NovitatesAMERICAN MUSEUM PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET NEW YORK, N.Y. 10024 U.S.A. NUMBER 2646 APRIL 21, 1978

RICHARD G. VAN GELDER A Review of Canid Classification

AMERICAN MUSEUM

Novitates PUBLISHED BY THE AMERICAN MUSEUM OF NATURAL HISTORY CENTRAL PARK WEST AT 79TH STREET, NEW YORK, N.Y. 10024 Number 2646, pp. 1-10 April 21, 1978

A Review of Canid Classification

RICHARD G. VAN GELDER'

ABSTRACT A review of the varied classifications of Recent tereutes, Cuon, Lycaon, and Otocyon), and one members of the family Canidae involving mor- polytypic genus, Canis, with eight subgenera (Canis, phological, numerical, and reproductive analyses Dusicyon, Pseudalopex, Lycalopex, Cerdocyon, suggests that the taxonomy should be arranged with Atelocynus, Vulpes, and Alopex). six monotypic genera (Chrysocyon, Speothos, Nyc- INTRODUCTION In the past few years reclassifications of the the recognition of Cerdocyon and Atelocynus as family Canidae have suggested different com- valid genera. They had been included in Dusi- positions of the genera. The generic allocations cyon by Simpson (1945). Stains (1967, 1975) of the canids had been fairly stable for a num- recognized Dasycyon, a genus named by Krum- ber of years. A dozen genera and 40 or fewer biegel (1949, 1953). species were usually recognized. Lately, the South American canids have Most studies followed Simpson's (1945) ar- been studied by Langguth (1969, 1975), and the rangement of three subfamilies: Caninae, Simo- entire family was subjected to a numerical anal- cyoninae, and Otocyoninae. In the subfamily ysis by Clutton-Brock, Corbet, and Hills Caninae, Simpson had put eight genera: Canis, (1976). The conclusions of these studies are Alopex, Vulpes, Fennecus, Urocyon, Nyc- different, and in attempting to recommend tax- tereutes, Dusicyon, and Chrysocyon. His sub- onomic changes that seemed warranted on the family Simocyoninae contained the genera basis of intergeneric hybridization, I (Van Speothos, Cuon, and Lycaon. Otocyon was the Gelder, 1977) found it necessary to attempt a sole representative of the subfamily Oto- consistent arrangement of the genera that dif- cyoninae. fers from that of Langguth (1975) and Clutton- Until 1969, subsequent compilations and Brock, Corbet, and Hills (1976) or any other classifications had, in general, continued to use current classification. The present paper is an these genera, the most consistent change being analysis of these different classifications.

'Curator, Department of Mammalogy, the American Museum of Natural History.

Copyright (©) The American Museum of Natural History 1978 ISSN 0003-0082 / Price $1.00 2 AMERICAN MUSEUM NOVITATES NO. 2646

ACKNOWLEDGMENTS Cerdocyon H. Smith, 1839. One species: C. thous. I am indebted to Drs. Sydney Anderson, Dusicyon H. Smith, 1839. Nine species: D. Guy Musser, and Karl F. Koopman, and es- australis, culpaeus, culpaeolus, griseus, pecially to Dr. Alfredo Langguth for comments fulvipes, gymnocercus, sechurae, vetulus, and criticism. and inca. Fennecus Desmarest, 1804. One species: F. DISCUSSION zerda. Urocyon Baird, 1858. Two species: U. cin- Other than the genera that Simpson (1945) ereoargenteus, and littoralis. included in the Caninae, there has been little Vulpes Bowdich, 1821. Ten species: V. dispute concerning the content of the genera of bengalensis, cana, chama, corsac, ferrilata, canids that Simpson had placed in the Simo- macrotis, pallida, rueppelli, velox, and vul- cyoninae and Otocyoninae. The alliance of pes. Speothos, Cuon, and Lycaon in a single sub- family had been questioned (Pocock, 1941; Additional clarification of these genera and Ellerman and Morrison-Scott, 1951; Todd, species can be made and, in fact, Stains evi- 1970), and Simpson himself (1945) admitted dently made some changes between his 1967 that it was a heterogeneous group. Clutton- and 1975 listings. Canis dingo is now generally Brock, Corbet, and Hills (1976) found that sub- regarded as a distinctive feral domestic dog. familial separations in the Canidae were not Canis familiaris is used for domestic dogs, al- warranted. They concluded that each of these though taxonomically it should probably be genera represented a monotypic, highly spe- synonymous with Canis lupus. In 1975 Stains cialized entity whose relationship to the others had dropped Dusicyon culpaeolus, evidently was distant, although perhaps closer to one an- following Langguth (1967). He did not regard other than to those genera that were usually Dusicyon fulvipes as specifically distinct from placed in the subfamily Caninae. Similarly, D. griseus. Dusicyon inca also was no longer Clutton-Brock, Corbet, and Hills did not recog- recognized by Stains (1975), again following nize the subfamily Otocyoninae for Otocyon. Langguth (1967). Stains listed only six species The distinction of each of these four genera of Dusicyon in 1975: D. australis, culpaeus, that were formerly split off from the Caninae is griseus, gymnocercus, sechurae, and vetulus. unquestioned. There seems also to be no ques- Stains (1975) listed Urocyon littoralis in his ac- tion about the recognition of Nyctereutes as a count of U. cinereoargenteus, pointing out that distinctive canid genus, although Frechkop these insular populations are regarded by some (1959) proposed that it belonged with the pro- as subspecies of U. cinereoargenteus, and by cyonids. Chrysocyon also is universally recog- others as subspecies of a separable species, U. nized as a clearcut genus. It is the other genera littoralis. within the Caninae that have received most of Stains (1967, 1975) persisted in recognizing the current attention and rearrangement that Dasycyon hagenbecki, known only from a sin- have resulted in the unequal classifications. gle skin. Cabrera (1958) and Hershkovitz (1961) For clarity in the following discussion, I have indicated that this genus and species is must define the extent of the genera that I shall most probably a domestic dog. mention. For these purposes I follow the list- For the South American canids, in addition ings of Stains (1967, 1975): to Urocyon, Chrysocyon, and Speothos, Cabrera (1958) recognized three genera: Dusi- Alopex Kaup, 1829. One species: A. lagopus. cyon, Atelocynus, and Cerdocyon. Simpson Atelocynus Cabrera, 1940. One species: A. mi- (1945) had put these in the genus Dusicyon and crotis. followed Osgood (1934) in subdividing it into Canis Linnaeus, 1758. Nine species: C. three subgenera: Dusicyon (including Pseuda- adustus, aureus, mesomelas, dingo, famil- lopex), Cerdocyon (including Atelocynus), and iaris, latrans, lupus, niger, and simensis. Lycalopex. Cabrera (1958) divided his genus 1978 VAN GELDER: CANID CLASSIFICATION 3

Dusicyon into two subgenera: Dusicyon, and At the end of 1975, the species in the genus Lycalopex. The latter contained only Dusicyon Canis included C. adustus, aureus, mesomelas, vetulus. familiaris, latrans, lupus, and niger, in the subgenus Canis; C. simensis in the subgenus Langguth (1969) arranged the South Ameri- Simenia (Ellerman, Morrison-Scott, and Hay- can canids at quite different levels. His classifi- man, 1953); C. australis in the subgenus Dusi- cation, which excluded Urocyon, had cyon; and C. culpaeus, gymnocercus, griseus; Chrysocyon, Dusicyon, and Cerdocyon as gen- and sechurae in the subgenus Pseudalopex. era. The first two were monotypic. Dusicyon In 1976 Clutton-Brock, Corbet, and Hills was divided into two subgenera, Dusicyon for published a classification based on a numerical D. australis alone, and Pseudalopex for D. analysis of the family Canidae. For the South culpaeus, gymnocercus, and griseus. His genus American canids they recognized only Chryso- Cerdocyon was divided into three subgenera: cyon and Speothos as monotypic genera, put Cerdocyon, Atelocynus, and Speothos. Urocyon in Vulpes, and included all the other Subsequently (1975), Langguth changed his South American canids in the genus Dusicyon, classification of the South American canines. more or less returning these species to the ar- He gave generic rank to the differentiated rangement (without subgenera) of Simpson in kinds, Cerdocyon, Speothos, Lycalopex, and 1945. Atelocynus, and he-placed the Patagonian can- Excluding the placement of Urocyon in Vul- ids, formerly called Dusicyon, in the genus pes by Clutton-Brock, Corbet, and Hills (1976), Canis. He recognized the subgenus Dusicyon which was a novel proposal-discussed later in for the Falkland wolf, C. australis only, and the present paper-the alternatives for the put the other species (culpaeus, gymnocercus, classification of South American canids at pres- and sechurae), in a second subgenus of Canis, ent are: Pseudalopex. Clutton-Brock, Corbet, and Hills, Langguth, 1976 1975 culpaeus valid species, placed in the genus valid species, placed in the genus Dusicyon Canis, subgenus Pseudalopex culpaeolus possibly conspecific with culpaeus, skin is culpaeus, skull is gymno- but needs study cercus; mismatch (Langguth, 1967) gymnocercus possibly conspecific with culpaeus, valid species, placed in the genus but needs study Canis, subgenus Pseudalopex inca possibly conspecific with culpaeus, skin is gymnocercus skull is culpaeus; but needs study mismatch (Langguth, 1967) griseus valid species, placed in the genus valid species, placed in the genus Dusicyon Canis, subgenus Pseudalopex fulvipes possibly conspecific with griseus, but conspecific with griseus, valid as a needs more study subspecies sechurae valid species, placed in the genus valid species, placed in the genus Dusicyon Canis, subgenus Pseudalopex vetulus valid species, placed in the genus valid species, placed in the genus Dusicyon Lycalopex thous valid species, placed in the genus valid species, placed in the genus Dusicyon Cerdocyon microtis valid species, placed in the genus valid species, placed in the genus Dusicyon Atelocynus australis valid species, placed in the genus valid species, placed in the genus Dusicyon Canis, subgenus Dusicyon brachyurus valid species, placed in the genus valid species, placed in the genus Chrysocyon Chrysocyon venaticus valid species, placed in the genus valid species, placed in the genus Speothos Speothos 4 AMERICAN MUSEUM NOVITATES NO. 2646

Langguth's reason for assigning the Patago- cluded Atelocynus within the genus Dusicyon; nian canids to Canis was based on the high they have not included Dusicyon within Canis. degree of morphological similarity, especially Some of the individual species, especially Du- of the skulls. He pointed out the difficulty of sicyon australis, culpaeolus, and A. microtis distinguishing, for example, a skull of Canis show a closer similarity to species of Canis adustus from Canis (Pseudalopex) culpaeus. (adustus, mesomelas, or aureus) than they do He believed that the degree of difference be- to other species placed in Dusicyon. Some idea tween these two species did not warrant generic of the subjectivity Clutton-Brock and col- separation. Both Canis and the former genus leagues employed in drawing their conclusions Dusicyon have maintained what he called a is evident in places in their text. For example, basic, generalized canid pattern (Langguth they suggested the inclusion of D. australis in 1975). Canis and of Canis simensis, mesomelas, and Although Clutton-Brock, Corbet, and Hills adustus in Dusicyon as a logical but ineffective (1976) came to a different taxonomic conclu- means of distinguishing the two genera. They sion from that of Langguth, they essentially seemed to be making rather weak excuses for confirmed Langguth's merging of Dusicyon the maintenance of Dusicyon as a genus in face with Canis. In their tables of systematic posi- of their own data to the contrary. tion of various species (percentage similarity to An additional datum for the inclusion of Du- "near neighbours") they show, for example, sicyon in Canis was based on the report by that Dusicyon australis is more similar to Canis Krieg (1925) of a case of hybridization between familiaris than is Canis aureus, that Dusicyon a South American canid called Pseudalopex gymnocercus is more similar to Canis meso- and a domestic dog. The species involved was melas than that jackal is to Canis adustus or C. most likely D. gymnocercus (see Van Gelder, aureus. In the case of Canis simensis, Clutton- 1977), although others (Gray, 1972; Chiarelli, Brock, Corbet, and Hills showed that the five 1975) have listed it as Cerdocyon thous. nearest neighbors are Dusicyon. Too few data are available from the canids, In their table I of the mean similarities be- especially the South American ones, for any tween and within genera of the existing (not conclusions to be drawn concerning their rela- their proposed) classification, Clutton-Brock, tionships on the basis of immunological or mo- Corbet, and Hills's data again show the prox- lecular studies (Seal, 1975). Similarly, the imity of Dusicyon to Canis. The mean intra- limited data from karyology offer no real help generic similarity of Canis is given as 83.9; the in establishing relationships. These have been similarity between Canis and Dusicyon is 83.4. summarized by Chiarelli (1975), who pointed There are a number of inconsistencies in the out that there was a "close resemblance of the conclusions drawn by these authors from their karyotypes of Chrysocyon, Atelocynus, Dusi- phenetic analysis, as can be seen by reconstruc- cyon, and Speothos with those of Canis." tion of a portion of their table I entitled "Mean On the basis of the various works cited similarities between and within genera of the above, especially that of Clutton-Brock, Cor- existing classification" (below). bet, and Hills (1976), the inclusion of a number According to these data, Atelocynus is not of the species formerly called Dusicyon within so close to Dusicyon as Dusicyon is to Canis, the genus Canis is warranted. Langguth (1975) but Clutton-Brock, Corbet, and Hills have in- had already proposed this, and the phenetic

genus Canis Dusicyon Atelocynus Cerdocyon Chrysocyon Speothos Canis 83.9 Dusicyon 83.4 90.5 Atelocynus 79.3 82.2 Cerdocyon 79.4 84.8 86.1 * Chrysocyon 69.4 71.4 73.4 68.6 * Speothos 61.9 63.3 68.2 60.0 53.8 * 1978 VAN GELDER: CANID CLASSIFICATION 5 analysis of Clutton-Brock, Corbet, and Hills should be noted, however, that the similarity of certainly suggests it, despite the reluctance by Dusicyon to Vulpes is even greater, but Clut- these investigators to propose it. ton-Brock, Corbet, and Hills did not seem to The decisions made by Clutton-Brock, Cor- consider this to any great extent. In the lists of bet, and Hills concerning Dusicyon and Canis near neighbors for the various species of Vul- may have been influenced by the extensive pes, Clutton-Brock, Corbet, and Hills show merger of species into the genus Canis that that Vulpes corsac, ferrilata, rueppelli, pallida, would result if they applied equally objective zerda, and chama each have at least one spe- standards to the relationship of Vulpes to cies of Dusicyon more similar to each of them Canis. In this instance, however, no one had than some other members of the genus Vulpes recently proposed the synonymizing of these are. two genera as Langguth had done for Dusicyon Clutton-Brock, Corbet, and Hills implied and Canis. that their line of separation between genera is The genera Alopex, Vulpes, Urocyon, and at the 80 percent level of similarity. From their Fennecus have been relatively stable for many table I of mean similarites, the level of 80 years. Except for Vulpes, they are essentially percent functions solely to separate Alopex monotypic genera. Urocyon littoralis is some- from Vulpes. One wonders whether the 0.8 times recognized as a valid species for the insu- percent difference between 79.2 and 80.0 is of lar populations of U. cinereoargenteus, sufficient biological and taxonomic significance whereas others consider them as subspecies of to warrant this. In their figure 5a (two-dimen- U. cinereoargenteus. Clutton-Brock, Corbet, sional plot of Caninae using principal coordi- and Hills did not include U. littoralis in their nates algorithm) the distance of Alopex from analysis. Vulpes vulpes seems less than Atelocynus or Alopex has been considered congeneric with Cerdocyon are from the nearest Dusicyon. Clut- Vulpes. Bobrinskii (1965) regarded it as a sub- ton-Brock, Corbet, and Hills, as already men- genus of Vulpes. Youngman (1975) also re- tioned, considered Atelocynus and Cerdocyon garded the Arctic fox as a Vulpes. On the basis to be congeneric with Dusicyon. To me, the of the production of viable hybrids from most significant data concerning the relation- crosses of A. lagopus and V. vulpes, I consid- ships of Alopex to Vulpes is the ability of the ered the two to be congeneric (Van Gelder, Arctic foxes to hybridize with Vulpes vulpes 1977). Clutton-Brock, Corbet, and Hills, and to produce not only viable offspring, but however, found Alopex to be the most distinc- fertile ones. Alopex has the highest number of tive of the foxes and retained it as a genus. chromosome arms of any of the canids, Funda- The phenetic analysis of Clutton-Brock, mental Number, 88. Vulpes vulpes has 72. Corbet, and Hills caused them to include Fen- They have respectively, 25 and 18-20 pairs of necus and Urocyon within the genus Vulpes. somatic chromosomes (Chiarelli, 1975) but de- Here again, the data suggest that their conclu- spite these differences, they are genetically sions were more subjective than objective. compatible. I have discussed elsewhere (Van Below is an extract of their table of mean sim- Gelder, 1977) the reasons why Alopex and Vul- ilarities between and within genera of the exist- pes should be considered congeneric. Clutton- ing classification: Brock, Corbet, and Hills stated that Alopex "lies close to the genus Vulpes but . . . it is the genus Vulpes Dusicyon Alopex Fennecus Urocyon Vulpes 86.9 most aberrant of the foxes" as their reason for Dusicyon 86.0 90.5 retaining it as a genus. Todd (1970) found no Alopex 79.2 79.0 * reason for recognizing Alopex as a genus. Fennecus 85.1 83.6 78.2 * The inclusion of Fennecus in Vulpes is more Urocyon 85.0 84.5 74.4 82.5 * strongly founded, according to the phenetic analysis. Fennecus has a similarity to Vulpes These data show a high degree of similarity even greater than that of Alopex to Vulpes, but of both Fennecus and Urocyon to Vulpes. It as I have earlier mentioned, the similarity of 6 AMERICAN MUSEUM NOVITATES NO. 2646

Dusicyon to Vulpes is still closer than that of species of Vulpes that have been reported Fennecus. The suggestion that Fennecus be in- (Chiarelli, 1975) also have a Fundamental corporated in Vulpes does not seem to have Number of 72, but vary from 60 to 34 in their been made before. Ellerman and Morrison- diploid number of chromosomes. Scott (1951) stated that "Pocock did not retain The final step in this analysis of the classifi- it [Fennecus] as a genus, but there seems little cation of the Canidae is the consideration of the doubt that it should be retained." I am, relationship of Vulpes to Canis and Vulpes to however, unable to find Pocock's statement to Dusicyon. Clutton-Brock, Corbet, and Hills's this effect in his 1941 work, the one presuma- figures of two-dimensional plots of the princi- bly cited by Ellerman and Morrison-Scott. pal coordinates algorithm (figures 2a, 3a, 4a, Clutton-Brock, Corbet, and Hills pointed out and 5a) showed, generally, that while the tradi- that Fennecus has nomenclatural priority over tional species of Canis seem separable from Vulpes, and Clutton-Brock and Corbet (1975) even the extended genus Vulpes (i.e., with Fen- have applied to the International Commission necus, Urocyon, and Alopex), Dusicyon fills an of Zoological Nomenclature for a decision to intermediate position and overlaps each of maintain the name Vulpes. these. Clutton-Brock, Corbet, and Hills's proposal The pertinent data on mean similarities be- to include Urocyon in Vulpes represented a tween and within these three genera (excluding change in the classification and nomenclature of Alopex, Fennecus, Urocyon, Atelocynus, and genus that had been stable for nearly a century. Cerdocyon) are extracted from Clutton-Brock, They suggested that one of the reasons for the Corbet, and Hills's table Ia: stability of Urocyon had been that no one had compared it with Vulpes, and that all attention genus Vulpes Canis Dusicyon to its affinities had been concerned with its Vulpes 86.9 possible relationship with Dusicyon. Their phe- Canis 78.0 83.9 netic analysis showed that the five nearest Dusicyon 86.0 83.4 90.5 neighbors of Urocyon are species of Vulpes (in order of proximity: V. bengalensis, velox, cor- The similarity of Dusicyon to Vulpes is, sac, rueppelli, and pallida). The level of differ- from these data, almost the same as the sim- ence of Urocyon from these is about the same ilarity of the species of Vulpes are to their own as Vulpes zerda is from its five nearest neigh- generalized genus. The same is true for the bors, and closer than V. cana or even V. vulpes similarity of Dusicyon to Canis relative to sim- is to its proximate species. From the phenetic ilarities of the species of Canis to their gener- data, it would seem that congeneracy between alized genus. Urocyon and Vulpes is warranted, especially if Species of Dusicyon show up as one or Fennecus is also lumped with Vulpes. Sim- more of the five nearest neighbors of species of ilarly, if Alopex is congeneric with Vulpes, then Vulpes in more than 25 percent of the cases, both Fennecus and Urocyon are equally deserv- whether Fennecus, Urocyon, or Alopex are in- ing of alliance in Vulpes. Todd (1970) found cluded in Vulpes or not. Canis does not show karyotypic similarity between Urocyon and up among the five closest relatives of any of Fennecus. the species of Vulpes, nor does any species of There is one report of hybridization between Vulpes appear as the nearest neighbor of any of Urocyon and Vulpes (Bezdek, 1944). It is based the members of the genus Canis that Clutton- on a furrier's skin, and there are no supporting Brock, Corbet, and Hills analyzed. However, data to reinforce its validity. However, if Vul- Dusicyon is represented as a nearest neighbor pes and Urocyon are congeneric, somewhat of species of Canis 45 percent of the time. more credence might be given to this report. Even if the bloodhound and dingo are excluded Chromosomally, Urocyon has a Fundamental from this analysis as conspecific with Canis Number of 70 (2n=66). Vulpes vulpes has 72 lupus, Dusicyon still represents more than 45 chromosome arms (2n= 34-38), and the other percent of the nearest neighbors. These data 1978 VAN GELDER: CANID CLASSIFICATION 7 serve to confirm the phenetic data that separate generic hybrids between Dusicyon and Canis Vulpes and Canis, but which show a bridge and between Vulpes and Canis that led me to between the two made by representatives of these conclusions and to analyze the other stud- Dusicyon. ies that were concerned with the relationships As Clutton-Brock, Corbet, and Hills put it, of these genera. Most students of both fossil "No objective analysis of the results of this and Recent canids acknowledge the similarities study would product these three genera [Canisl between these three genera, and, more than 30 Vulpes/Dusicyon] its presently composed." years ago, Simpson (1945) stated, "Despite They noted that Latigguth include a number of their world-wide distribution and an abundance species of Dusicyonr in Canis and they did not of well-distinguished, more or less local spe- believe that his conclusions were "greatly at cies, the recent canines are quite uniform in variance" with theirs. They stated, "if Dusi- structure, and it would be justified from many cyon were merged with Canis, it would be points of view to unite them all in a single difficult to argue that Vulpes should not be genus."9 treated likewise." Based on the analyses of Langguth (1975) Elsewhere (Van Gelder, 1977) I proposed for the South American canids, and those of the merger of both Dusicyon with Canis, fol- Clutton-Brock, Corbet, and Hills (1976) for all lowing Langguth, and the joining of Vulpes the canids, the most appropriate current classifi- with Canis as well. It was evidence from inter- cation for the family seems to be: Family Canidae Genus Canis Linnaeus, 1758 Subgenus Canis Linnaeus, 1758 Canis (Canis) lupus Linnaeus, 1758 Canis (Canis) latrans Say, 1823 Canis (Canis) rufus Audubon and Bachman, 18511 Canis (Canis) familiaris Linnaeus, 17582 Canis (Canis) aureus Linnaeus, 1758 Canis (Canis) adustus Sundevall, 1846 Canis (Canis) mesomelas Schreber, 1778 Canis (Canis) simensis Riippell, 1835 Subgenus Dusicyon H. Smith, 1839 Canis (Dusicyon) australis Kerr, 1792 Subgenus Pseudalopex Burmeister, 1856 Canis (Pseudalopex) culpaeus Molina, 1782 Canis (Pseudalopex) gymnocercus (Fischer, 1814) Canis (Pseudalopex) griseus Gray, 1837 Canis (Pseudalopex) sechurae Thomas, 1900 Subgenus Lycalopex Bunneister, 1854 Canis (Lycalopex) vetulus Lund, 1842 'Clutton-Brock, Corbet, and Hills (1976), Stains (1975) Goldman, 1944, Lawrence and Bossert, 1967, Paradiso, and others used Canis niger Bartram, 1791, for the "red 1968, and Paradiso and Nowak, 1972). Clutton-Brock, Cor- wolf." Bartram's names are on the Official Index of Re- bet, and Hills did not include specimens of Canis rufus in jected and Invalid Works in Zoological Nomenclature and their study. the correct name for this species is Canis ruifus Audubon 2This name is commonly and universally applied to and Bachman, 1851. See Nowak (1967), Paradiso (1968), domestic dogs that are believed to have been domesticated and Paradiso and Nowak (1972) for comments. Its status as from one or more subspecies of Canis lupus. What the a species, hybrid population between lupus and latrans, or status of domestic "species" should be in taxonomy is not hybrid population between lupus andfamiliaris have been resolved, and for the moment it seems best left alone. expressed (see McCarley, 1962, Nowak, 1970, Young and 8 AMERICAN MUSEUM NOVITATES NO. 2646

Subgenus Cerdocyon H. Smith, 1839 Canis (Cerdocyon) thous Linnaeus, 1766 Subgenus Atelocynus Cabrera, 1940 Canis (Atelocynus) microtis Sclater, 1882 Subgenus Vulpes Bowdich, 18211 Canis (Vulpes) vulpes Linnaeus, 1758 Canis (Vulpes) corsac Linnaeus, 1768 Canis (Vulpes) ferrilata (Hodgson, 1842) Canis (Vulpes) bengalensis Shaw, 1800 Canis (Vulpes) cana (Blanford, 1877) Canis (Vulpes) rueppelli Schinz, 1825 Canis (Vulpes) pallida Cretzschmar, 1826 Canis (Vulpes) zerda Zimmermann, 1780 Canis (Vulpes) chama A. Smith, 1833 Canis (Vulpes) velox Say, 1823 Canis (Vulpes) cinereoargenteus Schreber, 1775 Canis (Vulpes) littoralis (Baird, 1858)2 Subgenus Alopex Kaup, 1829 Canis (Alopex) lagopus Linnaeus, 1758 Genus Nyctereutes Temminck, 1839 Nyctereutes procyonoides (Gray, 1834) Genus Chrysocyon H. Smith, 1839 Chrysocyon brachyurus (Illiger, 1815) Genus Speothos Lund, 1839 Speothos venaticus (Lund, 1842) Genus Cuon Hodgson, 1838 Cuon alpinus (Pallas, 1811) Genus Lycaon Brookes, 1827 Lycaon pictus (Temminck, 1820) Genus Otocyon Muller, 1836 Otocyon megalotis (Desmarest, 1822) Unfortunately, Clutton-Brock, Corbet, and species may well have biased some of their Hills did not included Canis (Canis) rufus or conclusions. (Vulpes) littoralis in their study. It is also re- Elsewhere (Van Gelder, 1977), I have sug- grettable that they did not list the catalogue gested that species capable of hybridizing numbers of the specimens that they utilized in should not be placed in separate subgenera. their work. They seemed unaware that Lang- This suggestion was based on the same grounds guth (1967) had demonstrated that both Dusi- as the reasons for not considering genera capa- cyon culpaeolus and D. inca were based on ble of hybridizing: that the upper level of the mismatched skins and skulls of D. culpaeus species involves reproductive incompatability and D. gymnocercus. It is impossible to ascer- with other species. Of the canids on the preced- tain how many specimens Clutton-Brock, Cor- ing list, hybrids have been reported between bet, and Hills used in their analysis of the subgenera Vulpes and Alopex, between Vul- culpaeolus and inca, although it appears that only the holotypes were involved. However, 2Whether Canis (Vulpes) littoralis is a species closely the inclusion of these in their data-base as valid allied to C. (Vulpes) cinereoargenteus or whether its popu- lations are subspecies is largely a matter of opinion unsup- 'For the use of Vulpes see Clutton-Brock and Corbet ported by any recent studies. Clutton-Brock, Corbet, and (1975) and Clutton-Brock, Corbet, and Hills (1976). Hills (1976) did not include littoralis in their study. 1978 VAN GELDER: CANID CLASSIFICATION 9 pes and Canis, and between Pseudalopex and Clutton-Brock, Juliet, Gordon B. Corbet, and Canis (Gray, 1972; Van Gelder, 1977). An ap- Michael Hills propriate alternative would be to call each of 1976. A review of the family Canidae, with a the subgenera in the preceding list a "group," classification by numerical methods. Bull. using the subgeneric name for each of them as British Mus. (Nat. Hist.), Zool., vol. 29, pp. 117-199. in "Vulpes-group," or "Canis-group." Ellerman, J. R., and T. C. S. Morrison-Scott 1951. Checklist of Palaearctic and Indian mam- SUMMARY mals 1758 to 1946. London, British Mu- seum (Nat. Hist.). Current studies of Recent canids seem to Ellerman, J. R., T. C. S. Morrison-Scott, and R. confirm a closer relationship between a number W. Hayman of species and groups of species than pre- 1953. Southern African Mammals 1958 to 1951: viously thought. The taxonomic conclusions A reclassification. London, British Mu- presented in the publications of Clutton-Brock, seum (Nat. Hist.). Corbet, and Hills (1976) differ from those of Frechkop, Serge Langguth (1975), and the present paper is an 1959. De la position systematique du genre Nyc- analysis of these classifications and an attempt tereutes. Bull. Inst. Roy. Sci. Nat. Bel- to reconcile them. Chrysocyon, Speothos, Nyc- gique, vol. 35, pp. 1-20. tereutes, Cuon, Lycaon, and Otocyon are re- Gray, Annie P. 1972. Mammalian hybrids. A check-list with garded as monotypic genera, and the genus bibliography. Comm. Bur. Animal Breed- Canis is considered to be polytypic with eight ing and Genet. Edinburgh, Tech. Com- subgenera or groups for the taxa formerly con- mun. 10 (revised), pp. 1-262. sidered the genera Dusicyon, Pseudalopex, Hershkovitz, Philip Lycalopex, Cerdocyon, Atelocynus, Vulpes, and 1951. On the South American small-eared zorro Alopex. Atelocynus microtis Sclater (Canidae). Fieldiana Zool., vol. 39, pp. 505-523. LITERATURE CITED Krumbiegel, I. von. 1949. Der Andenwolf-ein neuentdecktes Gross- Bezdek, Hubert tier. Unschau Wiss. Tech., vol. 19, pp. 1944. A red-gray fox hybrid. Jour. Mammal., 590-591. vol. 25, pp. 90. 1953. Der "Andenwolf', Dasycyon hagenbecki Bobrinskii, N. A. (Krumbeigel, 1949). Saugetierk. Mitt., 1965. Order Camivora, In N. A. Bobrinskii, vol. 1, pp. 97-104. B. A. Kuznetzov, and A. P. Kuzyakin Langguth, Alredo [Key to the mammals of the USSR]. 1967. Sobre la identidad de Dusicyon culpaeolus Moscow, Sovietskaya Nauka (in Russian). (Thomas) y de Dusicyon inca (Thomas). Cabrera, Angel Neotropica, vol. 13, pp. 21-28. 1957 [1958]. Catalogo de los mamiferos de Amer- 1969. Die Suidamerikansichen Canidae under be- ica del sur, I. Rev. Mus. Argent. Cienc. sonderer Beruicksichtigung des Mahnen- Nat. Bernardino Rivadavia (Cienc. Zool.), wolfes Chrysocyon brachyurus Illiger. vol. 4, pp. 1-307. Zeitschr. fur wissenschaftliche Zoologie, Chiarelli, A. B. vol. 179, pp. 1-188. 1975. The chromosomes of the Canidae. In 1975. Ecology and evolution in the South Amer- M. W. Fox (ed.), The wild canids, their ican canids. In M. W. Fox (ed.), The systematics, behavioral ecology and evolu- wild canids, their systematics, behavioral tion. New York, Van Nostrand Reinhold ecology and evolution. New York, Van Co., pp. 40-53. Nostrand Reinhold Co., pp. 192-206. Clutton-Brock, Juliet, and Gordon B. Corbet Lawrence, Barbara, and Willian H. Bossert 1975. Vulpes Frisch, 1775 (Mammalia): Pro- 1967. Multiple character analysis of Canis lupus, posed conservation under the plenary latrans, and familiaris, with a discussion powers. Z. N. (S.) 977. Bull. Zool. No- of the relationship of Canis niger. Amer. menclature, vol. 32, pp. 110-112. Zool., vol. 7, pp. 223-232. 10 AMERICAN MUSEUM NOVITATES NO. 2646

McCarley, Howard classification of mammals. Bull. Amer. 1962. The taxonomic status of wild Canis (Can- Mus. Nat. Hist., vol. 85, pp. 1-350. idae) in the south central United States. Stains, Howard J. Southwestern Nat., vol. 7, pp. 227-235. 1967. Carnivores and pinnipeds. In S. Anderson Nowak, Ronald M. and J. K. Jones (eds.), Recent mammals 1967. The red wolf in Louisiana. Defenders of of the world, a synopsis of families. New Wildlife News, vol. 42, pp. 60-70. York, The Ronald Press Co., pp. Paradiso, John L. 325-354. 1968. Canids recently collected in east Texas 1975. Distribution and taxonomy of the Canidae. with comments on the taxonomy of the In M. W. Fox (ed.), The wild canids, red wolf. Amer. Midland Nat., vol. 80, their systematics, behavioral ecology and pp. 529-534. evolution. New York, Van Nostrand Rein- Paradiso, John L., and Ronald M. Nowak hold Co., pp. 3-26. 1972. Canis rufus. Mammalian Species, no. 22, Todd, N. B. pp. 1-4. 1970. Karyotypic fissioning and canid phylog- Pocock, R. I. eny. Jour. Theor. Biol., vol. 26, pp. 1941. The fauna of British India including 445-480. Ceylon and Burma. II. Mammalia, pp. Van Gelder, Richard G. i-xii+ 1-503, London: Taylor & Francis. 1977. Mammalian hybrids and generic limits. Ltd. Amer. Mus. Novitates, no. 2635, pp. Seal, U. S. 1-25. 1975. Molecular approaches to taxonomic prob- Young, Stanley P., and Edward A. Goldman lems in the Canidae. In M. W. Fox (ed), 1944. The wolves of North America. Wash- The wild canids, their systematics, be- ington, D. C., American Wildlife Inst. havioral ecology and evolution. New Youngman, Phillip M. York, Van Nostrand Reinhold Co., pp. 1975. Mammals of the Yukon Territory. Natl. 27-39. Mus. Nat. Sci., Natl. Mus. Canada, Publ. Simpson, George Gaylord Zool., no. 10, pp. 1-192. 1945. The principles of classification and a