OUP user menu

A Genetic Assessment of the Eastern Wolf (Canis lycaon) in Algonquin Provincial Park

Sonya K. Grewal, Paul. J. Wilson, Tabitha K. Kung, Karmi Shami, Mary T. Theberge, John B. Theberge, Bradley N. White
DOI: http://dx.doi.org/10.1644/1545-1542(2004)085<0625:AGAOTE>2.0.CO;2 625-632 First published online: 16 August 2004


Recent genetic data indicate that the eastern wolf is not a subspecies of the gray wolf (Canis lupus), but is a North American wolf more similar to the red wolf (C. rufus) and closely related to the coyote (C. latrans). The eastern wolf has been proposed as a separate species, C. lycaon. The largest protected area containing this wolf is Algonquin Provincial Park in Ontario, Canada, which is bounded to the south by areas containing the Tweed wolf or eastern coyote, a hybrid of the western coyote and eastern wolf. We assessed the relationships of animals in the park by using DNA profiles that comprised the genotype from 17 autosomal and 4 Y-linked microsatellite loci and the mitochondrial DNA control region. These profiles were used to establish maternity, paternity, and kin relationships for 102 wolves that were studied from 24 packs over a 12-year period. Genetic data do not support the hypothesis that a pack comprises an unrelated breeding pair and their offspring. There is evidence of frequent pack splitting, pack fusion, and adoption. Some unrelated individuals in the packs were identified as immigrants into the park. We found high levels of genetic structuring between the Tweed wolves to the southeast and the Algonquin Park wolves (RST = 0.114). Lower levels of genetic differentiation with animals to the north and west (RST = 0.057 and RST = 0.036) and high genetic diversity suggest that park animals are not an island population but the southern part of a larger metapopulation of C. lycaon.

Key words
  • Algonquin
  • Canis lupus
  • C. lycaon
  • C. rufus
  • metapopulation
  • structuring

Recently, genetic data (Wilson et al. 2000) have been used to suggest that the eastern timber wolf is not a subspecies of the Eurasian-evolved gray wolf (Canis lupus), but is the same species as or a close relative of the red wolf (C. rufus), found in the southern United States. Wilson et al. (2000) proposed that the eastern wolf be given the original taxonomic designation of C. lycaon (Miller 1912). Both C. rufus and C. lycaon are thought to have a common origin, evolving in North America, with the western coyote (C. latrans) diverging from that lineage 150,000–300,000 years ago. This close evolutionary relationship between C. lycaon and C. rufus is consistent with their ability to readily hybridize with western coyotes and the absence of hybridization between western and northern gray wolves and western coyotes (Wilson et al. 2000).

A significant number of protected packs of the eastern timber wolf are found in Algonquin Provincial Park, Ontario, Canada, which has been studied for about 40 years (Forbes and Theberge 1996a, 1996b; Pimlott et al. 1969; Theberge and Theberge 2001). However, a number of threats to the persistence of these packs have been identified, including human-caused mortalities when wolves leave the park and potential gene swamping from coyotes.

Before European settlement, gray wolves were thought to have occupied all of Ontario (Standfield 1970), preying primarily on larger ungulates, such as moose (Alces alces). In the 19th century, Algonquin Provincial Park underwent some significant changes as a result of logging activities, reducing gray wolf numbers and facilitating the arrival of white-tailed deer (Odocoileus hemionus) and the smaller eastern timber wolves (Franzmann and Schwartz 1997; Peterson 1955). At the beginning of the 20th century, western coyotes reached southern Ontario (Moore and Parker 1992) and hybridization with eastern wolves followed. This resulted in areas containing a diverse range of sizes of wolves and coyotes for which the term canid soup was coined (Standfield 1970).

The packs on the east side of Algonquin Provincial Park have been the subject of an intense study involving the radiocollaring of 150 wolves over 12 years. It has been estimated that in the winter there are 170–200 animals in 30–35 packs (Forbes and Theberge 1996a, 1996b). This estimate is lower than that for the 1960s of approximately 300 animals (Pimlott et al. 1969), which was at that time suggested to be the carrying capacity of the park. Algonquin wolves migrate 15–70 km (Forbes and Theberge 1995, 1996b) each winter to the deeryard (an area where deer congregate to feed in the winter) located 13 km southeast outside the park (Forbes and Theberge 1995). This migration appears to result in the high mortality observed (50 animals/year), of which 60% is due to human activity outside the park boundary.

Early studies on the dispersal and social structure of wolves were based primarily on limited observations on gray wolf packs in Alaska (Murie 1944; Rausch 1967) and the Northwest Territories (Fuller and Novakowski 1955), and on eastern timber wolves from Isle Royale (Jordan et al. 1967), Minnesota (Mech 1970; Olson 1938), and Ontario (Kolenosky and Johnston 1967; Pimlott et al. 1969). These data were used to formulate a generally accepted model of a wolf pack as a breeding pair (alpha male and alpha female) and their offspring. The subordinate animals or nonbreeding adults were thought to be offspring from the previous year. Little consideration was given to possible significant differences in pack structure between northern gray wolves and eastern timber wolves, and deviations from the model usually were considered to result from odd and unusual events. Recent studies have used more intense and long-term radiocollaring and tracking techniques (Forbes and Theberge 1996b; Fritts and Mech 1981; Van Ballenberghe et al. 1975) and genetic profiles (Forbes and Boyd 1997; Laikre and Ryman 1991; Lehman et al. 1992; Meier et al. 1995; Smith et al. 1997; Wayne et al. 1995). These studies suggest a more complex pack structure resulting from pack formation by wolves dispersing from nearby territories, the splitting of existing packs, and frequent pack mergers or adoptions.

Because Algonquin Provincial Park contains the most intensely studied packs of the eastern wolf, an objective of this study was to test previous hypotheses of pack composition. We set out to answer the following questions: Is a pack simply composed of an unrelated breeding pair and their offspring? If not, what is the origin of the unrelated nonbreeding or subordinate animals? Do the packs in Algonquin Park represent an island population? To address these questions, we used 3 different types of genetic markers: 17 autosomal microsatellite loci, the mitochondrial DNA control region, and 4 Y-chromosome microsatellite loci.

Materials and Methods

Samples and DNA extraction.—Over a 12-year field study (1987–1999_Forbes and Theberge 1995; Theberge and Theberge 2001), 150 animals from 35 packs in Algonquin Provincial Park (46°N, 79°W) were livetrapped and radiocollared (Fig. 1; American Care and Use Committee 1998). DNA was extracted from radiocollared animals and 5 nonradiocollared animals by methods described in Gulgich et al. (1994) from frozen organ samples (liver, heart, kidney, or muscle) or from whole blood obtained by venipuncture of individuals that were livetrapped and released. Genetic profiles of 102 samples were determined. Eighty-five of these animals represented 27 of the 35 packs. The remaining 17 animals were not assigned as belonging to a specific pack. The study focused primarily on the east side of Algonquin Provincial Park, with pack boundaries established from the radiocollaring data (Forbes and Theberge 1995). The presence or activity of a pack in a given year was defined by the presence of at least 1 collared and sampled individual from that pack. To estimate the proportion of individuals sampled from each pack (P) over the 12 years of study, the following equation was used: Embedded Image where Np equals the number of packs and N/12 is the average number of animals per pack. The number of animals recruited (births and migration) has been estimated at 30% per year (Theberge and Theberge 2001).

Fig. 1

Distribution of the 24 packs studied on the eastern side of Algonquin Provincial Park, Ontario, Canada (shaded area), between 1987 and 1999. Pack boundaries are indicated by ellipses. Ellipses are not statistically determined, but are an estimate of pack areas from previous radiocollaring data (Forbes and Theberge 1995; Theberge and Theberge 2001). The dotted ellipse represents the Foys pack beginning to dissolve as the existing Basin, Jocko, and Redpole packs extend their territories into part of the Foys area as the MacDonald pack begins to form. For a more accurate depiction of some territories, see Forbes and Theberge (1995). The star represents the Round Lake deeryard to which animals migrate in the winter.

Animals in regions surrounding Algonquin Provincial Park (Fig. 2) are grouped into the Frontenac Axis (n = 74), the Magnetawan region (n = 26), and northeastern Ontario upper-Great-Lakes region (n = 33). There also were samples from 2 groups in Quebec, 1 south of the Abitibi-Temiscamingue region (n = 13) and the other from La Verendrye Reserve (n = 13).

Fig. 2

Sampling location for the 6 groups of wolves sampled in Ontario and Quebec, Canada. Algonquin Provincial Park (n = 102) and La Verendrye Reserve (n = 13) sampling locations are shown as park boundaries (white, hatched). The Frontenac Axis (n = 74), the Magnetawan region (n = 26), and the Abitibi–Temiscamingue region (n = 13) are indicated by black outlines. The northeastern Ontario upper-Great-Lakes region (n = 33) is area between black dotted lines. The shaded area represents the current range of Canis lycaon in Ontario and Quebec, Canada.

Genetic markers.—A 343- to 347-base-pair product of the mitochondrial DNA control region was amplified and sequenced (Pilgrim et al. 1998; Wilson et al. 2000). Seventeen autosomal microsatellite loci (cxx.377, cxx.172, cxx.123, cxx.109, cxx.225, cxx.250, cxx.200, and cxx.204—Ostrander et al. 1993; cxx.147, cxx.253, cxx.383, cxx.410, cxx.442, cxx.606, and cxx.2010— Ostrander et al. 1995; cph11—Fredholm and Wintero 1995; and c.2202—Francisco et al. 1996) were genotyped. Four Y-chromosome microsatellite loci were amplified by using 2 primer sets (MS34 and MS41) characterized by Olivier et al. (1999). Haplotypes were established by using the combination of alleles at the 4 loci amplified. All microsatellites were genotyped by using conditions described in Wilson et al. (2000).

Statistical analysis.—To assess genetic variation, allele frequencies, expected heterozygosity (HE), observed heterozygosity (HO), and allelic diversity (A), we used the software program Cervus (Marshall et al. 1998). FIS (Weir and Cockerham 1984) was calculated by using the program Genetix 4.02 (Belkhir et al. 1999).

Three data sets were employed to assess parent–offspring relationships. First, genetic exclusion of a putative parent was determined if the parent did not share at least 1 allele at each of the 17 loci with the offspring. Second, a mother was excluded if her mtDNA haplotype was not identical to that of her putative offspring, and third, a father was excluded if his Y-chromosome haplotype was not identical to that of his putative son.

To assess kin relatedness allele frequencies were calculated for 17 microsatellite loci. To avoid bias, the genotypes of all known offspring were omitted from the allele-frequency calculation. By using KINSHIP 1.2 software (Queller and Goodnight 1989), 3 simulations were performed: 1,000 randomly generated pairs of unrelated individuals (r = 0), 1,000 pairs of half siblings (r = 0.25), and 1,000 pairs of full siblings (r = 0.5). From the simulation data, the mean and SD of unrelated, half-sibling, and full-sibling distributions were calculated. Relatedness values can range from −1.0 to 1.0, in which a negative value means that the dyads share fewer alleles than the average population. By using the software program STATISTICA (StatSoft, Tulsa, Oklahoma), a confidence level of 95% was calculated for each distribution to classify dyads. Individuals with r-values > 0.518 were identified as full siblings. Unless specific relationships were being tested, individuals with r-values > 0.238 were identified as being related (the upper 95% confidence level for unrelated). The index weights each allele inversely by its frequency in the population, so rare alleles are given a relatively higher weighting.

Genetic structuring (ϕ) between populations based on mitochondrial DNA and Y-chromosome haplotypes was estimated by using the analysis of molecular variance program AMOVA version 1.55 (Excoffier et al. 1992). Genetic structuring between populations based on microsatellite data was estimated by using RST (Slatkin 1995) and FST (Weir and Cockerham 1984). Levels of significance for the pairwise RST values were calculated after 1,000 bootstraps and permutations of the data by using the computer program RST CALC (Goodman 1997). Theta was calculated for each population pair and its significance was assessed with 1,000 permutations of the data by using the program GENETIX 4.02 (Belkhir et al. 1999). Immigrants into Algonquin Park were identified by using the software program STRUCTURE (Pritchard et al. 2000). This provides a conservative minimum estimate because it does not always identify immigrants from areas with similar allele frequencies. Confidence levels of 95% were used to determine if an animal was an immigrant.


We estimated that approximately 445 different wolves would have been present in the 27 Algonquin Park packs over the span of 12 years. Therefore, the 102 samples collected represent approximately 23% of the individuals present in the 27 packs over that time period.

Only 15 of these 27 packs had 3 or more members sampled. Based on individuals living concurrently in a pack, we assessed maternity, paternity, and kin relationships. The number of individuals sampled in each of the packs fluctuated considerably over the 12 years. As packs dissolved, others formed or expanded into their territory. For example, as the Foys pack (Fig. 1) began to dissolve in the early 1990s, the existing Basin, Jocko, and Redpole packs extended their territories into part of the Foys area and the MacDonald pack began to form.

The number of alleles present at the autosomal microsatellite loci in the park ranged from 4 at locus cxx.204 to 25 at locus c.2202. Within the mitochondrial control region, 8 haplotypes (C1, C9, C13, C14, C16, C17, C19, and C22; Table 1) were found in the park. One (C22) was of gray wolf origin and was found in only 4 full siblings, 2 (Cl and C9) were of eastern wolf origin, and 5 (C13, C14, C16, C17, and C19) were of western coyote origin. Seven Y-chromosome haplotypes (AA, BB, CC, CD, CE, DC, and EF) were identified (Table 2). The 2 frequent haplotypes AA and BB appeared to be of eastern timber wolf origin, whereas none appeared to be of western coyote origin (Shami 2002).

View this table:
Table 1

Number of individual wolves with each mitochondrial control region haplotype in Algonquin Provincial Park, Ontario, Canada, and surrounding areas.

HaplotypeAlgonquin Provincial ParkFrontenac AxisMagnetawan regionNortheastern OntarioAbitibi–Temiscamingue regionLa Verendrye Reserve
View this table:
Table 2

Variation in the Y chromosome in wolves in Algonquin Provincial Park, Ontario, Canada, and surrounding areas.

HaplotypeAlgonquin Provincial ParkFrontenac AxisMagnetawan regionNortheastern OntarioAbitibi–Temiscamingue regionLa Verendrye Reserve
% of males in population sampled503635275446

Based on the kinship assessment, we identified breeders (adults that have a putative offspring), nonbreeders (or subordinates that have no identified offspring), and offspring. We only identified 3 packs (Jocko, MacDonald, and ZigZag) that contained relationships consistent with the hypothesis that packs primarily are an unrelated pair and their offspring. In each of these packs, at least 1 parent but no nonbreeding adults were identified, although the number of samples in 2 of the 3 packs was small.

Putative parent-offspring relationships were identified in 7 other packs; however, in addition to the breeder, these had at least 1 additional adult present. Some packs in a given year had as many as 3 adults of the same sex present. Furthermore, 5 packs were identified with no putative parent–offspring relationships, and all but 1 pack had at least 2 adults of the same sex in a given year. Overall, the members of a pack were variable, and examination of these data shows that packs in the park are rarely an unrelated pair and their offspring.

To assess relationships of nonbreeding adults present in a pack, we focused on 2 categories of packs. In the packs where no putative parent–offspring relationships were identified, we found that all adults were unrelated to each other. Similarly, in packs with an identified breeder, the additional adults in that pack were most often unrelated (46%) to that breeder. Less often, nonbreeding adults were identified as offspring (33%) of the breeder or related to the breeder as either siblings (13%) or grandparents (8%).

We also examined whether male or female offspring were more likely to remain in a pack, by assessing whether Y-chromosome or mitochondrial haplotypes were more persistent in a pack over generations. We looked at 7 packs that had 1 or more members sampled for at least 10 of the 12 years (Table 3). Five of the packs had a single persistent Y haplotype (Table 3). Among the females in these packs, the number of mitochondrial haplotypes ranged from 1 (ZigZag and Redpole packs) to 4 in the Basin pack. The remaining 2 packs (Jackpine and Travers) each had 3 Y haplotypes present and 1 mitochondrial haplotype in the females (Table 3). Although the Y haplotype appears to be more persistent than the mitochondrial haplotype, the persistent Y haplotypes are most often the common AA haplotype, which occurs in 58% of the population.

View this table:
Table 3

Mitochondrial DNA and Y-chromosome haplotypes found in packs with at least 1 individual wolf sampled in a given year over 10 of the 12 years of study.

PacksY haplotypeNo. individualsMtDNA haplotypeNo. individuals
Jackpine (n = 9)AA4C143
Basin (n = 7)AA2C12
Jocko (n = 7)AA2C11
Travers (n = 7)AA1C142
Redpole (n = 4)BB3C192
Military (n = 3)AA1C141
ZigZag (n = 3)AA1C142

To assess potential gene swamping from eastern coyotes and genetic structuring among Algonquin Park animals and animals in the surrounding regions, we examined the origin of the animals in the park. By using the software program STRUCTURE (Pritchard et al. 2000), we identified 5 immigrants into the park. Three appeared to originate from the Frontenac Axis and 1 from the Magnetawan region. These immigrants had genotypes indicating the presence of significant coyote genetic material. Two of the immigrants were among the 5 smallest animals found in the park. The 5th immigrant appeared to originate from north of the park and showed a higher content of more gray wolf alleles. There was no evidence that any immigrants reproduced in their respective packs.

We also examined levels of genetic differentiation between the park animals and those in surrounding regions. By using ϕSt in the program AMOVA, mitochodrial and Y-chromosome haplotype structuring among the regions was identified (Table 4). By using the mitochondrial and Y haplotypes, animals from the northeastern Ontario upper-Great-Lakes region, the Abitibi–Temiscamingue region, and La Verendrye appear to be more similar to each other than to those in Algonquin Park, the Magnetawan region, and the Frontenac Axis. To further assess structuring, we calculated both RSt and FSt by using the data at 8 microsatellite loci (Table 5). Similar relative values were found with RST and FST. With the exception of the Magnetawan region, animals from the Frontenac Axis show higher levels of genetic differentiation than animals in other areas, including Algonquin Provincial Park and those in Quebec. Lower levels of structuring are apparent across the western and northern borders of Algonquin Park between the northeastern Ontario upper-Great-Lakes region and its surroundings, which include the Abitibi-Temiscamingue region and La Verendrye Reserve, as well as the Magnetawan region and Algonquin Provincial Park. These data are supported by the low FIS values and the high levels of allelic diversity (A) and observed heterozygosity (HO).

View this table:
Table 4

ϕST values for Y haplotype (above diagonal) and mitochondrial haplotype (below diagonal) for each pairwise comparison of canids from Algonquin Provincial Park, Ontario, Canada, and surrounding regions.

AreaFrontenac AxisAlgonquin Provincial ParkMagnetawan regionNortheastern OntarioAbitibi–Temiscamingue regionLa Verendrye Reserve
Frontenac Axis0.0568−0.00540.11140.14160.2321
Algonquin Provincial Park0.0594−0.05780.13310.13560.3250
Magnetawan region0.00900.05830.04930.05450.2441
Northeastern Ontario0.23540.13340.2282−0.05690.0733
La Verendrye Reserve0.32920.25360.36510.04240.2664
View this table:
Table 5

RST (above diagonal) and FSt (below diagonal) values for each pairwise comparison of canids from Algonquin Provincial Park, Ontario, Canada, and surrounding areas.

AreaFrontenac AxisAlgonquin Provincial ParkMagnetawan regionNortheastern OntarioAbitibi–Temiscamingue regionLa Verendrye Reserve
Frontenac Axis0.1140.0400.1020.1610.190
Algonquin Provincial Park0.0550.0570.0360.0700.067
Magnetawan region0.0240.0210.0160.0470.066
Northeastern Ontario0.0760.0720.0520.0250.018
La Verendrye Reserve0.0910.0510.0430.0120.017


In this study, we examined the conventional view of relationships among individuals in a wolf pack. We tested the hypothesis that a pack of eastern wolves predominantly is composed of an unrelated pair of breeding adults and their offspring by examining the DNA profiles of > 100 animals from Algonquin Provincial Park. In most packs, which had a known male or female breeder, at least 1 additional nonbreeding, unrelated adult of the same sex as the breeder was identified. In packs where no breeder was identified, at least 2 unrelated adults of the same sex were identified. The hypothesis is therefore rejected, and examination of the data clearly shows that a pack of eastern wolves in Algonquin Provincial Park is not simply an unrelated pair and their offspring.

Most nonbreeding adults in the packs studied appear to be unrelated to the breeder and to other nonbreeding adults within the pack. Many of the unrelated adults were found in peripheral packs (Fig. 1) such as Travers, Northeast, and Pretty. In addition to not being related to other pack members living at the same time, few had relationships with animals in surrounding packs. Two members of the Travers pack were identified as immigrants into the park. Other unrelated nonbreeders appear to have originated from packs that dissolved as others expanded into overlapping territories. In some of these packs, 3 generations of animals were identified. These data are consistent with studies on eastern wolves of Minnesota that suggest many pack members often originate through pack adoptions or fusions and through pack splitting or budding (Fritts and Mech 1981; Lehman et al. 1992; Mech and Nelson 1990).

A possible explanation for the complex pack structures identified in Algonquin Park is the high mortality observed when animals leave the park (Forbes and Theberge 1995). This mortality primarily is a consequence of long-range excursions by wolves outside park boundaries in search of deer during the winter. Seasonal migrations or extraterritorial movements by wolves have been reported in a number of studies (Carbyn 1981; Messier 1985; Parker 1973; Peterson et al. 1984; Van Ballenberghe 1983); however, the consequence of these movements on pack structure is poorly understood. In populations where harvesting and human-related mortality are high (Rausch 1967; Van Ballenberghe et al. 1975), increased recruitment and acceptance of unfamiliar wolves into packs has been observed (Van Ballenberghe et al. 1983). Adoptions also have been identified in nonharvested gray wolf populations, such as those at Denali National Park, Alaska, where wolves were accepted into long-established packs as well as into new packs with a single generation of pups (Meier et al. 1995).

We suggest that most of the western coyote mitochondrial DNA haplotypes currently found in the park animals probably entered the park after wolf culls that occurred up to the 1960s when smaller Tweed wolves may have flourished in the region. Given a low deer density in the park, it is likely that smaller animals are selected against and this is the basis of the present restricted immigration into the area.

In contrast to the southeast, animals from the northeastern Ontario upper-Great-Lakes region and Quebec's Abitibi–Temiscamingue region and La Verendrye Reserve are more genetically similar to animals from Algonquin Park, supporting the suggestion that the park represents the southern edge of a larger metapopulation. The connectivity to the north is manifested by the moderately high levels of heterozygosity and allelic diversity identified within Algonquin Provincial Park. The data are consistent with the suggestion that the park animals do not represent an island population.


This research was funded by grants provided by the World Wildlife Fund of Canada, the Ontario Ministry of Natural Resources, the Max Bell Foundation to B. N. White and J. B. Theberge, and by a National Sciences and Engineering Research Council grant to B. N. White. Care and treatment of animals was in accordance with the principles outlined in the Canadian Council of Animal Care guide.


  • Associate Editor was Robert D. Bradley.

Literature Cited

View Abstract