The author presents the results of an exhaustive study of existing cranial specimens of wild Canis in the eastern United States, and concludes that Canis rufus, the red wolf, is a distinct species rather than a hybrid of the gray wolf (Canis lupus) and the coyote (Canis latrans), as has been posited by others. Further, he suggests that Canis lupus lycaon, the eastern timber wolf, may in fact have resulted from the hybridization of the red wolf and the gray wolf. This study should go a long way to putting the debate to rest, as Nowak is an unimpeachable authority on wolf evolution -- JM
Assessment was made of all available cranial specimens of wild Canis dating since the Blancan and prior to AD 1918 in the region east of the Great Plains and south of the Prairie Peninsula, Lakes Erie and Ontario, and the St. Lawrence River. The small wolf C. priscolatrans (=C. edwardii) of the early Irvingtonian seems unrelated to the modern red wolf (C. rufus), but gave rise to a lineage including the larger C. armbrusteri and culminating in C. dirus of the late Rancholabrean. A small wolf, possibly a descendant of the Eurasian C. mosbachensis, did not reappear in the east until near the end of the Rancholabrean. At the same time, the coyote (C. latrans) disappeared from the east, not to return until the small wolf was extirpated in the 20th century. Fragmentary remains of the small wolf, dating from around 10,000 and 2,000-200 ybp, show continuity with 14 complete, mostly modern, eastern skulls. Multivariate analysis indicates those 14 represent a well-defined species, C. rufus, distinct from large series of the western gray wolf (C. lupus) and coyote. There is no evidence that the red wolf originated as a hybrid of the latter two species, though early specimens from central Texas suggest it began to interbreed with C. latrans by about 1900. Three long-recognized red wolf subspecies appear valid: C.r. floridanus, Maine toFlorida; C.r. gregoryi, south-central United States; and C.r. rufus, central and coastal Texas, southern Louisiana, and probably now represented in the captive/reintroduced populations. The subspecies C. lupus lycaeon of southeastern Ontario and southern Quebec is statistically intermediate to C. rufus and western C. lupus, and may have resulted from natural hybridization of those two species. Such could explain how the red and gray wolf differ so sharply where their ranges meet in the west but morphologically approach one another in the east.
Two species of wild Canis, the gray wolf (C. lupus Linnaeus) and the coyote (C. latrans Say), occurred in most of the northern and western North America in historical times. The situation is less clear in this study's region of interest: east of the Great Plains and south of the Prairie Peninsula, Lakes Erie and Ontario, and the St. Lawrence River. A third species, the red wolf (C. rufus Audubon and Bachman), reportedly inhabited much of that region, but the full extent of its range and the nature of its relationship with the gray wolf, coyote, and fossil species has never been resolved. By the early 1900s, people had extirpated wolves from that part of the region of interest east of the Mississippi River. Wolves persisted in an ever-shrinking area between the Mississippi and Great Plains but were gone from the wild by about 1980. Their disappearance, together with human habitat modification, opened a niche for the coyote, which now occupies nearly the entire east. The coyote evidently hybridized with some remnant wolf populations, further confusing the situation. A few individuals were removed from the last population that appeared morphologically close to original C. rufus, in southeastern Texas and southern Louisiana, and were used to found extant captive and reintroduced populations. Whether those populations will be maintained and used in further conservation efforts is in question, in part because of controversy regarding their systematic background.
Adubon and Bachman (1851) named rufus as a subspecies of C. lupus and assigned it a range centering in south-central Texas. They regarded another subspecies, then called ater and subsequently designated floridanus, or, incorrectly, niger, to occur east to Florida and Kentucky. They treated C. latrans as a separate species, restricted to the west. Their basic view held for another century .... Goldman (1937, 1944) was the first to combine rufus with other southeastern wolves to form a single species distinct from C. lupus; the latter was thought to comprise all other North American wolves. That position was supported by most later assessments of modern and fossil material. However, based on a multivariate analysis of skulls, Lawrence and Bossert (1967, 1975) concluded that the original wolf populations of the southeast were not more than subspecifically distinct from C. lupus.
Some analyses of mitochondrial and nuclear DNA have suggested that C. rufus is not a valid species or subspecies but is the product of hybridization, most likely within historical times, between C. lupus and C. latrans. That there indeed has been hybridization between southeastern wolf populations and the coyote long has been recognized. However, such hybridization generally was considered a modern phenomenon that contributed to the demise, not the origin, of the red wolf. In any case, hybrid derivation of rufus has not been supported by morphometric analysis, by observation of living animals, or by some geneticists who have reviewed the issue.
Recently, Wilson et al. (2000), also using DNA analysis, suggested that rufus and the subspecies C. lupus lycaon constitute a single species, independent of all other C. lupus and C. latrans. Goldman (1944) had assigned lycaon a large range in eastern North America, but more recent studies argue that it was restricted to extreme southeastern Ontario, extreme southern Quebec, and possibly some adjacent parts of the northeastern United States. Like rufus, lycaon evidently has been affected through hybridization with C. latrans.
There never has been a detailed study centering on the oldest available series of eastern Canis, and their relationship to one another and to known series of C. lupus and C. latrans. Both morphological and molecular analyses have been limited by the scarcity of specimens representing original wolf populations in the region of interest, especially those present before the modern invasion of, and hybridization with, the coyote. The following assessment covers all available material from the region dating from before that invasion -- modern, archeological, and paleontological. Specimens of lycaon, from the restricted version of its range, as delineated above, also have been considered.
This study relies entirely on cranial and dental morphology. Most remains of eastern Canis dating prior to 1800 -- those from paleontological or archeological sites -- are fragmentary. Many consist only of one or a few teeth and/or small sections of bone. Such material is of limited value, as the various species of Canis show considerable overlap in size and other characters. However, univariate and bivariate analysis, or evaluation of dental cusps and cingula, sometimes helps determine whether fragments are within the range of variation shown by a population defined by more substantive material. If a series of complete skulls indicates presence of a given species, a series of chronologically or geographically proximal fragments, showing no significant difference in those characters that can be evaluated, may offer reasonable evidence of the presence of the same species.
....All skulls used in multivariate analyses were considered to have reached full size, which occurs at about one year of age in small species of Canis and by about two years of age in large species. For that part of the region of interest east of the Mississippi River there are only nine known complete skulls of all wild Canis (none of them C. latrans) dating from the end of the Pleistocene to AD 1917....
....For specimens lacking data, assignment to sex was based primarily on size in relation to individuals of known sex in the same series. Both of the known females are smaller than the remaining six skulls from east of the Mississippi. Of those six, the smallest, from Fern Cave, Alabama, is known to be a male because of the presence of a baculum with the skeleton. It thus would not be unreasonable to treat the remaining five as males, even though sex (male) was recorded only for the specimen from near Cherokee, Alabama. The specimen from Miami, Florida, is the largest of the group and is larger than any known female wolf from the region of interest. Some postcranial elements are with the skull from Moosehead Lake, Maine; the pelvis was examined by A.J. Bezuidenhout (1999) whose opinion, based mainly on evidence of penile crura attachment to the ischium, is that the specimen is a male. Goldman's (1944) table of measurements listed one of the skulls from Pennsylvania as a male, the other as a female .... However, T. Daeschler (1999) reported that sex had not been recorded for either specimen and that no postcranial elements are known. Moreover, review of Goldman's table indicates that many of his assignments to sex were based on judgment, not recorded data.
Of the seven specimens taken 1898-1905 in northeastern Louisiana and used in multivariate analysis, two lacked recorded sex, but both are much larger than two known females from the same area and period. Those two females closely resemble one another, as well as the two females from east of the Mississippi, in size and other characters. Of the five skulls taken in 1900 in Calhoun County, Texas, and used in multivariate analysis, four lack sex data, but the one known male is smaller than all the others, and the one known female from the same area is smaller still ....
The sex recorded for two specimens assessed by this study is considered erroneous. A reported female, from upper Michigan, is among the three largest skulls known from that area and is here included as a male. A reported male from southern Quebec is smaller than all males and nearly all females collected in that region in 1905-1933 and is here regarded as a female. In addition, the specimen collected near Avery Island, Louisiana in 1919, and questionably listed as a female by Nowak (1979), has been reassessed and treated as a male in this study.
Some chronological flexibility was used in selecting the series thought to represent the original wolf populations of the region of interest, such depending in part on the period in which C. latrans moved into different parts of the east, replacing or hybridizing with the native wolves. The skull collected in 1917 in Alabama seems acceptable, as the coyote is not known to have crossed the Mississippi River, south of the Prairie Peninsula, until the 1960s. Of the other five skulls from east of the Mississippi and used in multivariate analysis, four are known to date from 1870 and one, from Fern Cave, Alabama, probably does. The seven skulls taken 1898-1905 from northeastern Louisiana date from about 50 years before the coyote is known to have entered Louisiana. The small series collected in 1900 from Calhoun County, coastal Texas, is much closer geographically to the range of C. latrans, but coyotes and other wild Canis are not known from that area and time....
....The specimens from the region of interest were compared with six series of male C. lupus from the western conterminous United States. Those samples represent six nominal subspecies, though all were placed in the synonymy of C. lupus nubilis by Nowak (1995). Three of the samples are the available series of C. lupus that are geographically most proximal to the western edge of the region of interest, and just west of those series are the other three samples. The six series therefore may help determine whether C. lupus tends to grade morphometrically towards the eastern wolf as the two approach geographically. Also used for comparison was a series of 10 male lycaon collected 1905-1933 from southeastern Ontario and Quebec, long before C. latrans became established in that region, and a series of 16 male C. lupus taken before 1966 from the upper peninsula of Michigan, when the original population was still present (that population was extirpated shortly thereafter, though C. lupus recently was reestablished in upper Michigan through migration from Minnesota.) Each of these wolf samples comprises every available fully developed male from the involved area and period; no other selectivity was used. The comparative series of coyotes contained 96 male C. latrans lestes from Colorado and Idaho; that subspecies is considered a relatively large coyote....
....There is no evidence of the presence of a small wolf in the later Irvingtonian or early Rancholabrean. Small wolves seem to disappear from eastern North America following the extinction of C. priscolatrans and do not reappear until the terminal Pleistocene, around 15,000-10,000 ybp, a gap of nearly 1 million years. There have been a few reports of a small wolf in the east during that interval but they are not valid....
If there is no definitive evidence of a small wolf in the east for nearly a million years prior to the latest Rancholabrean, where was the progenitor of the modern wolf population of the region? It has been suggested that C. priscolatrans, or a close relative, extended its range to Eurasia via the Bering Land Bridge, and there evolved into extant C. lupus, which subsequently reinvaded North America. C. etruscus Major, a small wolf of the European early Pleistocene, may be an Old World counterpart of C. priscolatrans and the ancestor of C. lupus. There was, however, an intermediate stage in the transition from priscolatrans/etruscus to lupus. That stage is represented by C. mosbachensis Soergel, a modest-sized wolf of the Eurasian Pleistocene. Five mandibles and two maxillary fragments of that species ... were examined. They were collected in the Lake Baikal region of south-central Siberia and dated from about 700,000 ybp, which would be a period corresponding to the mid- to late Irvingtonian of North America .... Two mandibles and one maxillary fragment from the Cripple Creek Sump of Alaska were examined. They date from about the same period and appear to represent the same population as the Siberian material .... Since the Alaskan and Siberian specimens are close to the size of the later material from eastern North America, and smaller than western C. lupus, and since wolf evolution generally involves progression from smaller to larger size, C. mosbachensis seems a logical candidate for the ancestor of modern North American (as well as Eurasian) wolves.
Although the late Irvingtonian population in Alaska and Siberia may represent the progenitor of modern wolves, there is little evidence that it moved to the south of the glaciated region before the late Rancholabrean. Western North America appears to have much the same evolutionary sequence of Canis that occurred in the east. C. lepophagus and C. priscolatrans were widespread, respectively, in the Blancan and early Irvingtonian, and C. dirus was abundant in the late Rancholabrean. Unlike eastern North America, the west has yielded no substantial collections of middle Irvingtonian to early Rancholabrean Canis. Three fragments from Hay Springs and Mullen, Nebraska, which do date from that period, were assigned to C. lupus, and one from Rushville, Nebraska, was referred to C. armbrusteri. In retrospect, it is questionable whether a population identical to modern C. lupus was then present to the south of the glaciers. Each of the Nebraska specimens is considerably larger than C. mosbachensis .... It is unlikely that C. lupus could then have been so far removed, morphologically and geographically, from its progenitor species. Perhaps the Nebraska material represents a population of C. armbrusteri that was beginning to develop in the direction of C. dirus; both the latter species and C. lupus have some parallel differences from C. armbrusteri. In any case, C. lupus is not otherwise known from North America before the late Rancholabrean, when for a time it evidently was sympatric with C. dirus.
By the terminal Pleistocene, the east had been reoccupied by a wolf differing from C. dirus, C. armbrusteri, and C. latrans. Assessment of later specimens indicates continuity of this small eastern wolf. There is no significant difference in size ... of paleontological fragments dating about 10,000 ybp, archeological fragments dating 2,000-200 ybp, and complete, mostly modern skulls. However, there is a significant difference ... between each of those samples and western C. lupus and C. latrans. That the latter two species overlap at all with the other samples may be attributable in part to inclusion of both sexes.
The number of early mandibular fragments that has been recovered east of the Mississippi is considerably greater than that of maxillary fragments. Of the latter, eight ... from Ohio, Pennsylvania, New York, Maryland, and North Carolina ... seem closely related .... In the eight complete, mostly modern skulls from east of the Mississippi, including two females, range is 21.8-25.2 mm for P4 and 20.5-22.0 mm for M1. The ninth eastern maxillary fragment from Frontenac Island, Cayuga County, western New York, dates from at least 5,000 ybp. At 27.5 ,, in P4 length and 24.4 in M1 diameter, it is the size of a large C. lupus and may represent an early movement of that species into the northeast.
The Frontenac Island specimen is the only indication that, at least for a time, two different wolf species (C. lupus and C. rufus) occurred in parts of the east subsequent to the Pleistocene, as was suggested by earlier studies. However, measurements of the other eastern material show no more variation than do those of other wolf populations. Moreover, the diagnostic teeth of the fragmentary specimens are approximately the same size as those in the series of complete eastern skulls.
The complete skulls can be assessed through multivariate procedures to determine whether they represent a single species that differs from other well-defined species. One such analysis ... deals with the six available skulls of fully developed males dating prior to 1918 and collected in the region of interest. Those six specimens are compared as a group with the six groups of C. lupus from western North America, including the three samples geographically most proximal to the eastern group. The series of C. lupus all overlap one another and are not similar morphologically to the eastern material, even as they approach the latter geographically. The eastern specimens group together separate from all the western specimens and, although they were taken from a much larger region than were any of the western samples, they show less statistical variation than do most of the latter.
A second analysis compares the same six specimens, as a group, with the six samples of western C. lupus, combined as one group, and the series of 96 western male C. latrans. In addition, the seven skulls of males, taken just west of the Mississippi in northeastern Louisiana in 1898-1905, and the one specimen from Garvin County, Oklahoma, were tested as individuals against the three defined series. Again, the six eastern skulls form a statistical group distinct from C. lupus and also far separated from C. latrans. The Louisiana and Oklahoma specimens fall within or near the range of variation of the six skulls from farther east.
The 14 pre-1918 skulls of eastern male wolves evidently represent a statistically well-defined species. They were combined in a single group and compared to western C. lupus, likewise in a single group. Another series tested in that analysis comprised the 10 oldest available specimens of male C. lupus lycaon from southeastern Ontario and extreme southern Quebec. Those 10 skulls were taken 1905-1933 and are unlikely to represent hybridization with C. latrans, which apparently subsequently affected lycaon in southeastern Canada. The coyote was first reported in southeastern Ontario in 1919, but that record was only 100 km northeast of Detroit, Michigan, and the species did not become established farther east until the 1940s. Also used in the analysis was the series of C. lupus collected from 1905 to 1965 in the upper peninsula of Michigan.
The analysis once again shows complete and wide statistical separation between the series of southeastern wolves and western C. lupus. Lycaon is intermediate to those populations, just about filling the statistical gap between them. While lycaon does not overlap with the southeastern wolves and only slightly with western C. lupus, it does show more affinity to the sample from upper Michigan. That sample is from an area geographically between the range of lycaon and that of the Minnesota/Isle Royale sample of western C. lupus. A single specimen collected prior to 1855 in the Adirondacks of northern New York, was tested as an individual against the four series used in the analysis. Its statistical position falls among those of lycaon from just to the northwest and it may be part of the same population. It was not included in any of the series, as it may not be fully developed, but it probably would not have grown much larger. Its general size is about that of males from the southeastern series and smaller than that of any male lycaon or another C. lupus assessed in this study. In proportion and dental characters it more closely resembles C. lupus than it does the southeastern series. In those respects it contrasts sharply with an 1863 specimen from Maine, which has the slender proportions and well-sculptured molars typical of the southeastern wolves.
The morphological approach seen in the northeast is not evident as the wolf populations of the southeast geographically converge with western C. lupus. The original situation on the western edge of the range of the southeastern wolves, especially in central Texas, is partly obscured by hybridization with C. latrans, but prior studies have shown C. lupus to be statistically well removed from the affected population. There is one old series that may represent the unmodified southeastern wolf near the western extremity of its range: the five males collected in 1900 in Calhoun County on the south Texas Gulf coast. A single female, morphologically compatible with those males, also was taken there, but no other specimens -- identified as C. rufus, C. latrans, or any other wild Canis -- are known from that area and period. The Calhoun County males were compared to three other groups of males: the six specimens from east of the Mississippi and dating prior to 1918, the seven taken 1898-1905 in northeastern Louisiana, and western C. latrans. The Calhoun County group was found to occupy a statistical position separate from those of the more easterly series but not intermediate to those two and C. latrans. Visually, the Calhoun County specimens are smaller and narrower than the other two and have relatively smaller teeth.
Considered alone, the statistical distributions ... may suggest that the Calhoun County skulls represent a separate species, intergrading neither with more easterly wolf populations nor C. latrans. Both the Calhoun County and northeastern Louisiana series date from about a century ago and, unfortunately, there are no contemporary samples of males from geographically intermediate areas, and hence no direct evidence for or against intergradation. A series of males was collected from 1919 to 1943 farther north along the Texas coast, in the Big Thicket area of inland southeastern Texas, and in southern Louisiana and Mississippi.
The 1919-1943 specimens ... fill most of the statistical gap between the Calhoun County and more easterly series. The Big Thicket and south Mississippi skulls fall close to the northeast Louisiana sample, while the Texas coastal and south Louisiana skulls are nearer to the early Calhoun County series. The pre-1869 skull from south-central Oklahoma also was tested in this analysis and falls precisely between the northeast Louisiana and east-of-the-Mississippi series.
The 1919-1943 series shows no statistical tendency to approach or blend with C. latrans; such a tendency might have been expected if that series had experienced substantive introgression from the latter species. however, hybridization with C. latrans most certainly had begun at the western edge of the range of C. rufus earlier in the century. Five males, collected 1899-1906 in Edwards and Kerr counties, central Texas, are statistically intermediate to C. latrans and coastal wolves, and may represent initial hybridization between the two species.
A much larger series from central Texas was collected 1915-1918 and shows a complete statistical blending of C. latrans and the southeastern wolf, though the entire involved population seemingly was extirpated by human agency shortly thereafter. Additional hybrid populations formed in the Ozark region and pushed to the south and east in ensuing decades. Nonetheless, a largely unmodified wolf population apparently persisted in extreme southeastern Texas and southern Louisiana until the 1970s. The close morphometric resemblance of that population, and of 14 individuals removed therefrom the begin the existing captive/reintroduced population, to the original southeastern wolf population has been documented (Nowak, 1979, 1992). This study centers on the original, not current, status of eastern wolves. However, to briefly update the current situation, six recently collected individuals (seen at Alligator River National Wildlife Refuge, Dare County, North Carolina, 17 April 1999) were compared to the three series of old southeastern wolves and to C. latrans. Those six are males that were born in the wild and raised to maturity in the population reintroduced in North Carolina, and were the only available specimens meeting those criteria. The six skulls are statistically close to the century-old series from Calhoun County, though there is some approach to the old northeast Louisiana series.
The view that the red wolf is a primitive species, closely related or identical to C. priscolatrans of the early Irvingtonian, is not supported by this study. It is conceivable that a small wolf, descended from C. priscolatrans, persisted in the eastern forests after the period and has not yet shown up in the fossil record, though that now seems unlikely. Available material suggests an archaic New World evolutionary sequence, from the Blancan C. lepophagus (or some related small species) through the early Irvingtonian C. priscolatrans, to the late Irvingtonian and early Rancholabrean C. armbrusteri, and finally culminating in the late Rancolabrean C. dirus .... [T]he outline given here may be an oversimplification, as there probably were additional species involved and some phases of the sequence that extended beyond eastern North America. In particular, one or two species of South America may represent part of the transition from C. armbrusteri to C. dirus. With the extinction of the dire wolf, about 8,000 ybp, the entire New World wolf line terminated.
The New World evolutionary progression was primarily in the direction of increasing size, from a relatively small to a relatively large species of Canis. There was little or no overlap in the size of the species present during any one period .... A coyote-sized species reappeared in the east, in the form of modern C. latrans, only when the wolf line had grown much larger, becoming the species C. armbrusteri in the late Irvingtonian to early Rancholabrean. After C. priscolatrans disappeared, nearly a million years passed before a small wolf again was evident in the east. This wolf could have been a direct descendant of the Eurasian C. mosbachensis. The above data suggest that C. mosbachensis was considerably smaller than most modern wolves. Indeed, Kurten (1968) reported C. mosbachensis to be about the size of C. lupus pallipes Sykes, the living wolf of southwestern Asia, which Nowak (1995) indicated is smaller than C. rufus. Kurten (1968) noted that C. mosbachensis was still present in Europe during a period corresponding to the early Rancholabrean of North America and that transition to the larger C. lupus did not occur until the time of the late Rancholabrean. Such a background offers the possibility that C. mosbachensis gave rise to small wolves that entered North America, became isolated by glaciation, and developed into the modern eastern populations. Eurasian mosbachensis then may have evolved into C. lupus and subsequent invasions of North America may have led to the modern subspecific differentiation of the latter species. Ecological space for the modern small wolf, as well as the modern small coyote, became available in the east only when the archaic New World wolf line had grown to enormous size, becoming C. dirus, in the late Rancholabrean. Both the small coyote and large wolf disappeared from the east at the end of the Pleistocene but the new small wolf persisted until the 20th century. Remarkably, when that wolf was itself extirpated in our own times, the coyote again occupied the east.
Kurten (1974) regarded C. priscolatrans as part of the coyote line leading from C. lepophagus to modern C. latrans. This study suggests that while C. lepophagus may have given rise to both C. priscolatrans and C. latrans, the latter represents a separate line of descent. It may have remained primarily in the west, moving eastward only when the niche for a smaller Canis sufficiently widened. A large subspecies of C. latrans was abundantly sympatric with C. dirus at Rancho La Brea and some other western Rancholabrean sites. Replacement of C. dirus by C. lupus may be a factor in the smaller size of modern C. latrans in the west. Eastern coyotes are not well represented in the fossil record ....The influx of a small wolf, apparently late in the Rancholabrean, may have been a factor in the ultimate disappearance of the coyote from most of the east.
From the end of the Pleistocene until the mid-20th century, the coyote was absent from the region of interest, except for a zone of perhaps 100-200 km extending eastward and southward from the prairies. Archeological records indicate that C. latrans, at least at times, occurred as far as southern Indiana, southern Missouri, and northwestern Arkansas. In 1919-1925, a series of C. latrans was collected in and around the St. Francois Mountains of southeastern Missouri, together with a large series apparently representing an isolated wolf population. Previous study showed no morphometric overlap between the two series, hence demonstrating that each represented a separate species and that hybridization had not yet developed in that area. All recent specimens from that period and earlier, and from farther south and east within the zone of interest, are wolves.
It again is emphasized that there are only six known complete and fully developed skulls of male wild Canis dating prior to 1918 and collected in the region of interest .... Using multivariate analysis, those six skulls group together and have a statistical distribution completely separate from the extensive overlapping distributions of six samples of western C. lupus, including the samples most proximal to the eastern series. Those six specimens also are completely removed from the statistical distribution of C. latrans. Seven additional complete skulls of males from northeastern Louisiana, taken 1898-1905, and one collected prior to 1869 in south-central Oklahoma, have about the same statistical distribution as the six more easterly specimens, when compared to C. lupus and C. latrans. Such an arrangement argues for recognition of the eastern sample as a distinct species that is appropriately named C. rufus. Moreover, that species, as defined by the complete eastern skulls, has diagnostic measurements that are statistically identical to those of two series of fragmentary specimens, collected in the same region and dating from around 10,000 ybp and from 2,000-200 ybp. Although not conclusive, this evidence supports the view that C. rufus has continuously occupied the east since the terminal Pleistocene and that it is the only species of wild Canis that was present in most of the region.
None of the archeological or modern specimens, either examined as part of this study or reported by others, indicates the presence of C. latrans, or hybridization between C. latrans and another wild species in the southeast between 10,000 and 100 ybp. Subsequently, C. latrans did recolonize the southeast and begin to hybridize with the native wolf, C. rufus. This hybridization process signaled the end of that species, not its beginning. This study thus is not in agreement with some of the recent DNA analyses, particularly the suggestion that C. rufus originated from hybridization between C. latrans and C. lupus, probably within the last 250 years as a result of environmental disruption by European colonists.
Surprisingly, this investigation provides evidence that another named kind of eastern wolf did have a hybrid origin. The oldest available series of C. lupus lycaon from the northeast is statistically intermediate to C. rufus and western C. lupus. The morphological similarity of rufus and lycaon has long been recognized and has confused the systematic status of eastern Canis. Recognition of lycaon -- specifically the population of extreme southeastern Ontario and southern Quebec, and possibly northern New York -- as a hybrid provides a solution to the problem of why the red and gray wolf differ so much along most of the line where their ranges meet, but resemble one another so closely in the northeast.
That solution also is in keeping with mitochondrial DNA analyses indicating that C. rufus and lycaon contain genetic sequences similarly divergent from C. latrans and differing from those of C. lupus. Wilson et al. (2000) interpreted those analyses to mean that rufus and lycaon form a single species, independent of C. lupus and C. latrans, that would appropriately be known as C. lycaon. However, another plausible interpretation is that the modern range of lycaon originally had been occupied by C. lupus, which had moved in from the west following the retreat of the glaciers at the end of the Pleistocene, as attested by the large specimen from Frontenac Island in western New York. C. rufus would also then have moved northward to the southern edge of that area. That newly available and perhaps unstable habitat may have been conducive to hybridization.
The predominant flow of introgression probably was from rufus to lycaon, with consequent modification of the latter, rather than the reverse. Although lycaon is statistically near rufus, available material shows no definite overlap, whereas lycaon shows greater statistical similarity with C. lupus, especially the original population of northern Michigan, which is geographically most proximal. Lycaon also does not substantively differ ecologically and behaviorally from other gray wolf subspecies. Lycaon may still be regarded as a subspecies of C. lupus, but one that perhaps developed as a result of natural hybridization with C. rufus. While lycaon thus would represent an early wave of C. lupus that had invaded North America and become isolated by glaciation, it and the other gray wolf subspecies suggested by previous study should for now be retained.
....The multivariate analyses reinforce Goldman's (1944) designation of three subspecies of C. rufus. However, some rearrangement of subspecific lines is advisable, based on assessment of the oldest available material. C.r. floridanus apparently occurred all along the Atlantic coast, from Maine to Florida, and inland to Ohio and northern Alabama. C.r. gregoryi was found throughout the lower Mississippi Valley, including some country along the Ohio and Red Rivers, and in the Big Thicket area of eastern Texas. C. r. rufus occupied the Texas coast and probably central Texas, though it is known from the latter area only as a hybrid with C. latrans. There is no conclusive evidence that it originally occurred farther north. Although Goldman reported its range to reach Oklahoma, southwestern Missouri, and northwestern Arkansas, the pertinent specimens probably express the spread of hybridization with C. latrans and hence smaller size. The oldest known Oklahoma skull, the pre-1869 specimen from Garvin County, shows statistical affinity to gregoryi/floridanus.
....Several specimens from central Texas show that C. rufus had begun to hybridize with C. latrans by 1900. The same process was evident not long afterwards in eastern Oklahoma and western Arkansas. The further spread of hybridization and of coyote-like animals through the southeast is well documented. However, statistical evidence also shows that the population of wild Canis that persisted in southeastern Texas and southern Louisiana until the 1970s, as well as individuals removed from that area and used to found the existing captive/reintroduced population, were similar to the original red wolf. Available specimens indicate that the breeding population now established in the wild in North Carolina retains the morphometric characterization of C. rufus rufus.
Figure 1: Distribution of North American Canis.
Figure 2: Distribution of eastern Canis.
Figure 3: Bivariate analysis comparing measurements of transverse diameter and anteroposterior length of M1 in individual specimens of fossil eastern Canis.
Figure 4: Length (in millimeters) of M1 in eastern fossil Canis.
Figure 5: Length (in millimeters) of m1 in six samples of both sexes of Canis.
Table 1: Results of the Tukey-Kramer Comparison Test of length of m1 in the six groups of Canis shown in Figure 5.
Figure 6: Statistical distribution of seven groups of American male Canis, plotted on the first and second canonical variables.
Figure 7: Statistical distribution of three groups and certain individuals of North American male Canis, plotted on the first and second canonical variables.
Figure 8: Statistical distribution of four groups and one individual of North American male Canis, plotted on the first and second canonical variables.
Figure 9: Statistical distribution of four groups and one individual of North American male Canis, plotted on the first and second canonical variables.
Figure 10: Statistical distribution of four groups and certain individuals of North American male Canis, plotted on the first and second canonical variables.
Appendix 1: Paleontological and Archeological Specimens Examined.
Appendix 2: Specimens Used in Multivariate Analyses.