ex- amples of the two types overlap rather broadly. REDUCTIONISM One of the central arguments about theories of the mechanisms of macroevolution is whether they are reducible to mechanisms of microevolution. Stanley (1975) states that macroevolution is not re- ducible to microevolution and I claim that it is. This is one of the fundamental distinctions between quantum and synthetic theories of major evolution- ary change. A consequence of these two positions is that in the reductionistic synthetic approach, no mechanisms of evolutionary change need be pro- posed other than those needed to explain micro- evolutionary events. In the nonreductionistic quan- tum approach, at least one additional mechanism of evolutionary change unique to macroevolutionary events must be postulated over and above those needed to explain the microevolutionary phenom- ena. The additional major evolutionary mecha- nism(s) must first be postulated and then tested somehow by observations. Moreover, it must be shown that these unique macroevolutionary mech- anisms are really not reducible to known microevo- lutionary mechanisms. Before this conflict on the reduction of major evo- lutionary explanation can be solved, it is neces- sary to clarify the meaning and usage of the concept of reduction of scientific theories. I accept the ap- proach to reductionism advocated by Ernest Nagel (1961; Chapter 11). Reductionism can be intersci- ence or intrascience; we are here concerned with an intrascience reductionism because we remain within the limits of evolutionary biology. The ques- tion is whether a large change in evolution is simply the consequence of a cumulative series of small changes and hence explainable by the mechanisms governing the small changes, or whether the large change involves at least one step different from the microevolutionary changes. If the former is correct, then the explanation must include consideration of levels of organization, which is intimately associ- ated with the concept of reductionism. Nagel argues that reduction of theories to be suc- cessful must follow a rigid set of rules. Both the theory to be reduced and the reducing theory must be carefully stated. It is not sufficient to claim that macroevolutionary explanation can or cannot be reduced to microevolutionary explanation. One must state the details of macroevolution — the the- ory to be reduced — and the details of microevolu- tionary explanation — the reducing theory. Hence the statements by many advocates of the synthetic theory that major evolutionary change can be ex- plained by the mechanisms of microevolution is correct, but grossly inadequate. Such a statement can neither be defended or argued against because of its vagueness. The second point mentioned by Nagel is that all phenomena explained by the reduced theory must be explained by the reducing theory for a successful reduction. If there are any phenomena that were explained by the reduced theory that cannot be ex- plained by the reducing theory, then some addition- al theory is needed and the reduction is not suc- cessful. Nagel goes on to argue that for strong reduction all parts of the reduced theory must be 22 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 13 derivable from the reducing theory. This require- ment is not essential for weak reduction. The usefulness of reduction of theories is a matter quite apart from whether a particular theory is re- ducible to another. It is generally accepted that re- duction will be useful if the reducing theory can explain many phenomena other than those covered by the reduced theory and if the outcome of the reduction is a more generalized theory uniting many aspects of the particular science or sciences. Reductionism, especially intrascience reduction- ism, is closely associated with levels of organization and with the question of the emergence of new properties with increasing levels of organization. It is not clear whether the contrast of microevolution and macroevolution fits into the typical pattern of hierarchical organization, but it does fit many of the characteristics of this pattern. Nagel shows that the common statement that the “whole is greater than the sum of its parts” is dependent upon knowing how to add the component parts and whether the details of organization must be included in the ad- dition. In consideration of whether macroevolutionary explanation is reducible to microevolutionary mechanisms, an important factor is how the indi- vidual small steps are added together to obtain a single major change. It is obvious that this addition must be done in a particular manner. The small steps cannot be added randomly or together in a single unit. Rather they must be added sequentially in a chronological series. Thus, the change at any point in time sets the stage for the next change and so forth. Features that arise or are modified at one point in time provide the foundation for the next evolutionary change in features. Macroevolution is thus not just a summation of many small changes, but a sequential summation of many small changes added together in their exact chronological series. Special care will be given below to stating pre- cisely and completely macroevolutionary explana- tion as the theory to be reduced and microevolu- tionary explanation as the reducing theory. Moreover, care will be given to providing reference to the observational and experimental bases used to test the microevolutionary explanation. Atten- tion will also be given to the proper sequential sum- mation of the small changes that add up to a major evolutionary modification. Lack of attention to these facets of reductionism in the past has been a major failure on the part of advocates of the syn- thetic theory as well as of the quantum theory of macroevolution. THEORIES AND THEIR TESTING To be successful, theories or explanation of ma- jor evolutionary change must satisfy two major re- quirements. They must provide a causal explana- tion of the phenomena associated with large scale modifications and they must be testable against ob- servations. I would like to consider each in turn. A theory explaining macroevolutionary change, be it the origin and specialization of new features or the adaptive radiation of a new taxon, is a causal explanation similar to theories of speciation, adap- tive modification, generation of new genetical vari- ation, and others. Although the time period in- volved with major changes and hence in the explanation may be great, covering tens of thou- sands or even millions of years, the theory is not a historical explanation. I am not concerned with ana- lyzing and explaining the evolutionary aspects of a particular feature or of a particular taxon. Rather, I wish to present a causal explanation of macroevo- lution, which can be used in historical explanations of individual cases. The fact that a causal explanation of macroevo- lution includes a sequential analysis of steps does not make it a historical analysis. Numerous other causal explanations, for example, geographical spe- ciation, depend upon a sequence of events arranged in proper chronological order. The second point is that any successful causal explanation of macroevolution, being a scientific theory, must be tested against experiments and ob- servations. Formulation of any explanation brings with it the obligation of demonstrating how the the- ory is to he tested and if possible to provide some tests. Development of an explanation that depends upon a vague or untestable mechanism weakens that theory considerably. Perhaps the greatest weakness of past attempts to provide explanations, be they synthetic or quantum, of macroevolution has been the failure to formulate convincing testing procedures. Procedures are frequently proposed to test theories of macroevolution, but upon close scrutiny, these tests prove to be inadequate. An 1979 BOCK— SYNTHETIC EXPLANATION OE MACROEVOLUTION 23 example is the tests of the theory of punctuated equilibria by the fossil record (Gould and Eldredge, 1977: 120). The nature of the fossil record, espe- cially the scale of resolution of time and geographic distribution, is simply inadequate to distinguish be- tween punctuated equilibria and conflicting theo- ries. Testing of macroevolutionary theories depends upon formulating a proper argument-chain of pre- dications, secondary theories, and connecting links, and finally the experiments and observations serv- ing as empirical tests. It is generally accepted that theories of major evolutionary change cannot be tested directly against experiments and observa- tion. This is generally valid, but we may have far more evidence available from experimental work and from observations of animal and plant breeders than generally suspected. Special care will be given to outlining the pro- cedures for testing the synthetic theory of macro- evolution. The steps in the argument-chain, the sec- ondary hypotheses, the basic links, and the final observations will all be clarified. What types of ob- servations that can and cannot be used to test this explanation will be pointed out. CONFLICTING EXPLANATIONS OF MACROEVOLUTION Introduction Numerous theories have been advocated to ex- plain the phenomena of major evolutionary modi- fications. All of these need not be discussed be- cause most have not attracted much attention. I will concentrate only upon the two major sets, which are in direct conflict with each other. Although con- siderable variation exists between the individual theories included in each of these sets, the basic agreements are more important. The first set is that of quantum theories, which are characterized by advocating a single-step jump or saltation at some point in the major change. The evolutionary mech- anism involved in this jump is one other than adap- tive change under the control of natural selection arising from the external environment. The second set is that of synthetic theories in which the expla- nation of major changes is reduced to mechanisms acting on the microevolutionary level. These theo- ries exclude any distinctive saltations, do not in- voke any special mechanisms of evolutionary change other than those operating at the microevo- lutionary level, and depend upon adaptive change throughout. All change is under the control of se- lection arising from the external environment. Quantum Theories I will discuss only those quantum theories ad- vocated since the origins of the synthetic theory of evolution in the early 1930’s. However, it should be emphasized that the history of ideas of quantum evolutionary changes date back to the period of idealistic morphology in the second half of the nine- teenth century and have their roots in pre-Darwin- ian typology. By a quantum theory, 1 mean one that depends upon a single saltation of a magnitude greater than the evolutionary change observed in microevolu- tionary modifications as studied by animal and plant breeders and by populations geneticists. The mech- anism controlling this saltatory jump is either a non- selective one (not natural selection arising from the external environment) or a type of selection dis- tinctive from natural selection. Frequently these theories depend upon the existence of a threshold or of a selective bottleneck. The concept of typogenesis or typostrophism of Schindewolf ( 1936, 1950:206) is a clear consequence of the concepts of typology and the idea of “bau- plan" of groups developed directly from the con- cepts of idealistic morphology (Bock and von Wah- lert, 1963; Reif, 1975). This theory depends upon the concept that the characteristics of a taxon are expressed in its bauplan or type, that a type bound- ary delimits a group and that a distinct gap sepa- rates the bauplan of one group from that of another. Hence the evolutionary change from one bauplan to another must be a jump over this gap — over the type boundary of one group to that of another. This change must be of a different type than the evolu- tionary change within the limits of a bauplan. Oth- erwise there could not have been a type jump. The saltation is not an adaptive change controlled by selection; it is left as a vague evolutionary mecha- nism. A similar theory is offered by Goldschmidt (1940:184-395). He argues that the small genetical changes studied by geneticists are insufficient to re- sult in a major change even if many are added to- gether. Rather he suggests that the large changes 24 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 13 result from the occurrence of systemic mutations ( = macromutations) leading to a new type of organ- ism that is adapted to a new set of environmental conditions. These organisms resulting from system- ic mutations have been dubbed “hopeful monsters” by other workers. The role of selection in Gold- schmidt’s theory is vague. He says (p. 396) that accumulation of micromutants by selection has been “ruled out,” and that selection can act on the new form after the systemic mutation. But he no- where discusses the magnitude of the modifications resulting from the systemic mutations and what are the environmental changes that can act as selection. In the absence of details on this point, it cannot be assumed that the selection discussed by Gold- schmidt corresponds to selection acting on micro- genetical changes or that it can be related to envi- ronments in the real world. It is interesting that Goldschmidt cites the Drepanididae (1940:214-215) as an example supporting his concept of macro- mutation by systemic mutations, but omits many of the details of geographic variation in species as well as possible intermediate stages represented by ex- tant species. The concept of “quantum evolution” was devel- oped by Simpson (1944:206-217) and has been widely cited as the synthetic theory explanation of macroevolution in sharp contrast to ideas of typo- genesis and systemic mutations. However, careful reading of Simpson’s text reveals that his concepts are basically more similar to those of earlier salta- tory theories than to a reductionary synthetic ex- planation of macroevolution. Quantum evolution depends upon a shift of a phyletic lineage through “discontinuities or essentially instable ecological zones” that lie between major adaptive zones. The start of quantum evolution is an inadaptive phase when the lineage enters the discontinuity. Exactly what is meant by “an inadaptive phase,” how long it exists and whether selection is acting during this period is not clear. The only conclusion that can be reached is that selection is not acting during the “inadaptive phase” otherwise this phase can not be so considered. The instable ecological zone is re- garded as a threshold through which the phyletic lineage must pass quickly or become extinct. These ideas are repeated by Simpson in his later book (1953:389-393) but with some changes. He says that “populations making a quantum shift do not lose adaptation” and “that the direction of the change is adaptive” (p. 391) but later says that “No inter- mediate stage persisted, because intermediate stages were less efficient (i.e., relatively inadap- tive)” (p. 392). If these shifts were fully adaptive, then it is not clear what is the distinction between quantum evolution and regular phyletic evolution. The concept of quantum evolution as expressed by Simpson involves a period where selection is not acting, depends upon passing over a threshold in the intermediate unstable ecological zone, and in- volves (always?) a key mutation. These expressions sound very similar to the concepts of Schinderwolf and of Goldschmidt in spite of differences in word- ing. Many of the concepts, such as the inadaptive phase and the ecologically unstable zone, are left vague and unconnected to natural phenomena. Quantum evolution is discussed as an evolutionary mechanism, yet it is not tested nor are testing pro- cedures clearly indicated. The concept of punctuated equilibria was postu- lated by Eldredge (1971) and developed more fully by Eldredge and Gould (1972) and discussed again by Gould and Eldredge ( 1977, see for references to other papers). They wished to examine the concept of slow evolutionary rates as the primary mode of evolution and to introduce the concept of allopatric speciation into paleontological thinking. Certainly the role of speciation has been neglected (although not totally) in discussions of macroevolutionary change and clearly many, probably most, major evolutionary modifications are more rapid than be- lieved by many workers. These concepts were al- ready discussed by Simpson (1944, 1953) and were a central part of my earlier analysis (Bock, 1970) of the role of microevolutionary events in macroevo- lution. Unfortunately, in the formulation of their concept of punctuated equilibria, Eldredge and Gould shifted from considering whether slow changes were the only or predominate mode of phy- letic change to discussing mechanisms of evolution- ary change. The result was a series of assertions which makes their concept one of quantum evolu- tion and hence unacceptable. The major problem is that the concept of phyletic gradualism (the term for slow uniform rate of evo- lutionary change) has been synonymized for phy- letic evolution. This is shown in Eldredge (1971:156- 157) and clearly in Eldredge and Gould (1972:87- 90; by the equating of Kellogg’s [1975] discussion of phyletic evolution with phyletic gradualism, pp. 126-128; and by their statement in the abstract “If, as we predict, the punctuational tempo is prevalent, then speciation — not phyletic evolution — must be the dominant mode of evolution.” p. 115). Part of 1979 BOCK— SYNTHETIC EXPLANATION OE MACROEVOLUTION 25 this problem may arise from a combination of their considerations of phyletic speciation, multiplication of species (geographic speciation) and phyletic evo- lution (1972) where discussions of phyletic evolu- tion and phyletic speciation appear to be interwo- ven. The consequence is that when they argue against (to the point of denying) phyletic gradualism as a dominant mode of evolution, they do the same for phyletic evolution. The outcome of this confusion of two concepts, which are completely independent of one another, is that they must reject the synthetic (reductionistic) explanation of macroevolution (for example, Gould and Eldredge, 1977:139-145). They accept the ar- gument of Stanley (1975), which is developed di- rectly from their concept of punctuated equilibrium. Stanley proposes a “rectangular model” for major evolutionary change in which the speciation shifts are regarded to be absolutely different from the evo- lutionary changes between successive speciations. He introduces a new evolutionary mechanism “species selection,” which is not described in suf- ficient detail and for which no procedures are pro- vided for testing. Stanley states in his discussion (1975:650) that “The reductionist view that evolu- tion can ultimately be understood in terms of ge- netics and molecular biology is clearly in error. We must turn not to population genetic studies of es- tablished species, but to studies of speciation and extinction in order to decipher the higher-level pro- cess that governs the general course of evolution.” Thus, the concept of punctuated equilibria can be summarized as major evolutionary changes that are the consequence of a series of speciation, not phy- letic evolution, in which the mechanism of species selection is important. Eurther, the explanation of these macroevolutionary changes cannot be re- duced to the evolutionary mechanisms operating at the microevolutionary level. Illustration of this change by a rectangular diagram underscores the quantum nature of this explanation. Eurther, Gould (1977r/, \911b, 1977c) clearly relates the ideas ad- vocated in punctuated equilibria with Gold- schmidt’s concept of systemic mutations. A curious approach to quantum evolution has been developed as an outcome of mathematical to- pological theory of catastrophe (Thom, 1975; Dob- son, 1975; Zeeman, 1976; Dobson and Hallan, 1977) developed by Rene Thom. Sussmann (1976) has presented arguments against the implications of this theory as developed in the above cited papers, and Kolata ( 1977) has commented on this devel- opment (see also letters to the editor, 1977, Science, 196: 1268-1270). This explanation can be discounted at this time because the possible evolutionary cor- relations to the mathematical theory have not been worked out, but it is of interest because it demon- strates how attractive the idea of quantum evolu- tionary steps is to many workers. Frazzetta ( 1970, 1975) discussed major evolution- ary change from the viewpoint of a functional mor- phologist, one of the few to do so. Although he does not deny the possibility of major changes resulting from a series of small evolutionary modifications, he believes that macroevolution can also occur by other processes, possibly by the appearance of sys- temic mutations as advocated by Goldschmidt. Many of the points discussed by Frazzetta, such as the need to examine interactions between struc- tures and the importance of somatic modifications (= physiological adaptation), are important and have been ignored by most evolutionists. As stressed by Frazzetta, analysis of evolutionary modification of morphological systems is far more complicated than appreciated by most workers ad- vocating macroevolutionary explanation. Yet 1 am not convinced that he presents a compelling argu- ment why all macroevolutionary explanation can- not be reduced to microevolutionary explanation, albeit the microevolutionary explanation will be more complex than usually assumed. A central issue in several recent quantum theo- ries of macroevolution is rapid speciation in which most of the evolutionary change takes place prior to the complete development of intrinsic isolating mechanisms and the sympatry of the newly ap- peared species. This factor is critical, for example, to the theory of punctuated equilibria. Carson’s ( 1975) ideas on the genetics of speciation have been cited in this connection. He argues that “microevo- lutionary events that lead to adaptations, however, do not appear to yield new species as a necessary or even a directly correlated consequence of the adaptation process” (p. 83). Carson suggests that the genetical system of a species is divided into two parts, the open and the closed. The open system responds readily to selection and is the portion of the genotype that has been studied by geneticists. The closed system is not affected by selection under usual conditions and does not yield easily to Men- delian analysis. Under normal circumstances, gene flow does not affect the closed system. Rather, the closed system is modified during speciation with a forced reorganization of the closed variability sys- 26 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 13 tern by a series of catastrophic stochastic genetic events. This occurs during a period when selection is not acting with a resulting population flush and crash (see his Fig. 2, p. 89). After the crash, a few founder individuals are left from which the new population is generated under natural selection. Carson is vague on the possible duration of this period of no selection nor does he give reasons to justify the existence of such periods. He does not postulate any new evolutionary mechanisms as the stochastic genetic events occur by known mecha- nisms of crossing over and other types of genetic recombination. Yet he is vague on why such events cannot take place during periods of normal selec- tion. Because it includes a period of no selection, Carson’s concept of speciation belongs to the class of quantum theories and has to be so used (for ex- ample, Gould and Eldredge, 1977). The major difficulty with the theory proposed by Carson is that he does not stipulate how the concept of the closed variability system and the mechanism of its change can be tested by empirical observa- tions. Also, the procedures by which this concept can be tested are difficult to envision because of the stipulation that the closed genetic system is not af- fected by selection during normal conditions, but changes only during speciation and because of the claim that speciation occurs only as a result of an alteration in the closed genetic system. These in- terlocking claims form a closed circularity devoid of any means of test by independent observation. Non-Darwinian evolution as postulated by King and Jukes (1969) and other workers has been in- voked by advocates of quantum explanation of macroevolution. Although the theory of non-Dar- winian evolution is not a quantum theory, it shares with these theories the notion of a mechanism di- recting evolutionary modification that is nonselec- tive. Unfortunately, details of the causal mecha- nism for this nonselective change have been left vague and certainly have not been tested. Other workers in biochemical evolution (for example, Goodman et al., 1975; Goodman, 1976) have argued against the claim of non-Darwinian evolution and state that evolutionary changes in proteins can be explained by selection. Another theory which does not belong to the class of quantum theories, but should be mentioned here is that of internal selection (Stebbins, 1974:123; Gutmann, 1977:645; Dullemeijer, 1974; Dullemeijer and Barel, 1977). This concept can be interpreted as a form of selection arising from the internal environment of the organism separately and independently of natural selection arising from the external environment. Such a concept has not been clearly formulated nor properly tested. More- over, the interpretation of a selection independent of natural selection arising from the external envi- ronment may not be the intent of Gutmann (personal communication) and possibly not of the other work- ers. Rather, the concept of internal selection refers to the interrelationships and interactions between features of an individual organism, including all of the mechanisms that serve to keep these features in proper functional balance with one another as required in a viable organism. The various mecha- nisms on internal adjustment of somatic features and their role in evolution have been considered by few workers (for example, Bock and von Wahlert, 1965; Frazzetta, 1975; Gutmann, 1977) but it is clear that these factors are of prime significance to major evolutionary modification. It is incorrect, however, to term these mechanisms of internal adjustment as internal selection or to imply that these mechanisms are not under the control of natural selection. Although the emphasis and the mechanisms differ in the above cited and other quantum explanations of macroevolution, the same thread of ideas runs through all. Most important is that they involve a jump or discontinuity at some point, usually during the presympatric phase of speciation, in which a break occurs in the adaptive modification of fea- tures and hence of the population. This break co- incides with a period in which natural selection re- sulting from the interaction of the external environment with the organism is not operating. The length of this period is not specified. Often evo- lutionary modification during this saltation is claimed to be under the control of a nonselective evolutionary mechanism, but this mechanism is not outlined clearly and/or not tested against empirical observations. These theories fail, in my opinion, for several rea- sons. The major one is that the needed evolutionary mechanism is not described clearly, or tested; sometimes the indicated mechanism appears incap- able of proper testing. Support for the claimed pe- riod of no selection is not provided. Although most advocates of quantum theories claim, often tacitly, that macroevolution cannot be reduced to micro- evolutionary mechanisms, they do not provide the needed support for this claim. To say that macro- evolution is or is not reducible to microevolution is not sufficient; the claim must be documented. Most 1979 BOCK— SYNTHETIC EXPLANATION OE MACROEVOLUTION 27 of the quantum theories do not consider the need to decompose major evolutionary changes into a proper chronological sequence of steps and to show how these individual steps are summarized; rather the component parts of a macroevolutionary mod- ification, if considered at all, are usually treated in some unordered fashion. Lastly, macroevolution usually involves modifications of structural features of organisms. Yet most of the discussions of quan- tum changes exclude consideration of the total bi- ology of these structural features, for example, functional and ecological morphology, and the de- tails of the complex interactions involved in the internal adjustment of somatic features. Synthetic Theories The synthetic or reductionistic explanation of macroevolutionary change is based on the postulate that all major evolutionary modifications in features and taxa are fully understandable in terms of evo- lutionary mechanisms at the microevolutionary level. No new causal evolutionary mechanisms are needed. Large scale modifications are adaptive throughout in that the entire shift is under the con- trol of natural selection arising from the external environment and acting on individuals of the evolv- ing population. Thus, evolutionary phenomena, from the smallest to the largest changes, can be explained by the same unified theory of evolution. This is the basic belief of many evolutionary biol- ogists since the early 1930’s, but close reading of most authors reveals that their statements were un- supported statements — articles of faith. Quite prob- ably these repeated unsatisfactory explanations provided the impetus for repeated formulation of alternative, usually quantum, theories. Macroevolutionary modifications in features and in taxa are those of a general magnitude character- ized by difference, expressed taxonomically, of the generic level or higher. Thus, the appearance of a new bone, a new articulation, feathers or hair, mod- ifications in the feeding apparatus to permit taking of different food and many others, would all be major evolutionary changes. As stated earlier, the minimum level of major evolutionary modification is not essential to the discussion. Most workers would place it at a level of difference greater than that observed between genera, others use a level greater than that between species, and still others would accept species level differences as the mini- mum macroevolutionary change. The synthetic ar- gument to be developed below could be applied equally well whatever level is accepted. Explanation of all macroevolutionary phenomena is fully reducible, in the strictest sense, to the known mechanisms of evolutionary change at the microevolutionary level. To be specific, these microevolutionary mechanisms are: a) Those of phyletic evolution, which are two mechanisms acting simultaneously (that is, every generation) and are namely — the production of ge- netically based phenotypic variation and natural se- lection arising from the interaction between the or- ganism and the external environment. These are the mechanisms of evolutionary change in populations, which have been studied by populational geneticists (for example, Dobzhansky, 1970, and earlier) and by animal and plant breeders. These are the evo- lutionary mechanisms that can be tested directly by experiments and by direct observations of known phyletic changes, for example, the history of breeds of dogs, pigeons, goldfish, wheat, corn and a host of other forms of domesticated plants and animals. b) The mechanisms of speciation — the multipli- cation of species — as discussed by Mayr (1942, 1963) and many other workers. Clearly, much disagreement exists on many as- pects of these evolutionary mechanisms and it would be necessary to specify exactly which concept one accepts. There are, for example, a number of con- cepts of geographical speciation (for example, Car- son’s concept of the closed variability system and its alteration during speciation), which 1 do not ac- cept. Much argument exists on the extent and role of gene flow. But these disagreements do not affect the claim about reduction of macroevolutionary ex- planation made above. Proper explanation of macroevolution events de- pends upon correct chronological summation of the individual small modification. If this summation is not done properly, then the remainder of the expla- nation will dissipate. I must emphasize that macroevolution is viewed as a sequence of microevolutionary events, not as a sequence of species level changes. The latter im- plies a distinction between evolution at the species level and evolution above the species level or transspecific evolution which I reject. Such a dis- tinction would suggest that phyletic evolution would transcend the species boundary, which is er- roneous (see below). The term “transspecific evo- lution’’ should be dropped. The entire period of major change is adaptive. 28 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 13 being under the control of natural selection at all times. No periods of inadaptiveness exist nor do any periods exist in which natural selection does not operate. However, the synthetic theory does not specify that all evolutionary change of all indi- vidual features must be adaptive. Clearly nonadap- tive evolutionary modifications of individual fea- tures occur as a result of pleiotrophic relationships between features. Such change may be quite com- mon and indeed is responsible for the evolution of a whole class of features, namely the evolution of intrinsic isolating mechanisms. The only exceptions to the generalization that phyletic evolution of a single lineage must be adaptive (under the control of natural selection) are those modifications associ- ated with founders as proposed by Mayr (1942). However, these changes would be of minor mag- nitude and would soon come under the control of natural selection. It is improbable to a vanishing degree that the whole or a major part of the phe- notypic shift in a macroevolutionary change results from genetic drift. The synthetic theory does not specify rates of evolutionary change. Certainly there is no upper limit on the rate of change other than that imposed by no rates of microevolutionary modification. And these rates of change can vary with periods of rapid modification intermeshed with periods of low change. Almost certainly rates of major change can be far faster than generally believed and the rates during the origin and early development of a major feature or new taxon are rapid, followed by much slower rates, but these ideas are not new, having been expressed by Simpson (1944). Nor am I in disagreement with the ideas of Eldredge and Gould on rates of evolutionary change expressed in their concept of punctuated equilibria. Arguments on rates of macroevolutionary change are independent to a large degree of the mechanisms of change re- sponsible for these rates. SPECIES AND PHYLETIC LINEAGES The species and its evolution is central to the dispute between the several major theories of mac- roevolution. I will consider only species in sexually reproducing organisms and accept the biological species concept as advocated by Mayr (1942, 1963:19). A species is composed of groups of ac- tually or potentially interbreeding natural popula- tions, which are reproductively isolated from other such groups. The reproductive gap between species is central and related to preservation of coadaptive complexes of genes. Thus, the species is a genetical unit and an ecological unit with its members forming a reproductive community that shares a common environment. Intrinsic isolating mecha- nisms serve to preserve the integrity of each species with respect to other sympatric species. The species concept is most objective as a non- dimensional rather than a multidimensional concept (Mayr, 1963:17-19). Thus, sympatric species are most objective with clear-cut limits and separations from one another. As one progresses geographically and chronologically further and further away from a single point, the species distinctiveness becomes more and more vague. Hence, sympatric species are distinct, but allopatric forms become increas- ingly indistinct with the extreme case being ring species in which the two terminal forms overlap and coexist as species without interbreeding. The unity of the species, which is held together by a common gene pool (and gene flow) and by a similar ecolog- ical interrelationship, breaks down as the popula- tions are separated by greater and greater geograph- ical distances and more and more ecogeographical barriers (Eig. 1). A similar breakdown occurs in the unity of the species as one traces it chronologically, generation by generation both forwards and back- wards, from a particular point in time. A species comprised of a series of interbreeding populations today is simply not the same as its ancestor 100 generations ago. The biological species concept is a nondimen- sional one, but is often applied multidimensionally over a broader geographical space and over a longer temporal period for practical purposes. Thus, I re- ject the concept of phyletic species except for prac- tical uses in paleontology. The use of the species concept by most phylogenetic systematists as the phyletic segment from one speciation (splitting point) to the next is simply not the same as the biological species concept. A phyletic lineage is the temporal continuum formed by a species (a group of actually or poten- tially interbreeding populations) reproducing itself generation by generation through time (Fig. 2). A phyletic lineage may remain as a single lineage over long periods of time or it may split into two or more 1979 BOCK— SYNTHETIC EXPLANATION OE MACROEVOLUTION 29 Fig. I. — Schematic diagram to illustrate the geographic range of a species with several large main populations and a number of small isolates. Some of the peripheral populations were formed by dispersion (indicated by arrows) and perhaps developed from a few founder individuals. Other populations resulted from contraction of the species. Gene flow within populations is indicated by the criss- crossing arrows. different lineages (= speciate) from time to time. The phenotypic characteristics of the members of a phyletic lineage may remain the same for long periods of time or they may change with respect to time (= phyletic evolution). In any case, whether the phyletic lineage remains single or splits or whether it remains unchanged or modifies through time, no species limits exist between any temporal segments of a phyletic lineage. No matter how much phyletic evolution occurs in a phyletic lineage and no matter how different ancestral and descen- dent populations may appear, no species bound- aries will be crossed as one traces a phyletic lin- eage; hence, transspecific evolution has not occurred. A cross-section of a phyletic lineage at any point in time is a species. However, cross-sections of the same phyletic lineage at different points in time are not different species nor are they the same species. These are simply different cross-sections of the same phyletic lineage at different times; one would he ancestral to the chronologically later one. The phyletic lineage is what is usually implied when the term phyletic species is used. 1 advocate the former term because it avoids confusion with the concept of species and because the relationship between the phyletic lineage and the species is clear. Because the biological species concept is a non- dimensional one, it is not possible to speak of the age of a species, or of the origin of a species, or of the life and death of a species. It is meaningless to speak of evolution within the limits of a species and to contrast this mode of evolution with transspecific evolution or with evolution beyond the bounds of a species. 30 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 13 MECHANISMS OE EVOLUTIONARY CHANGE It is possible to speak of many different types of evolutionary change and to formulate many mech- anisms of evolution. These certainly exist and must be detailed if we are to comprehend the full scope of evolutionary biology. Yet it is possible to sum- marize all evolutionary change into two major types and to correlate these with the concepts of the phy- letic lineage and of the species just discussed. Phyletic Evolution Phyletic evolution is change in a phyletic lineage with respect to time (Eig. 2). No mention should be made of a minimum time limit because this would become arbitrary. Thus phyletic evolution could be the change seen from one generation to the next. Only modifications that occur in an individual dur- ing its lifetime should be excluded from evolution- ary change. Phyletic evolution does not have to be Fig. 2. — Schematic diagram to illustrate the concept of a phyletic lineage, which is a species reproducing itself generation after- generation through time. Change in the species with respect to time is phyletic evolution. A cross-section through the phyletic lineage at any point in time is a species. Cross-sections at dif- ferent points in time are neither the same i?pecies nor different species (no species boundary separates them) but simply differ- ent cross-sections of the same phyletic lineage. specified as hereditary change because this would preclude labeling many modifications (for example, those observed in the fossil record) as evolutionary because it would not be possible to demonstrate that they are hereditary. In phyletic evolution, a descendent cross-section of a phyletic lineage would differ from an ancestral cross-section. Phyletic evolution can occur without speciation (splitting of the phyletic lineage) and, at least in theory, it would be possible to have drastic modification in the characteristics of members of a phyletic lineage by phyletic evolution without any speciation. The mechanisms by which phyletic evolution oc- cur can be summarized into two types — namely: a) the formation of genetically based phenotypic vari- ation generation by generation; and, b) the action of natural selection arising from the interaction be- tween the individual organisms and their external environment. The formation of genetically based phenotypic variations is the accidental or chance based factor in phyletic evolution (Mayr, 1962). This genetically based individual variation in a population results from a number of mechanisms of which genetical recombination of all types (crossing-over, inver- sion, translocation, segregation) is the most impor- tant. These are the mechanisms which produce ge- netic combinations from the existing genetic material in the gene pool. Gene flow is the next most important mechanism, but far less important than recombination. It results in new genetic ma- terial in the population, but not new genetic mate- rial in the species. Mutations are the least important source for the production of generation by genera- tion genetic variation; it is the source of new genetic material in the phyletic lineage. Natural selection is the design factor in evolution and results from the interaction of individual organ- isms with their external environment. I am here concerned with the mechanism of natural selection not with its result of changes in gene frequencies in the gene pool (the usual definition of natural se- lection in population genetics). Selection can only result from the action of the external environment on the individual. It cannot arise from the "internal environment" or from the "genetic environment” nor can one speak about a distinct and separate form of "internal selection." Natural selection acts only on the phenotypes of individual organisms and 1979 BOCK— SYNTHETIC EXPLANATION OE MACROEVOLUTION 31 (U O’ c o x: O >^ w O c o o > Ll) c o £ < Te Ts T4 T3 T2 Ti Fig. 3. — Schematic diagram to illustrate that populations of the same species can undergo different amounts of phyletic evolution (indicated by the differential points on the vertical arrows) during the same time period. can distinguish only between varying phenotypes. The concept of selection is closely associated with that of adaptation (see below). Both mechanisms of the production of genetically based phenotypic variation and of natural selection are required to have phyletic evolution. Both op- erate generation by generation and only in the pres- ent. Evolutionary change is the result of a chrono- logical summation of the generation by generation simultaneous action of these mechanisms, not an average over a long period. Both mechanisms must be considered simultaneously as the creative mech- anism of evolution, neither one or the other is suf- ficient by itself. The rate of phyletic evolutionary change depends upon the combined action of these two factors. Rapid evolution cannot occur by mechanisms pro- ducing genetically based phenotypic variation alone or by strong directional selection by itself. Of spe- cial interest in explanation of macroevolution is the source of the needed directional selection. It is possible to have varying rates of phyletic evolution in different populations of the same species (Eig. 3). The result will be variation in the amount of evolution in the sublineages of the same phyletic lineage. Speciation Speciation is the multiplication of species or the splitting of an original phyletic lineage into two or more lineages (Fig. 4). Speciation can only occur with the accompanying phyletic evolution in, at least, one of the two separated phyletic lineages. The essential aspect of speciation is the mechanism whereby intrinsic isolating mechanisms evolve in the two newly split lineages at a time when mem- bers of the two lineages would still be able to in- terbreed with one another. Speciation, as I will use it, is only the multipli- cation of species. It is not phyletic speciation, which is a misnomer for phyletic evolution. One of the major sources of confusion is: What is the fundamental aspect of the mechanism of spe- ciation? Most workers are not clear about this and have confused many aspects of phyletic evolution in the notion of speciation. Is it the evolution of 32 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 13 Fig. 4. — Schematic diagram to illustrate the relationship between speciation (splitting of a phyletic lineage into two lineages) and phyletic evolution. Speciation requires the presence of an ex- ternal isolating barrier and of phyletic evolution in at least one lineage. Species A and B are distinct and separate from one another, but each is not different from the ancestral species com- mon to both phyletic lineages. intrinsic isolating mechanism? Is it the evolution of ecological, behavioral, and other differences be- tween species that permit them to coexist? Is it the evolution of genetic differences that distinguish species? Moreover, most workers are not clear about the onset and about the completion of the spe- ciation mechanism. Is speciation over when the two newly evolved species become sympatric without gene flow between them? Or does speciation con- tinue for some time after the new species become sympatric? I will accept the view that the essential charac- teristic of speciation is the evolution of intrinsic isolating mechanisms and of the ecological, behav- ioral, and other differences that permit the newly evolved species to coexist ecologically and without interbreeding. Clearly, speciation involves phyletic evolution, hut I do not regard phyletic evolution as synonymous with speciation. Moreover, it is clear that the pattern and rate of phyletic evolution may differ in a small peripheral isolate as compared to a large central population or in a small isolate formed by a few founder individuals as compared to one formed by remnants left by a contracting species (Mayr, 1954), but these factors are aspects of phyletic evolution contributing to speciation, not speciation itself. Failure to separate these evolu- tionary mechanisms and to specify exactly what is meant by speciation has led to much confusion in macroevolutionary explanation. The onset of speciation frequently occurs prior to the appearance of the external ecogeographical barrier and continues long after this barrier disap- pears and the two species are able to reinvade each others’ range and coexist sympatrically (Fig. 5). Thus, I will separate speciation into two portions — the allopatric period and the neosympatric period. Populations of a species may start to diverge be- fore the appearance of an external barrier that splits the phyletic lineage into two separate sublineages. Hence at the onset of the allopatric phase of spe- ciation the two sublineages may or may not be dif- ferent from one another. The ecogeographical bar- rier (I will consider only allopatric or geographic speciation) splits the original single phyletic lineage into two and prevents members of the two sublin- eages from interbreeding during a period in which they could do so. The geographical barrier prevents gene flow between the two populations representing the split phyletic lineages. During the allopatric phase, the two populations will undergo separate phyletic evolution and will start to diverge from one another because each lineage is under the control of a different pattern of formation of genetically based phenotypical variation and of natural selec- tion (Fig. 6). The rate of change in each population will depend upon a number of factors, including the size of the population, whether it was founded by a few individuals, and the nature of the environment and hence selection. Intrinsic isolating mechanisms may evolve during this period. If so, these isolating mechanisms appear fortuitously, both in their na- ture and time of appearance, as a pleiotrophic con- sequence of other evolutionary changes. Evolution of intrinsic isolating mechanisms is not under the control of selection favoring the evolution of iso- lating mechanisms. During the allopatric phase of speciation, a cer- tain and quite variable amount of divergence occurs between the two lineages. Generally, the amount of evolutionary divergence that occurs during this pe- riod is a minor amount of the total divergence be- tween two sympatric and fully evolved species. 1979 BOCK— SYNTHETIC EXPLANATION OF MACROEVOLUTION 33 Fig. 5. — Schematic diagram to show the relationships of populations during speciation. The ancestral population (A) may have a period of subspeciation (B) before the appearance of a geographical-ecological barrier that separates two populations (C); this is the start of the aliopatric phase of speciation. At this time no gene flow exists between the isolated populations (A and C). After the external barrier disappears, the two species are able to reinvade the geographic range of each other and to coexist if intrinsic isolating mechanisms exist (D); this is the start of the neosympatric phase of speciation. If sufficient ecological differences evolve between the species, geographic overlap can continue until the two species are broadly sympatric (E). 34 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 13 Fig. 6. — Schematic diagram to show divergence of the two phyletic lineages during speciation. The stages indicated by letters along the left edge correspond to those in Fig. Rates of evolutionary change and resulting divergence are low during the period of subspeciation and somewhat higher during the allopatric phase after the appearance of the geographical ecological barrier. The rate of divergence increases sharply after the geographic barrier disappears and the two species become sympatric. Selection forcing the divergence during the sympatric phase arises from exclusionary species interactions between the two species. Speciation comes to an end when this period of rapid divergence terminates. 1979 BOCK— SYNTHETIC EXPLANATION OE MACROEVOLUTION 35 When the geographical barrier disappears, the two populations can expand their range and become sympatric as species if two conditions are met. The first is that the intrinsic isolating mechanisms evolved during the allopatric phase are 100% effec- tive. This means that there is no gene flow between the two forms even if there is some hybridization. The second is that the two forms are sufficiently different ecologically that they are able to coexist. They need not be completely different or even largely different ecologically. All that is essential is that they differ somewhat so that each can invade the range of the other. At the time of the breakdown of the external geo- graphic barrier and the establishment of the initial overlap between the two species, the allopatric pe- riod ends and the sympatric or neosympatric period begins. This is not the end of the speciation process although most evolutionary biologists end their dis- cussion at this point. For example, Mayr ( 1963) has almost no discussion of the events during the neo- sympatric period of speciation, which leaves a ma- jor gap between the genetics and ecology of specia- tion and the role of the species in transspecific evolution in his treatment of species and their evo- lution. After the two newly evolved species become sympatric, they are able to interact with one another and thereby exert strong mutual selection on one another. The exclusionary species interac- tion (Bock, 1972) of the now sympatric species are of two types, namely (a) ecological competition, and (b) reproductive interference. The first type of interaction results from the fact that newly evolved sister species are frequently similar ecologically and still share many parts of their environment (see Lack, 1944, 1947, and elsewhere). The resulting competition will result in mutual selection on both species, which generally results in divergence of the feeding and other structures associated with the ecological competition. The second type of inter- action results from the nature of the intrinsic iso- lating mechanism. When the two species overlap geographically, they must have 100% efficient in- trinsic isolating mechanisms or else gene flow will commence between the two forms, which will lead to a breakdown of the distinctiveness of the two species. (1 reject the notion that the isolating mech- anisms can be less than 100% effective and that selection during the neosympatric period can im- prove these isolating mechanisms from less than 100% to 100% effective.) Yet intrinsic isolating mechanisms can be 100% effective and vary greatly in their reproductive cost. By reproduction cost, I mean the percentage of a particular breeding period that an individual wastes because it attempts to breed with a member of another species. Such a wastage of time will reduce the number of offspring that individual could have. If one examines the clas- sification of isolating mechanisms presented by Mayr (1963:92), these mechanisms are arranged from high reproductive cost at the bottom of the list (F, hybrid zygote fully viable, but sterile) to low reproductive cost at the top (seasonal, habitat, and ethological isolation). Selection will favor isolating mechanisms of lower reproductive cost; hence dur- ing the neosympatric period selection arising from species interaction would select for new isolating mechanisms of lower reproductive cost than those existing at the time of initial sympatry. Although this selection is for isolating mechanisms, it is not to improve the efficiency of the isolating mechanism but to reduce reproductive cost. Evolution is from an isolating mechanism that is 100% effective but has high reproductive cost to one that is 100% ef- fective but has low reproductive cost. There has been no improvement in the isolating mechanism, but in reproductive cost. This selection will favor the evolution of courtship displays, more elaborate species-specific recognition characters (color, horns, plumes, song, and others) and temporal separation of breeding seasons. Features acted upon by the mutual selection from the two species under exclusionary species inter- action will diverge rapidly and considerably. These are often features associated with feeding (ecolog- ical competition) and with species specific recog- nition (reproductive cost). They are generally those features which distinguish sympatric species most readily especially in contrast to allopatric, closely related species. Selection arising from these exclu- sionary interactions are frequently the strongest known natural selection and presumably cause the most rapid evolutionary change. 1 have postulated (Bock, 1972) that most of the divergence between sister species results from the mutual selection aris- ing from exclusionary species interaction during the neosympatric period of speciation. This is after the two sister species are able to reinvade each other’s range after the breakdown of the geographic barrier and after the perfection of the intrinsic isolating mechanism (100% effective). This assertion is in direct variance with that made in some quantum theories (for example, rectilinear model of macro- 36 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 13 evolution) which assumes that the major change occurs prior to the breakdown of the external bar- rier and the onset of sympatry of the sister species. It can be tested by comparing the amount of diver- gence observed between allopatric sister species and fully sympatric sister species in closely related taxa (that is, members of the same genus or closely related genera). A rough survey of such species in birds supports the hypothesis that most evolution- ary divergence between sister species takes place during the sympatric portion of speciation. Speciation would be terminated when the exclu- sionary species interaction and the resultant mutual selection force between the sister species ap- proaches the zero level. Quite possibly one or both of the phyletic lineages of these sister species could have split again and entered a new cycle of specia- tion before the termination of the original cycle. Speciation and Macroevolution Macroevolution is simply a large amount of phy- letic evolutionary change. It is a summation of a number of microevolutionary phyletic events, often in a relatively short time. Speciation, per se, has nothing to do with major evolutionary change if one regards speciation as the mechanisms permitting the multiplication of species — evolution of intrinsic isolating mechanisms. Yet the phenomenon of spe- ciation, especially repeated speciations, is an im- portant factor in macroevolution because the selec- tion forces arising from species interactions during the neosympatric phase of speciation is an impor- tant driving force in macroevolution (Bock, 1970, 1972). Thus the relationship between speciation and macroevolution or the role of the species in mac- roevolution is that a major driving force for macro- evolutionary evolution comes from the selection forces arising from the exclusionary species inter- actions between sister species during the neosym- patric phase of speciation (as well as between other sympatric species). Another important driving force is coevolutionary interactions for which repeated speciations are less important. Thus, the more spe- ciations (repeated cycles of speciation) and the more species which interact, both exclusionary and coevolutionary, the stronger and the longer in time will be the driving directional selection force re- quired for the phyletic evolution that will result in a major evolutionary change. Although this assertion may sound like that of punctuated equilibria or rectilinear evolution, it is very different. In the punctuated equilibria model. macroevolution is regarded to be the conse- quence of a repeated series of speciations. That spe- ciation is the essential evolutionary mechanism. Phyletic evolution, which is synonymized with phy- letic gradualism, is regarded to be insignificant to the point of being nonexistent. In the punctuated equilibria model, most of the change is assumed to occur during the allopatric period of speciation. In the synthetic model advocated here, most of the change is postulated to occur during the neosym- patric period (plus the change resulting from coevo- lutionary interaction which occurs continuously). Unity of the Genotype At this point a digression must be followed to consider the concept of the genotype and of the gene pool of the species as discussed by Mayr (1963:263-296, and elsewhere). Closely associated with this concept is that of genetic revolution (Mayr, 1954, 1959, 1963). The basic concept is the cohesion of the genotype of the individual and of the gene pool of the species. The adaptive value of individual genes is not an absolute intrinsic prop- erty of each gene, but is dependent upon “coad- aptive” interaction between the gene and the re- mainder of the genotype. The adaptive value of the gene varies according to the genotype. Cohesion of the gene pool of a population or the interbreeding populations of a species is dependent upon the mul- tiple patterns of gene flow throughout the species (Mayr, 1954, 1959, 1963). It is the phenomenon of gene flow that is the critical factor. It should be noted that many recent evolutionists reject gene flow as an important evolutionary factor, including some workers who accept fully Mayr’s concept of genetic revolution. Mayr argues that elimination of gene flow result- ing from isolation of a population will have a great effect on the consequence of selection on that pop- ulation because the genes affected by selection are acting against a different genetical background and will have a different adaptive value. The effect of the disruptive effect of isolation will be greatest for a new population established by a few founder in- dividuals. It would be next greatest for a small iso- late resulting from the appearance of a barrier that cut it off from the remainder of the species. This frequently happens when the range of the species shrinks leaving isolates in pockets of favorable hab- itat. The effect is smallest for a species with a large range that was divided into two subequal segments. In such examples, divergence would be slowest, but 1979 BOCK— SYNTHETIC EXPLANATION OF MACROEVOLUTION 37 it would still occur as shown by the many examples of eastern and western species or well-marked sub- species of North American forest birds (for exam- ple, Colaptes, Contopus, Cyanocitta, Dendroica, Oporornis, Icterus, Pheiicticus, Passerina — Mayr and Short, 1970). Mayr stresses the importance of selection acting on the genetical modification resulting from the dis- ruption of the gene flow. He argues that the con- sequence of selection acting on genes whose adap- tive value has been altered because of the changed genetical background may result in a major change which he termed a genetical revolution. Such alter- ation in adaptive value of genes would be greatest in populations originating from a few founders. Car- son’s (1975) model has similar elements in that he includes a population crash that results in a few founder individuals from which the subsequent pop- ulations originate. In Mayr’s model the founders are considered as individuals that have invaded a new area. Problems exist with the concept of genetic revolution because other workers have deempha- sized the importance of selection and have argued that speciation must be associated with a genetical modification without the action of selection. The general notion is that this genetical revolution has occurred completely during the allopatric phase of speciation prior to the reestablishment of sympatry. This implies that evolutionary change as expressed in the phenotype would have occurred prior to the reestablishment of sympatry of the sister species. The concept of the unity of the genotype and of the gene pool, the concept of variable adaptive val- ue of genes depending upon the genotypic back- ground, and the concept of genetic revolution are all sound and are supported by a considerable mass of observations. However, a number of aspects of genetic revolution and its bearing on macroevolu- tion have not been discussed by Mayr or by other workers who have accepted these ideas in their de- velopment of quantum theories. One of the most important is what are the types of phenotypic features that are usually affected by genetical revolutions. Are these parts of the feeding apparatus, the locomotory system, or external fea- tures which may be associated with species-specific recognition characters? To be sure, many examples of distinctive subspecies or allopatric species exist whose evolution is well explained by the concept of genetic revolution. But the features that have been modified are usually not features characteris- tic of major evolutionary change. What is needed at this point are surveys of the features affected by genetic revolutions, not only of taxa that support this concept. The second problem is that Mayr quite rightly argues that his concept of genetical revolution places emphasis on the essential and central role of natural selection which is too often ignored. Mayr points out that the genetically based phenotypic variation in the founder population available for se- lection differs greatly from that in the main popu- lation because of the disruption of gene flow. What is not discussed is the source of this selection and the time at which this selection acts. Clearly, in many examples, this selection has acted during the allopatric period of speciation because the taxa being compared are still allopatric. But, one must examine the features modified and consider the pos- sible environmental source of the selection. Disrup- tion of gene flow and evolutionary change during the allopatric period may provide the needed ge- netically based phenotypic variation for the change during the sympatric phase. Selection may arise largely from species interactions, both exclusionary (from forms other than the sister species) and co- evolutionary, during the allopatric period as noted by Mayr (1954) who points out that particularly the biotic environment of an isolate may differ from the rest of the species. Nothing in the concepts of dis- ruption of gene flow and of genetical revolution as advocated by Mayr would conflict with the concept that the major selection forces arise from species interactions and that most of this selection acts dur- ing the sympatric phase of speciation after the in- trinsic isolating mechanisms are completely effec- tive. COMPARISON Most of the testing of models of macroevolution- ary explanation is by comparison and interpretation of characteristics of different forms. Moreover, one of the important conceptual steps in the synthetic model is dependent upon proper interpretation of comparison and the extrapolation from one type of comparison to another. This conceptual step is not limited to the synthetic model, but exists in all ex- 38 BULLETIN CARNEGIE MUSEUM OE NATURAL HISTORY NO. 13 Horizontal Paradaptive differences Fig. 7. — Schematic diagram to show the difference between hor- izontal comparisons and vertical comparisons. Vertical compar- isons are those between members of the same phyletic lineage, for example, between A and B, or A and C, or A and D, or E and F, along the time axis. Horizontal comparisons are those between members of different phyletic lineages, for example, between members of lineages A-B, A-C, A-D, and E-F, no matter if the forms being compared are or are not at the same time level (taken from Bock, 1967:Fig. 1). Fig. 8. — Schematic diagram to show the pattern of multiple path- ways of evolution of perching feet in birds. Evolution of the four different arrangements of the toes from the ancestral configu- ration was under the control of the same selection force for a more efficient perching foot. Differences observed in vertical comparisons are adaptive, whereas horizontal differences are paradaptive with respect to the selection force controlling the evolution of these adaptations (taken from Bock, 1967:Fig. 2). planatory models of macroevolution. Little discus- sion exists on principles of biological comparison, especially on the theoretical level; I will refer main- ly to comments in my earlier papers (Bock, 1967, 1969, 1977). Not all comparisons in biology are the same and the interpretations reached on the basis of a partic- ular comparison cannot be extrapolated simply to all others. Not all comparisons are between mem- bers of different species, excluding the case of com- parison between conspecific individuals. Although comparisons can be made for many diverse pur- poses and with many goals in mind, they can be divided into two major categories — horizontal and vertical (Eig. 7). This dichotomy does not exhaust the possible classifications of types of comparison; it is one that is of particular relevance to the de- velopment of macroevolutionary theories. Horizontal comparisons are those across phyletic lineages — between members of different phyletic lineages and hence, between members of different species. They can be between species at the same point in time or at different points in time so long as they are between different phyletic lineages. Comparisons between conspecific individuals at the same time period would be horizontal. Vertical comparisons are those within a phyletic lineage — between members of the same phyletic lin- eage at different points in time. These are compar- isons between ancestral and descendent cross-sec- tions of the same phyletic lineage and hence, are not between different species. It is not possible to extrapolate simply from hor- izontal comparisons to vertical comparisons. Hor- izontal similarities are not the same as vertical sim- ilarities and horizontal differences are not equal to vertical differences. The origins of these two types of similarities and of these two types of differences may be quite different and hence will require di- verse explanations. Eor example, differences ob- served in a horizontal classification may have noth- ing to do with adaptation and are designated as paradaptive (Bock, 1967; and see below) yet the vertical difference in each phyletic lineage may be adaptive. The conclusion offered earlier that the evolution of features is not adaptive if their hori- zontal differences are not adaptive is simply in error. The difficulties in extrapolating interpretations from a horizontal to a vertical comparison and vice versa is the consequence of the simultaneous action of two evolutionary mechanisms in phyletic evolu- tion. These are, of course, the production of genet- ically based phenotypic variation, which is chance- based, and the action of natural selection, which is a design mechanism. These mechanisms act in the phyletic evolution of every lineage. Hence, given 1979 BOCK— SYNTHETIC EXPLANATION OF MACROEVOLUTION 39 the existence of a particular selection force acting on a species, it is not possible to predict the future course of phyletic evolution and the resulting ad- aptation, if any, because it is not possible to predict the outcome of the chance-based genetical mecha- nisms. The concept of multiple pathways of evolution (or adaption) stems directly from the action of these two evolutionary mechanisms and their conse- quences as interpreted in horizontal and vertical comparisons (Bock, 1959; Bock and deW. Miller, 1959). Different adaptive answers (Fig. 8) may ap- pear and evolve in several lineages under the con- trol of the same selection force as shown by the evolution of different perching foot types in birds (Bock and deW. Miller, 1959). Associated with this idea is the concept of paradaptation (Bock, 1967). Paradaptive differences are ones between different multiple adaptive answers and are the consequence of the chance-based genetical mechanisms (Fig. 8). The conceptual step, which will be necessary in the development of the synthetic explanation of macroevolution, is to formulate a (pseudo)phylogeny of steps leading to a major change using horizontal comparisons of closely related forms (for example, congeneric species) and to interpret the differences as adaptive steps. Then this horizontal sequence must be transposed to a vertical sequence which is the key conceptual step. This is one that is difficult to support by empirical observations as one re- quires known phylogenies (that is, those of domes- ticated animals and plants). Moreover, it is a con- ceptual step that falls in the realm of a link or bridge between theories. A worker is free to reject or ac- cept it. However, if rejected, then much of the basis for developing and testing explanatory models, both quantum and synthetic, of macroevolutionary change is eliminated. In developing the synthetic model and in dis- cussing supporting examples, 1 will give particular attention to this conceptual jump between horizon- tal and vertical comparisons. ADAPTATION The key to all explanatory models of macroevo- lution is the concept of biological adaptation. What is an adaptation and how are individual adaptations ascertained? How do adaptive features evolve? What is meant by adaptive evolution of features and by adaptive evolution of a population under the overall notion of adaptive phyletic evolution? Is the adaptive evolution involved in the origin of a new major feature or of a new major taxon (adaptive radiation) different qualitatively from adaptive evo- lution on the microlevel? The concept of biological adaptation has always been used to designate features of an organism, which operate well in the particular environment of that organism. Hence wings of most birds are ad- aptations for aerial flight, whereas the wings of pen- guins are adaptations for underwater flight. The concept of adaptation long predates ideas of biolog- ical evolution and indeed the attempt to provide a scientific explanation for adaptation led to the for- mulation by Darwin of the concept of organic evo- lution by natural selection. An adaptation is a fea- ture of the organism. Individual features or complexes of features are adapted to particular components of the organism’s environment. It is almost of no interest to inquire whether a whole organism is adapted to its environment — it must be. otherwise it would be dead. The questions of inter- est are what is the adaptive significance of individ- ual features and how each adapted feature contrib- utes to the survival or to the fitness of the organism. An adaptation is, thus, a feature of the organism, which interacts operationally with some factor of its environment so that the individual survives and reproduces. Stress is placed on the organism sur- viving as an individual because it cannot otherwise reproduce. However, adaptations cannot be judged only with respect to survival of the individual; it must survive and reproduce. Adaptations must be judged with respect to a particular environment and always on a probability basis with respect to pres- ent (and possibly past) environmental conditions, but never against future factors. The environment is the external environment, be it biotic or physical. Hence the concept of adaptation is defined and in- dividual adaptations are judged with respect to se- lection forces arising from the external environment and acting on the organism. Adaptation does not designate operational relationships between parts of the organism or an operational relationship of a fea- ture to the “internal environment.’’ Notions such as “the internal environment,’’ or “the genetical environment,’’ or “internal selection’’ are mislead- ing to the extreme and should be abandoned. Mus- 40 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 13 ORGANISM ENVIRONMENT Feature = the Adaptation —..^depends upon Umwelt Form Faculty- Function) ^ o o w O o Biological Synerg Selection Role Force determines Environmental Factor . Species diversity in the Phanerozoic: an inter- pretation. Paleobiology 2:289-297. Saunders, H. L. 1968. Marine benthic diversity: a compara- tive study. Amer. Nat., 102:243-282. ScHOPF, T. J. M. 1974. Permo-Triassic extinctions: relation to sea-floor spreading. J. Geol., 82:129-143. Sepkoski, J. J., Jr. 1976. Species diversity in the Phanerozoic: species-area effects. Paleobiology, 2:298-303. SiMBERLOFF, D. S. 1972. Properties of the rarefaction diversity measurement. Amer. Nat., 106:414^18. . 1974. Permo-Triassic extinctions: effects of area and biotic equilibrium. J. Geol., 82:267-274. Simpson, G. G. I960. The history of life. Pp. 1 17-180, in Evo- lution after Darwin (S. Tax, ed.), Univ. Chicago Press, I: viii -I- 1-629. SoRGENFREi, T. 1958. Molluscan assemblages from marine Middle Miocene of South Jutland and their environments. Geol. Survey Denmark, 2nd ser., 79:1-503. Valentine, J. W. 1973. Evolutionary paleoecology of the marine biosphere. Prentice-Hall, Inc., Englewood Cliffs, 511 pp. Van Valen, L. I973u. A new evolutionary law. Evol. Theory, 1:1-30. . 19736. Are categories in different phyla comparable? Taxon, 22:333-373. TETRAPOD MONOPHYLY: A PHYLOGENETIC ANALYSIS Eugene S. Gaffney Department of Vertebrate Paleontology, The American Museum of Natural History, Central Park West at 79th Street, New York, New York 10024, and Department of Geological Sciences, Columbia University, New York, New York INTRODUCTION “What we really need is more fossils.” This oft- repeated statement usually follows, both in print and in meetings, arguments concerning the relation- ships of primitive tetrapods. But we do have more fossils; in fact, we have more fossils than ever be- fore and, because of new discoveries and newly prepared “old” specimens, we have over the past fifty years conspicuously enlarged the diversity of extinct forms as well as increased our knowledge of their structure. Yet Panchen (1975:290) writes “. . . the story of the early evolution of tetrapod vertebrae is certainly less simple and less certain than it appeared to be even less than ten years ago . . . .” Although his comment is directed toward one particular morphologic region, I think it char- acterizes feelings about tetrapod phylogeny in gen- eral. For example, Olson (1971:583) in a review of early tetrapod evolution stated ”... the use of as- sumed phylogenies as bases for deductions con- cerning almost all of the problems of tetrapod origins have introduced such a mixture of objective and subjective analyses of data that dispassionate assessments are difficult and too infrequently en- countered.” Olson lays the blame on “assumed phylogenies” but I will argue here that the general lack of success in developing objective ideas about relationships is due to basic problems in method- ology. Tattersall and Eldredge (1977) analyze a similar situation in studies of hominid evolution. They sug- gest that ”... phylogenetic hypotheses can be for- mulated at three different levels of complexity, each successively further removed from the basic data available” (p. 204). The first, and most objective, is the cladogram, a methodology advocated here and explained below. The second level, called phy- logenetic trees by Nelson (manuscript) and Tatter- sall and Eldredge (1977) involves the recognition of ancestor-descendant sequences, even though this requires the addition of untestable assumptions and suppositions. The third and least objective type of hypothesis is the scenario, which is essentially a tree further encumbered by assumptions and spec- ulations about selective forces, niches, key inno- vations, radiations, adaptive zones, and other no- tions that have become the hallmark of explanation couched in the jargon of the Synthetic Theory. The elements of a cladogram are usually implicit in a scenario and might be abstracted from it, but “. . . as things stand, the diverse components of scena- rios are seldom separable, and much of the reason- ing that goes into their construction is circular: the many elements involved feed back upon each other in an extremely intricate way” (Tattersall and Eld- redge, 1977:205). “There is no methodology at all for the formulation of a scenario, with all its varied aspects of evolutionary relationship, time, adapta- tion, ecology, and so forth. In devising a scenario one is limited only by the bounds of one’s imagi- nation and by the credulity of one’s audience . . .” (Tattersall and Eldredge, 1977:207). My own examination of literature on the “fish- amphibian transition” suggests to me that nearly all hypotheses concerning this subject have been for- mulated at the scenario level with little or no effort to separate adaptive speculation from testable phy- logenetic hypotheses. Representative examples of scenarios are Szarski (1962), Schaeffer (1965), and Thomson ( 1966c?)- This, rather than the use of “as- sumed phylogenies,” is the source of the problem referred to by Olson above (1971:583). It is impos- sible to choose objectively among contradictory scenarios and the result of difficult and tedious work in exploration for and preparation of fossils is often simply the generation of new scenarios. It is my feeling that the prominence of scenario creation has had a particularly subtle effect even on critical workers not easily swayed by ad hoc hy- potheses. For example, Carroll (1969:427), whose work is clearly oriented towards phylogeny recon- struction per se rather than adaptive scenarios of the sort seen in Olson (1966, 1976), nonetheless de- scribes the standard “parallelism scenario” even though he does not accept it: “From our knowledge 92 1979 GAFFNEY— TETRAPOD MONOPHYLY 93 of the reptilian-mammalian transition, we know that extensive parallel development of very similar structures and character complexes may occur in- dependently in numerous lineages. It is conceivable that limnoscelids, solenodonsaurids and romeriids evolved independently from a group with the anat- omy similar to that of Gephyrostegus. Similar se- lective pressure could produce the reptilian condi- tion in each. In the absence of any evidence to the contrary, solenodonsaurids and limnoscelids may be included as reptiles since they have developed a typically reptilian palatal structure and atlas-axis complex.” This insistence on the ad hoc recognition of par- allelism or convergence is unfortunate because it requires information that cannot be obtained in an objective system. It would be wonderful to have testable hypotheses of “selection pressure” and “adaptive zones” concerning Devonian tetrapods and to be able to relate these to morphologic pa- rameters. Ideas of this sort must be relegated to the “scenario” level of explanation and should not be used as tests of phylogenies. In fact, the only way to argue for the presence of parallelism or conver- gence is the demonstration that two alternative phy- logenies are still viable after testing. As Cracraft (1972:387) has said: “A judgement of convergence must be based on an a priori assumption of rela- tionship. The fact that two taxa . . . might show a number of morphological differences cannot serve as an argument for nonrelationship. The latter can only be proposed once a relationship has been dem- onstrated between one of these taxa and a third taxon.” An argument for parallelism or conver- gence is simply an argument (often unstated) in fa- vor of one phylogeny over an alternative one. Another aspect of scenario creation particularly evident in early tetrapod studies is the “derivation- ist” approach to character distributions. Because of the emphasis on ancestor recognition rather than the testing of monophyletic groups, much effort has been put into how the rhipidistian fin (or vertebral column, skull, or whatever) might be changed into a tetrapod foot. This sort of argument was probably of significance in the early days of evolution versus creation debates when series of recent forms or fos- sils were arranged with hypothetical intermediates in order to show how changes might take place be- tween greatly divergent morphologies. However, this has become an argument for homology of struc- tures, and, therefore, phylogenetic relationship. 1 think a more empirical approach, clearly related to a philosophic framework of science within which ideas may be criticized and judged, will lead to greater progress in understanding the history of or- ganisms. Such an approach begins with phylogeny reconstruction of the sort advocated here, but this is not to say that scenarios have no place in pa- leontology. We will always be interested in mech- anisms and processes of change as well as the ge- ometry of it, and even though 1 see little hope of developing testable hypotheses of this sort in the near future, it is truly impossible to test hypotheses that are not formulated at all. My intention in this paper is to take two hypoth- eses that although commonly accepted often create considerable debate, and test them using shared derived characters. In this procedure 1 hope to point out problem areas and alternative views also worthy of test. I also hope that this treatment will serve as a basis for further studies of tetrapod re- lationships. It should not be construed that I think this study in any way “settles the question” of tet- rapod monophyly nor necessarily provides a com- plete review of all the characters pertinent to this hypothesis. 1 have used all the characters that in my opinion are pertinent but this is based primarily on a study of the literature. I hope that new studies of specimens will suggest the use of new characters that can be used to test monophyly of the Tetra- poda. METHODOLOGY There are currently a number of competing methods of phy- logeny reconstruction, and I have based my choice of method on a particular view of the philosophy of science. Some system- atists argue that philosophic discussions are fruitless, a waste of time, and irrelevant; that the interesting and important job is to “do the work." There is a growing feeling among systematists, however, that objective choices can be made but that they are dependent on a basic philosophy of science that sets limits and objectives. A philosophy of science is like any other intellectual subject matter; a certain amount of thoughtful inquiry is essential to its acquisition. Perhaps the most common view of science, held by both sci- entists and non-scientists alike, is the Baconian view, or induc- tion. In this philosophy, science is supposed to be a fact gath- ering operation, which proceeds until a theory or generalization emerges from the accumulations and finally a proving or dem- 94 BULLETIN CARNEGIE MUSEUM OF NATURAL HISTORY NO. 13 onstrating experiment or observation proves the truth of the gen- eralization. It may then take its place in the firmament of ac- quired knowledge and the seeker of truth moves on to the frontiers of the unknown to gather new data. This process is usually called induction and it has been criticized philosophically at least since Hume’s day, and more recently scientists such as Einstein and Medawar (Magee, 1973) have argued that in reality induction does not describe the manner in which the acquisition of knowledge progresses. In phylogeny reconstruction, I am continually impressed by the falseness of the inductive method when workers state that there are not enough fossils to “prove” a phylogeny, or that phylogeny cannot be done because of an “incomplete" fossil record, both required elements of the in- ductive approach. One of the primary difficulties with the inductive or Baconian philosophy of science is that it allows or even requires the rec- ognition of truth. Most scientists are too skeptical to accept this, and even though they may use induction, they will argue that their conclusions are expressable only in terms of probability. One conclusion is said to be more probable than the next or with increasingly greater quantities of data, one’s conclusions become more probable. But how can “more versus less” probable be dealt with objectively'? How can one identify a “more probable” conclusion unless one knows the yardstick of comparison, that is, the “true” conclusion? The problem, as has been pointed out by philosophers for some time, is that no matter how many observations have been made that are consistent with a generalization, another obser- vation that is inconsistent with the generalization is always pos- sible. In other words, no number of consistent observations can prove the truth of a generalization. An alternative view of sci- ence and the acquisition of knowledge is the hypothetico-deduc- tive method, or, as Karl Popper, a noted exponent of this phi- losophy has characterized it, conjectures and refutations. This philosophy recognizes that although we may have found the truth in our hypotheses and generalizations, we cannot identify it. We can, however, identify hypotheses or ideas that are in- consistent with observations, and this is the focal point of Pop- per’s view of science. Although we can never know when we have found the truth, we do know when we are wrong. In this philosophy, all ideas and explanations in science are advanced us improvable hypotheses that are intended to be submitted to rigorous attempts at falsification. Falsification may be defined as the result of a test of a hypothesis in which one (or more) of the expectations (predictions) of the hypothesis is shown to be in- consistent with observations. Observation, in this sense, refers to a lower level hypothesis (that is, a more specific or less general hypothesis) that is not itself being questioned or analyzed at this time but that is susceptible to test. Those hypotheses that sur- vive repeated attempts at falsification are the most useful for further work. Nothing, however, is permanently removed from criticism, nor accepted as true. Karl Popper (1968«, l968/>; see also Magee, 1973, 1974) has been most responsible for a modern development of the hypo- thetico-deductive view of science, hut the method of advancing ideas or conjectures and then attempting to test them by exper- iment and observation is an old one. The application of the hy- pothetico-deductive method to systematics is relatively recent, at least in a formalized explicit sense, although Ghiselin (1969) has argued that it was essentially the method used by Darwin. In any case, some recent systematic work has emphasized the use of the hypothetico-deductive philosophy (Bock, 1973; Bonde, 1974, 1975; Miles, 1973, 1975; Platnick and Gaffney, 1977; Wiley, 1975). Because Wiley ( 1976) has an excellent review of the method of phylogenetic reconstruction used here, only a short summary is given below. Hypothesis The sirr.plest hypothesis of relationship is the statement — two taxa are more closely related to each other than either is to a third. An equivalent statement would be — two taxa have an ancestor in common that neither has in common with a third taxon. Much of the confusion about cladism can be dispelled by understanding this statement of the hypothesis. A three-taxon statement of this sort is best expressed in the form of a diagram, called a cladogram. Cladograms do not indicate the positions of ancestral taxa, even though they may be present, and each line does not necessarily imply the existence of a separate lineage (see Nelson, manuscript; Tattersall and Eldredge, 1977, for cladograms versus trees). A cladogram is meant to be a hypoth- esis of monophyly and an ancestor and its descendant is just as much a monophyletic group as two descendants from a common ancestor. The recognition of ancestors requires a further set of assumptions that I am not willing to make. In any case, despite the incongruence with tradition, ancestors are not particularly important in phylogeny reconstruction. Each of the three taxa in the three-taxon hypothesis should be monophyletic. That is, they are not monophyletic by assumption but they should be susceptible to test for monophyly. However, it is not necessary that these tests for individual monophyly of the three taxa be completed before testing the three-taxon state- ment as a whole. When one of the taxa in a three-taxon statement is hypothesized as being non-monophyletic, it changes the hy- pothesis, making the original hypothesis logically irrelevant rath- er than false per se. In any case, it is clear that a hypothesis of relationships, such as the one advanced below for Tetrapoda, that does not include tests for monophyly of its constituent taxa, is less satisfactory than one that does. Test The use of cladistic techniques to test hypotheses of mono- phyly rests on the following basic hypothesis (or axiom, an as- sumption at this level of hypothesis formation but still subject to falsification; see Popper, I968 I?' . fr> • ^ CO*2 < ~ o m 'dl m ^ — to £ CO ■ ■ £ ^ tn - .niiiSNi NviNosHims saidvaan libraries Smithsonian institution NoiiniiisNi NviNOSHims ^ ^ 2 T ^ 2 r CO 2 CO RAR I ES SMITHSONIAN INSTITUTION ‘^'^0'-LniliSNI_ NVINOSHimS^^SB I d Vd 8 ll^LI B RAR I Es‘^SMITHSONIAN^IN ~ ^ CO — ^^ _ ON IAN mi O X^ost^ j 2 J ^ — J ^ — niliSNI NVINOSHillNS S3ldVdail libraries SMITHSONIAN INSTITUTION NOIifUllSNI~^NVINOSHiltVS^ S *~'V ^ ^ ^ *” 2 f— - o P Z o . o £ F CD o) > ■ t- , - - cn 'Ca_DC^ in PI CO • ; c^ ‘■- — CO — CO £ ^ /yj 2 ONIAN VARIES SMITHSONIAN INSTITUTION NOlifUliSNI NVINOSHillMS SOIdVdail L I B R A R I E S SMITHSONIAN " V ^ ^ V ^2 2: CO 2: ... CO ^ % c'’ ^ CO in CO /•= i S \^4^wJJ i H z \ _ > 2 2 — CO 2 CO 12 C0^2 7f\ 1 niiiSNi_NviN0SHiiws SBiMvaan libraries Smithsonian institution NoiiniiisNi nvinoshiiws s 2 vK 2 CO — (O — (O Ox ^ \ >v til 2 >CX I.t _ _ ^ ,OliSg>^ o CD 2 ->j ^ ^ ^ ^ ^ *> iL VARIES SMITHSONIAN INSTITUTION NOlifUliSNI NVlNOSHillAIS SHIdVdail LIBRARIES SMITHSONIAN l^ ^ ^ ^ 2 r- 2: r- 2 r- U'^oX P . r- O _ \v O O yJn,. — y — CO £ CO ''■’ £ j- pj^i niliSNI NVINOSHilWS saiavdan libraries SMITHSONIAN institution NOUniliSNI NVINOSHimS S'-^ ^ CO 2 ■• CO ■ /A -y /rt V/X w HilWS o ■ ' W Z Q “ ' o o 2 ' '^ _J "Z _1 2 _j ^ _i ^^’''hES SMITHSONIAN^INSTITUTION NOlirUlISNI NVINOSHimS S3 I Hvy a n^U BRAR 1 ES^SMITHSONIAN^INSTIT ^ ° i m i i ° . . m Xo^o^ w m X 5 t JTION ~ ^ — (/} \ ± w _ (o (/JlSni NviNosHiiws SBiavaaii libraries Smithsonian institution NoiiniiiSNi nvinoshii^ms S3iai e z w z -.• CO z V- ^ 2: CO - ? '• - - ■- 2 r i. 2 2 ^'•'£2_dC>X j, 2 ^ > 5 NOaj^c^ ygi-ic/) 2 2 r- Z r- 2 - - ,IES SMITHSONIAN INSTITUTION NOliniliSNI NVINOSHIIIAIS SSIHVaan LIBRARIES SMITHSONIAN INSTIT -s ^ ^ ^ Z r. CO 2 to 2 . CO . — « 33 ^ „ m ^ m X 15 m XcVoc;x ^ m U T 1 0 N “* CO _ ^ (/) _ CO g.SNi NViNOSHims SBiavaan libraries smithsonian institution NoiiniiisNi NvmosHims S3ia ? <.*■ ^ 2 ,v- W z -.< s i I ^ y ^ /4 -» / ° life' ^ ^ fit" 'HI ° ^ 2 i Vo^ioc^ > i ^ i > 'aailto 2 CO *2 W ‘ 2 CO 2 ilES SMITHSONIAN INSTITUTION NOliniliSNI NVINOSHIIWS SBIdVaail LIBRARIES SMITHSONIAN INSTH UTION 2 — I 2 '■' _J 2 -J 2 ^ 2SNI NVINOSHillNS S3iavaail libraries SMITHSONIAN INSTITUTION NOliniliSNI NVINOSHIIIAIS S3ld^ C r- V z r- z I- z ^ ■- 2: > (.£: ^<1 K > f^r /Han CO CO 2 ES SMITHSONIAN INSTITUTION NOliniliSNI NVINOSHIIWS S3IHVHan LIBRARIES SMITHSONIAN INSTIl 15 2 »CO Z CO Z y SH 2